diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,61221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.22916262964117512,
+  "eval_steps": 500,
+  "global_step": 10200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666668e-08,
+      "loss": 6.4772,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333336e-07,
+      "loss": 6.488,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 6.5587,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 6.4569,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 6.4478,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 6.5591,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.666666666666667e-07,
+      "loss": 6.5574,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 6.4541,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 6.3646,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 6.2758,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.333333333333334e-07,
+      "loss": 6.1429,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 6.2242,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.666666666666668e-07,
+      "loss": 5.7594,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.333333333333334e-07,
+      "loss": 5.7263,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 5.6062,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 5.6439,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1333333333333334e-06,
+      "loss": 4.6878,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.6595,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2666666666666669e-06,
+      "loss": 4.5771,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 4.4155,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.3013,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4666666666666669e-06,
+      "loss": 4.221,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5333333333333334e-06,
+      "loss": 3.0617,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.8669,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 2.7921,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7333333333333336e-06,
+      "loss": 2.7499,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.6792,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 2.6345,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9333333333333336e-06,
+      "loss": 2.6288,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.5876,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0666666666666666e-06,
+      "loss": 2.4993,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 2.3902,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-06,
+      "loss": 2.351,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 2.3061,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 2.2755,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.296,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.466666666666667e-06,
+      "loss": 2.2536,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5333333333333338e-06,
+      "loss": 2.2745,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-06,
+      "loss": 2.2561,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 2.2541,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7333333333333336e-06,
+      "loss": 2.2196,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.2253,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.866666666666667e-06,
+      "loss": 2.2276,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 2.202,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 2.2098,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.066666666666667e-06,
+      "loss": 2.2349,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.133333333333334e-06,
+      "loss": 2.2185,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.2159,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.266666666666667e-06,
+      "loss": 2.1916,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 2.2114,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.2161,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 2.2029,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5333333333333335e-06,
+      "loss": 2.2204,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.2048,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 2.22,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 2.1985,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.1914,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 2.178,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9333333333333335e-06,
+      "loss": 2.2216,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.2073,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.066666666666667e-06,
+      "loss": 2.191,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 2.196,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.1938,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 2.2107,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 2.1731,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-06,
+      "loss": 2.2043,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4666666666666665e-06,
+      "loss": 2.1933,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 2.195,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.1901,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 2.1844,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7333333333333335e-06,
+      "loss": 2.189,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.194,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.866666666666667e-06,
+      "loss": 2.199,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.933333333333334e-06,
+      "loss": 2.1813,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 2.2026,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 2.2022,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.133333333333334e-06,
+      "loss": 2.1961,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-06,
+      "loss": 2.1739,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2666666666666665e-06,
+      "loss": 2.1845,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 2.18,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.1951,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 2.1891,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.533333333333334e-06,
+      "loss": 2.1745,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.1764,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 2.1637,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 2.1962,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.8e-06,
+      "loss": 2.1863,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 2.1986,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.933333333333335e-06,
+      "loss": 2.1811,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 2.182,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.066666666666667e-06,
+      "loss": 2.18,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 2.1824,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 2.1712,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.266666666666668e-06,
+      "loss": 2.1917,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 2.1591,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.1911,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.466666666666667e-06,
+      "loss": 2.1866,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.533333333333334e-06,
+      "loss": 2.1772,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 2.1599,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 2.1735,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.733333333333334e-06,
+      "loss": 2.1954,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.1685,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.866666666666667e-06,
+      "loss": 2.1978,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 2.1799,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-06,
+      "loss": 2.1739,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 2.1871,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.133333333333334e-06,
+      "loss": 2.1555,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 2.1809,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.266666666666668e-06,
+      "loss": 2.1725,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 2.1687,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-06,
+      "loss": 2.1697,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 2.1827,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.533333333333334e-06,
+      "loss": 2.171,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 2.1662,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 2.183,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 2.1652,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.163,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 2.1655,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.933333333333334e-06,
+      "loss": 2.1566,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.1605,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.066666666666667e-06,
+      "loss": 2.1604,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 2.1701,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.2e-06,
+      "loss": 2.1776,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 2.1775,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 2.1458,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.148,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.466666666666668e-06,
+      "loss": 2.1575,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 2.1833,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.6e-06,
+      "loss": 2.1689,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 2.1765,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.733333333333333e-06,
+      "loss": 2.1638,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.8e-06,
+      "loss": 2.1531,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.866666666666668e-06,
+      "loss": 2.1706,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 2.173,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 2.1602,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 2.1788,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.133333333333335e-06,
+      "loss": 2.1638,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 2.1776,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.266666666666667e-06,
+      "loss": 2.1516,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 2.1547,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.4e-06,
+      "loss": 2.1688,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 2.1579,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.533333333333334e-06,
+      "loss": 2.1709,
+      "step": 143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.1726,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 2.1592,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 2.164,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 2.1693,
+      "step": 147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 2.1603,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.933333333333334e-06,
+      "loss": 2.1662,
+      "step": 149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 2.1666,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0066666666666666e-05,
+      "loss": 2.15,
+      "step": 151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 2.1497,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.02e-05,
+      "loss": 2.1459,
+      "step": 153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0266666666666668e-05,
+      "loss": 2.1644,
+      "step": 154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0333333333333335e-05,
+      "loss": 2.1504,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.04e-05,
+      "loss": 2.1562,
+      "step": 156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0466666666666668e-05,
+      "loss": 2.1227,
+      "step": 157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0533333333333333e-05,
+      "loss": 2.1938,
+      "step": 158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 2.1655,
+      "step": 159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 2.1552,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0733333333333333e-05,
+      "loss": 2.1426,
+      "step": 161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 2.145,
+      "step": 162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0866666666666667e-05,
+      "loss": 2.1821,
+      "step": 163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 2.1795,
+      "step": 164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 2.1662,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1066666666666669e-05,
+      "loss": 2.1838,
+      "step": 166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1133333333333334e-05,
+      "loss": 2.1483,
+      "step": 167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 2.1564,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1266666666666668e-05,
+      "loss": 2.1521,
+      "step": 169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 2.142,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.14e-05,
+      "loss": 2.1562,
+      "step": 171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 2.1372,
+      "step": 172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1533333333333334e-05,
+      "loss": 2.1473,
+      "step": 173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.16e-05,
+      "loss": 2.1655,
+      "step": 174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 2.1497,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 2.1759,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.18e-05,
+      "loss": 2.1566,
+      "step": 177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.186666666666667e-05,
+      "loss": 2.1607,
+      "step": 178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1933333333333335e-05,
+      "loss": 2.1667,
+      "step": 179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 2.1636,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.206666666666667e-05,
+      "loss": 2.1395,
+      "step": 181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2133333333333335e-05,
+      "loss": 2.156,
+      "step": 182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.22e-05,
+      "loss": 2.1624,
+      "step": 183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 2.1761,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2333333333333334e-05,
+      "loss": 2.1745,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 2.172,
+      "step": 186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2466666666666667e-05,
+      "loss": 2.161,
+      "step": 187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 2.1634,
+      "step": 188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 2.1428,
+      "step": 189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 2.1465,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2733333333333336e-05,
+      "loss": 2.1597,
+      "step": 191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 2.1598,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2866666666666667e-05,
+      "loss": 2.1622,
+      "step": 193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2933333333333334e-05,
+      "loss": 2.1552,
+      "step": 194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 2.154,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 2.1676,
+      "step": 196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3133333333333334e-05,
+      "loss": 2.1504,
+      "step": 197
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 2.1554,
+      "step": 198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3266666666666668e-05,
+      "loss": 2.1529,
+      "step": 199
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 2.1538,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 2.1419,
+      "step": 201
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3466666666666668e-05,
+      "loss": 2.1587,
+      "step": 202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3533333333333333e-05,
+      "loss": 2.1612,
+      "step": 203
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 2.1567,
+      "step": 204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3666666666666667e-05,
+      "loss": 2.1432,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3733333333333335e-05,
+      "loss": 2.1539,
+      "step": 206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.38e-05,
+      "loss": 2.1749,
+      "step": 207
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 2.1393,
+      "step": 208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3933333333333334e-05,
+      "loss": 2.1541,
+      "step": 209
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-05,
+      "loss": 2.1713,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4066666666666669e-05,
+      "loss": 2.1516,
+      "step": 211
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 2.127,
+      "step": 212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 2.1582,
+      "step": 213
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4266666666666668e-05,
+      "loss": 2.1526,
+      "step": 214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4333333333333334e-05,
+      "loss": 2.1693,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 2.1478,
+      "step": 216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4466666666666668e-05,
+      "loss": 2.1451,
+      "step": 217
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4533333333333335e-05,
+      "loss": 2.1584,
+      "step": 218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.46e-05,
+      "loss": 2.1679,
+      "step": 219
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 2.1482,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4733333333333335e-05,
+      "loss": 2.1493,
+      "step": 221
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.48e-05,
+      "loss": 2.1653,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4866666666666668e-05,
+      "loss": 2.1723,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 2.1515,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 2.1418,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5066666666666668e-05,
+      "loss": 2.1583,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5133333333333335e-05,
+      "loss": 2.1478,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 2.1609,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5266666666666667e-05,
+      "loss": 2.1484,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 2.1549,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-05,
+      "loss": 2.1456,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 2.1429,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5533333333333333e-05,
+      "loss": 2.1479,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 2.1549,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5666666666666667e-05,
+      "loss": 2.1561,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 2.1595,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.58e-05,
+      "loss": 2.1381,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.586666666666667e-05,
+      "loss": 2.1599,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5933333333333336e-05,
+      "loss": 2.1485,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.1376,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.606666666666667e-05,
+      "loss": 2.1627,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6133333333333334e-05,
+      "loss": 2.1465,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.62e-05,
+      "loss": 2.15,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 2.1675,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6333333333333335e-05,
+      "loss": 2.1572,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.64e-05,
+      "loss": 2.1393,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.646666666666667e-05,
+      "loss": 2.1499,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 2.1567,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.66e-05,
+      "loss": 2.1388,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 2.1541,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6733333333333335e-05,
+      "loss": 2.1406,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 2.1598,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.686666666666667e-05,
+      "loss": 2.161,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6933333333333336e-05,
+      "loss": 2.1378,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7e-05,
+      "loss": 2.1603,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 2.1415,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7133333333333334e-05,
+      "loss": 2.1562,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.72e-05,
+      "loss": 2.1475,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.726666666666667e-05,
+      "loss": 2.1528,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 2.1487,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 2.1581,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7466666666666667e-05,
+      "loss": 2.1865,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7533333333333337e-05,
+      "loss": 2.1432,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-05,
+      "loss": 2.1431,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7666666666666668e-05,
+      "loss": 2.1659,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7733333333333335e-05,
+      "loss": 2.1418,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 2.1217,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 2.1338,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7933333333333333e-05,
+      "loss": 2.1789,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 2.1423,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8066666666666668e-05,
+      "loss": 2.1317,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 2.1616,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 2.1343,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.826666666666667e-05,
+      "loss": 2.1318,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 2.153,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 2.1504,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8466666666666667e-05,
+      "loss": 2.1369,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 2.1526,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.86e-05,
+      "loss": 2.1445,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 2.1657,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8733333333333336e-05,
+      "loss": 2.1497,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.88e-05,
+      "loss": 2.1499,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.886666666666667e-05,
+      "loss": 2.1332,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 2.1513,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9e-05,
+      "loss": 2.1586,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9066666666666668e-05,
+      "loss": 2.1537,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9133333333333335e-05,
+      "loss": 2.1529,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 2.1499,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.926666666666667e-05,
+      "loss": 2.1543,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 2.1422,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-05,
+      "loss": 2.1517,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 2.1324,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9533333333333335e-05,
+      "loss": 2.1544,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 2.1491,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9666666666666666e-05,
+      "loss": 2.1601,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 2.1568,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-05,
+      "loss": 2.1444,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9866666666666667e-05,
+      "loss": 2.1403,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9933333333333334e-05,
+      "loss": 2.1429,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 2.1407,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999999997475076e-05,
+      "loss": 2.1523,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999999989900303e-05,
+      "loss": 2.1467,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999999772756813e-05,
+      "loss": 2.1565,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999999596012113e-05,
+      "loss": 2.1428,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999999368768925e-05,
+      "loss": 2.1528,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999999091027255e-05,
+      "loss": 2.1377,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999998762787104e-05,
+      "loss": 2.157,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999998384048475e-05,
+      "loss": 2.1541,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999997954811364e-05,
+      "loss": 2.1367,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999999747507578e-05,
+      "loss": 2.1376,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999999694484172e-05,
+      "loss": 2.1279,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999996364109185e-05,
+      "loss": 2.1389,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999999573287819e-05,
+      "loss": 2.1525,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999995051148726e-05,
+      "loss": 2.1681,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99999943189208e-05,
+      "loss": 2.1552,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999993536194416e-05,
+      "loss": 2.1663,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999992702969577e-05,
+      "loss": 2.1519,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999991819246293e-05,
+      "loss": 2.1415,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999990885024558e-05,
+      "loss": 2.1394,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999989900304386e-05,
+      "loss": 2.137,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999988865085776e-05,
+      "loss": 2.1441,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999998777936874e-05,
+      "loss": 2.1249,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999986643153276e-05,
+      "loss": 2.1426,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999998545643939e-05,
+      "loss": 2.1431,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999984219227097e-05,
+      "loss": 2.1602,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999982931516392e-05,
+      "loss": 2.1472,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999998159330729e-05,
+      "loss": 2.1284,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999980204599792e-05,
+      "loss": 2.1586,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999997876539391e-05,
+      "loss": 2.1333,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999997727568965e-05,
+      "loss": 2.1395,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999975735487012e-05,
+      "loss": 2.1385,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999974144786015e-05,
+      "loss": 2.1356,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999997250358666e-05,
+      "loss": 2.1513,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999970811888957e-05,
+      "loss": 2.1499,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999969069692916e-05,
+      "loss": 2.1475,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999967276998542e-05,
+      "loss": 2.1568,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999996543380585e-05,
+      "loss": 2.1585,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999963540114845e-05,
+      "loss": 2.1624,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999996159592554e-05,
+      "loss": 2.1567,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999959601237938e-05,
+      "loss": 2.1505,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999957556052055e-05,
+      "loss": 2.1459,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999955460367897e-05,
+      "loss": 2.1306,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999953314185482e-05,
+      "loss": 2.1376,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999951117504816e-05,
+      "loss": 2.1408,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999994887032591e-05,
+      "loss": 2.151,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999946572648774e-05,
+      "loss": 2.1497,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999944224473423e-05,
+      "loss": 2.1666,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999941825799864e-05,
+      "loss": 2.1489,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999939376628116e-05,
+      "loss": 2.1476,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999936876958184e-05,
+      "loss": 2.1356,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999934326790086e-05,
+      "loss": 2.1577,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999931726123832e-05,
+      "loss": 2.1473,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999929074959435e-05,
+      "loss": 2.1489,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999926373296913e-05,
+      "loss": 2.1485,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999923621136278e-05,
+      "loss": 2.158,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999920818477538e-05,
+      "loss": 2.1542,
+      "step": 356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999917965320713e-05,
+      "loss": 2.1502,
+      "step": 357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999915061665814e-05,
+      "loss": 2.1383,
+      "step": 358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999991210751286e-05,
+      "loss": 2.1531,
+      "step": 359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999990910286186e-05,
+      "loss": 2.1717,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999906047712835e-05,
+      "loss": 2.1718,
+      "step": 361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999902942065795e-05,
+      "loss": 2.1641,
+      "step": 362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999899785920762e-05,
+      "loss": 2.1325,
+      "step": 363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999989657927775e-05,
+      "loss": 2.1343,
+      "step": 364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999989332213677e-05,
+      "loss": 2.1582,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999890014497844e-05,
+      "loss": 2.1328,
+      "step": 366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999886656360986e-05,
+      "loss": 2.1379,
+      "step": 367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999883247726212e-05,
+      "loss": 2.1434,
+      "step": 368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999879788593543e-05,
+      "loss": 2.1442,
+      "step": 369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999876278962992e-05,
+      "loss": 2.1295,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999872718834583e-05,
+      "loss": 2.1614,
+      "step": 371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999986910820833e-05,
+      "loss": 2.1498,
+      "step": 372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999986544708425e-05,
+      "loss": 2.1387,
+      "step": 373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999861735462362e-05,
+      "loss": 2.1435,
+      "step": 374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999857973342687e-05,
+      "loss": 2.1518,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999854160725243e-05,
+      "loss": 2.1769,
+      "step": 376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999850297610045e-05,
+      "loss": 2.1583,
+      "step": 377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999846383997122e-05,
+      "loss": 2.1523,
+      "step": 378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999842419886483e-05,
+      "loss": 2.1473,
+      "step": 379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999983840527816e-05,
+      "loss": 2.158,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999983434017216e-05,
+      "loss": 2.1717,
+      "step": 381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999830224568512e-05,
+      "loss": 2.1516,
+      "step": 382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999826058467234e-05,
+      "loss": 2.1434,
+      "step": 383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999982184186835e-05,
+      "loss": 2.1374,
+      "step": 384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999817574771876e-05,
+      "loss": 2.1599,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999981325717784e-05,
+      "loss": 2.1459,
+      "step": 386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999980888908626e-05,
+      "loss": 2.1509,
+      "step": 387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999804470497158e-05,
+      "loss": 2.1607,
+      "step": 388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999800001410555e-05,
+      "loss": 2.1714,
+      "step": 389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999979548182648e-05,
+      "loss": 2.1655,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999790911744947e-05,
+      "loss": 2.1427,
+      "step": 391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999786291165986e-05,
+      "loss": 2.1551,
+      "step": 392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999781620089615e-05,
+      "loss": 2.149,
+      "step": 393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999776898515863e-05,
+      "loss": 2.1468,
+      "step": 394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999977212644475e-05,
+      "loss": 2.1578,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99997673038763e-05,
+      "loss": 2.1303,
+      "step": 396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999976243081054e-05,
+      "loss": 2.1565,
+      "step": 397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999975750724749e-05,
+      "loss": 2.1552,
+      "step": 398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999752533187182e-05,
+      "loss": 2.1399,
+      "step": 399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999747508629634e-05,
+      "loss": 2.1323,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999742433574877e-05,
+      "loss": 2.1289,
+      "step": 401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999737308022934e-05,
+      "loss": 2.1507,
+      "step": 402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999732131973827e-05,
+      "loss": 2.1555,
+      "step": 403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999972690542759e-05,
+      "loss": 2.1509,
+      "step": 404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999721628384245e-05,
+      "loss": 2.1714,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999971630084382e-05,
+      "loss": 2.1375,
+      "step": 406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999710922806338e-05,
+      "loss": 2.1302,
+      "step": 407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999705494271834e-05,
+      "loss": 2.1645,
+      "step": 408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999700015240323e-05,
+      "loss": 2.1432,
+      "step": 409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999694485711848e-05,
+      "loss": 2.1566,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999688905686426e-05,
+      "loss": 2.1621,
+      "step": 411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999968327516409e-05,
+      "loss": 2.1378,
+      "step": 412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999677594144865e-05,
+      "loss": 2.1581,
+      "step": 413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999967186262878e-05,
+      "loss": 2.158,
+      "step": 414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999966608061587e-05,
+      "loss": 2.1392,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999660248106157e-05,
+      "loss": 2.1536,
+      "step": 416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999654365099673e-05,
+      "loss": 2.1371,
+      "step": 417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999964843159645e-05,
+      "loss": 2.1534,
+      "step": 418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999964244759651e-05,
+      "loss": 2.1495,
+      "step": 419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999636413099895e-05,
+      "loss": 2.1641,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999963032810663e-05,
+      "loss": 2.1161,
+      "step": 421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999962419261674e-05,
+      "loss": 2.1435,
+      "step": 422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999618006630266e-05,
+      "loss": 2.1461,
+      "step": 423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999611770147232e-05,
+      "loss": 2.1434,
+      "step": 424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999605483167672e-05,
+      "loss": 2.159,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999959914569162e-05,
+      "loss": 2.1643,
+      "step": 426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999592757719102e-05,
+      "loss": 2.1307,
+      "step": 427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999586319250156e-05,
+      "loss": 2.1614,
+      "step": 428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999579830284812e-05,
+      "loss": 2.1641,
+      "step": 429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99995732908231e-05,
+      "loss": 2.1525,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999566700865064e-05,
+      "loss": 2.1624,
+      "step": 431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999560060410723e-05,
+      "loss": 2.157,
+      "step": 432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999955336946012e-05,
+      "loss": 2.1382,
+      "step": 433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999546628013283e-05,
+      "loss": 2.1183,
+      "step": 434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999539836070254e-05,
+      "loss": 2.1712,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999953299363106e-05,
+      "loss": 2.1507,
+      "step": 436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999526100695736e-05,
+      "loss": 2.1363,
+      "step": 437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999951915726432e-05,
+      "loss": 2.1473,
+      "step": 438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999512163336846e-05,
+      "loss": 2.1362,
+      "step": 439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999505118913347e-05,
+      "loss": 2.1623,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999498023993863e-05,
+      "loss": 2.1526,
+      "step": 441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999490878578426e-05,
+      "loss": 2.1583,
+      "step": 442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999483682667072e-05,
+      "loss": 2.1582,
+      "step": 443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999947643625984e-05,
+      "loss": 2.1475,
+      "step": 444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999469139356767e-05,
+      "loss": 2.1653,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999461791957886e-05,
+      "loss": 2.1632,
+      "step": 446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999945439406324e-05,
+      "loss": 2.165,
+      "step": 447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999446945672857e-05,
+      "loss": 2.1641,
+      "step": 448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999439446786783e-05,
+      "loss": 2.1558,
+      "step": 449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999431897405053e-05,
+      "loss": 2.1588,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999424297527705e-05,
+      "loss": 2.1615,
+      "step": 451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999941664715478e-05,
+      "loss": 2.1425,
+      "step": 452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999408946286312e-05,
+      "loss": 2.1564,
+      "step": 453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999401194922342e-05,
+      "loss": 2.1676,
+      "step": 454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999393393062913e-05,
+      "loss": 2.1574,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999385540708055e-05,
+      "loss": 2.1638,
+      "step": 456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999377637857816e-05,
+      "loss": 2.1581,
+      "step": 457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999369684512233e-05,
+      "loss": 2.1742,
+      "step": 458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999361680671347e-05,
+      "loss": 2.1587,
+      "step": 459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999353626335198e-05,
+      "loss": 2.1535,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999345521503827e-05,
+      "loss": 2.1447,
+      "step": 461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999933736617727e-05,
+      "loss": 2.1574,
+      "step": 462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999329160355578e-05,
+      "loss": 2.1551,
+      "step": 463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999320904038784e-05,
+      "loss": 2.1598,
+      "step": 464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999312597226935e-05,
+      "loss": 2.1566,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999930423992007e-05,
+      "loss": 2.1538,
+      "step": 466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999929583211823e-05,
+      "loss": 2.1367,
+      "step": 467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999287373821462e-05,
+      "loss": 2.1656,
+      "step": 468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999278865029805e-05,
+      "loss": 2.1365,
+      "step": 469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999270305743303e-05,
+      "loss": 2.163,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999261695962e-05,
+      "loss": 2.1371,
+      "step": 471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999253035685937e-05,
+      "loss": 2.1569,
+      "step": 472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999924432491516e-05,
+      "loss": 2.1531,
+      "step": 473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999235563649715e-05,
+      "loss": 2.1364,
+      "step": 474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999922675188964e-05,
+      "loss": 2.1576,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999217889634987e-05,
+      "loss": 2.1688,
+      "step": 476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999208976885797e-05,
+      "loss": 2.1614,
+      "step": 477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999920001364211e-05,
+      "loss": 2.1679,
+      "step": 478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999190999903982e-05,
+      "loss": 2.1519,
+      "step": 479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999181935671447e-05,
+      "loss": 2.1652,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999172820944563e-05,
+      "loss": 2.1584,
+      "step": 481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999163655723365e-05,
+      "loss": 2.1511,
+      "step": 482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999154440007908e-05,
+      "loss": 2.1525,
+      "step": 483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999145173798228e-05,
+      "loss": 2.1949,
+      "step": 484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999135857094383e-05,
+      "loss": 2.1586,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999126489896414e-05,
+      "loss": 2.1508,
+      "step": 486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999911707220437e-05,
+      "loss": 2.1812,
+      "step": 487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999107604018296e-05,
+      "loss": 2.156,
+      "step": 488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999098085338245e-05,
+      "loss": 2.1612,
+      "step": 489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999908851616426e-05,
+      "loss": 2.1525,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999907889649639e-05,
+      "loss": 2.1689,
+      "step": 491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999069226334686e-05,
+      "loss": 2.1569,
+      "step": 492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999059505679196e-05,
+      "loss": 2.1657,
+      "step": 493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999904973452997e-05,
+      "loss": 2.1647,
+      "step": 494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999039912887058e-05,
+      "loss": 2.1354,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999030040750506e-05,
+      "loss": 2.1527,
+      "step": 496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999902011812036e-05,
+      "loss": 2.1613,
+      "step": 497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999010144996682e-05,
+      "loss": 2.1697,
+      "step": 498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999000121379516e-05,
+      "loss": 2.1678,
+      "step": 499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998990047268913e-05,
+      "loss": 2.161,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998979922664918e-05,
+      "loss": 2.1618,
+      "step": 501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999896974756759e-05,
+      "loss": 2.1779,
+      "step": 502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999895952197698e-05,
+      "loss": 2.1523,
+      "step": 503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999894924589314e-05,
+      "loss": 2.1465,
+      "step": 504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998938919316116e-05,
+      "loss": 2.1706,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998928542245962e-05,
+      "loss": 2.1651,
+      "step": 506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998918114682733e-05,
+      "loss": 2.1559,
+      "step": 507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998907636626485e-05,
+      "loss": 2.1746,
+      "step": 508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999889710807726e-05,
+      "loss": 2.1751,
+      "step": 509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998886529035126e-05,
+      "loss": 2.1652,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998875899500124e-05,
+      "loss": 2.1815,
+      "step": 511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999886521947231e-05,
+      "loss": 2.1492,
+      "step": 512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998854488951744e-05,
+      "loss": 2.1761,
+      "step": 513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998843707938472e-05,
+      "loss": 2.1475,
+      "step": 514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998832876432556e-05,
+      "loss": 2.171,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998821994434044e-05,
+      "loss": 2.1699,
+      "step": 516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998811061942997e-05,
+      "loss": 2.1544,
+      "step": 517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999880007895947e-05,
+      "loss": 2.1664,
+      "step": 518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999878904548351e-05,
+      "loss": 2.1845,
+      "step": 519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998777961515182e-05,
+      "loss": 2.1477,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998766827054536e-05,
+      "loss": 2.1784,
+      "step": 521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998755642101633e-05,
+      "loss": 2.1673,
+      "step": 522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998744406656526e-05,
+      "loss": 2.1782,
+      "step": 523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998733120719274e-05,
+      "loss": 2.1733,
+      "step": 524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998721784289933e-05,
+      "loss": 2.16,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999871039736856e-05,
+      "loss": 2.1897,
+      "step": 526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998698959955212e-05,
+      "loss": 2.1556,
+      "step": 527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998687472049947e-05,
+      "loss": 2.1769,
+      "step": 528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998675933652824e-05,
+      "loss": 2.1722,
+      "step": 529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99986643447639e-05,
+      "loss": 2.1937,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998652705383235e-05,
+      "loss": 2.1744,
+      "step": 531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998641015510892e-05,
+      "loss": 2.1532,
+      "step": 532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998629275146918e-05,
+      "loss": 2.1699,
+      "step": 533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998617484291384e-05,
+      "loss": 2.1898,
+      "step": 534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998605642944342e-05,
+      "loss": 2.1658,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998593751105856e-05,
+      "loss": 2.1651,
+      "step": 536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998581808775986e-05,
+      "loss": 2.1731,
+      "step": 537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998569815954787e-05,
+      "loss": 2.1559,
+      "step": 538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998557772642327e-05,
+      "loss": 2.1732,
+      "step": 539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998545678838663e-05,
+      "loss": 2.1675,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998533534543857e-05,
+      "loss": 2.1875,
+      "step": 541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999852133975797e-05,
+      "loss": 2.1769,
+      "step": 542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998509094481063e-05,
+      "loss": 2.1778,
+      "step": 543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998496798713195e-05,
+      "loss": 2.1811,
+      "step": 544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998484452454434e-05,
+      "loss": 2.1692,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999847205570484e-05,
+      "loss": 2.1683,
+      "step": 546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998459608464474e-05,
+      "loss": 2.1734,
+      "step": 547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998447110733403e-05,
+      "loss": 2.2125,
+      "step": 548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998434562511688e-05,
+      "loss": 2.1909,
+      "step": 549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999842196379939e-05,
+      "loss": 2.1759,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999840931459657e-05,
+      "loss": 2.1723,
+      "step": 551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99983966149033e-05,
+      "loss": 2.1833,
+      "step": 552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998383864719644e-05,
+      "loss": 2.1728,
+      "step": 553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998371064045658e-05,
+      "loss": 2.1876,
+      "step": 554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998358212881413e-05,
+      "loss": 2.172,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999834531122697e-05,
+      "loss": 2.1912,
+      "step": 556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998332359082398e-05,
+      "loss": 2.1822,
+      "step": 557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999831935644776e-05,
+      "loss": 2.1765,
+      "step": 558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998306303323123e-05,
+      "loss": 2.18,
+      "step": 559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998293199708555e-05,
+      "loss": 2.189,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998280045604115e-05,
+      "loss": 2.2023,
+      "step": 561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998266841009877e-05,
+      "loss": 2.1808,
+      "step": 562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998253585925903e-05,
+      "loss": 2.1733,
+      "step": 563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999824028035226e-05,
+      "loss": 2.1614,
+      "step": 564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999822692428902e-05,
+      "loss": 2.1773,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998213517736244e-05,
+      "loss": 2.1632,
+      "step": 566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998200060694008e-05,
+      "loss": 2.1679,
+      "step": 567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999818655316237e-05,
+      "loss": 2.1925,
+      "step": 568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998172995141406e-05,
+      "loss": 2.1935,
+      "step": 569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999815938663118e-05,
+      "loss": 2.1641,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998145727631765e-05,
+      "loss": 2.1727,
+      "step": 571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998132018143224e-05,
+      "loss": 2.1807,
+      "step": 572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998118258165627e-05,
+      "loss": 2.1817,
+      "step": 573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998104447699053e-05,
+      "loss": 2.1826,
+      "step": 574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998090586743558e-05,
+      "loss": 2.1849,
+      "step": 575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999807667529922e-05,
+      "loss": 2.1977,
+      "step": 576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999806271336611e-05,
+      "loss": 2.2127,
+      "step": 577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998048700944296e-05,
+      "loss": 2.176,
+      "step": 578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998034638033853e-05,
+      "loss": 2.1857,
+      "step": 579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998020524634842e-05,
+      "loss": 2.1884,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998006360747345e-05,
+      "loss": 2.1951,
+      "step": 581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997992146371426e-05,
+      "loss": 2.1942,
+      "step": 582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999797788150716e-05,
+      "loss": 2.2034,
+      "step": 583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997963566154622e-05,
+      "loss": 2.1954,
+      "step": 584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997949200313878e-05,
+      "loss": 2.165,
+      "step": 585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997934783985005e-05,
+      "loss": 2.1947,
+      "step": 586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997920317168074e-05,
+      "loss": 2.1719,
+      "step": 587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997905799863157e-05,
+      "loss": 2.2011,
+      "step": 588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999789123207033e-05,
+      "loss": 2.1991,
+      "step": 589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997876613789664e-05,
+      "loss": 2.1849,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997861945021236e-05,
+      "loss": 2.1986,
+      "step": 591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997847225765118e-05,
+      "loss": 2.2,
+      "step": 592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997832456021388e-05,
+      "loss": 2.1908,
+      "step": 593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997817635790113e-05,
+      "loss": 2.1746,
+      "step": 594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997802765071372e-05,
+      "loss": 2.1898,
+      "step": 595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997787843865242e-05,
+      "loss": 2.1861,
+      "step": 596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997772872171794e-05,
+      "loss": 2.1931,
+      "step": 597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997757849991107e-05,
+      "loss": 2.18,
+      "step": 598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997742777323255e-05,
+      "loss": 2.1911,
+      "step": 599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999772765416832e-05,
+      "loss": 2.17,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999771248052637e-05,
+      "loss": 2.1947,
+      "step": 601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997697256397485e-05,
+      "loss": 2.1915,
+      "step": 602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997681981781745e-05,
+      "loss": 2.1965,
+      "step": 603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999766665667922e-05,
+      "loss": 2.1977,
+      "step": 604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999765128108999e-05,
+      "loss": 2.2116,
+      "step": 605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997635855014138e-05,
+      "loss": 2.2195,
+      "step": 606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999762037845174e-05,
+      "loss": 2.1978,
+      "step": 607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999760485140287e-05,
+      "loss": 2.1964,
+      "step": 608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997589273867605e-05,
+      "loss": 2.1901,
+      "step": 609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999757364584603e-05,
+      "loss": 2.1703,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999755796733822e-05,
+      "loss": 2.1832,
+      "step": 611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997542238344258e-05,
+      "loss": 2.1943,
+      "step": 612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999752645886422e-05,
+      "loss": 2.2226,
+      "step": 613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997510628898182e-05,
+      "loss": 2.1946,
+      "step": 614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997494748446233e-05,
+      "loss": 2.1918,
+      "step": 615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997478817508448e-05,
+      "loss": 2.1827,
+      "step": 616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999746283608491e-05,
+      "loss": 2.2029,
+      "step": 617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997446804175694e-05,
+      "loss": 2.1821,
+      "step": 618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997430721780886e-05,
+      "loss": 2.1887,
+      "step": 619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997414588900566e-05,
+      "loss": 2.1704,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997398405534815e-05,
+      "loss": 2.1693,
+      "step": 621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997382171683716e-05,
+      "loss": 2.191,
+      "step": 622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999736588734735e-05,
+      "loss": 2.1935,
+      "step": 623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99973495525258e-05,
+      "loss": 2.1834,
+      "step": 624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997333167219143e-05,
+      "loss": 2.1924,
+      "step": 625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999731673142747e-05,
+      "loss": 2.1803,
+      "step": 626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999730024515086e-05,
+      "loss": 2.2075,
+      "step": 627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997283708389398e-05,
+      "loss": 2.1939,
+      "step": 628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999726712114317e-05,
+      "loss": 2.1934,
+      "step": 629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999725048341225e-05,
+      "loss": 2.1739,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999723379519673e-05,
+      "loss": 2.2276,
+      "step": 631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997217056496693e-05,
+      "loss": 2.1674,
+      "step": 632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997200267312223e-05,
+      "loss": 2.2015,
+      "step": 633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997183427643404e-05,
+      "loss": 2.197,
+      "step": 634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999716653749032e-05,
+      "loss": 2.21,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997149596853058e-05,
+      "loss": 2.185,
+      "step": 636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997132605731705e-05,
+      "loss": 2.1861,
+      "step": 637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997115564126345e-05,
+      "loss": 2.1754,
+      "step": 638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997098472037065e-05,
+      "loss": 2.2089,
+      "step": 639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999708132946395e-05,
+      "loss": 2.1839,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999706413640709e-05,
+      "loss": 2.196,
+      "step": 641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999704689286657e-05,
+      "loss": 2.1928,
+      "step": 642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999702959884247e-05,
+      "loss": 2.1952,
+      "step": 643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997012254334888e-05,
+      "loss": 2.1833,
+      "step": 644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996994859343902e-05,
+      "loss": 2.1967,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999697741386961e-05,
+      "loss": 2.1927,
+      "step": 646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996959917912095e-05,
+      "loss": 2.1927,
+      "step": 647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996942371471445e-05,
+      "loss": 2.172,
+      "step": 648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996924774547744e-05,
+      "loss": 2.2011,
+      "step": 649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996907127141093e-05,
+      "loss": 2.2024,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996889429251573e-05,
+      "loss": 2.1888,
+      "step": 651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999687168087927e-05,
+      "loss": 2.1878,
+      "step": 652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999685388202428e-05,
+      "loss": 2.2096,
+      "step": 653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999683603268669e-05,
+      "loss": 2.2036,
+      "step": 654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996818132866595e-05,
+      "loss": 2.1884,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999680018256408e-05,
+      "loss": 2.1977,
+      "step": 656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996782181779232e-05,
+      "loss": 2.1901,
+      "step": 657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999676413051215e-05,
+      "loss": 2.1862,
+      "step": 658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996746028762922e-05,
+      "loss": 2.1995,
+      "step": 659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999672787653164e-05,
+      "loss": 2.1836,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99967096738184e-05,
+      "loss": 2.19,
+      "step": 661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996691420623283e-05,
+      "loss": 2.2061,
+      "step": 662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996673116946387e-05,
+      "loss": 2.2201,
+      "step": 663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996654762787808e-05,
+      "loss": 2.1989,
+      "step": 664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999663635814763e-05,
+      "loss": 2.1836,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996617903025956e-05,
+      "loss": 2.1932,
+      "step": 666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996599397422876e-05,
+      "loss": 2.193,
+      "step": 667
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996580841338476e-05,
+      "loss": 2.206,
+      "step": 668
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999656223477286e-05,
+      "loss": 2.209,
+      "step": 669
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996543577726118e-05,
+      "loss": 2.1887,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996524870198343e-05,
+      "loss": 2.2178,
+      "step": 671
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999650611218963e-05,
+      "loss": 2.193,
+      "step": 672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996487303700074e-05,
+      "loss": 2.1848,
+      "step": 673
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999646844472977e-05,
+      "loss": 2.1865,
+      "step": 674
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996449535278813e-05,
+      "loss": 2.1752,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996430575347298e-05,
+      "loss": 2.1995,
+      "step": 676
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996411564935327e-05,
+      "loss": 2.1916,
+      "step": 677
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996392504042987e-05,
+      "loss": 2.189,
+      "step": 678
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996373392670373e-05,
+      "loss": 2.2112,
+      "step": 679
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999635423081759e-05,
+      "loss": 2.1888,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999633501848473e-05,
+      "loss": 2.1793,
+      "step": 681
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996315755671894e-05,
+      "loss": 2.2026,
+      "step": 682
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996296442379175e-05,
+      "loss": 2.2076,
+      "step": 683
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999627707860667e-05,
+      "loss": 2.1918,
+      "step": 684
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996257664354482e-05,
+      "loss": 2.1929,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99962381996227e-05,
+      "loss": 2.2262,
+      "step": 686
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996218684411433e-05,
+      "loss": 2.2218,
+      "step": 687
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996199118720772e-05,
+      "loss": 2.2048,
+      "step": 688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999617950255082e-05,
+      "loss": 2.1887,
+      "step": 689
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996159835901675e-05,
+      "loss": 2.2042,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996140118773436e-05,
+      "loss": 2.1951,
+      "step": 691
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99961203511662e-05,
+      "loss": 2.2065,
+      "step": 692
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996100533080067e-05,
+      "loss": 2.2158,
+      "step": 693
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996080664515143e-05,
+      "loss": 2.1901,
+      "step": 694
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996060745471526e-05,
+      "loss": 2.198,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999604077594931e-05,
+      "loss": 2.1953,
+      "step": 696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996020755948605e-05,
+      "loss": 2.2017,
+      "step": 697
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996000685469506e-05,
+      "loss": 2.215,
+      "step": 698
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995980564512116e-05,
+      "loss": 2.1988,
+      "step": 699
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995960393076536e-05,
+      "loss": 2.2141,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995940171162874e-05,
+      "loss": 2.1946,
+      "step": 701
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995919898771222e-05,
+      "loss": 2.1759,
+      "step": 702
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999589957590169e-05,
+      "loss": 2.1884,
+      "step": 703
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999587920255438e-05,
+      "loss": 2.2252,
+      "step": 704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995858778729387e-05,
+      "loss": 2.2025,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995838304426823e-05,
+      "loss": 2.2077,
+      "step": 706
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995817779646786e-05,
+      "loss": 2.191,
+      "step": 707
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995797204389387e-05,
+      "loss": 2.2022,
+      "step": 708
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999577657865472e-05,
+      "loss": 2.1876,
+      "step": 709
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995755902442898e-05,
+      "loss": 2.2281,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995735175754015e-05,
+      "loss": 2.1997,
+      "step": 711
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999571439858819e-05,
+      "loss": 2.2133,
+      "step": 712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995693570945516e-05,
+      "loss": 2.2068,
+      "step": 713
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995672692826106e-05,
+      "loss": 2.2098,
+      "step": 714
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995651764230057e-05,
+      "loss": 2.2254,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999563078515748e-05,
+      "loss": 2.2083,
+      "step": 716
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999560975560848e-05,
+      "loss": 2.2104,
+      "step": 717
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995588675583168e-05,
+      "loss": 2.2215,
+      "step": 718
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999556754508164e-05,
+      "loss": 2.1983,
+      "step": 719
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995546364104012e-05,
+      "loss": 2.2048,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995525132650388e-05,
+      "loss": 2.1936,
+      "step": 721
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995503850720875e-05,
+      "loss": 2.2134,
+      "step": 722
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995482518315577e-05,
+      "loss": 2.1949,
+      "step": 723
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999546113543461e-05,
+      "loss": 2.218,
+      "step": 724
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995439702078073e-05,
+      "loss": 2.2154,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999541821824608e-05,
+      "loss": 2.2391,
+      "step": 726
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995396683938736e-05,
+      "loss": 2.2033,
+      "step": 727
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995375099156157e-05,
+      "loss": 2.2003,
+      "step": 728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999535346389844e-05,
+      "loss": 2.2155,
+      "step": 729
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995331778165706e-05,
+      "loss": 2.24,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995310041958054e-05,
+      "loss": 2.2049,
+      "step": 731
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995288255275603e-05,
+      "loss": 2.218,
+      "step": 732
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995266418118456e-05,
+      "loss": 2.2383,
+      "step": 733
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995244530486728e-05,
+      "loss": 2.2324,
+      "step": 734
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995222592380526e-05,
+      "loss": 2.2006,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995200603799967e-05,
+      "loss": 2.2222,
+      "step": 736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995178564745155e-05,
+      "loss": 2.221,
+      "step": 737
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99951564752162e-05,
+      "loss": 2.2112,
+      "step": 738
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995134335213225e-05,
+      "loss": 2.2103,
+      "step": 739
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995112144736332e-05,
+      "loss": 2.2286,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995089903785633e-05,
+      "loss": 2.2141,
+      "step": 741
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995067612361243e-05,
+      "loss": 2.2276,
+      "step": 742
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995045270463275e-05,
+      "loss": 2.2368,
+      "step": 743
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999502287809184e-05,
+      "loss": 2.2066,
+      "step": 744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995000435247054e-05,
+      "loss": 2.2329,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994977941929028e-05,
+      "loss": 2.2052,
+      "step": 746
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994955398137877e-05,
+      "loss": 2.2324,
+      "step": 747
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999493280387371e-05,
+      "loss": 2.2402,
+      "step": 748
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999491015913665e-05,
+      "loss": 2.2151,
+      "step": 749
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99948874639268e-05,
+      "loss": 2.2203,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999486471824429e-05,
+      "loss": 2.2622,
+      "step": 751
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999484192208922e-05,
+      "loss": 2.2196,
+      "step": 752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999481907546171e-05,
+      "loss": 2.2082,
+      "step": 753
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999479617836188e-05,
+      "loss": 2.2217,
+      "step": 754
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999477323078984e-05,
+      "loss": 2.2249,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994750232745708e-05,
+      "loss": 2.2087,
+      "step": 756
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99947271842296e-05,
+      "loss": 2.2107,
+      "step": 757
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994704085241633e-05,
+      "loss": 2.2177,
+      "step": 758
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994680935781924e-05,
+      "loss": 2.2149,
+      "step": 759
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994657735850585e-05,
+      "loss": 2.2227,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994634485447738e-05,
+      "loss": 2.2002,
+      "step": 761
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99946111845735e-05,
+      "loss": 2.2088,
+      "step": 762
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999458783322799e-05,
+      "loss": 2.2242,
+      "step": 763
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994564431411324e-05,
+      "loss": 2.2271,
+      "step": 764
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999454097912362e-05,
+      "loss": 2.2151,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994517476364993e-05,
+      "loss": 2.2054,
+      "step": 766
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994493923135568e-05,
+      "loss": 2.221,
+      "step": 767
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999447031943546e-05,
+      "loss": 2.2269,
+      "step": 768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994446665264795e-05,
+      "loss": 2.2183,
+      "step": 769
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994422960623684e-05,
+      "loss": 2.2054,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994399205512246e-05,
+      "loss": 2.2087,
+      "step": 771
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999437539993061e-05,
+      "loss": 2.2005,
+      "step": 772
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994351543878885e-05,
+      "loss": 2.2208,
+      "step": 773
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994327637357202e-05,
+      "loss": 2.2297,
+      "step": 774
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994303680365673e-05,
+      "loss": 2.2155,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994279672904425e-05,
+      "loss": 2.2204,
+      "step": 776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999425561497358e-05,
+      "loss": 2.2417,
+      "step": 777
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994231506573255e-05,
+      "loss": 2.2123,
+      "step": 778
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999420734770357e-05,
+      "loss": 2.2154,
+      "step": 779
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994183138364654e-05,
+      "loss": 2.2248,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994158878556624e-05,
+      "loss": 2.2227,
+      "step": 781
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994134568279607e-05,
+      "loss": 2.2353,
+      "step": 782
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999411020753372e-05,
+      "loss": 2.2015,
+      "step": 783
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999408579631909e-05,
+      "loss": 2.2119,
+      "step": 784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994061334635838e-05,
+      "loss": 2.2229,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999403682248409e-05,
+      "loss": 2.2155,
+      "step": 786
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999401225986397e-05,
+      "loss": 2.2409,
+      "step": 787
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993987646775598e-05,
+      "loss": 2.2229,
+      "step": 788
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993962983219102e-05,
+      "loss": 2.2369,
+      "step": 789
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993938269194608e-05,
+      "loss": 2.2295,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993913504702234e-05,
+      "loss": 2.2385,
+      "step": 791
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993888689742113e-05,
+      "loss": 2.2414,
+      "step": 792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993863824314365e-05,
+      "loss": 2.2107,
+      "step": 793
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993838908419118e-05,
+      "loss": 2.2156,
+      "step": 794
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993813942056495e-05,
+      "loss": 2.2298,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993788925226626e-05,
+      "loss": 2.2316,
+      "step": 796
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993763857929633e-05,
+      "loss": 2.2282,
+      "step": 797
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999373874016565e-05,
+      "loss": 2.2097,
+      "step": 798
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993713571934793e-05,
+      "loss": 2.2175,
+      "step": 799
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993688353237196e-05,
+      "loss": 2.1717,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999366308407299e-05,
+      "loss": 2.2235,
+      "step": 801
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993637764442294e-05,
+      "loss": 2.2335,
+      "step": 802
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993612394345242e-05,
+      "loss": 2.2297,
+      "step": 803
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993586973781957e-05,
+      "loss": 2.2135,
+      "step": 804
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999356150275257e-05,
+      "loss": 2.2271,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993535981257213e-05,
+      "loss": 2.2103,
+      "step": 806
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993510409296012e-05,
+      "loss": 2.2369,
+      "step": 807
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993484786869092e-05,
+      "loss": 2.2021,
+      "step": 808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993459113976586e-05,
+      "loss": 2.2198,
+      "step": 809
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999343339061863e-05,
+      "loss": 2.2328,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993407616795345e-05,
+      "loss": 2.23,
+      "step": 811
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993381792506864e-05,
+      "loss": 2.2421,
+      "step": 812
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993355917753315e-05,
+      "loss": 2.2139,
+      "step": 813
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999332999253483e-05,
+      "loss": 2.2368,
+      "step": 814
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993304016851546e-05,
+      "loss": 2.2298,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993277990703587e-05,
+      "loss": 2.2105,
+      "step": 816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993251914091085e-05,
+      "loss": 2.217,
+      "step": 817
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993225787014173e-05,
+      "loss": 2.2338,
+      "step": 818
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993199609472987e-05,
+      "loss": 2.2475,
+      "step": 819
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999317338146765e-05,
+      "loss": 2.2231,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993147102998307e-05,
+      "loss": 2.2192,
+      "step": 821
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999312077406508e-05,
+      "loss": 2.2167,
+      "step": 822
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993094394668104e-05,
+      "loss": 2.2307,
+      "step": 823
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999306796480751e-05,
+      "loss": 2.2083,
+      "step": 824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993041484483444e-05,
+      "loss": 2.2329,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993014953696027e-05,
+      "loss": 2.2255,
+      "step": 826
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992988372445395e-05,
+      "loss": 2.2187,
+      "step": 827
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992961740731688e-05,
+      "loss": 2.2128,
+      "step": 828
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999293505855503e-05,
+      "loss": 2.2479,
+      "step": 829
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999290832591557e-05,
+      "loss": 2.2303,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992881542813432e-05,
+      "loss": 2.212,
+      "step": 831
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999285470924875e-05,
+      "loss": 2.2328,
+      "step": 832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992827825221673e-05,
+      "loss": 2.1923,
+      "step": 833
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999280089073232e-05,
+      "loss": 2.2179,
+      "step": 834
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999277390578084e-05,
+      "loss": 2.2187,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992746870367365e-05,
+      "loss": 2.2315,
+      "step": 836
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992719784492023e-05,
+      "loss": 2.2286,
+      "step": 837
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999269264815497e-05,
+      "loss": 2.2228,
+      "step": 838
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992665461356322e-05,
+      "loss": 2.2217,
+      "step": 839
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992638224096227e-05,
+      "loss": 2.2064,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992610936374824e-05,
+      "loss": 2.2283,
+      "step": 841
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992583598192244e-05,
+      "loss": 2.2079,
+      "step": 842
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992556209548635e-05,
+      "loss": 2.2176,
+      "step": 843
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992528770444125e-05,
+      "loss": 2.235,
+      "step": 844
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999250128087886e-05,
+      "loss": 2.2305,
+      "step": 845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992473740852975e-05,
+      "loss": 2.2406,
+      "step": 846
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992446150366608e-05,
+      "loss": 2.2111,
+      "step": 847
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992418509419903e-05,
+      "loss": 2.2095,
+      "step": 848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992390818012993e-05,
+      "loss": 2.2345,
+      "step": 849
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992363076146026e-05,
+      "loss": 2.2212,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999233528381914e-05,
+      "loss": 2.2227,
+      "step": 851
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992307441032465e-05,
+      "loss": 2.243,
+      "step": 852
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992279547786152e-05,
+      "loss": 2.2486,
+      "step": 853
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992251604080345e-05,
+      "loss": 2.2232,
+      "step": 854
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992223609915172e-05,
+      "loss": 2.2372,
+      "step": 855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999219556529079e-05,
+      "loss": 2.2539,
+      "step": 856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992167470207326e-05,
+      "loss": 2.2443,
+      "step": 857
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999213932466493e-05,
+      "loss": 2.2285,
+      "step": 858
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992111128663744e-05,
+      "loss": 2.2129,
+      "step": 859
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992082882203912e-05,
+      "loss": 2.249,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999205458528557e-05,
+      "loss": 2.2565,
+      "step": 861
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992026237908863e-05,
+      "loss": 2.2318,
+      "step": 862
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991997840073937e-05,
+      "loss": 2.2141,
+      "step": 863
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991969391780937e-05,
+      "loss": 2.2177,
+      "step": 864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999194089303e-05,
+      "loss": 2.234,
+      "step": 865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999191234382128e-05,
+      "loss": 2.235,
+      "step": 866
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991883744154908e-05,
+      "loss": 2.2144,
+      "step": 867
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991855094031036e-05,
+      "loss": 2.2508,
+      "step": 868
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991826393449808e-05,
+      "loss": 2.2152,
+      "step": 869
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991797642411375e-05,
+      "loss": 2.2253,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999176884091587e-05,
+      "loss": 2.227,
+      "step": 871
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991739988963445e-05,
+      "loss": 2.2425,
+      "step": 872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999171108655425e-05,
+      "loss": 2.2274,
+      "step": 873
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999168213368842e-05,
+      "loss": 2.2322,
+      "step": 874
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999165313036611e-05,
+      "loss": 2.2724,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991624076587465e-05,
+      "loss": 2.2476,
+      "step": 876
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991594972352628e-05,
+      "loss": 2.2066,
+      "step": 877
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991565817661752e-05,
+      "loss": 2.2406,
+      "step": 878
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991536612514978e-05,
+      "loss": 2.2099,
+      "step": 879
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991507356912456e-05,
+      "loss": 2.2333,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991478050854337e-05,
+      "loss": 2.2415,
+      "step": 881
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991448694340762e-05,
+      "loss": 2.2244,
+      "step": 882
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991419287371884e-05,
+      "loss": 2.2351,
+      "step": 883
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991389829947855e-05,
+      "loss": 2.2463,
+      "step": 884
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991360322068815e-05,
+      "loss": 2.23,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999133076373492e-05,
+      "loss": 2.2433,
+      "step": 886
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991301154946313e-05,
+      "loss": 2.2173,
+      "step": 887
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991271495703153e-05,
+      "loss": 2.2345,
+      "step": 888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991241786005584e-05,
+      "loss": 2.2673,
+      "step": 889
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999121202585375e-05,
+      "loss": 2.2345,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999118221524781e-05,
+      "loss": 2.2447,
+      "step": 891
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991152354187916e-05,
+      "loss": 2.2315,
+      "step": 892
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999112244267421e-05,
+      "loss": 2.2533,
+      "step": 893
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991092480706848e-05,
+      "loss": 2.2248,
+      "step": 894
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991062468285982e-05,
+      "loss": 2.2419,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991032405411764e-05,
+      "loss": 2.2392,
+      "step": 896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999100229208434e-05,
+      "loss": 2.2476,
+      "step": 897
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999097212830387e-05,
+      "loss": 2.2605,
+      "step": 898
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990941914070503e-05,
+      "loss": 2.2575,
+      "step": 899
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999091164938439e-05,
+      "loss": 2.2302,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999088133424569e-05,
+      "loss": 2.2571,
+      "step": 901
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990850968654544e-05,
+      "loss": 2.2392,
+      "step": 902
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990820552611118e-05,
+      "loss": 2.2376,
+      "step": 903
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990790086115557e-05,
+      "loss": 2.2229,
+      "step": 904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999075956916802e-05,
+      "loss": 2.2484,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999072900176866e-05,
+      "loss": 2.2357,
+      "step": 906
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999069838391763e-05,
+      "loss": 2.2201,
+      "step": 907
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990667715615082e-05,
+      "loss": 2.2085,
+      "step": 908
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990636996861176e-05,
+      "loss": 2.2474,
+      "step": 909
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990606227656067e-05,
+      "loss": 2.2263,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990575407999907e-05,
+      "loss": 2.2387,
+      "step": 911
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990544537892855e-05,
+      "loss": 2.2248,
+      "step": 912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990513617335064e-05,
+      "loss": 2.2532,
+      "step": 913
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999048264632669e-05,
+      "loss": 2.2334,
+      "step": 914
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999045162486789e-05,
+      "loss": 2.2392,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990420552958824e-05,
+      "loss": 2.2319,
+      "step": 916
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990389430599645e-05,
+      "loss": 2.2648,
+      "step": 917
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990358257790513e-05,
+      "loss": 2.241,
+      "step": 918
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999032703453158e-05,
+      "loss": 2.2562,
+      "step": 919
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990295760823008e-05,
+      "loss": 2.238,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990264436664958e-05,
+      "loss": 2.2364,
+      "step": 921
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999023306205758e-05,
+      "loss": 2.2298,
+      "step": 922
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999020163700104e-05,
+      "loss": 2.2458,
+      "step": 923
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990170161495497e-05,
+      "loss": 2.2309,
+      "step": 924
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99901386355411e-05,
+      "loss": 2.2166,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999010705913802e-05,
+      "loss": 2.2294,
+      "step": 926
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990075432286407e-05,
+      "loss": 2.2507,
+      "step": 927
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990043754986427e-05,
+      "loss": 2.221,
+      "step": 928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990012027238238e-05,
+      "loss": 2.258,
+      "step": 929
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989980249042e-05,
+      "loss": 2.2341,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989948420397876e-05,
+      "loss": 2.2262,
+      "step": 931
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998991654130602e-05,
+      "loss": 2.2388,
+      "step": 932
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989884611766598e-05,
+      "loss": 2.2465,
+      "step": 933
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989852631779776e-05,
+      "loss": 2.2355,
+      "step": 934
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989820601345706e-05,
+      "loss": 2.2432,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989788520464555e-05,
+      "loss": 2.2558,
+      "step": 936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998975638913648e-05,
+      "loss": 2.2204,
+      "step": 937
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989724207361653e-05,
+      "loss": 2.2575,
+      "step": 938
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989691975140228e-05,
+      "loss": 2.2065,
+      "step": 939
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998965969247237e-05,
+      "loss": 2.243,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998962735935824e-05,
+      "loss": 2.2396,
+      "step": 941
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989594975798008e-05,
+      "loss": 2.2317,
+      "step": 942
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998956254179183e-05,
+      "loss": 2.24,
+      "step": 943
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989530057339873e-05,
+      "loss": 2.2385,
+      "step": 944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989497522442306e-05,
+      "loss": 2.2645,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989464937099282e-05,
+      "loss": 2.2215,
+      "step": 946
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989432301310974e-05,
+      "loss": 2.2452,
+      "step": 947
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989399615077548e-05,
+      "loss": 2.2349,
+      "step": 948
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989366878399162e-05,
+      "loss": 2.2271,
+      "step": 949
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998933409127599e-05,
+      "loss": 2.2289,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998930125370819e-05,
+      "loss": 2.2251,
+      "step": 951
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998926836569593e-05,
+      "loss": 2.2347,
+      "step": 952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989235427239374e-05,
+      "loss": 2.239,
+      "step": 953
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989202438338693e-05,
+      "loss": 2.2559,
+      "step": 954
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998916939899405e-05,
+      "loss": 2.2567,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989136309205616e-05,
+      "loss": 2.2399,
+      "step": 956
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989103168973555e-05,
+      "loss": 2.2306,
+      "step": 957
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989069978298036e-05,
+      "loss": 2.2544,
+      "step": 958
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998903673717922e-05,
+      "loss": 2.2267,
+      "step": 959
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989003445617284e-05,
+      "loss": 2.2364,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988970103612396e-05,
+      "loss": 2.2466,
+      "step": 961
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988936711164715e-05,
+      "loss": 2.2373,
+      "step": 962
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998890326827442e-05,
+      "loss": 2.2487,
+      "step": 963
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998886977494167e-05,
+      "loss": 2.2216,
+      "step": 964
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998883623116664e-05,
+      "loss": 2.2666,
+      "step": 965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99888026369495e-05,
+      "loss": 2.249,
+      "step": 966
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998876899229042e-05,
+      "loss": 2.229,
+      "step": 967
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998873529718957e-05,
+      "loss": 2.2406,
+      "step": 968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988701551647117e-05,
+      "loss": 2.2394,
+      "step": 969
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998866775566323e-05,
+      "loss": 2.2618,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988633909238085e-05,
+      "loss": 2.2463,
+      "step": 971
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998860001237185e-05,
+      "loss": 2.229,
+      "step": 972
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988566065064698e-05,
+      "loss": 2.2266,
+      "step": 973
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988532067316796e-05,
+      "loss": 2.253,
+      "step": 974
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988498019128323e-05,
+      "loss": 2.2388,
+      "step": 975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998846392049944e-05,
+      "loss": 2.2373,
+      "step": 976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998842977143033e-05,
+      "loss": 2.2529,
+      "step": 977
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988395571921165e-05,
+      "loss": 2.246,
+      "step": 978
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998836132197211e-05,
+      "loss": 2.2559,
+      "step": 979
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998832702158334e-05,
+      "loss": 2.2223,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988292670755035e-05,
+      "loss": 2.2648,
+      "step": 981
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998825826948736e-05,
+      "loss": 2.2408,
+      "step": 982
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988223817780494e-05,
+      "loss": 2.2134,
+      "step": 983
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998818931563461e-05,
+      "loss": 2.2342,
+      "step": 984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998815476304988e-05,
+      "loss": 2.2438,
+      "step": 985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998812016002648e-05,
+      "loss": 2.221,
+      "step": 986
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988085506564586e-05,
+      "loss": 2.2566,
+      "step": 987
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998805080266437e-05,
+      "loss": 2.2243,
+      "step": 988
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988016048326014e-05,
+      "loss": 2.2615,
+      "step": 989
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987981243549684e-05,
+      "loss": 2.2328,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998794638833556e-05,
+      "loss": 2.2516,
+      "step": 991
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998791148268382e-05,
+      "loss": 2.2657,
+      "step": 992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998787652659464e-05,
+      "loss": 2.2606,
+      "step": 993
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987841520068195e-05,
+      "loss": 2.2473,
+      "step": 994
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987806463104657e-05,
+      "loss": 2.2272,
+      "step": 995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987771355704215e-05,
+      "loss": 2.2432,
+      "step": 996
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987736197867032e-05,
+      "loss": 2.2484,
+      "step": 997
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987700989593297e-05,
+      "loss": 2.2521,
+      "step": 998
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987665730883184e-05,
+      "loss": 2.2728,
+      "step": 999
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998763042173687e-05,
+      "loss": 2.2417,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987595062154533e-05,
+      "loss": 2.2458,
+      "step": 1001
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987559652136354e-05,
+      "loss": 2.2619,
+      "step": 1002
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998752419168251e-05,
+      "loss": 2.2491,
+      "step": 1003
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998748868079318e-05,
+      "loss": 2.2494,
+      "step": 1004
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987453119468542e-05,
+      "loss": 2.2387,
+      "step": 1005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987417507708778e-05,
+      "loss": 2.2447,
+      "step": 1006
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998738184551407e-05,
+      "loss": 2.2441,
+      "step": 1007
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987346132884592e-05,
+      "loss": 2.2342,
+      "step": 1008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998731036982053e-05,
+      "loss": 2.2376,
+      "step": 1009
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998727455632206e-05,
+      "loss": 2.247,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987238692389364e-05,
+      "loss": 2.2216,
+      "step": 1011
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987202778022628e-05,
+      "loss": 2.2381,
+      "step": 1012
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998716681322203e-05,
+      "loss": 2.2552,
+      "step": 1013
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987130797987747e-05,
+      "loss": 2.2384,
+      "step": 1014
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987094732319967e-05,
+      "loss": 2.2638,
+      "step": 1015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998705861621887e-05,
+      "loss": 2.2357,
+      "step": 1016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987022449684635e-05,
+      "loss": 2.2524,
+      "step": 1017
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986986232717452e-05,
+      "loss": 2.243,
+      "step": 1018
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99869499653175e-05,
+      "loss": 2.255,
+      "step": 1019
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986913647484964e-05,
+      "loss": 2.2278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998687727922002e-05,
+      "loss": 2.2523,
+      "step": 1021
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998684086052286e-05,
+      "loss": 2.2593,
+      "step": 1022
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986804391393665e-05,
+      "loss": 2.2221,
+      "step": 1023
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998676787183262e-05,
+      "loss": 2.2527,
+      "step": 1024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998673130183991e-05,
+      "loss": 2.244,
+      "step": 1025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986694681415715e-05,
+      "loss": 2.2381,
+      "step": 1026
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986658010560225e-05,
+      "loss": 2.2472,
+      "step": 1027
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986621289273624e-05,
+      "loss": 2.262,
+      "step": 1028
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986584517556096e-05,
+      "loss": 2.2503,
+      "step": 1029
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986547695407827e-05,
+      "loss": 2.2551,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986510822829004e-05,
+      "loss": 2.248,
+      "step": 1031
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986473899819812e-05,
+      "loss": 2.2435,
+      "step": 1032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998643692638044e-05,
+      "loss": 2.2463,
+      "step": 1033
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986399902511074e-05,
+      "loss": 2.2271,
+      "step": 1034
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986362828211897e-05,
+      "loss": 2.2782,
+      "step": 1035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99863257034831e-05,
+      "loss": 2.2497,
+      "step": 1036
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998628852832487e-05,
+      "loss": 2.2478,
+      "step": 1037
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986251302737393e-05,
+      "loss": 2.2169,
+      "step": 1038
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998621402672086e-05,
+      "loss": 2.2367,
+      "step": 1039
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986176700275456e-05,
+      "loss": 2.2391,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986139323401372e-05,
+      "loss": 2.2477,
+      "step": 1041
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99861018960988e-05,
+      "loss": 2.236,
+      "step": 1042
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986064418367916e-05,
+      "loss": 2.2458,
+      "step": 1043
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986026890208923e-05,
+      "loss": 2.2431,
+      "step": 1044
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985989311622006e-05,
+      "loss": 2.2818,
+      "step": 1045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985951682607353e-05,
+      "loss": 2.2559,
+      "step": 1046
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985914003165155e-05,
+      "loss": 2.2487,
+      "step": 1047
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985876273295603e-05,
+      "loss": 2.2454,
+      "step": 1048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998583849299889e-05,
+      "loss": 2.2422,
+      "step": 1049
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985800662275198e-05,
+      "loss": 2.2767,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985762781124728e-05,
+      "loss": 2.2606,
+      "step": 1051
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985724849547666e-05,
+      "loss": 2.2437,
+      "step": 1052
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99856868675442e-05,
+      "loss": 2.2317,
+      "step": 1053
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985648835114535e-05,
+      "loss": 2.2612,
+      "step": 1054
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985610752258852e-05,
+      "loss": 2.2632,
+      "step": 1055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985572618977347e-05,
+      "loss": 2.2526,
+      "step": 1056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985534435270208e-05,
+      "loss": 2.2693,
+      "step": 1057
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985496201137632e-05,
+      "loss": 2.2379,
+      "step": 1058
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985457916579813e-05,
+      "loss": 2.232,
+      "step": 1059
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985419581596944e-05,
+      "loss": 2.2575,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985381196189214e-05,
+      "loss": 2.2655,
+      "step": 1061
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985342760356823e-05,
+      "loss": 2.2554,
+      "step": 1062
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998530427409996e-05,
+      "loss": 2.2673,
+      "step": 1063
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985265737418824e-05,
+      "loss": 2.2455,
+      "step": 1064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985227150313603e-05,
+      "loss": 2.2495,
+      "step": 1065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985188512784504e-05,
+      "loss": 2.266,
+      "step": 1066
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985149824831706e-05,
+      "loss": 2.2466,
+      "step": 1067
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998511108645542e-05,
+      "loss": 2.2479,
+      "step": 1068
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985072297655827e-05,
+      "loss": 2.2479,
+      "step": 1069
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985033458433133e-05,
+      "loss": 2.2447,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984994568787533e-05,
+      "loss": 2.2362,
+      "step": 1071
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984955628719217e-05,
+      "loss": 2.2289,
+      "step": 1072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984916638228392e-05,
+      "loss": 2.2473,
+      "step": 1073
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984877597315243e-05,
+      "loss": 2.2405,
+      "step": 1074
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998483850597998e-05,
+      "loss": 2.2579,
+      "step": 1075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998479936422279e-05,
+      "loss": 2.2579,
+      "step": 1076
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984760172043877e-05,
+      "loss": 2.2503,
+      "step": 1077
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984720929443432e-05,
+      "loss": 2.2405,
+      "step": 1078
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984681636421657e-05,
+      "loss": 2.2499,
+      "step": 1079
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984642292978752e-05,
+      "loss": 2.2366,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998460289911492e-05,
+      "loss": 2.249,
+      "step": 1081
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984563454830347e-05,
+      "loss": 2.2495,
+      "step": 1082
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984523960125247e-05,
+      "loss": 2.2361,
+      "step": 1083
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984484414999806e-05,
+      "loss": 2.2393,
+      "step": 1084
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998444481945423e-05,
+      "loss": 2.261,
+      "step": 1085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984405173488724e-05,
+      "loss": 2.2617,
+      "step": 1086
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984365477103477e-05,
+      "loss": 2.2728,
+      "step": 1087
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984325730298698e-05,
+      "loss": 2.2468,
+      "step": 1088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984285933074585e-05,
+      "loss": 2.2462,
+      "step": 1089
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984246085431337e-05,
+      "loss": 2.249,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998420618736916e-05,
+      "loss": 2.256,
+      "step": 1091
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984166238888253e-05,
+      "loss": 2.2542,
+      "step": 1092
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984126239988817e-05,
+      "loss": 2.2242,
+      "step": 1093
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984086190671056e-05,
+      "loss": 2.235,
+      "step": 1094
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998404609093517e-05,
+      "loss": 2.2468,
+      "step": 1095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998400594078136e-05,
+      "loss": 2.2465,
+      "step": 1096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983965740209836e-05,
+      "loss": 2.2666,
+      "step": 1097
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998392548922079e-05,
+      "loss": 2.2599,
+      "step": 1098
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998388518781443e-05,
+      "loss": 2.2676,
+      "step": 1099
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983844835990965e-05,
+      "loss": 2.2734,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983804433750594e-05,
+      "loss": 2.2219,
+      "step": 1101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998376398109352e-05,
+      "loss": 2.2647,
+      "step": 1102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983723478019953e-05,
+      "loss": 2.2765,
+      "step": 1103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983682924530093e-05,
+      "loss": 2.2649,
+      "step": 1104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983642320624144e-05,
+      "loss": 2.2729,
+      "step": 1105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998360166630231e-05,
+      "loss": 2.267,
+      "step": 1106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983560961564802e-05,
+      "loss": 2.2529,
+      "step": 1107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998352020641182e-05,
+      "loss": 2.2744,
+      "step": 1108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983479400843572e-05,
+      "loss": 2.2594,
+      "step": 1109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983438544860262e-05,
+      "loss": 2.2571,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99833976384621e-05,
+      "loss": 2.2494,
+      "step": 1111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983356681649292e-05,
+      "loss": 2.273,
+      "step": 1112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9983315674422042e-05,
+      "loss": 2.2556,
+      "step": 1113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998327461678056e-05,
+      "loss": 2.2612,
+      "step": 1114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998323350872505e-05,
+      "loss": 2.2338,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9983192350255723e-05,
+      "loss": 2.2484,
+      "step": 1116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998315114137279e-05,
+      "loss": 2.2623,
+      "step": 1117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998310988207645e-05,
+      "loss": 2.2416,
+      "step": 1118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9983068572366916e-05,
+      "loss": 2.2556,
+      "step": 1119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9983027212244395e-05,
+      "loss": 2.2653,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99829858017091e-05,
+      "loss": 2.24,
+      "step": 1121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982944340761238e-05,
+      "loss": 2.2526,
+      "step": 1122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982902829401015e-05,
+      "loss": 2.2666,
+      "step": 1123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982861267628647e-05,
+      "loss": 2.241,
+      "step": 1124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998281965544434e-05,
+      "loss": 2.2491,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982777992848302e-05,
+      "loss": 2.249,
+      "step": 1126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982736279840746e-05,
+      "loss": 2.2546,
+      "step": 1127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982694516421884e-05,
+      "loss": 2.2394,
+      "step": 1128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998265270259193e-05,
+      "loss": 2.2338,
+      "step": 1129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998261083835108e-05,
+      "loss": 2.2567,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982568923699566e-05,
+      "loss": 2.2436,
+      "step": 1131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982526958637587e-05,
+      "loss": 2.2555,
+      "step": 1132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982484943165356e-05,
+      "loss": 2.2637,
+      "step": 1133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982442877283084e-05,
+      "loss": 2.2365,
+      "step": 1134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982400760990992e-05,
+      "loss": 2.2407,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982358594289285e-05,
+      "loss": 2.2728,
+      "step": 1136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982316377178178e-05,
+      "loss": 2.2569,
+      "step": 1137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982274109657884e-05,
+      "loss": 2.2375,
+      "step": 1138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982231791728616e-05,
+      "loss": 2.2537,
+      "step": 1139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998218942339059e-05,
+      "loss": 2.2567,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982147004644016e-05,
+      "loss": 2.2833,
+      "step": 1141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982104535489107e-05,
+      "loss": 2.2708,
+      "step": 1142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982062015926085e-05,
+      "loss": 2.2453,
+      "step": 1143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982019445955164e-05,
+      "loss": 2.2521,
+      "step": 1144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981976825576553e-05,
+      "loss": 2.2462,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981934154790466e-05,
+      "loss": 2.2495,
+      "step": 1146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981891433597126e-05,
+      "loss": 2.247,
+      "step": 1147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981848661996744e-05,
+      "loss": 2.2452,
+      "step": 1148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981805839989536e-05,
+      "loss": 2.2752,
+      "step": 1149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998176296757572e-05,
+      "loss": 2.2575,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998172004475551e-05,
+      "loss": 2.2326,
+      "step": 1151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998167707152913e-05,
+      "loss": 2.2519,
+      "step": 1152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981634047896784e-05,
+      "loss": 2.2449,
+      "step": 1153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99815909738587e-05,
+      "loss": 2.266,
+      "step": 1154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981547849415092e-05,
+      "loss": 2.2425,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998150467456618e-05,
+      "loss": 2.2536,
+      "step": 1156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998146144931218e-05,
+      "loss": 2.2658,
+      "step": 1157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981418173653305e-05,
+      "loss": 2.2416,
+      "step": 1158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981374847589784e-05,
+      "loss": 2.2636,
+      "step": 1159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998133147112183e-05,
+      "loss": 2.2461,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981288044249664e-05,
+      "loss": 2.2319,
+      "step": 1161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981244566973502e-05,
+      "loss": 2.2392,
+      "step": 1162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981201039293564e-05,
+      "loss": 2.2461,
+      "step": 1163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981157461210073e-05,
+      "loss": 2.2604,
+      "step": 1164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981113832723247e-05,
+      "loss": 2.2439,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981070153833305e-05,
+      "loss": 2.2461,
+      "step": 1166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981026424540474e-05,
+      "loss": 2.2534,
+      "step": 1167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980982644844968e-05,
+      "loss": 2.2612,
+      "step": 1168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980938814747007e-05,
+      "loss": 2.2383,
+      "step": 1169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980894934246818e-05,
+      "loss": 2.2561,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998085100334462e-05,
+      "loss": 2.2608,
+      "step": 1171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980807022040637e-05,
+      "loss": 2.2661,
+      "step": 1172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998076299033509e-05,
+      "loss": 2.2421,
+      "step": 1173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980718908228197e-05,
+      "loss": 2.2648,
+      "step": 1174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980674775720185e-05,
+      "loss": 2.2647,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980630592811274e-05,
+      "loss": 2.248,
+      "step": 1176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998058635950169e-05,
+      "loss": 2.2452,
+      "step": 1177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980542075791655e-05,
+      "loss": 2.2599,
+      "step": 1178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980497741681396e-05,
+      "loss": 2.2669,
+      "step": 1179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998045335717113e-05,
+      "loss": 2.2571,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998040892226109e-05,
+      "loss": 2.2686,
+      "step": 1181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998036443695149e-05,
+      "loss": 2.2486,
+      "step": 1182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980319901242562e-05,
+      "loss": 2.2683,
+      "step": 1183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998027531513453e-05,
+      "loss": 2.2696,
+      "step": 1184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980230678627616e-05,
+      "loss": 2.2484,
+      "step": 1185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998018599172205e-05,
+      "loss": 2.2682,
+      "step": 1186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980141254418056e-05,
+      "loss": 2.2603,
+      "step": 1187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980096466715856e-05,
+      "loss": 2.252,
+      "step": 1188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998005162861568e-05,
+      "loss": 2.2272,
+      "step": 1189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980006740117757e-05,
+      "loss": 2.2532,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979961801222307e-05,
+      "loss": 2.2468,
+      "step": 1191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997991681192956e-05,
+      "loss": 2.2553,
+      "step": 1192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979871772239744e-05,
+      "loss": 2.2593,
+      "step": 1193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997982668215309e-05,
+      "loss": 2.2694,
+      "step": 1194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979781541669817e-05,
+      "loss": 2.2438,
+      "step": 1195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979736350790157e-05,
+      "loss": 2.2613,
+      "step": 1196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997969110951434e-05,
+      "loss": 2.2394,
+      "step": 1197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979645817842597e-05,
+      "loss": 2.2515,
+      "step": 1198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979600475775148e-05,
+      "loss": 2.2476,
+      "step": 1199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979555083312227e-05,
+      "loss": 2.2732,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979509640454068e-05,
+      "loss": 2.2612,
+      "step": 1201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997946414720089e-05,
+      "loss": 2.2657,
+      "step": 1202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997941860355293e-05,
+      "loss": 2.2566,
+      "step": 1203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979373009510418e-05,
+      "loss": 2.2629,
+      "step": 1204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979327365073583e-05,
+      "loss": 2.2457,
+      "step": 1205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979281670242655e-05,
+      "loss": 2.2786,
+      "step": 1206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979235925017864e-05,
+      "loss": 2.2667,
+      "step": 1207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997919012939944e-05,
+      "loss": 2.2625,
+      "step": 1208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997914428338762e-05,
+      "loss": 2.2742,
+      "step": 1209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997909838698263e-05,
+      "loss": 2.2513,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99790524401847e-05,
+      "loss": 2.2657,
+      "step": 1211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997900644299407e-05,
+      "loss": 2.2596,
+      "step": 1212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978960395410965e-05,
+      "loss": 2.2636,
+      "step": 1213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978914297435624e-05,
+      "loss": 2.2597,
+      "step": 1214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978868149068273e-05,
+      "loss": 2.2537,
+      "step": 1215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978821950309147e-05,
+      "loss": 2.2441,
+      "step": 1216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978775701158482e-05,
+      "loss": 2.2577,
+      "step": 1217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997872940161651e-05,
+      "loss": 2.2641,
+      "step": 1218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978683051683466e-05,
+      "loss": 2.2504,
+      "step": 1219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978636651359582e-05,
+      "loss": 2.2664,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997859020064509e-05,
+      "loss": 2.2643,
+      "step": 1221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997854369954023e-05,
+      "loss": 2.2557,
+      "step": 1222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997849714804524e-05,
+      "loss": 2.2412,
+      "step": 1223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997845054616034e-05,
+      "loss": 2.2515,
+      "step": 1224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978403893885782e-05,
+      "loss": 2.259,
+      "step": 1225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997835719122179e-05,
+      "loss": 2.2648,
+      "step": 1226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978310438168608e-05,
+      "loss": 2.2492,
+      "step": 1227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978263634726464e-05,
+      "loss": 2.251,
+      "step": 1228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978216780895603e-05,
+      "loss": 2.2523,
+      "step": 1229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978169876676254e-05,
+      "loss": 2.2506,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997812292206866e-05,
+      "loss": 2.2819,
+      "step": 1231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978075917073054e-05,
+      "loss": 2.255,
+      "step": 1232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978028861689674e-05,
+      "loss": 2.2523,
+      "step": 1233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997798175591876e-05,
+      "loss": 2.2303,
+      "step": 1234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997793459976055e-05,
+      "loss": 2.2787,
+      "step": 1235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977887393215278e-05,
+      "loss": 2.2338,
+      "step": 1236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977840136283183e-05,
+      "loss": 2.2656,
+      "step": 1237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997779282896451e-05,
+      "loss": 2.2272,
+      "step": 1238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997774547125949e-05,
+      "loss": 2.2649,
+      "step": 1239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977698063168364e-05,
+      "loss": 2.2642,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977650604691373e-05,
+      "loss": 2.247,
+      "step": 1241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997760309582876e-05,
+      "loss": 2.2349,
+      "step": 1242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997755553658076e-05,
+      "loss": 2.2653,
+      "step": 1243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977507926947615e-05,
+      "loss": 2.2462,
+      "step": 1244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977460266929562e-05,
+      "loss": 2.2312,
+      "step": 1245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997741255652685e-05,
+      "loss": 2.2467,
+      "step": 1246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997736479573971e-05,
+      "loss": 2.2528,
+      "step": 1247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997731698456839e-05,
+      "loss": 2.2454,
+      "step": 1248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997726912301313e-05,
+      "loss": 2.2534,
+      "step": 1249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977221211074166e-05,
+      "loss": 2.2492,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977173248751752e-05,
+      "loss": 2.2486,
+      "step": 1251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977125236046118e-05,
+      "loss": 2.2587,
+      "step": 1252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977077172957513e-05,
+      "loss": 2.2426,
+      "step": 1253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997702905948618e-05,
+      "loss": 2.226,
+      "step": 1254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976980895632358e-05,
+      "loss": 2.2707,
+      "step": 1255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976932681396294e-05,
+      "loss": 2.2726,
+      "step": 1256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976884416778226e-05,
+      "loss": 2.2558,
+      "step": 1257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976836101778405e-05,
+      "loss": 2.2639,
+      "step": 1258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997678773639707e-05,
+      "loss": 2.2631,
+      "step": 1259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976739320634467e-05,
+      "loss": 2.2518,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976690854490846e-05,
+      "loss": 2.2498,
+      "step": 1261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997664233796644e-05,
+      "loss": 2.257,
+      "step": 1262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99765937710615e-05,
+      "loss": 2.2678,
+      "step": 1263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976545153776274e-05,
+      "loss": 2.255,
+      "step": 1264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976496486111002e-05,
+      "loss": 2.2886,
+      "step": 1265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976447768065935e-05,
+      "loss": 2.2719,
+      "step": 1266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976398999641317e-05,
+      "loss": 2.2678,
+      "step": 1267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976350180837395e-05,
+      "loss": 2.27,
+      "step": 1268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997630131165441e-05,
+      "loss": 2.2515,
+      "step": 1269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976252392092618e-05,
+      "loss": 2.2596,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976203422152254e-05,
+      "loss": 2.2641,
+      "step": 1271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997615440183358e-05,
+      "loss": 2.255,
+      "step": 1272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976105331136835e-05,
+      "loss": 2.2693,
+      "step": 1273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976056210062266e-05,
+      "loss": 2.2685,
+      "step": 1274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976007038610122e-05,
+      "loss": 2.2422,
+      "step": 1275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975957816780656e-05,
+      "loss": 2.2589,
+      "step": 1276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997590854457411e-05,
+      "loss": 2.2637,
+      "step": 1277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975859221990738e-05,
+      "loss": 2.2401,
+      "step": 1278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975809849030784e-05,
+      "loss": 2.2558,
+      "step": 1279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99757604256945e-05,
+      "loss": 2.2533,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975710951982135e-05,
+      "loss": 2.2601,
+      "step": 1281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975661427893942e-05,
+      "loss": 2.2469,
+      "step": 1282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997561185343017e-05,
+      "loss": 2.2775,
+      "step": 1283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975562228591063e-05,
+      "loss": 2.2756,
+      "step": 1284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997551255337688e-05,
+      "loss": 2.2364,
+      "step": 1285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975462827787866e-05,
+      "loss": 2.2679,
+      "step": 1286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975413051824275e-05,
+      "loss": 2.2853,
+      "step": 1287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975363225486358e-05,
+      "loss": 2.2619,
+      "step": 1288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975313348774368e-05,
+      "loss": 2.2709,
+      "step": 1289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975263421688553e-05,
+      "loss": 2.2572,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997521344422917e-05,
+      "loss": 2.2635,
+      "step": 1291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975163416396467e-05,
+      "loss": 2.248,
+      "step": 1292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975113338190698e-05,
+      "loss": 2.2762,
+      "step": 1293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997506320961212e-05,
+      "loss": 2.2396,
+      "step": 1294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997501303066098e-05,
+      "loss": 2.2471,
+      "step": 1295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974962801337533e-05,
+      "loss": 2.2428,
+      "step": 1296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974912521642032e-05,
+      "loss": 2.2598,
+      "step": 1297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974862191574736e-05,
+      "loss": 2.2552,
+      "step": 1298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997481181113589e-05,
+      "loss": 2.256,
+      "step": 1299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997476138032576e-05,
+      "loss": 2.2698,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974710899144594e-05,
+      "loss": 2.256,
+      "step": 1301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974660367592646e-05,
+      "loss": 2.2521,
+      "step": 1302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974609785670172e-05,
+      "loss": 2.2402,
+      "step": 1303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997455915337743e-05,
+      "loss": 2.2481,
+      "step": 1304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974508470714673e-05,
+      "loss": 2.2715,
+      "step": 1305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974457737682158e-05,
+      "loss": 2.2437,
+      "step": 1306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974406954280142e-05,
+      "loss": 2.2492,
+      "step": 1307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974356120508878e-05,
+      "loss": 2.2516,
+      "step": 1308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974305236368626e-05,
+      "loss": 2.2516,
+      "step": 1309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974254301859642e-05,
+      "loss": 2.2642,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974203316982183e-05,
+      "loss": 2.2445,
+      "step": 1311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997415228173651e-05,
+      "loss": 2.2424,
+      "step": 1312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997410119612287e-05,
+      "loss": 2.2657,
+      "step": 1313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997405006014154e-05,
+      "loss": 2.2807,
+      "step": 1314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973998873792758e-05,
+      "loss": 2.2275,
+      "step": 1315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973947637076794e-05,
+      "loss": 2.2443,
+      "step": 1316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973896349993903e-05,
+      "loss": 2.2381,
+      "step": 1317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973845012544343e-05,
+      "loss": 2.2462,
+      "step": 1318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997379362472838e-05,
+      "loss": 2.2518,
+      "step": 1319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973742186546267e-05,
+      "loss": 2.2675,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973690697998263e-05,
+      "loss": 2.251,
+      "step": 1321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973639159084634e-05,
+      "loss": 2.2559,
+      "step": 1322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973587569805636e-05,
+      "loss": 2.2296,
+      "step": 1323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973535930161527e-05,
+      "loss": 2.2648,
+      "step": 1324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997348424015257e-05,
+      "loss": 2.2401,
+      "step": 1325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973432499779033e-05,
+      "loss": 2.2566,
+      "step": 1326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973380709041166e-05,
+      "loss": 2.2439,
+      "step": 1327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997332886793924e-05,
+      "loss": 2.2509,
+      "step": 1328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997327697647351e-05,
+      "loss": 2.2541,
+      "step": 1329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973225034644243e-05,
+      "loss": 2.2534,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973173042451695e-05,
+      "loss": 2.2487,
+      "step": 1331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973120999896138e-05,
+      "loss": 2.2501,
+      "step": 1332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973068906977824e-05,
+      "loss": 2.2502,
+      "step": 1333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973016763697025e-05,
+      "loss": 2.2758,
+      "step": 1334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972964570054e-05,
+      "loss": 2.265,
+      "step": 1335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997291232604901e-05,
+      "loss": 2.2413,
+      "step": 1336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972860031682323e-05,
+      "loss": 2.2681,
+      "step": 1337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972807686954205e-05,
+      "loss": 2.2562,
+      "step": 1338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972755291864917e-05,
+      "loss": 2.2514,
+      "step": 1339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972702846414723e-05,
+      "loss": 2.2757,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997265035060389e-05,
+      "loss": 2.2289,
+      "step": 1341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997259780443268e-05,
+      "loss": 2.2561,
+      "step": 1342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972545207901362e-05,
+      "loss": 2.2585,
+      "step": 1343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99724925610102e-05,
+      "loss": 2.2574,
+      "step": 1344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997243986375946e-05,
+      "loss": 2.2431,
+      "step": 1345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972387116149406e-05,
+      "loss": 2.2467,
+      "step": 1346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972334318180308e-05,
+      "loss": 2.2603,
+      "step": 1347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972281469852432e-05,
+      "loss": 2.2694,
+      "step": 1348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972228571166044e-05,
+      "loss": 2.2529,
+      "step": 1349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972175622121407e-05,
+      "loss": 2.2589,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972122622718797e-05,
+      "loss": 2.2424,
+      "step": 1351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972069572958476e-05,
+      "loss": 2.2419,
+      "step": 1352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972016472840708e-05,
+      "loss": 2.2472,
+      "step": 1353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997196332236577e-05,
+      "loss": 2.2718,
+      "step": 1354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971910121533925e-05,
+      "loss": 2.2582,
+      "step": 1355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971856870345446e-05,
+      "loss": 2.2694,
+      "step": 1356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971803568800597e-05,
+      "loss": 2.2422,
+      "step": 1357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997175021689965e-05,
+      "loss": 2.2528,
+      "step": 1358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971696814642875e-05,
+      "loss": 2.2577,
+      "step": 1359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971643362030536e-05,
+      "loss": 2.2592,
+      "step": 1360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997158985906291e-05,
+      "loss": 2.262,
+      "step": 1361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971536305740264e-05,
+      "loss": 2.269,
+      "step": 1362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971482702062866e-05,
+      "loss": 2.252,
+      "step": 1363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971429048030996e-05,
+      "loss": 2.261,
+      "step": 1364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997137534364491e-05,
+      "loss": 2.2708,
+      "step": 1365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971321588904895e-05,
+      "loss": 2.2284,
+      "step": 1366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971267783811214e-05,
+      "loss": 2.249,
+      "step": 1367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971213928364134e-05,
+      "loss": 2.2658,
+      "step": 1368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971160022563938e-05,
+      "loss": 2.269,
+      "step": 1369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971106066410892e-05,
+      "loss": 2.2452,
+      "step": 1370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971052059905268e-05,
+      "loss": 2.2497,
+      "step": 1371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997099800304734e-05,
+      "loss": 2.2545,
+      "step": 1372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970943895837386e-05,
+      "loss": 2.2511,
+      "step": 1373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997088973827567e-05,
+      "loss": 2.246,
+      "step": 1374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970835530362468e-05,
+      "loss": 2.2521,
+      "step": 1375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997078127209806e-05,
+      "loss": 2.2315,
+      "step": 1376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970726963482714e-05,
+      "loss": 2.2496,
+      "step": 1377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970672604516706e-05,
+      "loss": 2.2591,
+      "step": 1378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997061819520031e-05,
+      "loss": 2.2523,
+      "step": 1379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99705637355338e-05,
+      "loss": 2.2456,
+      "step": 1380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970509225517453e-05,
+      "loss": 2.2546,
+      "step": 1381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970454665151543e-05,
+      "loss": 2.2545,
+      "step": 1382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997040005443635e-05,
+      "loss": 2.264,
+      "step": 1383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997034539337214e-05,
+      "loss": 2.2605,
+      "step": 1384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970290681959196e-05,
+      "loss": 2.2725,
+      "step": 1385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970235920197793e-05,
+      "loss": 2.2547,
+      "step": 1386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970181108088206e-05,
+      "loss": 2.2644,
+      "step": 1387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997012624563072e-05,
+      "loss": 2.2528,
+      "step": 1388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99700713328256e-05,
+      "loss": 2.2557,
+      "step": 1389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970016369673128e-05,
+      "loss": 2.2465,
+      "step": 1390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969961356173587e-05,
+      "loss": 2.2599,
+      "step": 1391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969906292327245e-05,
+      "loss": 2.2656,
+      "step": 1392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969851178134386e-05,
+      "loss": 2.2491,
+      "step": 1393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996979601359529e-05,
+      "loss": 2.2688,
+      "step": 1394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996974079871023e-05,
+      "loss": 2.2405,
+      "step": 1395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996968553347949e-05,
+      "loss": 2.2482,
+      "step": 1396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969630217903347e-05,
+      "loss": 2.2672,
+      "step": 1397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996957485198208e-05,
+      "loss": 2.2643,
+      "step": 1398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996951943571597e-05,
+      "loss": 2.2484,
+      "step": 1399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969463969105294e-05,
+      "loss": 2.2403,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969408452150332e-05,
+      "loss": 2.2484,
+      "step": 1401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996935288485137e-05,
+      "loss": 2.2529,
+      "step": 1402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996929726720868e-05,
+      "loss": 2.2472,
+      "step": 1403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996924159922255e-05,
+      "loss": 2.2668,
+      "step": 1404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996918588089326e-05,
+      "loss": 2.2567,
+      "step": 1405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969130112221088e-05,
+      "loss": 2.2828,
+      "step": 1406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996907429320632e-05,
+      "loss": 2.257,
+      "step": 1407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969018423849236e-05,
+      "loss": 2.2875,
+      "step": 1408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968962504150115e-05,
+      "loss": 2.2321,
+      "step": 1409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968906534109243e-05,
+      "loss": 2.2271,
+      "step": 1410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99688505137269e-05,
+      "loss": 2.2406,
+      "step": 1411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968794443003376e-05,
+      "loss": 2.2678,
+      "step": 1412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968738321938942e-05,
+      "loss": 2.2611,
+      "step": 1413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996868215053389e-05,
+      "loss": 2.2409,
+      "step": 1414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996862592878851e-05,
+      "loss": 2.2365,
+      "step": 1415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968569656703068e-05,
+      "loss": 2.2707,
+      "step": 1416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996851333427786e-05,
+      "loss": 2.2684,
+      "step": 1417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996845696151317e-05,
+      "loss": 2.2701,
+      "step": 1418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996840053840928e-05,
+      "loss": 2.2743,
+      "step": 1419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968344064966475e-05,
+      "loss": 2.267,
+      "step": 1420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968287541185045e-05,
+      "loss": 2.2636,
+      "step": 1421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968230967065266e-05,
+      "loss": 2.2882,
+      "step": 1422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968174342607434e-05,
+      "loss": 2.2609,
+      "step": 1423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968117667811827e-05,
+      "loss": 2.2469,
+      "step": 1424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996806094267874e-05,
+      "loss": 2.2466,
+      "step": 1425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968004167208444e-05,
+      "loss": 2.2523,
+      "step": 1426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967947341401245e-05,
+      "loss": 2.2616,
+      "step": 1427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967890465257415e-05,
+      "loss": 2.2427,
+      "step": 1428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967833538777247e-05,
+      "loss": 2.257,
+      "step": 1429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996777656196103e-05,
+      "loss": 2.2721,
+      "step": 1430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996771953480905e-05,
+      "loss": 2.2378,
+      "step": 1431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967662457321597e-05,
+      "loss": 2.2633,
+      "step": 1432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967605329498953e-05,
+      "loss": 2.257,
+      "step": 1433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967548151341412e-05,
+      "loss": 2.2531,
+      "step": 1434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967490922849265e-05,
+      "loss": 2.2439,
+      "step": 1435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996743364402279e-05,
+      "loss": 2.2561,
+      "step": 1436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967376314862288e-05,
+      "loss": 2.2881,
+      "step": 1437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967318935368045e-05,
+      "loss": 2.2591,
+      "step": 1438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996726150554035e-05,
+      "loss": 2.2667,
+      "step": 1439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996720402537949e-05,
+      "loss": 2.2523,
+      "step": 1440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996714649488576e-05,
+      "loss": 2.2572,
+      "step": 1441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967088914059448e-05,
+      "loss": 2.2746,
+      "step": 1442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967031282900848e-05,
+      "loss": 2.2436,
+      "step": 1443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966973601410245e-05,
+      "loss": 2.2385,
+      "step": 1444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966915869587935e-05,
+      "loss": 2.241,
+      "step": 1445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966858087434205e-05,
+      "loss": 2.2748,
+      "step": 1446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966800254949355e-05,
+      "loss": 2.2528,
+      "step": 1447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996674237213367e-05,
+      "loss": 2.2491,
+      "step": 1448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966684438987445e-05,
+      "loss": 2.2399,
+      "step": 1449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966626455510972e-05,
+      "loss": 2.2687,
+      "step": 1450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966568421704543e-05,
+      "loss": 2.2607,
+      "step": 1451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966510337568454e-05,
+      "loss": 2.2871,
+      "step": 1452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966452203102995e-05,
+      "loss": 2.262,
+      "step": 1453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996639401830846e-05,
+      "loss": 2.2723,
+      "step": 1454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966335783185143e-05,
+      "loss": 2.2452,
+      "step": 1455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996627749773334e-05,
+      "loss": 2.2735,
+      "step": 1456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996621916195334e-05,
+      "loss": 2.2359,
+      "step": 1457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966160775845448e-05,
+      "loss": 2.2618,
+      "step": 1458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966102339409948e-05,
+      "loss": 2.2689,
+      "step": 1459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966043852647142e-05,
+      "loss": 2.283,
+      "step": 1460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996598531555732e-05,
+      "loss": 2.2617,
+      "step": 1461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965926728140786e-05,
+      "loss": 2.2418,
+      "step": 1462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965868090397827e-05,
+      "loss": 2.2794,
+      "step": 1463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965809402328743e-05,
+      "loss": 2.2585,
+      "step": 1464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965750663933827e-05,
+      "loss": 2.2747,
+      "step": 1465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996569187521338e-05,
+      "loss": 2.2194,
+      "step": 1466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965633036167698e-05,
+      "loss": 2.2468,
+      "step": 1467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965574146797077e-05,
+      "loss": 2.2703,
+      "step": 1468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965515207101817e-05,
+      "loss": 2.2582,
+      "step": 1469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996545621708221e-05,
+      "loss": 2.2632,
+      "step": 1470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965397176738556e-05,
+      "loss": 2.2268,
+      "step": 1471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996533808607116e-05,
+      "loss": 2.2605,
+      "step": 1472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965278945080307e-05,
+      "loss": 2.251,
+      "step": 1473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996521975376631e-05,
+      "loss": 2.2531,
+      "step": 1474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996516051212946e-05,
+      "loss": 2.2601,
+      "step": 1475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996510122017006e-05,
+      "loss": 2.2531,
+      "step": 1476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99650418778884e-05,
+      "loss": 2.2522,
+      "step": 1477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964982485284793e-05,
+      "loss": 2.2462,
+      "step": 1478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996492304235953e-05,
+      "loss": 2.2616,
+      "step": 1479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964863549112914e-05,
+      "loss": 2.2773,
+      "step": 1480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964804005545247e-05,
+      "loss": 2.2695,
+      "step": 1481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964744411656826e-05,
+      "loss": 2.2464,
+      "step": 1482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964684767447956e-05,
+      "loss": 2.2681,
+      "step": 1483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964625072918932e-05,
+      "loss": 2.2457,
+      "step": 1484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996456532807006e-05,
+      "loss": 2.2426,
+      "step": 1485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964505532901643e-05,
+      "loss": 2.2468,
+      "step": 1486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964445687413984e-05,
+      "loss": 2.2515,
+      "step": 1487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964385791607377e-05,
+      "loss": 2.2601,
+      "step": 1488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964325845482135e-05,
+      "loss": 2.2547,
+      "step": 1489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964265849038547e-05,
+      "loss": 2.2354,
+      "step": 1490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964205802276934e-05,
+      "loss": 2.2699,
+      "step": 1491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964145705197582e-05,
+      "loss": 2.2388,
+      "step": 1492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996408555780081e-05,
+      "loss": 2.256,
+      "step": 1493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996402536008691e-05,
+      "loss": 2.2482,
+      "step": 1494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996396511205619e-05,
+      "loss": 2.2581,
+      "step": 1495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996390481370895e-05,
+      "loss": 2.2482,
+      "step": 1496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963844465045507e-05,
+      "loss": 2.2519,
+      "step": 1497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996378406606615e-05,
+      "loss": 2.2498,
+      "step": 1498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963723616771197e-05,
+      "loss": 2.2538,
+      "step": 1499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963663117160945e-05,
+      "loss": 2.262,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963602567235703e-05,
+      "loss": 2.2867,
+      "step": 1501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963541966995776e-05,
+      "loss": 2.2522,
+      "step": 1502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963481316441468e-05,
+      "loss": 2.2686,
+      "step": 1503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996342061557309e-05,
+      "loss": 2.237,
+      "step": 1504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963359864390943e-05,
+      "loss": 2.2582,
+      "step": 1505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996329906289534e-05,
+      "loss": 2.2467,
+      "step": 1506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963238211086583e-05,
+      "loss": 2.2421,
+      "step": 1507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996317730896498e-05,
+      "loss": 2.2652,
+      "step": 1508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996311635653084e-05,
+      "loss": 2.2424,
+      "step": 1509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996305535378447e-05,
+      "loss": 2.2441,
+      "step": 1510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996299430072618e-05,
+      "loss": 2.2176,
+      "step": 1511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962933197356273e-05,
+      "loss": 2.2521,
+      "step": 1512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962872043675063e-05,
+      "loss": 2.2585,
+      "step": 1513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996281083968286e-05,
+      "loss": 2.247,
+      "step": 1514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962749585379966e-05,
+      "loss": 2.258,
+      "step": 1515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962688280766697e-05,
+      "loss": 2.2475,
+      "step": 1516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962626925843358e-05,
+      "loss": 2.2564,
+      "step": 1517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962565520610263e-05,
+      "loss": 2.262,
+      "step": 1518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962504065067717e-05,
+      "loss": 2.2521,
+      "step": 1519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962442559216036e-05,
+      "loss": 2.2559,
+      "step": 1520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962381003055524e-05,
+      "loss": 2.261,
+      "step": 1521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99623193965865e-05,
+      "loss": 2.2542,
+      "step": 1522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996225773980927e-05,
+      "loss": 2.2621,
+      "step": 1523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962196032724145e-05,
+      "loss": 2.2667,
+      "step": 1524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962134275331437e-05,
+      "loss": 2.257,
+      "step": 1525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996207246763146e-05,
+      "loss": 2.2703,
+      "step": 1526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962010609624524e-05,
+      "loss": 2.2648,
+      "step": 1527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996194870131094e-05,
+      "loss": 2.2416,
+      "step": 1528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996188674269103e-05,
+      "loss": 2.2594,
+      "step": 1529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961824733765094e-05,
+      "loss": 2.2966,
+      "step": 1530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961762674533453e-05,
+      "loss": 2.2457,
+      "step": 1531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961700564996413e-05,
+      "loss": 2.2453,
+      "step": 1532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961638405154296e-05,
+      "loss": 2.2691,
+      "step": 1533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961576195007415e-05,
+      "loss": 2.2621,
+      "step": 1534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996151393455608e-05,
+      "loss": 2.2392,
+      "step": 1535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961451623800607e-05,
+      "loss": 2.2598,
+      "step": 1536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996138926274131e-05,
+      "loss": 2.2577,
+      "step": 1537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996132685137851e-05,
+      "loss": 2.2602,
+      "step": 1538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996126438971251e-05,
+      "loss": 2.262,
+      "step": 1539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961201877743635e-05,
+      "loss": 2.2597,
+      "step": 1540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99611393154722e-05,
+      "loss": 2.2516,
+      "step": 1541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961076702898516e-05,
+      "loss": 2.2492,
+      "step": 1542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961014040022906e-05,
+      "loss": 2.2613,
+      "step": 1543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960951326845678e-05,
+      "loss": 2.2458,
+      "step": 1544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960888563367157e-05,
+      "loss": 2.2172,
+      "step": 1545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960825749587657e-05,
+      "loss": 2.2507,
+      "step": 1546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996076288550749e-05,
+      "loss": 2.2401,
+      "step": 1547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996069997112698e-05,
+      "loss": 2.2644,
+      "step": 1548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960637006446444e-05,
+      "loss": 2.2558,
+      "step": 1549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99605739914662e-05,
+      "loss": 2.2524,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996051092618656e-05,
+      "loss": 2.2682,
+      "step": 1551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960447810607852e-05,
+      "loss": 2.2545,
+      "step": 1552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960384644730386e-05,
+      "loss": 2.2564,
+      "step": 1553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960321428554487e-05,
+      "loss": 2.2674,
+      "step": 1554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996025816208047e-05,
+      "loss": 2.2731,
+      "step": 1555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960194845308662e-05,
+      "loss": 2.2606,
+      "step": 1556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960131478239373e-05,
+      "loss": 2.2284,
+      "step": 1557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996006806087293e-05,
+      "loss": 2.2588,
+      "step": 1558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9960004593209648e-05,
+      "loss": 2.2839,
+      "step": 1559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959941075249852e-05,
+      "loss": 2.2583,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995987750699386e-05,
+      "loss": 2.2565,
+      "step": 1561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959813888441993e-05,
+      "loss": 2.2309,
+      "step": 1562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959750219594577e-05,
+      "loss": 2.2157,
+      "step": 1563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959686500451924e-05,
+      "loss": 2.2685,
+      "step": 1564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959622731014368e-05,
+      "loss": 2.2608,
+      "step": 1565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995955891128222e-05,
+      "loss": 2.2518,
+      "step": 1566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959495041255808e-05,
+      "loss": 2.2611,
+      "step": 1567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959431120935453e-05,
+      "loss": 2.2394,
+      "step": 1568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959367150321477e-05,
+      "loss": 2.2542,
+      "step": 1569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959303129414205e-05,
+      "loss": 2.2514,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959239058213962e-05,
+      "loss": 2.2691,
+      "step": 1571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959174936721067e-05,
+      "loss": 2.2677,
+      "step": 1572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959110764935848e-05,
+      "loss": 2.2658,
+      "step": 1573
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959046542858624e-05,
+      "loss": 2.2452,
+      "step": 1574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958982270489723e-05,
+      "loss": 2.2419,
+      "step": 1575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995891794782947e-05,
+      "loss": 2.2914,
+      "step": 1576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958853574878186e-05,
+      "loss": 2.2472,
+      "step": 1577
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958789151636203e-05,
+      "loss": 2.2437,
+      "step": 1578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995872467810384e-05,
+      "loss": 2.2391,
+      "step": 1579
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958660154281423e-05,
+      "loss": 2.2381,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958595580169276e-05,
+      "loss": 2.2586,
+      "step": 1581
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958530955767736e-05,
+      "loss": 2.2609,
+      "step": 1582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995846628107712e-05,
+      "loss": 2.2353,
+      "step": 1583
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995840155609775e-05,
+      "loss": 2.237,
+      "step": 1584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958336780829967e-05,
+      "loss": 2.2824,
+      "step": 1585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995827195527409e-05,
+      "loss": 2.2613,
+      "step": 1586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958207079430444e-05,
+      "loss": 2.2469,
+      "step": 1587
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995814215329936e-05,
+      "loss": 2.2565,
+      "step": 1588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958077176881162e-05,
+      "loss": 2.2641,
+      "step": 1589
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958012150176188e-05,
+      "loss": 2.2517,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957947073184754e-05,
+      "loss": 2.2551,
+      "step": 1591
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957881945907197e-05,
+      "loss": 2.2411,
+      "step": 1592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995781676834384e-05,
+      "loss": 2.2491,
+      "step": 1593
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957751540495017e-05,
+      "loss": 2.2517,
+      "step": 1594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995768626236106e-05,
+      "loss": 2.2351,
+      "step": 1595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995762093394229e-05,
+      "loss": 2.2334,
+      "step": 1596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957555555239042e-05,
+      "loss": 2.2478,
+      "step": 1597
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957490126251646e-05,
+      "loss": 2.2583,
+      "step": 1598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957424646980433e-05,
+      "loss": 2.2553,
+      "step": 1599
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957359117425734e-05,
+      "loss": 2.2472,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957293537587875e-05,
+      "loss": 2.2452,
+      "step": 1601
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957227907467193e-05,
+      "loss": 2.2468,
+      "step": 1602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957162227064012e-05,
+      "loss": 2.2688,
+      "step": 1603
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957096496378673e-05,
+      "loss": 2.2797,
+      "step": 1604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957030715411502e-05,
+      "loss": 2.253,
+      "step": 1605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956964884162833e-05,
+      "loss": 2.2387,
+      "step": 1606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956899002633e-05,
+      "loss": 2.2475,
+      "step": 1607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995683307082233e-05,
+      "loss": 2.2499,
+      "step": 1608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956767088731165e-05,
+      "loss": 2.2433,
+      "step": 1609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956701056359832e-05,
+      "loss": 2.2303,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956634973708664e-05,
+      "loss": 2.2493,
+      "step": 1611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956568840777996e-05,
+      "loss": 2.2622,
+      "step": 1612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956502657568164e-05,
+      "loss": 2.2613,
+      "step": 1613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99564364240795e-05,
+      "loss": 2.2623,
+      "step": 1614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956370140312338e-05,
+      "loss": 2.2575,
+      "step": 1615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956303806267012e-05,
+      "loss": 2.2348,
+      "step": 1616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956237421943864e-05,
+      "loss": 2.247,
+      "step": 1617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956170987343218e-05,
+      "loss": 2.2553,
+      "step": 1618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956104502465417e-05,
+      "loss": 2.2521,
+      "step": 1619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956037967310797e-05,
+      "loss": 2.2467,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955971381879692e-05,
+      "loss": 2.2409,
+      "step": 1621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955904746172435e-05,
+      "loss": 2.2501,
+      "step": 1622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995583806018937e-05,
+      "loss": 2.2464,
+      "step": 1623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955771323930827e-05,
+      "loss": 2.2556,
+      "step": 1624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955704537397143e-05,
+      "loss": 2.2624,
+      "step": 1625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955637700588662e-05,
+      "loss": 2.2437,
+      "step": 1626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955570813505717e-05,
+      "loss": 2.2695,
+      "step": 1627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955503876148645e-05,
+      "loss": 2.2573,
+      "step": 1628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955436888517786e-05,
+      "loss": 2.251,
+      "step": 1629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955369850613478e-05,
+      "loss": 2.2568,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955302762436054e-05,
+      "loss": 2.2645,
+      "step": 1631
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955235623985866e-05,
+      "loss": 2.2505,
+      "step": 1632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995516843526324e-05,
+      "loss": 2.2282,
+      "step": 1633
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995510119626852e-05,
+      "loss": 2.2571,
+      "step": 1634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955033907002045e-05,
+      "loss": 2.2576,
+      "step": 1635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954966567464154e-05,
+      "loss": 2.2597,
+      "step": 1636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995489917765519e-05,
+      "loss": 2.2577,
+      "step": 1637
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954831737575496e-05,
+      "loss": 2.2539,
+      "step": 1638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954764247225406e-05,
+      "loss": 2.2718,
+      "step": 1639
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995469670660526e-05,
+      "loss": 2.2462,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954629115715403e-05,
+      "loss": 2.2521,
+      "step": 1641
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995456147455618e-05,
+      "loss": 2.2686,
+      "step": 1642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995449378312792e-05,
+      "loss": 2.2538,
+      "step": 1643
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995442604143098e-05,
+      "loss": 2.2449,
+      "step": 1644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954358249465692e-05,
+      "loss": 2.2701,
+      "step": 1645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954290407232403e-05,
+      "loss": 2.2611,
+      "step": 1646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954222514731452e-05,
+      "loss": 2.2722,
+      "step": 1647
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954154571963186e-05,
+      "loss": 2.2374,
+      "step": 1648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954086578927945e-05,
+      "loss": 2.2576,
+      "step": 1649
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954018535626073e-05,
+      "loss": 2.2486,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953950442057915e-05,
+      "loss": 2.2474,
+      "step": 1651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953882298223814e-05,
+      "loss": 2.2271,
+      "step": 1652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995381410412411e-05,
+      "loss": 2.2409,
+      "step": 1653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953745859759155e-05,
+      "loss": 2.2372,
+      "step": 1654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995367756512929e-05,
+      "loss": 2.2678,
+      "step": 1655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953609220234854e-05,
+      "loss": 2.2349,
+      "step": 1656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953540825076204e-05,
+      "loss": 2.23,
+      "step": 1657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953472379653677e-05,
+      "loss": 2.2585,
+      "step": 1658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995340388396762e-05,
+      "loss": 2.2592,
+      "step": 1659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953335338018384e-05,
+      "loss": 2.248,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995326674180631e-05,
+      "loss": 2.2499,
+      "step": 1661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953198095331744e-05,
+      "loss": 2.2739,
+      "step": 1662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995312939859503e-05,
+      "loss": 2.2342,
+      "step": 1663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995306065159652e-05,
+      "loss": 2.2644,
+      "step": 1664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952991854336566e-05,
+      "loss": 2.247,
+      "step": 1665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952923006815504e-05,
+      "loss": 2.2554,
+      "step": 1666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995285410903369e-05,
+      "loss": 2.2515,
+      "step": 1667
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995278516099147e-05,
+      "loss": 2.2561,
+      "step": 1668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952716162689188e-05,
+      "loss": 2.2485,
+      "step": 1669
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952647114127198e-05,
+      "loss": 2.2438,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952578015305844e-05,
+      "loss": 2.277,
+      "step": 1671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952508866225478e-05,
+      "loss": 2.2211,
+      "step": 1672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995243966688645e-05,
+      "loss": 2.2325,
+      "step": 1673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952370417289108e-05,
+      "loss": 2.2489,
+      "step": 1674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952301117433798e-05,
+      "loss": 2.2637,
+      "step": 1675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952231767320875e-05,
+      "loss": 2.2368,
+      "step": 1676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952162366950693e-05,
+      "loss": 2.2661,
+      "step": 1677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995209291632359e-05,
+      "loss": 2.2361,
+      "step": 1678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952023415439927e-05,
+      "loss": 2.2354,
+      "step": 1679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951953864300052e-05,
+      "loss": 2.2358,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951884262904316e-05,
+      "loss": 2.264,
+      "step": 1681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951814611253072e-05,
+      "loss": 2.2527,
+      "step": 1682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995174490934667e-05,
+      "loss": 2.2344,
+      "step": 1683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995167515718546e-05,
+      "loss": 2.2493,
+      "step": 1684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99516053547698e-05,
+      "loss": 2.2766,
+      "step": 1685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951535502100036e-05,
+      "loss": 2.2718,
+      "step": 1686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951465599176526e-05,
+      "loss": 2.2388,
+      "step": 1687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951395645999618e-05,
+      "loss": 2.2489,
+      "step": 1688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995132564256967e-05,
+      "loss": 2.2683,
+      "step": 1689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951255588887033e-05,
+      "loss": 2.2565,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951185484952063e-05,
+      "loss": 2.2523,
+      "step": 1691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951115330765113e-05,
+      "loss": 2.2669,
+      "step": 1692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951045126326533e-05,
+      "loss": 2.2636,
+      "step": 1693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995097487163668e-05,
+      "loss": 2.2651,
+      "step": 1694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950904566695914e-05,
+      "loss": 2.2555,
+      "step": 1695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950834211504586e-05,
+      "loss": 2.244,
+      "step": 1696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995076380606305e-05,
+      "loss": 2.2415,
+      "step": 1697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995069335037166e-05,
+      "loss": 2.2388,
+      "step": 1698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995062284443078e-05,
+      "loss": 2.2444,
+      "step": 1699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995055228824076e-05,
+      "loss": 2.2523,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950481681801955e-05,
+      "loss": 2.2322,
+      "step": 1701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995041102511472e-05,
+      "loss": 2.2323,
+      "step": 1702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995034031817942e-05,
+      "loss": 2.2484,
+      "step": 1703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950269560996405e-05,
+      "loss": 2.2439,
+      "step": 1704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950198753566038e-05,
+      "loss": 2.2395,
+      "step": 1705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995012789588867e-05,
+      "loss": 2.247,
+      "step": 1706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995005698796466e-05,
+      "loss": 2.2678,
+      "step": 1707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949986029794373e-05,
+      "loss": 2.247,
+      "step": 1708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994991502137816e-05,
+      "loss": 2.2777,
+      "step": 1709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994984396271638e-05,
+      "loss": 2.2504,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99497728538094e-05,
+      "loss": 2.2325,
+      "step": 1711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949701694657567e-05,
+      "loss": 2.2589,
+      "step": 1712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994963048526125e-05,
+      "loss": 2.2571,
+      "step": 1713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949559225620802e-05,
+      "loss": 2.2601,
+      "step": 1714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949487915736586e-05,
+      "loss": 2.2556,
+      "step": 1715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949416555608964e-05,
+      "loss": 2.2478,
+      "step": 1716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949345145238292e-05,
+      "loss": 2.2659,
+      "step": 1717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949273684624932e-05,
+      "loss": 2.2331,
+      "step": 1718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949202173769247e-05,
+      "loss": 2.2284,
+      "step": 1719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99491306126716e-05,
+      "loss": 2.2357,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949059001332345e-05,
+      "loss": 2.2636,
+      "step": 1721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994898733975185e-05,
+      "loss": 2.2611,
+      "step": 1722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948915627930473e-05,
+      "loss": 2.2421,
+      "step": 1723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994884386586858e-05,
+      "loss": 2.253,
+      "step": 1724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994877205356653e-05,
+      "loss": 2.2449,
+      "step": 1725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948700191024684e-05,
+      "loss": 2.2432,
+      "step": 1726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948628278243412e-05,
+      "loss": 2.2523,
+      "step": 1727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948556315223068e-05,
+      "loss": 2.255,
+      "step": 1728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948484301964024e-05,
+      "loss": 2.2402,
+      "step": 1729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948412238466642e-05,
+      "loss": 2.2838,
+      "step": 1730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994834012473128e-05,
+      "loss": 2.2384,
+      "step": 1731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948267960758304e-05,
+      "loss": 2.2289,
+      "step": 1732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948195746548083e-05,
+      "loss": 2.2372,
+      "step": 1733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994812348210098e-05,
+      "loss": 2.2619,
+      "step": 1734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994805116741736e-05,
+      "loss": 2.2413,
+      "step": 1735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947978802497587e-05,
+      "loss": 2.2362,
+      "step": 1736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947906387342023e-05,
+      "loss": 2.2575,
+      "step": 1737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947833921951042e-05,
+      "loss": 2.2546,
+      "step": 1738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947761406325004e-05,
+      "loss": 2.2519,
+      "step": 1739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947688840464274e-05,
+      "loss": 2.2548,
+      "step": 1740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994761622436922e-05,
+      "loss": 2.2547,
+      "step": 1741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947543558040213e-05,
+      "loss": 2.2149,
+      "step": 1742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947470841477617e-05,
+      "loss": 2.2356,
+      "step": 1743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947398074681797e-05,
+      "loss": 2.2505,
+      "step": 1744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994732525765312e-05,
+      "loss": 2.2442,
+      "step": 1745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994725239039196e-05,
+      "loss": 2.2553,
+      "step": 1746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947179472898678e-05,
+      "loss": 2.2434,
+      "step": 1747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947106505173645e-05,
+      "loss": 2.2406,
+      "step": 1748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994703348721723e-05,
+      "loss": 2.2702,
+      "step": 1749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946960419029802e-05,
+      "loss": 2.2504,
+      "step": 1750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994688730061173e-05,
+      "loss": 2.235,
+      "step": 1751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994681413196338e-05,
+      "loss": 2.259,
+      "step": 1752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946740913085128e-05,
+      "loss": 2.2471,
+      "step": 1753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946667643977337e-05,
+      "loss": 2.2476,
+      "step": 1754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994659432464038e-05,
+      "loss": 2.2499,
+      "step": 1755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994652095507463e-05,
+      "loss": 2.2691,
+      "step": 1756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994644753528045e-05,
+      "loss": 2.2384,
+      "step": 1757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994637406525822e-05,
+      "loss": 2.244,
+      "step": 1758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946300545008302e-05,
+      "loss": 2.2383,
+      "step": 1759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946226974531074e-05,
+      "loss": 2.2301,
+      "step": 1760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946153353826903e-05,
+      "loss": 2.2595,
+      "step": 1761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946079682896164e-05,
+      "loss": 2.235,
+      "step": 1762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994600596173923e-05,
+      "loss": 2.2445,
+      "step": 1763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945932190356468e-05,
+      "loss": 2.237,
+      "step": 1764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945858368748255e-05,
+      "loss": 2.2545,
+      "step": 1765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994578449691496e-05,
+      "loss": 2.2543,
+      "step": 1766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994571057485696e-05,
+      "loss": 2.256,
+      "step": 1767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994563660257463e-05,
+      "loss": 2.2384,
+      "step": 1768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945562580068336e-05,
+      "loss": 2.2534,
+      "step": 1769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994548850733846e-05,
+      "loss": 2.2243,
+      "step": 1770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994541438438537e-05,
+      "loss": 2.2376,
+      "step": 1771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994534021120944e-05,
+      "loss": 2.2408,
+      "step": 1772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945265987811054e-05,
+      "loss": 2.2259,
+      "step": 1773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945191714190573e-05,
+      "loss": 2.2341,
+      "step": 1774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945117390348384e-05,
+      "loss": 2.2527,
+      "step": 1775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945043016284856e-05,
+      "loss": 2.2498,
+      "step": 1776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944968592000365e-05,
+      "loss": 2.2254,
+      "step": 1777
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944894117495287e-05,
+      "loss": 2.2262,
+      "step": 1778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994481959277e-05,
+      "loss": 2.2417,
+      "step": 1779
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994474501782488e-05,
+      "loss": 2.2814,
+      "step": 1780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99446703926603e-05,
+      "loss": 2.2476,
+      "step": 1781
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994459571727664e-05,
+      "loss": 2.2373,
+      "step": 1782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944520991674275e-05,
+      "loss": 2.2533,
+      "step": 1783
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994444621585359e-05,
+      "loss": 2.2457,
+      "step": 1784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944371389814953e-05,
+      "loss": 2.2438,
+      "step": 1785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944296513558746e-05,
+      "loss": 2.2481,
+      "step": 1786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944221587085345e-05,
+      "loss": 2.2135,
+      "step": 1787
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944146610395134e-05,
+      "loss": 2.2384,
+      "step": 1788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944071583488483e-05,
+      "loss": 2.244,
+      "step": 1789
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994399650636578e-05,
+      "loss": 2.2327,
+      "step": 1790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943921379027394e-05,
+      "loss": 2.2253,
+      "step": 1791
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943846201473715e-05,
+      "loss": 2.2407,
+      "step": 1792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943770973705117e-05,
+      "loss": 2.2391,
+      "step": 1793
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994369569572198e-05,
+      "loss": 2.247,
+      "step": 1794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943620367524683e-05,
+      "loss": 2.2567,
+      "step": 1795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943544989113607e-05,
+      "loss": 2.1978,
+      "step": 1796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943469560489136e-05,
+      "loss": 2.2302,
+      "step": 1797
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994339408165165e-05,
+      "loss": 2.25,
+      "step": 1798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994331855260153e-05,
+      "loss": 2.2224,
+      "step": 1799
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943242973339154e-05,
+      "loss": 2.2438,
+      "step": 1800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943167343864903e-05,
+      "loss": 2.2451,
+      "step": 1801
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943091664179165e-05,
+      "loss": 2.2328,
+      "step": 1802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943015934282314e-05,
+      "loss": 2.2518,
+      "step": 1803
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942940154174743e-05,
+      "loss": 2.2459,
+      "step": 1804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942864323856827e-05,
+      "loss": 2.2551,
+      "step": 1805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942788443328952e-05,
+      "loss": 2.236,
+      "step": 1806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99427125125915e-05,
+      "loss": 2.2308,
+      "step": 1807
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942636531644854e-05,
+      "loss": 2.2349,
+      "step": 1808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942560500489395e-05,
+      "loss": 2.23,
+      "step": 1809
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942484419125513e-05,
+      "loss": 2.235,
+      "step": 1810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994240828755359e-05,
+      "loss": 2.235,
+      "step": 1811
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942332105774013e-05,
+      "loss": 2.2281,
+      "step": 1812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942255873787158e-05,
+      "loss": 2.2537,
+      "step": 1813
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942179591593417e-05,
+      "loss": 2.2343,
+      "step": 1814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942103259193174e-05,
+      "loss": 2.2324,
+      "step": 1815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994202687658682e-05,
+      "loss": 2.2473,
+      "step": 1816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941950443774723e-05,
+      "loss": 2.2342,
+      "step": 1817
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994187396075729e-05,
+      "loss": 2.2255,
+      "step": 1818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941797427534894e-05,
+      "loss": 2.2379,
+      "step": 1819
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994172084410793e-05,
+      "loss": 2.2224,
+      "step": 1820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941644210476777e-05,
+      "loss": 2.2603,
+      "step": 1821
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941567526641826e-05,
+      "loss": 2.234,
+      "step": 1822
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941490792603465e-05,
+      "loss": 2.2525,
+      "step": 1823
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994141400836208e-05,
+      "loss": 2.2429,
+      "step": 1824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941337173918052e-05,
+      "loss": 2.2244,
+      "step": 1825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941260289271786e-05,
+      "loss": 2.2577,
+      "step": 1826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941183354423656e-05,
+      "loss": 2.236,
+      "step": 1827
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941106369374054e-05,
+      "loss": 2.2614,
+      "step": 1828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941029334123368e-05,
+      "loss": 2.2184,
+      "step": 1829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994095224867199e-05,
+      "loss": 2.2574,
+      "step": 1830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9940875113020307e-05,
+      "loss": 2.2577,
+      "step": 1831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994079792716871e-05,
+      "loss": 2.2353,
+      "step": 1832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994072069111759e-05,
+      "loss": 2.2389,
+      "step": 1833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994064340486733e-05,
+      "loss": 2.2385,
+      "step": 1834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9940566068418328e-05,
+      "loss": 2.2536,
+      "step": 1835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9940488681770972e-05,
+      "loss": 2.2229,
+      "step": 1836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994041124492565e-05,
+      "loss": 2.2536,
+      "step": 1837
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994033375788276e-05,
+      "loss": 2.2492,
+      "step": 1838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9940256220642687e-05,
+      "loss": 2.2289,
+      "step": 1839
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994017863320582e-05,
+      "loss": 2.2336,
+      "step": 1840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994010099557256e-05,
+      "loss": 2.2518,
+      "step": 1841
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9940023307743297e-05,
+      "loss": 2.2512,
+      "step": 1842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939945569718418e-05,
+      "loss": 2.2337,
+      "step": 1843
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939867781498318e-05,
+      "loss": 2.2576,
+      "step": 1844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993978994308339e-05,
+      "loss": 2.227,
+      "step": 1845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939712054474025e-05,
+      "loss": 2.245,
+      "step": 1846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939634115670624e-05,
+      "loss": 2.2444,
+      "step": 1847
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939556126673568e-05,
+      "loss": 2.2563,
+      "step": 1848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993947808748326e-05,
+      "loss": 2.2317,
+      "step": 1849
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939399998100096e-05,
+      "loss": 2.2436,
+      "step": 1850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939321858524462e-05,
+      "loss": 2.2634,
+      "step": 1851
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993924366875676e-05,
+      "loss": 2.2676,
+      "step": 1852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939165428797376e-05,
+      "loss": 2.2599,
+      "step": 1853
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939087138646718e-05,
+      "loss": 2.2436,
+      "step": 1854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993900879830517e-05,
+      "loss": 2.221,
+      "step": 1855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938930407773135e-05,
+      "loss": 2.2414,
+      "step": 1856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938851967051003e-05,
+      "loss": 2.237,
+      "step": 1857
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938773476139174e-05,
+      "loss": 2.2318,
+      "step": 1858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993869493503804e-05,
+      "loss": 2.2312,
+      "step": 1859
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938616343748003e-05,
+      "loss": 2.2375,
+      "step": 1860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993853770226946e-05,
+      "loss": 2.2242,
+      "step": 1861
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938459010602803e-05,
+      "loss": 2.2361,
+      "step": 1862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993838026874843e-05,
+      "loss": 2.2527,
+      "step": 1863
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938301476706746e-05,
+      "loss": 2.2241,
+      "step": 1864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938222634478142e-05,
+      "loss": 2.2552,
+      "step": 1865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938143742063016e-05,
+      "loss": 2.2395,
+      "step": 1866
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993806479946177e-05,
+      "loss": 2.2415,
+      "step": 1867
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99379858066748e-05,
+      "loss": 2.2315,
+      "step": 1868
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937906763702504e-05,
+      "loss": 2.221,
+      "step": 1869
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937827670545286e-05,
+      "loss": 2.2216,
+      "step": 1870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937748527203537e-05,
+      "loss": 2.2193,
+      "step": 1871
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937669333677668e-05,
+      "loss": 2.2621,
+      "step": 1872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937590089968068e-05,
+      "loss": 2.2486,
+      "step": 1873
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937510796075144e-05,
+      "loss": 2.2497,
+      "step": 1874
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99374314519993e-05,
+      "loss": 2.2211,
+      "step": 1875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937352057740923e-05,
+      "loss": 2.2563,
+      "step": 1876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937272613300426e-05,
+      "loss": 2.2121,
+      "step": 1877
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937193118678203e-05,
+      "loss": 2.2392,
+      "step": 1878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937113573874662e-05,
+      "loss": 2.23,
+      "step": 1879
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99370339788902e-05,
+      "loss": 2.2553,
+      "step": 1880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993695433372522e-05,
+      "loss": 2.2414,
+      "step": 1881
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936874638380127e-05,
+      "loss": 2.2237,
+      "step": 1882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993679489285532e-05,
+      "loss": 2.2226,
+      "step": 1883
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99367150971512e-05,
+      "loss": 2.2395,
+      "step": 1884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936635251268174e-05,
+      "loss": 2.236,
+      "step": 1885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993655535520665e-05,
+      "loss": 2.2208,
+      "step": 1886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936475408967015e-05,
+      "loss": 2.2439,
+      "step": 1887
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936395412549688e-05,
+      "loss": 2.2427,
+      "step": 1888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993631536595507e-05,
+      "loss": 2.2642,
+      "step": 1889
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993623526918356e-05,
+      "loss": 2.2233,
+      "step": 1890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936155122235566e-05,
+      "loss": 2.216,
+      "step": 1891
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993607492511149e-05,
+      "loss": 2.2237,
+      "step": 1892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935994677811747e-05,
+      "loss": 2.2304,
+      "step": 1893
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935914380336728e-05,
+      "loss": 2.2497,
+      "step": 1894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993583403268685e-05,
+      "loss": 2.2405,
+      "step": 1895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935753634862508e-05,
+      "loss": 2.2221,
+      "step": 1896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935673186864117e-05,
+      "loss": 2.2161,
+      "step": 1897
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935592688692082e-05,
+      "loss": 2.2494,
+      "step": 1898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935512140346805e-05,
+      "loss": 2.2455,
+      "step": 1899
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935431541828694e-05,
+      "loss": 2.2344,
+      "step": 1900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935350893138162e-05,
+      "loss": 2.2455,
+      "step": 1901
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993527019427561e-05,
+      "loss": 2.2176,
+      "step": 1902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935189445241444e-05,
+      "loss": 2.24,
+      "step": 1903
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935108646036078e-05,
+      "loss": 2.2385,
+      "step": 1904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935027796659913e-05,
+      "loss": 2.2492,
+      "step": 1905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934946897113367e-05,
+      "loss": 2.2371,
+      "step": 1906
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993486594739684e-05,
+      "loss": 2.2545,
+      "step": 1907
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993478494751074e-05,
+      "loss": 2.2483,
+      "step": 1908
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934703897455486e-05,
+      "loss": 2.2376,
+      "step": 1909
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993462279723148e-05,
+      "loss": 2.2295,
+      "step": 1910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993454164683913e-05,
+      "loss": 2.2466,
+      "step": 1911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993446044627885e-05,
+      "loss": 2.2377,
+      "step": 1912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934379195551043e-05,
+      "loss": 2.2521,
+      "step": 1913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993429789465613e-05,
+      "loss": 2.2114,
+      "step": 1914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934216543594514e-05,
+      "loss": 2.2444,
+      "step": 1915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993413514236661e-05,
+      "loss": 2.2176,
+      "step": 1916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934053690972823e-05,
+      "loss": 2.2279,
+      "step": 1917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993397218941357e-05,
+      "loss": 2.2373,
+      "step": 1918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933890637689264e-05,
+      "loss": 2.2629,
+      "step": 1919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933809035800312e-05,
+      "loss": 2.2444,
+      "step": 1920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933727383747127e-05,
+      "loss": 2.2183,
+      "step": 1921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933645681530124e-05,
+      "loss": 2.22,
+      "step": 1922
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933563929149712e-05,
+      "loss": 2.2297,
+      "step": 1923
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933482126606305e-05,
+      "loss": 2.2466,
+      "step": 1924
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933400273900316e-05,
+      "loss": 2.2401,
+      "step": 1925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993331837103216e-05,
+      "loss": 2.2225,
+      "step": 1926
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993323641800225e-05,
+      "loss": 2.21,
+      "step": 1927
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933154414811e-05,
+      "loss": 2.2197,
+      "step": 1928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933072361458822e-05,
+      "loss": 2.234,
+      "step": 1929
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932990257946133e-05,
+      "loss": 2.2329,
+      "step": 1930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932908104273346e-05,
+      "loss": 2.2615,
+      "step": 1931
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932825900440877e-05,
+      "loss": 2.2503,
+      "step": 1932
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993274364644914e-05,
+      "loss": 2.2237,
+      "step": 1933
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932661342298553e-05,
+      "loss": 2.2483,
+      "step": 1934
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932578987989526e-05,
+      "loss": 2.2467,
+      "step": 1935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993249658352248e-05,
+      "loss": 2.2468,
+      "step": 1936
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993241412889783e-05,
+      "loss": 2.2337,
+      "step": 1937
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993233162411599e-05,
+      "loss": 2.2467,
+      "step": 1938
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993224906917738e-05,
+      "loss": 2.25,
+      "step": 1939
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932166464082415e-05,
+      "loss": 2.2488,
+      "step": 1940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993208380883151e-05,
+      "loss": 2.2206,
+      "step": 1941
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993200110342509e-05,
+      "loss": 2.2386,
+      "step": 1942
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931918347863567e-05,
+      "loss": 2.2139,
+      "step": 1943
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993183554214735e-05,
+      "loss": 2.2449,
+      "step": 1944
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931752686276878e-05,
+      "loss": 2.2224,
+      "step": 1945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931669780252557e-05,
+      "loss": 2.2329,
+      "step": 1946
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931586824074803e-05,
+      "loss": 2.2038,
+      "step": 1947
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931503817744037e-05,
+      "loss": 2.2256,
+      "step": 1948
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931420761260685e-05,
+      "loss": 2.2116,
+      "step": 1949
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931337654625156e-05,
+      "loss": 2.2355,
+      "step": 1950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931254497837876e-05,
+      "loss": 2.2306,
+      "step": 1951
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931171290899267e-05,
+      "loss": 2.2281,
+      "step": 1952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931088033809744e-05,
+      "loss": 2.2352,
+      "step": 1953
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931004726569728e-05,
+      "loss": 2.2435,
+      "step": 1954
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930921369179645e-05,
+      "loss": 2.2455,
+      "step": 1955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930837961639907e-05,
+      "loss": 2.2235,
+      "step": 1956
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930754503950946e-05,
+      "loss": 2.2335,
+      "step": 1957
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930670996113175e-05,
+      "loss": 2.2439,
+      "step": 1958
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993058743812702e-05,
+      "loss": 2.2218,
+      "step": 1959
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930503829992897e-05,
+      "loss": 2.2554,
+      "step": 1960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930420171711237e-05,
+      "loss": 2.2284,
+      "step": 1961
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930336463282454e-05,
+      "loss": 2.2119,
+      "step": 1962
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930252704706977e-05,
+      "loss": 2.2198,
+      "step": 1963
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930168895985228e-05,
+      "loss": 2.2336,
+      "step": 1964
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930085037117627e-05,
+      "loss": 2.2452,
+      "step": 1965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930001128104602e-05,
+      "loss": 2.2427,
+      "step": 1966
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992991716894657e-05,
+      "loss": 2.2401,
+      "step": 1967
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929833159643963e-05,
+      "loss": 2.2353,
+      "step": 1968
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99297491001972e-05,
+      "loss": 2.2136,
+      "step": 1969
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929664990606704e-05,
+      "loss": 2.2168,
+      "step": 1970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992958083087291e-05,
+      "loss": 2.2199,
+      "step": 1971
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929496620996227e-05,
+      "loss": 2.2231,
+      "step": 1972
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929412360977092e-05,
+      "loss": 2.23,
+      "step": 1973
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929328050815928e-05,
+      "loss": 2.2229,
+      "step": 1974
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992924369051316e-05,
+      "loss": 2.2255,
+      "step": 1975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929159280069215e-05,
+      "loss": 2.2389,
+      "step": 1976
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929074819484516e-05,
+      "loss": 2.2312,
+      "step": 1977
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928990308759496e-05,
+      "loss": 2.233,
+      "step": 1978
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928905747894573e-05,
+      "loss": 2.2342,
+      "step": 1979
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928821136890183e-05,
+      "loss": 2.2189,
+      "step": 1980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928736475746744e-05,
+      "loss": 2.2315,
+      "step": 1981
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992865176446469e-05,
+      "loss": 2.2266,
+      "step": 1982
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992856700304445e-05,
+      "loss": 2.2374,
+      "step": 1983
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928482191486446e-05,
+      "loss": 2.2349,
+      "step": 1984
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928397329791115e-05,
+      "loss": 2.2054,
+      "step": 1985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928312417958873e-05,
+      "loss": 2.242,
+      "step": 1986
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992822745599016e-05,
+      "loss": 2.2243,
+      "step": 1987
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928142443885398e-05,
+      "loss": 2.2281,
+      "step": 1988
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992805738164502e-05,
+      "loss": 2.2136,
+      "step": 1989
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927972269269456e-05,
+      "loss": 2.2201,
+      "step": 1990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992788710675914e-05,
+      "loss": 2.2358,
+      "step": 1991
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927801894114486e-05,
+      "loss": 2.2255,
+      "step": 1992
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927716631335942e-05,
+      "loss": 2.2419,
+      "step": 1993
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992763131842393e-05,
+      "loss": 2.219,
+      "step": 1994
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992754595537888e-05,
+      "loss": 2.2032,
+      "step": 1995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992746054220123e-05,
+      "loss": 2.2203,
+      "step": 1996
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927375078891402e-05,
+      "loss": 2.2331,
+      "step": 1997
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927289565449835e-05,
+      "loss": 2.2139,
+      "step": 1998
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992720400187696e-05,
+      "loss": 2.2565,
+      "step": 1999
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99271183881732e-05,
+      "loss": 2.2404,
+      "step": 2000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927032724339e-05,
+      "loss": 2.2242,
+      "step": 2001
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926947010374787e-05,
+      "loss": 2.2139,
+      "step": 2002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926861246280996e-05,
+      "loss": 2.2274,
+      "step": 2003
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926775432058054e-05,
+      "loss": 2.2269,
+      "step": 2004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926689567706398e-05,
+      "loss": 2.2299,
+      "step": 2005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926603653226464e-05,
+      "loss": 2.2135,
+      "step": 2006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926517688618687e-05,
+      "loss": 2.2293,
+      "step": 2007
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926431673883493e-05,
+      "loss": 2.2282,
+      "step": 2008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926345609021323e-05,
+      "loss": 2.2032,
+      "step": 2009
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992625949403261e-05,
+      "loss": 2.2264,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992617332891779e-05,
+      "loss": 2.2071,
+      "step": 2011
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926087113677293e-05,
+      "loss": 2.2297,
+      "step": 2012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926000848311562e-05,
+      "loss": 2.2183,
+      "step": 2013
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992591453282103e-05,
+      "loss": 2.2292,
+      "step": 2014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925828167206128e-05,
+      "loss": 2.2206,
+      "step": 2015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99257417514673e-05,
+      "loss": 2.2308,
+      "step": 2016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925655285604976e-05,
+      "loss": 2.2087,
+      "step": 2017
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925568769619598e-05,
+      "loss": 2.2051,
+      "step": 2018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925482203511593e-05,
+      "loss": 2.2348,
+      "step": 2019
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992539558728141e-05,
+      "loss": 2.2346,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992530892092948e-05,
+      "loss": 2.2411,
+      "step": 2021
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925222204456244e-05,
+      "loss": 2.2284,
+      "step": 2022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925135437862138e-05,
+      "loss": 2.2125,
+      "step": 2023
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925048621147598e-05,
+      "loss": 2.1772,
+      "step": 2024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9924961754313065e-05,
+      "loss": 2.2366,
+      "step": 2025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992487483735898e-05,
+      "loss": 2.2183,
+      "step": 2026
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9924787870285774e-05,
+      "loss": 2.2311,
+      "step": 2027
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9924700853093894e-05,
+      "loss": 2.2081,
+      "step": 2028
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9924613785783776e-05,
+      "loss": 2.2143,
+      "step": 2029
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992452666835586e-05,
+      "loss": 2.2092,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992443950081059e-05,
+      "loss": 2.2163,
+      "step": 2031
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99243522831484e-05,
+      "loss": 2.2288,
+      "step": 2032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9924265015369734e-05,
+      "loss": 2.222,
+      "step": 2033
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9924177697475032e-05,
+      "loss": 2.2061,
+      "step": 2034
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992409032946473e-05,
+      "loss": 2.2004,
+      "step": 2035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992400291133928e-05,
+      "loss": 2.2071,
+      "step": 2036
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923915443099115e-05,
+      "loss": 2.2369,
+      "step": 2037
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923827924744677e-05,
+      "loss": 2.1997,
+      "step": 2038
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992374035627641e-05,
+      "loss": 2.2304,
+      "step": 2039
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992365273769476e-05,
+      "loss": 2.2168,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923565069000162e-05,
+      "loss": 2.2219,
+      "step": 2041
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992347735019306e-05,
+      "loss": 2.2101,
+      "step": 2042
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923389581273907e-05,
+      "loss": 2.2184,
+      "step": 2043
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992330176224313e-05,
+      "loss": 2.2152,
+      "step": 2044
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923213893101187e-05,
+      "loss": 2.2345,
+      "step": 2045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923125973848512e-05,
+      "loss": 2.2159,
+      "step": 2046
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923038004485554e-05,
+      "loss": 2.2428,
+      "step": 2047
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922949985012757e-05,
+      "loss": 2.2266,
+      "step": 2048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992286191543056e-05,
+      "loss": 2.2118,
+      "step": 2049
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922773795739418e-05,
+      "loss": 2.2159,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922685625939768e-05,
+      "loss": 2.2172,
+      "step": 2051
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922597406032057e-05,
+      "loss": 2.2172,
+      "step": 2052
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992250913601673e-05,
+      "loss": 2.2297,
+      "step": 2053
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922420815894233e-05,
+      "loss": 2.2194,
+      "step": 2054
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922332445665015e-05,
+      "loss": 2.2295,
+      "step": 2055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922244025329518e-05,
+      "loss": 2.2174,
+      "step": 2056
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992215555488819e-05,
+      "loss": 2.2178,
+      "step": 2057
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992206703434148e-05,
+      "loss": 2.2136,
+      "step": 2058
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921978463689833e-05,
+      "loss": 2.2439,
+      "step": 2059
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921889842933697e-05,
+      "loss": 2.225,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921801172073516e-05,
+      "loss": 2.2156,
+      "step": 2061
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921712451109743e-05,
+      "loss": 2.2141,
+      "step": 2062
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921623680042823e-05,
+      "loss": 2.1883,
+      "step": 2063
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921534858873203e-05,
+      "loss": 2.1868,
+      "step": 2064
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921445987601336e-05,
+      "loss": 2.2274,
+      "step": 2065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921357066227667e-05,
+      "loss": 2.2098,
+      "step": 2066
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921268094752647e-05,
+      "loss": 2.2023,
+      "step": 2067
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921179073176726e-05,
+      "loss": 2.1996,
+      "step": 2068
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992109000150035e-05,
+      "loss": 2.2127,
+      "step": 2069
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992100087972397e-05,
+      "loss": 2.2087,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992091170784804e-05,
+      "loss": 2.2113,
+      "step": 2071
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9920822485873007e-05,
+      "loss": 2.2252,
+      "step": 2072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9920733213799323e-05,
+      "loss": 2.2321,
+      "step": 2073
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9920643891627438e-05,
+      "loss": 2.2152,
+      "step": 2074
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99205545193578e-05,
+      "loss": 2.2079,
+      "step": 2075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992046509699086e-05,
+      "loss": 2.2109,
+      "step": 2076
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992037562452708e-05,
+      "loss": 2.2159,
+      "step": 2077
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9920286101966898e-05,
+      "loss": 2.2277,
+      "step": 2078
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992019652931078e-05,
+      "loss": 2.2172,
+      "step": 2079
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9920106906559162e-05,
+      "loss": 2.2144,
+      "step": 2080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9920017233712513e-05,
+      "loss": 2.2204,
+      "step": 2081
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991992751077127e-05,
+      "loss": 2.2206,
+      "step": 2082
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99198377377359e-05,
+      "loss": 2.2049,
+      "step": 2083
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9919747914606854e-05,
+      "loss": 2.2078,
+      "step": 2084
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9919658041384576e-05,
+      "loss": 2.2163,
+      "step": 2085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991956811806953e-05,
+      "loss": 2.2129,
+      "step": 2086
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9919478144662166e-05,
+      "loss": 2.21,
+      "step": 2087
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991938812116294e-05,
+      "loss": 2.2197,
+      "step": 2088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99192980475723e-05,
+      "loss": 2.2087,
+      "step": 2089
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9919207923890708e-05,
+      "loss": 2.2201,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991911775011862e-05,
+      "loss": 2.2131,
+      "step": 2091
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9919027526256485e-05,
+      "loss": 2.2031,
+      "step": 2092
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918937252304763e-05,
+      "loss": 2.2124,
+      "step": 2093
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918846928263914e-05,
+      "loss": 2.2309,
+      "step": 2094
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991875655413438e-05,
+      "loss": 2.2105,
+      "step": 2095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918666129916636e-05,
+      "loss": 2.1999,
+      "step": 2096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918575655611122e-05,
+      "loss": 2.2015,
+      "step": 2097
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918485131218307e-05,
+      "loss": 2.1779,
+      "step": 2098
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991839455673864e-05,
+      "loss": 2.2077,
+      "step": 2099
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918303932172583e-05,
+      "loss": 2.2143,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918213257520592e-05,
+      "loss": 2.1982,
+      "step": 2101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918122532783126e-05,
+      "loss": 2.1987,
+      "step": 2102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991803175796064e-05,
+      "loss": 2.2207,
+      "step": 2103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917940933053596e-05,
+      "loss": 2.203,
+      "step": 2104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917850058062453e-05,
+      "loss": 2.215,
+      "step": 2105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917759132987665e-05,
+      "loss": 2.2041,
+      "step": 2106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917668157829695e-05,
+      "loss": 2.2077,
+      "step": 2107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917577132589002e-05,
+      "loss": 2.2126,
+      "step": 2108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991748605726605e-05,
+      "loss": 2.1863,
+      "step": 2109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917394931861288e-05,
+      "loss": 2.2056,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917303756375185e-05,
+      "loss": 2.2074,
+      "step": 2111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917212530808198e-05,
+      "loss": 2.2325,
+      "step": 2112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991712125516079e-05,
+      "loss": 2.1946,
+      "step": 2113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991702992943342e-05,
+      "loss": 2.2309,
+      "step": 2114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991693855362655e-05,
+      "loss": 2.2031,
+      "step": 2115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916847127740643e-05,
+      "loss": 2.1973,
+      "step": 2116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916755651776155e-05,
+      "loss": 2.2165,
+      "step": 2117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916664125733554e-05,
+      "loss": 2.2318,
+      "step": 2118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99165725496133e-05,
+      "loss": 2.2292,
+      "step": 2119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916480923415854e-05,
+      "loss": 2.233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991638924714168e-05,
+      "loss": 2.2059,
+      "step": 2121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916297520791245e-05,
+      "loss": 2.2236,
+      "step": 2122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916205744365003e-05,
+      "loss": 2.2458,
+      "step": 2123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916113917863427e-05,
+      "loss": 2.2152,
+      "step": 2124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916022041286975e-05,
+      "loss": 2.2113,
+      "step": 2125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991593011463611e-05,
+      "loss": 2.2047,
+      "step": 2126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915838137911303e-05,
+      "loss": 2.2205,
+      "step": 2127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915746111113012e-05,
+      "loss": 2.213,
+      "step": 2128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915654034241704e-05,
+      "loss": 2.2034,
+      "step": 2129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915561907297843e-05,
+      "loss": 2.2059,
+      "step": 2130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915469730281898e-05,
+      "loss": 2.2219,
+      "step": 2131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915377503194325e-05,
+      "loss": 2.201,
+      "step": 2132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915285226035602e-05,
+      "loss": 2.2026,
+      "step": 2133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991519289880619e-05,
+      "loss": 2.1907,
+      "step": 2134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991510052150655e-05,
+      "loss": 2.2296,
+      "step": 2135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915008094137155e-05,
+      "loss": 2.2115,
+      "step": 2136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991491561669847e-05,
+      "loss": 2.2019,
+      "step": 2137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991482308919096e-05,
+      "loss": 2.2247,
+      "step": 2138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9914730511615096e-05,
+      "loss": 2.2152,
+      "step": 2139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9914637883971345e-05,
+      "loss": 2.1973,
+      "step": 2140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991454520626017e-05,
+      "loss": 2.2185,
+      "step": 2141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9914452478482043e-05,
+      "loss": 2.2186,
+      "step": 2142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9914359700637432e-05,
+      "loss": 2.2128,
+      "step": 2143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9914266872726804e-05,
+      "loss": 2.2078,
+      "step": 2144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991417399475063e-05,
+      "loss": 2.2049,
+      "step": 2145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991408106670938e-05,
+      "loss": 2.2106,
+      "step": 2146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913988088603515e-05,
+      "loss": 2.2301,
+      "step": 2147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913895060433516e-05,
+      "loss": 2.2123,
+      "step": 2148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913801982199844e-05,
+      "loss": 2.2273,
+      "step": 2149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913708853902975e-05,
+      "loss": 2.2224,
+      "step": 2150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913615675543375e-05,
+      "loss": 2.2077,
+      "step": 2151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913522447121515e-05,
+      "loss": 2.2029,
+      "step": 2152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913429168637867e-05,
+      "loss": 2.1747,
+      "step": 2153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913335840092905e-05,
+      "loss": 2.1908,
+      "step": 2154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913242461487096e-05,
+      "loss": 2.1941,
+      "step": 2155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913149032820912e-05,
+      "loss": 2.1961,
+      "step": 2156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913055554094827e-05,
+      "loss": 2.1972,
+      "step": 2157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991296202530931e-05,
+      "loss": 2.1846,
+      "step": 2158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991286844646484e-05,
+      "loss": 2.196,
+      "step": 2159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9912774817561878e-05,
+      "loss": 2.1977,
+      "step": 2160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9912681138600907e-05,
+      "loss": 2.1884,
+      "step": 2161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9912587409582396e-05,
+      "loss": 2.2113,
+      "step": 2162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991249363050682e-05,
+      "loss": 2.2222,
+      "step": 2163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991239980137465e-05,
+      "loss": 2.2166,
+      "step": 2164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9912305922186358e-05,
+      "loss": 2.202,
+      "step": 2165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9912211992942425e-05,
+      "loss": 2.2026,
+      "step": 2166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9912118013643324e-05,
+      "loss": 2.1846,
+      "step": 2167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991202398428952e-05,
+      "loss": 2.2021,
+      "step": 2168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99119299048815e-05,
+      "loss": 2.2268,
+      "step": 2169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9911835775419734e-05,
+      "loss": 2.1815,
+      "step": 2170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9911741595904694e-05,
+      "loss": 2.2316,
+      "step": 2171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9911647366336863e-05,
+      "loss": 2.1935,
+      "step": 2172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991155308671671e-05,
+      "loss": 2.1958,
+      "step": 2173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991145875704472e-05,
+      "loss": 2.2272,
+      "step": 2174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9911364377321355e-05,
+      "loss": 2.1805,
+      "step": 2175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9911269947547103e-05,
+      "loss": 2.2095,
+      "step": 2176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991117546772244e-05,
+      "loss": 2.1976,
+      "step": 2177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991108093784784e-05,
+      "loss": 2.1908,
+      "step": 2178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910986357923782e-05,
+      "loss": 2.1825,
+      "step": 2179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991089172795074e-05,
+      "loss": 2.2079,
+      "step": 2180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910797047929194e-05,
+      "loss": 2.2068,
+      "step": 2181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910702317859628e-05,
+      "loss": 2.2232,
+      "step": 2182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910607537742513e-05,
+      "loss": 2.2018,
+      "step": 2183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910512707578325e-05,
+      "loss": 2.2049,
+      "step": 2184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910417827367554e-05,
+      "loss": 2.1965,
+      "step": 2185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991032289711067e-05,
+      "loss": 2.2013,
+      "step": 2186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910227916808155e-05,
+      "loss": 2.206,
+      "step": 2187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910132886460493e-05,
+      "loss": 2.1944,
+      "step": 2188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910037806068156e-05,
+      "loss": 2.1841,
+      "step": 2189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909942675631624e-05,
+      "loss": 2.1888,
+      "step": 2190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909847495151387e-05,
+      "loss": 2.1928,
+      "step": 2191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909752264627918e-05,
+      "loss": 2.1954,
+      "step": 2192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909656984061702e-05,
+      "loss": 2.189,
+      "step": 2193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909561653453216e-05,
+      "loss": 2.2133,
+      "step": 2194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909466272802946e-05,
+      "loss": 2.2169,
+      "step": 2195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909370842111368e-05,
+      "loss": 2.2172,
+      "step": 2196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909275361378968e-05,
+      "loss": 2.2216,
+      "step": 2197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990917983060623e-05,
+      "loss": 2.2,
+      "step": 2198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909084249793626e-05,
+      "loss": 2.1954,
+      "step": 2199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990898861894165e-05,
+      "loss": 2.2154,
+      "step": 2200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9908892938050784e-05,
+      "loss": 2.1927,
+      "step": 2201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9908797207121507e-05,
+      "loss": 2.2089,
+      "step": 2202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9908701426154304e-05,
+      "loss": 2.2009,
+      "step": 2203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9908605595149657e-05,
+      "loss": 2.195,
+      "step": 2204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9908509714108052e-05,
+      "loss": 2.185,
+      "step": 2205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9908413783029972e-05,
+      "loss": 2.1892,
+      "step": 2206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99083178019159e-05,
+      "loss": 2.2061,
+      "step": 2207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990822177076633e-05,
+      "loss": 2.2009,
+      "step": 2208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990812568958173e-05,
+      "loss": 2.2095,
+      "step": 2209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99080295583626e-05,
+      "loss": 2.2272,
+      "step": 2210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907933377109422e-05,
+      "loss": 2.1892,
+      "step": 2211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990783714582268e-05,
+      "loss": 2.1919,
+      "step": 2212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907740864502855e-05,
+      "loss": 2.1883,
+      "step": 2213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907644533150438e-05,
+      "loss": 2.1951,
+      "step": 2214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907548151765922e-05,
+      "loss": 2.1855,
+      "step": 2215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907451720349783e-05,
+      "loss": 2.2084,
+      "step": 2216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907355238902514e-05,
+      "loss": 2.1896,
+      "step": 2217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907258707424594e-05,
+      "loss": 2.1902,
+      "step": 2218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907162125916524e-05,
+      "loss": 2.1777,
+      "step": 2219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907065494378785e-05,
+      "loss": 2.1747,
+      "step": 2220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990696881281186e-05,
+      "loss": 2.2032,
+      "step": 2221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906872081216244e-05,
+      "loss": 2.1795,
+      "step": 2222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906775299592425e-05,
+      "loss": 2.1867,
+      "step": 2223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990667846794089e-05,
+      "loss": 2.2158,
+      "step": 2224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906581586262126e-05,
+      "loss": 2.1743,
+      "step": 2225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906484654556622e-05,
+      "loss": 2.1793,
+      "step": 2226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906387672824878e-05,
+      "loss": 2.1846,
+      "step": 2227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906290641067368e-05,
+      "loss": 2.1915,
+      "step": 2228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990619355928459e-05,
+      "loss": 2.1885,
+      "step": 2229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990609642747704e-05,
+      "loss": 2.2006,
+      "step": 2230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905999245645192e-05,
+      "loss": 2.1758,
+      "step": 2231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905902013789555e-05,
+      "loss": 2.2062,
+      "step": 2232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905804731910608e-05,
+      "loss": 2.1953,
+      "step": 2233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990570740000885e-05,
+      "loss": 2.1828,
+      "step": 2234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905610018084765e-05,
+      "loss": 2.1854,
+      "step": 2235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990551258613885e-05,
+      "loss": 2.1702,
+      "step": 2236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99054151041716e-05,
+      "loss": 2.1922,
+      "step": 2237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99053175721835e-05,
+      "loss": 2.1627,
+      "step": 2238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905219990175043e-05,
+      "loss": 2.1614,
+      "step": 2239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905122358146723e-05,
+      "loss": 2.1835,
+      "step": 2240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905024676099038e-05,
+      "loss": 2.2201,
+      "step": 2241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904926944032478e-05,
+      "loss": 2.1705,
+      "step": 2242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990482916194753e-05,
+      "loss": 2.2039,
+      "step": 2243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99047313298447e-05,
+      "loss": 2.185,
+      "step": 2244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990463344772447e-05,
+      "loss": 2.1951,
+      "step": 2245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990453551558735e-05,
+      "loss": 2.1798,
+      "step": 2246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904437533433815e-05,
+      "loss": 2.187,
+      "step": 2247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904339501264376e-05,
+      "loss": 2.1908,
+      "step": 2248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990424141907952e-05,
+      "loss": 2.1639,
+      "step": 2249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904143286879738e-05,
+      "loss": 2.1791,
+      "step": 2250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904045104665538e-05,
+      "loss": 2.1819,
+      "step": 2251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903946872437407e-05,
+      "loss": 2.1814,
+      "step": 2252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903848590195843e-05,
+      "loss": 2.1871,
+      "step": 2253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990375025794134e-05,
+      "loss": 2.1908,
+      "step": 2254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99036518756744e-05,
+      "loss": 2.1935,
+      "step": 2255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903553443395515e-05,
+      "loss": 2.18,
+      "step": 2256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903454961105186e-05,
+      "loss": 2.1959,
+      "step": 2257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903356428803907e-05,
+      "loss": 2.1548,
+      "step": 2258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903257846492176e-05,
+      "loss": 2.1835,
+      "step": 2259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903159214170496e-05,
+      "loss": 2.1863,
+      "step": 2260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903060531839356e-05,
+      "loss": 2.189,
+      "step": 2261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990296179949926e-05,
+      "loss": 2.2079,
+      "step": 2262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902863017150707e-05,
+      "loss": 2.1774,
+      "step": 2263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902764184794193e-05,
+      "loss": 2.1779,
+      "step": 2264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902665302430223e-05,
+      "loss": 2.19,
+      "step": 2265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902566370059288e-05,
+      "loss": 2.17,
+      "step": 2266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902467387681892e-05,
+      "loss": 2.1778,
+      "step": 2267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902368355298534e-05,
+      "loss": 2.1771,
+      "step": 2268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902269272909715e-05,
+      "loss": 2.182,
+      "step": 2269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902170140515934e-05,
+      "loss": 2.1773,
+      "step": 2270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902070958117695e-05,
+      "loss": 2.195,
+      "step": 2271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901971725715496e-05,
+      "loss": 2.1897,
+      "step": 2272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901872443309835e-05,
+      "loss": 2.2002,
+      "step": 2273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990177311090122e-05,
+      "loss": 2.1516,
+      "step": 2274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901673728490147e-05,
+      "loss": 2.1627,
+      "step": 2275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990157429607712e-05,
+      "loss": 2.1887,
+      "step": 2276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901474813662642e-05,
+      "loss": 2.1852,
+      "step": 2277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901375281247214e-05,
+      "loss": 2.1773,
+      "step": 2278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990127569883134e-05,
+      "loss": 2.1743,
+      "step": 2279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901176066415524e-05,
+      "loss": 2.1753,
+      "step": 2280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990107638400027e-05,
+      "loss": 2.2004,
+      "step": 2281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990097665158607e-05,
+      "loss": 2.2008,
+      "step": 2282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9900876869173445e-05,
+      "loss": 2.195,
+      "step": 2283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9900777036762885e-05,
+      "loss": 2.1965,
+      "step": 2284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99006771543549e-05,
+      "loss": 2.1888,
+      "step": 2285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9900577221949996e-05,
+      "loss": 2.1783,
+      "step": 2286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9900477239548674e-05,
+      "loss": 2.1796,
+      "step": 2287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990037720715144e-05,
+      "loss": 2.1609,
+      "step": 2288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99002771247588e-05,
+      "loss": 2.1857,
+      "step": 2289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990017699237126e-05,
+      "loss": 2.2187,
+      "step": 2290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990007680998932e-05,
+      "loss": 2.2011,
+      "step": 2291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899976577613495e-05,
+      "loss": 2.1831,
+      "step": 2292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899876295244284e-05,
+      "loss": 2.1793,
+      "step": 2293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98997759628822e-05,
+      "loss": 2.1734,
+      "step": 2294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899675580527742e-05,
+      "loss": 2.1773,
+      "step": 2295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989957514818142e-05,
+      "loss": 2.1835,
+      "step": 2296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899474665843744e-05,
+      "loss": 2.1682,
+      "step": 2297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989937413351522e-05,
+      "loss": 2.1609,
+      "step": 2298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989927355119635e-05,
+      "loss": 2.1795,
+      "step": 2299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989917291888765e-05,
+      "loss": 2.1928,
+      "step": 2300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989907223658963e-05,
+      "loss": 2.1748,
+      "step": 2301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898971504302788e-05,
+      "loss": 2.1733,
+      "step": 2302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898870722027633e-05,
+      "loss": 2.1715,
+      "step": 2303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898769889764687e-05,
+      "loss": 2.1975,
+      "step": 2304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898669007514448e-05,
+      "loss": 2.2212,
+      "step": 2305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989856807527743e-05,
+      "loss": 2.1867,
+      "step": 2306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989846709305414e-05,
+      "loss": 2.1775,
+      "step": 2307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989836606084509e-05,
+      "loss": 2.1742,
+      "step": 2308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989826497865079e-05,
+      "loss": 2.1979,
+      "step": 2309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898163846471748e-05,
+      "loss": 2.1908,
+      "step": 2310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898062664308477e-05,
+      "loss": 2.1776,
+      "step": 2311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989796143216149e-05,
+      "loss": 2.141,
+      "step": 2312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897860150031292e-05,
+      "loss": 2.1958,
+      "step": 2313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897758817918402e-05,
+      "loss": 2.169,
+      "step": 2314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897657435823327e-05,
+      "loss": 2.1675,
+      "step": 2315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989755600374658e-05,
+      "loss": 2.1812,
+      "step": 2316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989745452168867e-05,
+      "loss": 2.1936,
+      "step": 2317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897352989650113e-05,
+      "loss": 2.1832,
+      "step": 2318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897251407631424e-05,
+      "loss": 2.1783,
+      "step": 2319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989714977563311e-05,
+      "loss": 2.1776,
+      "step": 2320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897048093655692e-05,
+      "loss": 2.1692,
+      "step": 2321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896946361699675e-05,
+      "loss": 2.1766,
+      "step": 2322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989684457976558e-05,
+      "loss": 2.1805,
+      "step": 2323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896742747853917e-05,
+      "loss": 2.1875,
+      "step": 2324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98966408659652e-05,
+      "loss": 2.17,
+      "step": 2325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896538934099944e-05,
+      "loss": 2.1757,
+      "step": 2326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896436952258664e-05,
+      "loss": 2.1863,
+      "step": 2327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989633492044188e-05,
+      "loss": 2.1854,
+      "step": 2328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896232838650094e-05,
+      "loss": 2.1709,
+      "step": 2329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896130706883838e-05,
+      "loss": 2.1828,
+      "step": 2330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896028525143613e-05,
+      "loss": 2.1831,
+      "step": 2331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895926293429945e-05,
+      "loss": 2.1821,
+      "step": 2332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989582401174335e-05,
+      "loss": 2.1624,
+      "step": 2333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895721680084333e-05,
+      "loss": 2.1681,
+      "step": 2334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895619298453426e-05,
+      "loss": 2.1598,
+      "step": 2335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895516866851136e-05,
+      "loss": 2.1626,
+      "step": 2336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895414385277985e-05,
+      "loss": 2.1981,
+      "step": 2337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895311853734488e-05,
+      "loss": 2.1825,
+      "step": 2338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895209272221167e-05,
+      "loss": 2.1549,
+      "step": 2339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895106640738532e-05,
+      "loss": 2.1893,
+      "step": 2340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989500395928711e-05,
+      "loss": 2.1873,
+      "step": 2341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894901227867415e-05,
+      "loss": 2.1636,
+      "step": 2342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894798446479963e-05,
+      "loss": 2.186,
+      "step": 2343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894695615125278e-05,
+      "loss": 2.1785,
+      "step": 2344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989459273380388e-05,
+      "loss": 2.1825,
+      "step": 2345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894489802516284e-05,
+      "loss": 2.1881,
+      "step": 2346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894386821263012e-05,
+      "loss": 2.178,
+      "step": 2347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894283790044585e-05,
+      "loss": 2.1617,
+      "step": 2348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989418070886152e-05,
+      "loss": 2.1775,
+      "step": 2349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894077577714343e-05,
+      "loss": 2.1747,
+      "step": 2350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893974396603566e-05,
+      "loss": 2.1701,
+      "step": 2351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989387116552972e-05,
+      "loss": 2.1676,
+      "step": 2352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893767884493324e-05,
+      "loss": 2.1483,
+      "step": 2353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893664553494893e-05,
+      "loss": 2.1536,
+      "step": 2354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893561172534955e-05,
+      "loss": 2.1486,
+      "step": 2355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893457741614033e-05,
+      "loss": 2.1535,
+      "step": 2356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893354260732644e-05,
+      "loss": 2.1558,
+      "step": 2357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893250729891314e-05,
+      "loss": 2.1677,
+      "step": 2358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893147149090565e-05,
+      "loss": 2.1742,
+      "step": 2359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989304351833092e-05,
+      "loss": 2.1833,
+      "step": 2360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892939837612905e-05,
+      "loss": 2.1854,
+      "step": 2361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892836106937038e-05,
+      "loss": 2.1686,
+      "step": 2362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989273232630385e-05,
+      "loss": 2.1818,
+      "step": 2363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892628495713855e-05,
+      "loss": 2.1769,
+      "step": 2364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892524615167584e-05,
+      "loss": 2.1659,
+      "step": 2365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989242068466556e-05,
+      "loss": 2.1706,
+      "step": 2366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892316704208315e-05,
+      "loss": 2.1891,
+      "step": 2367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892212673796365e-05,
+      "loss": 2.1551,
+      "step": 2368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892108593430233e-05,
+      "loss": 2.1507,
+      "step": 2369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892004463110456e-05,
+      "loss": 2.1599,
+      "step": 2370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891900282837554e-05,
+      "loss": 2.1779,
+      "step": 2371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989179605261205e-05,
+      "loss": 2.1507,
+      "step": 2372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989169177243447e-05,
+      "loss": 2.1698,
+      "step": 2373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989158744230535e-05,
+      "loss": 2.175,
+      "step": 2374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891483062225206e-05,
+      "loss": 2.1745,
+      "step": 2375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891378632194574e-05,
+      "loss": 2.1819,
+      "step": 2376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989127415221397e-05,
+      "loss": 2.17,
+      "step": 2377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891169622283934e-05,
+      "loss": 2.176,
+      "step": 2378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891065042404988e-05,
+      "loss": 2.165,
+      "step": 2379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890960412577657e-05,
+      "loss": 2.1746,
+      "step": 2380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890855732802477e-05,
+      "loss": 2.1813,
+      "step": 2381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989075100307997e-05,
+      "loss": 2.1448,
+      "step": 2382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890646223410667e-05,
+      "loss": 2.1555,
+      "step": 2383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890541393795098e-05,
+      "loss": 2.1556,
+      "step": 2384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890436514233794e-05,
+      "loss": 2.1444,
+      "step": 2385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989033158472728e-05,
+      "loss": 2.1516,
+      "step": 2386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989022660527609e-05,
+      "loss": 2.1588,
+      "step": 2387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989012157588075e-05,
+      "loss": 2.1693,
+      "step": 2388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890016496541794e-05,
+      "loss": 2.1737,
+      "step": 2389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988991136725975e-05,
+      "loss": 2.1746,
+      "step": 2390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9889806188035155e-05,
+      "loss": 2.1727,
+      "step": 2391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988970095886853e-05,
+      "loss": 2.154,
+      "step": 2392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9889595679760416e-05,
+      "loss": 2.1443,
+      "step": 2393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988949035071134e-05,
+      "loss": 2.1604,
+      "step": 2394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988938497172183e-05,
+      "loss": 2.1296,
+      "step": 2395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9889279542792426e-05,
+      "loss": 2.1738,
+      "step": 2396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988917406392366e-05,
+      "loss": 2.1711,
+      "step": 2397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988906853511606e-05,
+      "loss": 2.1858,
+      "step": 2398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988896295637016e-05,
+      "loss": 2.1377,
+      "step": 2399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888857327686494e-05,
+      "loss": 2.1824,
+      "step": 2400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988875164906559e-05,
+      "loss": 2.1669,
+      "step": 2401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988864592050799e-05,
+      "loss": 2.1635,
+      "step": 2402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988854014201423e-05,
+      "loss": 2.177,
+      "step": 2403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888434313584835e-05,
+      "loss": 2.1849,
+      "step": 2404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888328435220342e-05,
+      "loss": 2.1419,
+      "step": 2405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888222506921288e-05,
+      "loss": 2.158,
+      "step": 2406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888116528688205e-05,
+      "loss": 2.1437,
+      "step": 2407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888010500521632e-05,
+      "loss": 2.1673,
+      "step": 2408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98879044224221e-05,
+      "loss": 2.1676,
+      "step": 2409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988779829439015e-05,
+      "loss": 2.1608,
+      "step": 2410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887692116426316e-05,
+      "loss": 2.1797,
+      "step": 2411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988758588853113e-05,
+      "loss": 2.1537,
+      "step": 2412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887479610705132e-05,
+      "loss": 2.1631,
+      "step": 2413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988737328294886e-05,
+      "loss": 2.1723,
+      "step": 2414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988726690526285e-05,
+      "loss": 2.171,
+      "step": 2415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887160477647638e-05,
+      "loss": 2.1491,
+      "step": 2416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988705400010376e-05,
+      "loss": 2.1583,
+      "step": 2417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886947472631757e-05,
+      "loss": 2.1621,
+      "step": 2418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886840895232168e-05,
+      "loss": 2.1733,
+      "step": 2419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886734267905527e-05,
+      "loss": 2.1509,
+      "step": 2420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886627590652372e-05,
+      "loss": 2.1469,
+      "step": 2421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886520863473245e-05,
+      "loss": 2.1759,
+      "step": 2422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886414086368685e-05,
+      "loss": 2.1687,
+      "step": 2423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988630725933923e-05,
+      "loss": 2.1598,
+      "step": 2424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988620038238542e-05,
+      "loss": 2.1788,
+      "step": 2425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988609345550779e-05,
+      "loss": 2.1887,
+      "step": 2426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885986478706888e-05,
+      "loss": 2.1666,
+      "step": 2427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885879451983253e-05,
+      "loss": 2.1529,
+      "step": 2428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988577237533742e-05,
+      "loss": 2.161,
+      "step": 2429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885665248769934e-05,
+      "loss": 2.1514,
+      "step": 2430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885558072281333e-05,
+      "loss": 2.1377,
+      "step": 2431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885450845872158e-05,
+      "loss": 2.169,
+      "step": 2432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885343569542957e-05,
+      "loss": 2.1429,
+      "step": 2433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885236243294265e-05,
+      "loss": 2.1483,
+      "step": 2434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885128867126626e-05,
+      "loss": 2.1465,
+      "step": 2435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988502144104058e-05,
+      "loss": 2.1468,
+      "step": 2436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988491396503667e-05,
+      "loss": 2.1594,
+      "step": 2437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988480643911545e-05,
+      "loss": 2.1533,
+      "step": 2438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884698863277444e-05,
+      "loss": 2.136,
+      "step": 2439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988459123752321e-05,
+      "loss": 2.153,
+      "step": 2440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884483561853284e-05,
+      "loss": 2.1509,
+      "step": 2441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884375836268213e-05,
+      "loss": 2.1617,
+      "step": 2442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988426806076854e-05,
+      "loss": 2.14,
+      "step": 2443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988416023535481e-05,
+      "loss": 2.1627,
+      "step": 2444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884052360027563e-05,
+      "loss": 2.165,
+      "step": 2445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883944434787353e-05,
+      "loss": 2.1337,
+      "step": 2446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883836459634717e-05,
+      "loss": 2.1631,
+      "step": 2447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98837284345702e-05,
+      "loss": 2.1751,
+      "step": 2448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988362035959435e-05,
+      "loss": 2.1482,
+      "step": 2449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9883512234707717e-05,
+      "loss": 2.1608,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988340405991084e-05,
+      "loss": 2.1638,
+      "step": 2451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9883295835204274e-05,
+      "loss": 2.1533,
+      "step": 2452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9883187560588553e-05,
+      "loss": 2.1745,
+      "step": 2453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9883079236064232e-05,
+      "loss": 2.1633,
+      "step": 2454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988297086163186e-05,
+      "loss": 2.1417,
+      "step": 2455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988286243729198e-05,
+      "loss": 2.165,
+      "step": 2456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9882753963045135e-05,
+      "loss": 2.1756,
+      "step": 2457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9882645438891885e-05,
+      "loss": 2.1585,
+      "step": 2458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988253686483277e-05,
+      "loss": 2.1696,
+      "step": 2459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9882428240868338e-05,
+      "loss": 2.1532,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9882319566999138e-05,
+      "loss": 2.1451,
+      "step": 2461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988221084322572e-05,
+      "loss": 2.1664,
+      "step": 2462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9882102069548635e-05,
+      "loss": 2.1513,
+      "step": 2463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988199324596843e-05,
+      "loss": 2.1456,
+      "step": 2464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881884372485652e-05,
+      "loss": 2.1609,
+      "step": 2465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881775449100855e-05,
+      "loss": 2.1479,
+      "step": 2466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988166647581459e-05,
+      "loss": 2.1641,
+      "step": 2467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98815574526274e-05,
+      "loss": 2.1687,
+      "step": 2468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988144837953984e-05,
+      "loss": 2.145,
+      "step": 2469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881339256552465e-05,
+      "loss": 2.1643,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881230083665822e-05,
+      "loss": 2.1512,
+      "step": 2471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881120860880462e-05,
+      "loss": 2.1689,
+      "step": 2472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881011588196936e-05,
+      "loss": 2.1474,
+      "step": 2473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880902265615797e-05,
+      "loss": 2.1552,
+      "step": 2474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98807928931376e-05,
+      "loss": 2.1414,
+      "step": 2475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880683470762892e-05,
+      "loss": 2.1329,
+      "step": 2476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880573998492227e-05,
+      "loss": 2.1654,
+      "step": 2477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880464476326158e-05,
+      "loss": 2.1692,
+      "step": 2478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988035490426524e-05,
+      "loss": 2.154,
+      "step": 2479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880245282310026e-05,
+      "loss": 2.1377,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988013561046107e-05,
+      "loss": 2.1675,
+      "step": 2481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880025888718923e-05,
+      "loss": 2.1564,
+      "step": 2482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987991611708414e-05,
+      "loss": 2.1426,
+      "step": 2483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879806295557275e-05,
+      "loss": 2.1472,
+      "step": 2484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879696424138885e-05,
+      "loss": 2.1352,
+      "step": 2485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987958650282952e-05,
+      "loss": 2.1618,
+      "step": 2486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879476531629742e-05,
+      "loss": 2.149,
+      "step": 2487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879366510540102e-05,
+      "loss": 2.1325,
+      "step": 2488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879256439561156e-05,
+      "loss": 2.1555,
+      "step": 2489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987914631869346e-05,
+      "loss": 2.1649,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879036147937573e-05,
+      "loss": 2.145,
+      "step": 2491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878925927294044e-05,
+      "loss": 2.1675,
+      "step": 2492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878815656763436e-05,
+      "loss": 2.1358,
+      "step": 2493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878705336346304e-05,
+      "loss": 2.1599,
+      "step": 2494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878594966043204e-05,
+      "loss": 2.1582,
+      "step": 2495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878484545854695e-05,
+      "loss": 2.1525,
+      "step": 2496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878374075781336e-05,
+      "loss": 2.1219,
+      "step": 2497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878263555823682e-05,
+      "loss": 2.1503,
+      "step": 2498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878152985982293e-05,
+      "loss": 2.1363,
+      "step": 2499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878042366257727e-05,
+      "loss": 2.1438,
+      "step": 2500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877931696650538e-05,
+      "loss": 2.1324,
+      "step": 2501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987782097716129e-05,
+      "loss": 2.1568,
+      "step": 2502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877710207790543e-05,
+      "loss": 2.1624,
+      "step": 2503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877599388538856e-05,
+      "loss": 2.1455,
+      "step": 2504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987748851940678e-05,
+      "loss": 2.1565,
+      "step": 2505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987737760039489e-05,
+      "loss": 2.1466,
+      "step": 2506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877266631503734e-05,
+      "loss": 2.1257,
+      "step": 2507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877155612733876e-05,
+      "loss": 2.1521,
+      "step": 2508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877044544085878e-05,
+      "loss": 2.1576,
+      "step": 2509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98769334255603e-05,
+      "loss": 2.1461,
+      "step": 2510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98768222571577e-05,
+      "loss": 2.1396,
+      "step": 2511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987671103887865e-05,
+      "loss": 2.1141,
+      "step": 2512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9876599770723698e-05,
+      "loss": 2.1489,
+      "step": 2513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9876488452693414e-05,
+      "loss": 2.1555,
+      "step": 2514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9876377084788356e-05,
+      "loss": 2.1461,
+      "step": 2515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987626566700909e-05,
+      "loss": 2.1517,
+      "step": 2516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9876154199356176e-05,
+      "loss": 2.1591,
+      "step": 2517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9876042681830178e-05,
+      "loss": 2.1571,
+      "step": 2518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875931114431663e-05,
+      "loss": 2.1344,
+      "step": 2519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875819497161183e-05,
+      "loss": 2.1322,
+      "step": 2520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875707830019317e-05,
+      "loss": 2.15,
+      "step": 2521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875596113006615e-05,
+      "loss": 2.1379,
+      "step": 2522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875484346123652e-05,
+      "loss": 2.1521,
+      "step": 2523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875372529370987e-05,
+      "loss": 2.1311,
+      "step": 2524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987526066274918e-05,
+      "loss": 2.1462,
+      "step": 2525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987514874625881e-05,
+      "loss": 2.1527,
+      "step": 2526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875036779900426e-05,
+      "loss": 2.1603,
+      "step": 2527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874924763674603e-05,
+      "loss": 2.1587,
+      "step": 2528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874812697581903e-05,
+      "loss": 2.1318,
+      "step": 2529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874700581622896e-05,
+      "loss": 2.1301,
+      "step": 2530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874588415798145e-05,
+      "loss": 2.1345,
+      "step": 2531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874476200108215e-05,
+      "loss": 2.1573,
+      "step": 2532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874363934553676e-05,
+      "loss": 2.1562,
+      "step": 2533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874251619135094e-05,
+      "loss": 2.1266,
+      "step": 2534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874139253853037e-05,
+      "loss": 2.1328,
+      "step": 2535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987402683870807e-05,
+      "loss": 2.1435,
+      "step": 2536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987391437370076e-05,
+      "loss": 2.1643,
+      "step": 2537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987380185883168e-05,
+      "loss": 2.1343,
+      "step": 2538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9873689294101392e-05,
+      "loss": 2.1325,
+      "step": 2539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9873576679510467e-05,
+      "loss": 2.1408,
+      "step": 2540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9873464015059477e-05,
+      "loss": 2.139,
+      "step": 2541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987335130074899e-05,
+      "loss": 2.1522,
+      "step": 2542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987323853657957e-05,
+      "loss": 2.1291,
+      "step": 2543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9873125722551785e-05,
+      "loss": 2.1193,
+      "step": 2544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9873012858666217e-05,
+      "loss": 2.1442,
+      "step": 2545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9872899944923425e-05,
+      "loss": 2.1321,
+      "step": 2546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9872786981323982e-05,
+      "loss": 2.1468,
+      "step": 2547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9872673967868458e-05,
+      "loss": 2.15,
+      "step": 2548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987256090455743e-05,
+      "loss": 2.1293,
+      "step": 2549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987244779139146e-05,
+      "loss": 2.156,
+      "step": 2550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987233462837112e-05,
+      "loss": 2.1271,
+      "step": 2551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987222141549699e-05,
+      "loss": 2.1529,
+      "step": 2552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987210815276963e-05,
+      "loss": 2.1359,
+      "step": 2553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987199484018962e-05,
+      "loss": 2.1439,
+      "step": 2554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9871881477757533e-05,
+      "loss": 2.1191,
+      "step": 2555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9871768065473937e-05,
+      "loss": 2.1609,
+      "step": 2556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9871654603339406e-05,
+      "loss": 2.1497,
+      "step": 2557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9871541091354512e-05,
+      "loss": 2.1346,
+      "step": 2558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987142752951983e-05,
+      "loss": 2.1327,
+      "step": 2559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9871313917835933e-05,
+      "loss": 2.1551,
+      "step": 2560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9871200256303395e-05,
+      "loss": 2.1453,
+      "step": 2561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987108654492279e-05,
+      "loss": 2.1291,
+      "step": 2562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870972783694693e-05,
+      "loss": 2.1212,
+      "step": 2563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870858972619675e-05,
+      "loss": 2.1283,
+      "step": 2564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870745111698312e-05,
+      "loss": 2.132,
+      "step": 2565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870631200931184e-05,
+      "loss": 2.1256,
+      "step": 2566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870517240318862e-05,
+      "loss": 2.1353,
+      "step": 2567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987040322986192e-05,
+      "loss": 2.142,
+      "step": 2568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870289169560934e-05,
+      "loss": 2.1471,
+      "step": 2569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870175059416485e-05,
+      "loss": 2.1408,
+      "step": 2570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870060899429144e-05,
+      "loss": 2.1506,
+      "step": 2571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869946689599487e-05,
+      "loss": 2.1477,
+      "step": 2572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869832429928095e-05,
+      "loss": 2.1456,
+      "step": 2573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869718120415545e-05,
+      "loss": 2.1452,
+      "step": 2574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986960376106241e-05,
+      "loss": 2.1242,
+      "step": 2575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986948935186927e-05,
+      "loss": 2.1302,
+      "step": 2576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869374892836703e-05,
+      "loss": 2.1351,
+      "step": 2577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869260383965287e-05,
+      "loss": 2.1419,
+      "step": 2578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869145825255597e-05,
+      "loss": 2.1403,
+      "step": 2579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869031216708213e-05,
+      "loss": 2.1441,
+      "step": 2580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986891655832372e-05,
+      "loss": 2.1425,
+      "step": 2581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986880185010269e-05,
+      "loss": 2.1096,
+      "step": 2582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9868687092045708e-05,
+      "loss": 2.1094,
+      "step": 2583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986857228415334e-05,
+      "loss": 2.1427,
+      "step": 2584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9868457426426185e-05,
+      "loss": 2.1343,
+      "step": 2585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986834251886481e-05,
+      "loss": 2.121,
+      "step": 2586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98682275614698e-05,
+      "loss": 2.145,
+      "step": 2587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986811255424173e-05,
+      "loss": 2.1514,
+      "step": 2588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986799749718119e-05,
+      "loss": 2.1269,
+      "step": 2589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867882390288754e-05,
+      "loss": 2.1409,
+      "step": 2590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867767233565005e-05,
+      "loss": 2.1293,
+      "step": 2591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867652027010527e-05,
+      "loss": 2.133,
+      "step": 2592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867536770625897e-05,
+      "loss": 2.1365,
+      "step": 2593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867421464411702e-05,
+      "loss": 2.1426,
+      "step": 2594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986730610836852e-05,
+      "loss": 2.1186,
+      "step": 2595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867190702496937e-05,
+      "loss": 2.1361,
+      "step": 2596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986707524679753e-05,
+      "loss": 2.1314,
+      "step": 2597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986695974127089e-05,
+      "loss": 2.1517,
+      "step": 2598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98668441859176e-05,
+      "loss": 2.144,
+      "step": 2599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9866728580738237e-05,
+      "loss": 2.1656,
+      "step": 2600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9866612925733387e-05,
+      "loss": 2.1361,
+      "step": 2601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9866497220903637e-05,
+      "loss": 2.1283,
+      "step": 2602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9866381466249565e-05,
+      "loss": 2.1492,
+      "step": 2603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9866265661771765e-05,
+      "loss": 2.1132,
+      "step": 2604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9866149807470813e-05,
+      "loss": 2.1271,
+      "step": 2605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98660339033473e-05,
+      "loss": 2.1316,
+      "step": 2606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9865917949401807e-05,
+      "loss": 2.1362,
+      "step": 2607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9865801945634923e-05,
+      "loss": 2.1364,
+      "step": 2608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986568589204723e-05,
+      "loss": 2.1311,
+      "step": 2609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986556978863932e-05,
+      "loss": 2.139,
+      "step": 2610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9865453635411773e-05,
+      "loss": 2.1352,
+      "step": 2611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986533743236518e-05,
+      "loss": 2.1223,
+      "step": 2612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9865221179500124e-05,
+      "loss": 2.1434,
+      "step": 2613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9865104876817195e-05,
+      "loss": 2.1494,
+      "step": 2614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986498852431698e-05,
+      "loss": 2.1421,
+      "step": 2615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864872122000067e-05,
+      "loss": 2.1245,
+      "step": 2616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864755669867038e-05,
+      "loss": 2.1306,
+      "step": 2617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986463916791849e-05,
+      "loss": 2.1471,
+      "step": 2618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864522616155008e-05,
+      "loss": 2.1494,
+      "step": 2619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864406014577175e-05,
+      "loss": 2.1364,
+      "step": 2620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986428936318559e-05,
+      "loss": 2.127,
+      "step": 2621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864172661980835e-05,
+      "loss": 2.1374,
+      "step": 2622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864055910963496e-05,
+      "loss": 2.1298,
+      "step": 2623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9863939110134173e-05,
+      "loss": 2.1415,
+      "step": 2624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986382225949345e-05,
+      "loss": 2.1416,
+      "step": 2625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986370535904192e-05,
+      "loss": 2.1215,
+      "step": 2626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9863588408780165e-05,
+      "loss": 2.1177,
+      "step": 2627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986347140870878e-05,
+      "loss": 2.1359,
+      "step": 2628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9863354358828362e-05,
+      "loss": 2.1393,
+      "step": 2629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9863237259139496e-05,
+      "loss": 2.137,
+      "step": 2630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9863120109642775e-05,
+      "loss": 2.1254,
+      "step": 2631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986300291033879e-05,
+      "loss": 2.1191,
+      "step": 2632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9862885661228133e-05,
+      "loss": 2.1429,
+      "step": 2633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9862768362311396e-05,
+      "loss": 2.1239,
+      "step": 2634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9862651013589168e-05,
+      "loss": 2.1236,
+      "step": 2635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9862533615062053e-05,
+      "loss": 2.1474,
+      "step": 2636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9862416166730628e-05,
+      "loss": 2.1385,
+      "step": 2637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9862298668595498e-05,
+      "loss": 2.1264,
+      "step": 2638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986218112065725e-05,
+      "loss": 2.1355,
+      "step": 2639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986206352291648e-05,
+      "loss": 2.1015,
+      "step": 2640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861945875373785e-05,
+      "loss": 2.1416,
+      "step": 2641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861828178029754e-05,
+      "loss": 2.1286,
+      "step": 2642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861710430884983e-05,
+      "loss": 2.1295,
+      "step": 2643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861592633940066e-05,
+      "loss": 2.138,
+      "step": 2644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98614747871956e-05,
+      "loss": 2.139,
+      "step": 2645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986135689065218e-05,
+      "loss": 2.1313,
+      "step": 2646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861238944310398e-05,
+      "loss": 2.1054,
+      "step": 2647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861120948170853e-05,
+      "loss": 2.106,
+      "step": 2648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986100290223414e-05,
+      "loss": 2.0994,
+      "step": 2649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860884806500853e-05,
+      "loss": 2.1253,
+      "step": 2650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986076666097159e-05,
+      "loss": 2.1049,
+      "step": 2651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986064846564695e-05,
+      "loss": 2.1219,
+      "step": 2652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860530220527524e-05,
+      "loss": 2.1193,
+      "step": 2653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860411925613915e-05,
+      "loss": 2.1315,
+      "step": 2654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860293580906716e-05,
+      "loss": 2.1116,
+      "step": 2655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860175186406532e-05,
+      "loss": 2.1181,
+      "step": 2656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860056742113954e-05,
+      "loss": 2.1302,
+      "step": 2657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9859938248029577e-05,
+      "loss": 2.0996,
+      "step": 2658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9859819704154008e-05,
+      "loss": 2.1258,
+      "step": 2659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985970111048784e-05,
+      "loss": 2.1326,
+      "step": 2660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9859582467031676e-05,
+      "loss": 2.1331,
+      "step": 2661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985946377378611e-05,
+      "loss": 2.1361,
+      "step": 2662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9859345030751746e-05,
+      "loss": 2.1209,
+      "step": 2663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9859226237929182e-05,
+      "loss": 2.1069,
+      "step": 2664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9859107395319018e-05,
+      "loss": 2.1336,
+      "step": 2665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985898850292185e-05,
+      "loss": 2.1096,
+      "step": 2666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9858869560738287e-05,
+      "loss": 2.1288,
+      "step": 2667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985875056876892e-05,
+      "loss": 2.1237,
+      "step": 2668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9858631527014357e-05,
+      "loss": 2.1227,
+      "step": 2669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98585124354752e-05,
+      "loss": 2.1143,
+      "step": 2670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9858393294152042e-05,
+      "loss": 2.1259,
+      "step": 2671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9858274103045493e-05,
+      "loss": 2.1071,
+      "step": 2672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9858154862156148e-05,
+      "loss": 2.1349,
+      "step": 2673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9858035571484616e-05,
+      "loss": 2.1243,
+      "step": 2674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857916231031498e-05,
+      "loss": 2.1095,
+      "step": 2675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857796840797393e-05,
+      "loss": 2.1194,
+      "step": 2676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857677400782904e-05,
+      "loss": 2.1048,
+      "step": 2677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857557910988638e-05,
+      "loss": 2.1272,
+      "step": 2678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857438371415197e-05,
+      "loss": 2.1258,
+      "step": 2679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857318782063185e-05,
+      "loss": 2.1274,
+      "step": 2680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857199142933202e-05,
+      "loss": 2.1117,
+      "step": 2681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857079454025858e-05,
+      "loss": 2.121,
+      "step": 2682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985695971534175e-05,
+      "loss": 2.1162,
+      "step": 2683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9856839926881492e-05,
+      "loss": 2.1092,
+      "step": 2684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985672008864568e-05,
+      "loss": 2.1172,
+      "step": 2685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985660020063493e-05,
+      "loss": 2.1318,
+      "step": 2686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9856480262849835e-05,
+      "loss": 2.1063,
+      "step": 2687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985636027529101e-05,
+      "loss": 2.1239,
+      "step": 2688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9856240237959054e-05,
+      "loss": 2.1206,
+      "step": 2689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9856120150854575e-05,
+      "loss": 2.1172,
+      "step": 2690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9856000013978186e-05,
+      "loss": 2.1298,
+      "step": 2691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9855879827330487e-05,
+      "loss": 2.1181,
+      "step": 2692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9855759590912083e-05,
+      "loss": 2.1221,
+      "step": 2693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985563930472359e-05,
+      "loss": 2.1286,
+      "step": 2694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9855518968765608e-05,
+      "loss": 2.1096,
+      "step": 2695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9855398583038745e-05,
+      "loss": 2.1295,
+      "step": 2696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985527814754361e-05,
+      "loss": 2.112,
+      "step": 2697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9855157662280816e-05,
+      "loss": 2.1182,
+      "step": 2698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9855037127250966e-05,
+      "loss": 2.1161,
+      "step": 2699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985491654245467e-05,
+      "loss": 2.1216,
+      "step": 2700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9854795907892534e-05,
+      "loss": 2.0996,
+      "step": 2701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985467522356517e-05,
+      "loss": 2.131,
+      "step": 2702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9854554489473193e-05,
+      "loss": 2.1283,
+      "step": 2703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9854433705617203e-05,
+      "loss": 2.0848,
+      "step": 2704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9854312871997817e-05,
+      "loss": 2.0934,
+      "step": 2705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9854191988615642e-05,
+      "loss": 2.1006,
+      "step": 2706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9854071055471284e-05,
+      "loss": 2.1338,
+      "step": 2707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853950072565363e-05,
+      "loss": 2.1029,
+      "step": 2708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853829039898483e-05,
+      "loss": 2.1097,
+      "step": 2709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853707957471263e-05,
+      "loss": 2.124,
+      "step": 2710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853586825284303e-05,
+      "loss": 2.1261,
+      "step": 2711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853465643338223e-05,
+      "loss": 2.0997,
+      "step": 2712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853344411633634e-05,
+      "loss": 2.1095,
+      "step": 2713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853223130171147e-05,
+      "loss": 2.1039,
+      "step": 2714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853101798951374e-05,
+      "loss": 2.0902,
+      "step": 2715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9852980417974926e-05,
+      "loss": 2.1156,
+      "step": 2716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985285898724242e-05,
+      "loss": 2.1156,
+      "step": 2717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9852737506754464e-05,
+      "loss": 2.1156,
+      "step": 2718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985261597651168e-05,
+      "loss": 2.1028,
+      "step": 2719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9852494396514673e-05,
+      "loss": 2.1152,
+      "step": 2720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985237276676406e-05,
+      "loss": 2.1041,
+      "step": 2721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985225108726046e-05,
+      "loss": 2.1113,
+      "step": 2722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9852129358004477e-05,
+      "loss": 2.0985,
+      "step": 2723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985200757899674e-05,
+      "loss": 2.0959,
+      "step": 2724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985188575023785e-05,
+      "loss": 2.1281,
+      "step": 2725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9851763871728424e-05,
+      "loss": 2.1193,
+      "step": 2726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9851641943469085e-05,
+      "loss": 2.1163,
+      "step": 2727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985151996546045e-05,
+      "loss": 2.127,
+      "step": 2728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9851397937703124e-05,
+      "loss": 2.1009,
+      "step": 2729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9851275860197732e-05,
+      "loss": 2.1091,
+      "step": 2730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985115373294489e-05,
+      "loss": 2.1388,
+      "step": 2731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9851031555945208e-05,
+      "loss": 2.1078,
+      "step": 2732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985090932919931e-05,
+      "loss": 2.1055,
+      "step": 2733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850787052707812e-05,
+      "loss": 2.1105,
+      "step": 2734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850664726471327e-05,
+      "loss": 2.1113,
+      "step": 2735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850542350490477e-05,
+      "loss": 2.1143,
+      "step": 2736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985041992476588e-05,
+      "loss": 2.1115,
+      "step": 2737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985029744929815e-05,
+      "loss": 2.1066,
+      "step": 2738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850174924087915e-05,
+      "loss": 2.1137,
+      "step": 2739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850052349135783e-05,
+      "loss": 2.1042,
+      "step": 2740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849929724442377e-05,
+      "loss": 2.1095,
+      "step": 2741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849807050008317e-05,
+      "loss": 2.0921,
+      "step": 2742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849684325834225e-05,
+      "loss": 2.1206,
+      "step": 2743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849561551920716e-05,
+      "loss": 2.1199,
+      "step": 2744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849438728268412e-05,
+      "loss": 2.1279,
+      "step": 2745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849315854877934e-05,
+      "loss": 2.1205,
+      "step": 2746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849192931749897e-05,
+      "loss": 2.1046,
+      "step": 2747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849069958884933e-05,
+      "loss": 2.0922,
+      "step": 2748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984894693628365e-05,
+      "loss": 2.1159,
+      "step": 2749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984882386394668e-05,
+      "loss": 2.1191,
+      "step": 2750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984870074187464e-05,
+      "loss": 2.1211,
+      "step": 2751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984857757006815e-05,
+      "loss": 2.1248,
+      "step": 2752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9848454348527833e-05,
+      "loss": 2.1289,
+      "step": 2753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9848331077254313e-05,
+      "loss": 2.1035,
+      "step": 2754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9848207756248213e-05,
+      "loss": 2.1085,
+      "step": 2755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9848084385510153e-05,
+      "loss": 2.094,
+      "step": 2756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984796096504076e-05,
+      "loss": 2.1136,
+      "step": 2757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984783749484065e-05,
+      "loss": 2.086,
+      "step": 2758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984771397491045e-05,
+      "loss": 2.0903,
+      "step": 2759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984759040525079e-05,
+      "loss": 2.1098,
+      "step": 2760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9847466785862287e-05,
+      "loss": 2.1094,
+      "step": 2761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9847343116745565e-05,
+      "loss": 2.1046,
+      "step": 2762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9847219397901253e-05,
+      "loss": 2.0889,
+      "step": 2763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9847095629329974e-05,
+      "loss": 2.0898,
+      "step": 2764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984697181103235e-05,
+      "loss": 2.0856,
+      "step": 2765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984684794300901e-05,
+      "loss": 2.0797,
+      "step": 2766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984672402526058e-05,
+      "loss": 2.1101,
+      "step": 2767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984660005778768e-05,
+      "loss": 2.0972,
+      "step": 2768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984647604059094e-05,
+      "loss": 2.1006,
+      "step": 2769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846351973670986e-05,
+      "loss": 2.1143,
+      "step": 2770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846227857028448e-05,
+      "loss": 2.0912,
+      "step": 2771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846103690663948e-05,
+      "loss": 2.0984,
+      "step": 2772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845979474578115e-05,
+      "loss": 2.1026,
+      "step": 2773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845855208771578e-05,
+      "loss": 2.1441,
+      "step": 2774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845730893244958e-05,
+      "loss": 2.1051,
+      "step": 2775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845606527998887e-05,
+      "loss": 2.079,
+      "step": 2776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845482113033995e-05,
+      "loss": 2.1063,
+      "step": 2777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845357648350907e-05,
+      "loss": 2.0954,
+      "step": 2778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845233133950256e-05,
+      "loss": 2.1081,
+      "step": 2779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845108569832666e-05,
+      "loss": 2.0929,
+      "step": 2780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844983955998766e-05,
+      "loss": 2.1087,
+      "step": 2781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984485929244919e-05,
+      "loss": 2.1014,
+      "step": 2782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984473457918456e-05,
+      "loss": 2.1125,
+      "step": 2783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844609816205516e-05,
+      "loss": 2.1188,
+      "step": 2784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844485003512678e-05,
+      "loss": 2.1004,
+      "step": 2785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844360141106683e-05,
+      "loss": 2.1027,
+      "step": 2786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984423522898816e-05,
+      "loss": 2.1295,
+      "step": 2787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844110267157738e-05,
+      "loss": 2.1084,
+      "step": 2788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984398525561605e-05,
+      "loss": 2.1115,
+      "step": 2789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843860194363723e-05,
+      "loss": 2.0805,
+      "step": 2790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843735083401392e-05,
+      "loss": 2.0952,
+      "step": 2791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984360992272969e-05,
+      "loss": 2.1148,
+      "step": 2792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843484712349247e-05,
+      "loss": 2.0914,
+      "step": 2793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843359452260698e-05,
+      "loss": 2.0982,
+      "step": 2794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843234142464673e-05,
+      "loss": 2.1036,
+      "step": 2795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843108782961805e-05,
+      "loss": 2.0947,
+      "step": 2796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842983373752725e-05,
+      "loss": 2.0901,
+      "step": 2797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842857914838072e-05,
+      "loss": 2.1082,
+      "step": 2798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842732406218475e-05,
+      "loss": 2.0974,
+      "step": 2799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842606847894568e-05,
+      "loss": 2.0793,
+      "step": 2800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842481239866988e-05,
+      "loss": 2.1285,
+      "step": 2801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842355582136365e-05,
+      "loss": 2.0892,
+      "step": 2802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842229874703336e-05,
+      "loss": 2.0821,
+      "step": 2803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842104117568538e-05,
+      "loss": 2.0994,
+      "step": 2804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98419783107326e-05,
+      "loss": 2.101,
+      "step": 2805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841852454196165e-05,
+      "loss": 2.0893,
+      "step": 2806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841726547959865e-05,
+      "loss": 2.099,
+      "step": 2807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841600592024334e-05,
+      "loss": 2.1167,
+      "step": 2808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841474586390208e-05,
+      "loss": 2.0929,
+      "step": 2809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841348531058125e-05,
+      "loss": 2.0991,
+      "step": 2810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841222426028722e-05,
+      "loss": 2.1019,
+      "step": 2811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841096271302635e-05,
+      "loss": 2.0984,
+      "step": 2812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98409700668805e-05,
+      "loss": 2.0855,
+      "step": 2813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984084381276296e-05,
+      "loss": 2.1105,
+      "step": 2814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840717508950647e-05,
+      "loss": 2.1081,
+      "step": 2815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840591155444196e-05,
+      "loss": 2.1076,
+      "step": 2816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984046475224425e-05,
+      "loss": 2.0879,
+      "step": 2817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984033829935145e-05,
+      "loss": 2.1096,
+      "step": 2818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840211796766428e-05,
+      "loss": 2.1056,
+      "step": 2819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840085244489825e-05,
+      "loss": 2.0918,
+      "step": 2820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983995864252228e-05,
+      "loss": 2.1117,
+      "step": 2821
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839831990864438e-05,
+      "loss": 2.0967,
+      "step": 2822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839705289516928e-05,
+      "loss": 2.1063,
+      "step": 2823
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98395785384804e-05,
+      "loss": 2.1019,
+      "step": 2824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839451737755487e-05,
+      "loss": 2.1058,
+      "step": 2825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839324887342832e-05,
+      "loss": 2.0862,
+      "step": 2826
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839197987243077e-05,
+      "loss": 2.1021,
+      "step": 2827
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983907103745686e-05,
+      "loss": 2.1007,
+      "step": 2828
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983894403798482e-05,
+      "loss": 2.1097,
+      "step": 2829
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838816988827605e-05,
+      "loss": 2.1018,
+      "step": 2830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838689889985856e-05,
+      "loss": 2.0897,
+      "step": 2831
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838562741460207e-05,
+      "loss": 2.1037,
+      "step": 2832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838435543251308e-05,
+      "loss": 2.1158,
+      "step": 2833
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838308295359797e-05,
+      "loss": 2.0992,
+      "step": 2834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983818099778632e-05,
+      "loss": 2.0951,
+      "step": 2835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838053650531516e-05,
+      "loss": 2.0995,
+      "step": 2836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837926253596033e-05,
+      "loss": 2.0888,
+      "step": 2837
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983779880698051e-05,
+      "loss": 2.0907,
+      "step": 2838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837671310685592e-05,
+      "loss": 2.0993,
+      "step": 2839
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983754376471192e-05,
+      "loss": 2.0986,
+      "step": 2840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983741616906014e-05,
+      "loss": 2.0989,
+      "step": 2841
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98372885237309e-05,
+      "loss": 2.1304,
+      "step": 2842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837160828724843e-05,
+      "loss": 2.1124,
+      "step": 2843
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837033084042613e-05,
+      "loss": 2.1013,
+      "step": 2844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836905289684855e-05,
+      "loss": 2.101,
+      "step": 2845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983677744565221e-05,
+      "loss": 2.0544,
+      "step": 2846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983664955194533e-05,
+      "loss": 2.0784,
+      "step": 2847
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836521608564858e-05,
+      "loss": 2.0963,
+      "step": 2848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983639361551144e-05,
+      "loss": 2.1063,
+      "step": 2849
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836265572785725e-05,
+      "loss": 2.1083,
+      "step": 2850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836137480388355e-05,
+      "loss": 2.0912,
+      "step": 2851
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983600933831998e-05,
+      "loss": 2.1016,
+      "step": 2852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983588114658125e-05,
+      "loss": 2.0807,
+      "step": 2853
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835752905172804e-05,
+      "loss": 2.0951,
+      "step": 2854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835624614095298e-05,
+      "loss": 2.0979,
+      "step": 2855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835496273349376e-05,
+      "loss": 2.1003,
+      "step": 2856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835367882935682e-05,
+      "loss": 2.089,
+      "step": 2857
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835239442854872e-05,
+      "loss": 2.0976,
+      "step": 2858
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983511095310759e-05,
+      "loss": 2.093,
+      "step": 2859
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983498241369449e-05,
+      "loss": 2.0989,
+      "step": 2860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983485382461621e-05,
+      "loss": 2.0806,
+      "step": 2861
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834725185873415e-05,
+      "loss": 2.0898,
+      "step": 2862
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834596497466743e-05,
+      "loss": 2.111,
+      "step": 2863
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834467759396846e-05,
+      "loss": 2.0967,
+      "step": 2864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834338971664374e-05,
+      "loss": 2.0917,
+      "step": 2865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983421013426998e-05,
+      "loss": 2.1017,
+      "step": 2866
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834081247214313e-05,
+      "loss": 2.0954,
+      "step": 2867
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833952310498025e-05,
+      "loss": 2.1036,
+      "step": 2868
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833823324121764e-05,
+      "loss": 2.0979,
+      "step": 2869
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833694288086188e-05,
+      "loss": 2.1034,
+      "step": 2870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983356520239194e-05,
+      "loss": 2.0942,
+      "step": 2871
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833436067039677e-05,
+      "loss": 2.1059,
+      "step": 2872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983330688203005e-05,
+      "loss": 2.0888,
+      "step": 2873
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833177647363713e-05,
+      "loss": 2.0966,
+      "step": 2874
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833048363041316e-05,
+      "loss": 2.0763,
+      "step": 2875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983291902906351e-05,
+      "loss": 2.1036,
+      "step": 2876
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832789645430955e-05,
+      "loss": 2.1007,
+      "step": 2877
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98326602121443e-05,
+      "loss": 2.084,
+      "step": 2878
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832530729204195e-05,
+      "loss": 2.0966,
+      "step": 2879
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832401196611305e-05,
+      "loss": 2.1009,
+      "step": 2880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832271614366272e-05,
+      "loss": 2.0708,
+      "step": 2881
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832141982469756e-05,
+      "loss": 2.0943,
+      "step": 2882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832012300922412e-05,
+      "loss": 2.0956,
+      "step": 2883
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831882569724894e-05,
+      "loss": 2.107,
+      "step": 2884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831752788877855e-05,
+      "loss": 2.0997,
+      "step": 2885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831622958381956e-05,
+      "loss": 2.0863,
+      "step": 2886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831493078237844e-05,
+      "loss": 2.0838,
+      "step": 2887
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831363148446187e-05,
+      "loss": 2.1029,
+      "step": 2888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831233169007628e-05,
+      "loss": 2.0903,
+      "step": 2889
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831103139922828e-05,
+      "loss": 2.0954,
+      "step": 2890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9830973061192452e-05,
+      "loss": 2.0736,
+      "step": 2891
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9830842932817146e-05,
+      "loss": 2.1054,
+      "step": 2892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983071275479757e-05,
+      "loss": 2.1044,
+      "step": 2893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9830582527134384e-05,
+      "loss": 2.1051,
+      "step": 2894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9830452249828242e-05,
+      "loss": 2.0834,
+      "step": 2895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9830321922879804e-05,
+      "loss": 2.0895,
+      "step": 2896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983019154628973e-05,
+      "loss": 2.1043,
+      "step": 2897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983006112005868e-05,
+      "loss": 2.097,
+      "step": 2898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9829930644187303e-05,
+      "loss": 2.0762,
+      "step": 2899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9829800118676266e-05,
+      "loss": 2.0858,
+      "step": 2900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9829669543526225e-05,
+      "loss": 2.092,
+      "step": 2901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982953891873784e-05,
+      "loss": 2.0805,
+      "step": 2902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9829408244311775e-05,
+      "loss": 2.103,
+      "step": 2903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9829277520248684e-05,
+      "loss": 2.1027,
+      "step": 2904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9829146746549228e-05,
+      "loss": 2.0725,
+      "step": 2905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9829015923214068e-05,
+      "loss": 2.0734,
+      "step": 2906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9828885050243868e-05,
+      "loss": 2.0881,
+      "step": 2907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9828754127639286e-05,
+      "loss": 2.0946,
+      "step": 2908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9828623155400982e-05,
+      "loss": 2.0792,
+      "step": 2909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9828492133529614e-05,
+      "loss": 2.1015,
+      "step": 2910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9828361062025852e-05,
+      "loss": 2.1003,
+      "step": 2911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9828229940890358e-05,
+      "loss": 2.0842,
+      "step": 2912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9828098770123784e-05,
+      "loss": 2.0828,
+      "step": 2913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9827967549726802e-05,
+      "loss": 2.0858,
+      "step": 2914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9827836279700073e-05,
+      "loss": 2.0917,
+      "step": 2915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9827704960044257e-05,
+      "loss": 2.0916,
+      "step": 2916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9827573590760015e-05,
+      "loss": 2.1085,
+      "step": 2917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9827442171848017e-05,
+      "loss": 2.0764,
+      "step": 2918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982731070330892e-05,
+      "loss": 2.0863,
+      "step": 2919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9827179185143396e-05,
+      "loss": 2.0923,
+      "step": 2920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98270476173521e-05,
+      "loss": 2.0859,
+      "step": 2921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9826915999935702e-05,
+      "loss": 2.0921,
+      "step": 2922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9826784332894866e-05,
+      "loss": 2.0899,
+      "step": 2923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982665261623026e-05,
+      "loss": 2.0856,
+      "step": 2924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9826520849942538e-05,
+      "loss": 2.0666,
+      "step": 2925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9826389034032378e-05,
+      "loss": 2.0874,
+      "step": 2926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982625716850044e-05,
+      "loss": 2.0865,
+      "step": 2927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982612525334739e-05,
+      "loss": 2.1013,
+      "step": 2928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9825993288573893e-05,
+      "loss": 2.0636,
+      "step": 2929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982586127418062e-05,
+      "loss": 2.099,
+      "step": 2930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9825729210168232e-05,
+      "loss": 2.0981,
+      "step": 2931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9825597096537403e-05,
+      "loss": 2.0842,
+      "step": 2932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982546493328879e-05,
+      "loss": 2.098,
+      "step": 2933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9825332720423066e-05,
+      "loss": 2.0958,
+      "step": 2934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9825200457940903e-05,
+      "loss": 2.0806,
+      "step": 2935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9825068145842963e-05,
+      "loss": 2.0842,
+      "step": 2936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982493578412992e-05,
+      "loss": 2.0956,
+      "step": 2937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9824803372802428e-05,
+      "loss": 2.1008,
+      "step": 2938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982467091186117e-05,
+      "loss": 2.0869,
+      "step": 2939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9824538401306816e-05,
+      "loss": 2.0901,
+      "step": 2940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9824405841140027e-05,
+      "loss": 2.1086,
+      "step": 2941
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9824273231361475e-05,
+      "loss": 2.0907,
+      "step": 2942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982414057197183e-05,
+      "loss": 2.0945,
+      "step": 2943
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982400786297176e-05,
+      "loss": 2.087,
+      "step": 2944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982387510436194e-05,
+      "loss": 2.0983,
+      "step": 2945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9823742296143036e-05,
+      "loss": 2.1066,
+      "step": 2946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982360943831572e-05,
+      "loss": 2.0917,
+      "step": 2947
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9823476530880666e-05,
+      "loss": 2.0882,
+      "step": 2948
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982334357383854e-05,
+      "loss": 2.075,
+      "step": 2949
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9823210567190013e-05,
+      "loss": 2.0937,
+      "step": 2950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9823077510935763e-05,
+      "loss": 2.0833,
+      "step": 2951
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9822944405076458e-05,
+      "loss": 2.0949,
+      "step": 2952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982281124961277e-05,
+      "loss": 2.1037,
+      "step": 2953
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982267804454537e-05,
+      "loss": 2.0889,
+      "step": 2954
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9822544789874936e-05,
+      "loss": 2.0858,
+      "step": 2955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9822411485602134e-05,
+      "loss": 2.0951,
+      "step": 2956
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9822278131727645e-05,
+      "loss": 2.0803,
+      "step": 2957
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9822144728252133e-05,
+      "loss": 2.1104,
+      "step": 2958
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982201127517628e-05,
+      "loss": 2.0775,
+      "step": 2959
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9821877772500754e-05,
+      "loss": 2.0786,
+      "step": 2960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9821744220226233e-05,
+      "loss": 2.0999,
+      "step": 2961
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982161061835339e-05,
+      "loss": 2.0919,
+      "step": 2962
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98214769668829e-05,
+      "loss": 2.094,
+      "step": 2963
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9821343265815436e-05,
+      "loss": 2.0727,
+      "step": 2964
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9821209515151676e-05,
+      "loss": 2.0748,
+      "step": 2965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9821075714892295e-05,
+      "loss": 2.1213,
+      "step": 2966
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9820941865037968e-05,
+      "loss": 2.0795,
+      "step": 2967
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982080796558937e-05,
+      "loss": 2.0814,
+      "step": 2968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982067401654718e-05,
+      "loss": 2.0776,
+      "step": 2969
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982054001791207e-05,
+      "loss": 2.0785,
+      "step": 2970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9820405969684722e-05,
+      "loss": 2.0781,
+      "step": 2971
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982027187186581e-05,
+      "loss": 2.0829,
+      "step": 2972
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982013772445601e-05,
+      "loss": 2.0865,
+      "step": 2973
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9820003527455997e-05,
+      "loss": 2.0694,
+      "step": 2974
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9819869280866457e-05,
+      "loss": 2.0612,
+      "step": 2975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9819734984688065e-05,
+      "loss": 2.0827,
+      "step": 2976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981960063892149e-05,
+      "loss": 2.0739,
+      "step": 2977
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9819466243567427e-05,
+      "loss": 2.0806,
+      "step": 2978
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981933179862654e-05,
+      "loss": 2.1019,
+      "step": 2979
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9819197304099512e-05,
+      "loss": 2.1201,
+      "step": 2980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9819062759987028e-05,
+      "loss": 2.0746,
+      "step": 2981
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9818928166289763e-05,
+      "loss": 2.074,
+      "step": 2982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9818793523008396e-05,
+      "loss": 2.082,
+      "step": 2983
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9818658830143604e-05,
+      "loss": 2.0897,
+      "step": 2984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9818524087696075e-05,
+      "loss": 2.0979,
+      "step": 2985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9818389295666486e-05,
+      "loss": 2.0968,
+      "step": 2986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9818254454055515e-05,
+      "loss": 2.088,
+      "step": 2987
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9818119562863844e-05,
+      "loss": 2.0832,
+      "step": 2988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9817984622092156e-05,
+      "loss": 2.101,
+      "step": 2989
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9817849631741132e-05,
+      "loss": 2.0628,
+      "step": 2990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9817714591811454e-05,
+      "loss": 2.0886,
+      "step": 2991
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9817579502303803e-05,
+      "loss": 2.0887,
+      "step": 2992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981744436321886e-05,
+      "loss": 2.0976,
+      "step": 2993
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9817309174557307e-05,
+      "loss": 2.0796,
+      "step": 2994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9817173936319832e-05,
+      "loss": 2.0811,
+      "step": 2995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9817038648507118e-05,
+      "loss": 2.0539,
+      "step": 2996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9816903311119838e-05,
+      "loss": 2.0778,
+      "step": 2997
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9816767924158685e-05,
+      "loss": 2.0854,
+      "step": 2998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9816632487624335e-05,
+      "loss": 2.0915,
+      "step": 2999
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9816497001517484e-05,
+      "loss": 2.086,
+      "step": 3000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9816361465838802e-05,
+      "loss": 2.0843,
+      "step": 3001
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9816225880588984e-05,
+      "loss": 2.0896,
+      "step": 3002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981609024576871e-05,
+      "loss": 2.0805,
+      "step": 3003
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815954561378667e-05,
+      "loss": 2.0762,
+      "step": 3004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815818827419537e-05,
+      "loss": 2.0727,
+      "step": 3005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815683043892008e-05,
+      "loss": 2.074,
+      "step": 3006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815547210796765e-05,
+      "loss": 2.0887,
+      "step": 3007
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815411328134496e-05,
+      "loss": 2.0813,
+      "step": 3008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815275395905888e-05,
+      "loss": 2.0717,
+      "step": 3009
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815139414111616e-05,
+      "loss": 2.0927,
+      "step": 3010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815003382752384e-05,
+      "loss": 2.0965,
+      "step": 3011
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9814867301828866e-05,
+      "loss": 2.0867,
+      "step": 3012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981473117134175e-05,
+      "loss": 2.0581,
+      "step": 3013
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9814594991291734e-05,
+      "loss": 2.0838,
+      "step": 3014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9814458761679497e-05,
+      "loss": 2.1043,
+      "step": 3015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9814322482505726e-05,
+      "loss": 2.0686,
+      "step": 3016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9814186153771113e-05,
+      "loss": 2.1033,
+      "step": 3017
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9814049775476347e-05,
+      "loss": 2.0828,
+      "step": 3018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9813913347622114e-05,
+      "loss": 2.0872,
+      "step": 3019
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98137768702091e-05,
+      "loss": 2.0922,
+      "step": 3020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9813640343238007e-05,
+      "loss": 2.0764,
+      "step": 3021
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9813503766709508e-05,
+      "loss": 2.0699,
+      "step": 3022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98133671406243e-05,
+      "loss": 2.0786,
+      "step": 3023
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9813230464983077e-05,
+      "loss": 2.0665,
+      "step": 3024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981309373978652e-05,
+      "loss": 2.0878,
+      "step": 3025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981295696503533e-05,
+      "loss": 2.0916,
+      "step": 3026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981282014073019e-05,
+      "loss": 2.0716,
+      "step": 3027
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98126832668718e-05,
+      "loss": 2.0886,
+      "step": 3028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9812546343460838e-05,
+      "loss": 2.0859,
+      "step": 3029
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9812409370498e-05,
+      "loss": 2.0691,
+      "step": 3030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9812272347983984e-05,
+      "loss": 2.0819,
+      "step": 3031
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9812135275919473e-05,
+      "loss": 2.069,
+      "step": 3032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811998154305168e-05,
+      "loss": 2.0861,
+      "step": 3033
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811860983141756e-05,
+      "loss": 2.0617,
+      "step": 3034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811723762429933e-05,
+      "loss": 2.0676,
+      "step": 3035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811586492170388e-05,
+      "loss": 2.0838,
+      "step": 3036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811449172363817e-05,
+      "loss": 2.0752,
+      "step": 3037
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981131180301091e-05,
+      "loss": 2.0752,
+      "step": 3038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811174384112365e-05,
+      "loss": 2.0744,
+      "step": 3039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811036915668876e-05,
+      "loss": 2.0871,
+      "step": 3040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9810899397681136e-05,
+      "loss": 2.092,
+      "step": 3041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9810761830149836e-05,
+      "loss": 2.0914,
+      "step": 3042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9810624213075673e-05,
+      "loss": 2.092,
+      "step": 3043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9810486546459343e-05,
+      "loss": 2.0828,
+      "step": 3044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9810348830301543e-05,
+      "loss": 2.0626,
+      "step": 3045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9810211064602965e-05,
+      "loss": 2.0686,
+      "step": 3046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981007324936431e-05,
+      "loss": 2.0668,
+      "step": 3047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980993538458626e-05,
+      "loss": 2.0845,
+      "step": 3048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980979747026953e-05,
+      "loss": 2.0767,
+      "step": 3049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98096595064148e-05,
+      "loss": 2.0654,
+      "step": 3050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9809521493022782e-05,
+      "loss": 2.0567,
+      "step": 3051
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980938343009416e-05,
+      "loss": 2.0863,
+      "step": 3052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9809245317629633e-05,
+      "loss": 2.0869,
+      "step": 3053
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980910715562991e-05,
+      "loss": 2.0567,
+      "step": 3054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9808968944095674e-05,
+      "loss": 2.0539,
+      "step": 3055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9808830683027634e-05,
+      "loss": 2.0781,
+      "step": 3056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9808692372426482e-05,
+      "loss": 2.0598,
+      "step": 3057
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9808554012292918e-05,
+      "loss": 2.1047,
+      "step": 3058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9808415602627637e-05,
+      "loss": 2.0847,
+      "step": 3059
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9808277143431347e-05,
+      "loss": 2.0657,
+      "step": 3060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980813863470474e-05,
+      "loss": 2.0657,
+      "step": 3061
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9808000076448516e-05,
+      "loss": 2.0743,
+      "step": 3062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9807861468663378e-05,
+      "loss": 2.0716,
+      "step": 3063
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980772281135002e-05,
+      "loss": 2.0821,
+      "step": 3064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980758410450915e-05,
+      "loss": 2.09,
+      "step": 3065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9807445348141464e-05,
+      "loss": 2.0882,
+      "step": 3066
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9807306542247663e-05,
+      "loss": 2.0929,
+      "step": 3067
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9807167686828448e-05,
+      "loss": 2.0979,
+      "step": 3068
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9807028781884518e-05,
+      "loss": 2.0642,
+      "step": 3069
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980688982741658e-05,
+      "loss": 2.0975,
+      "step": 3070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980675082342533e-05,
+      "loss": 2.0745,
+      "step": 3071
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9806611769911476e-05,
+      "loss": 2.0799,
+      "step": 3072
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9806472666875715e-05,
+      "loss": 2.0657,
+      "step": 3073
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9806333514318755e-05,
+      "loss": 2.0756,
+      "step": 3074
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9806194312241288e-05,
+      "loss": 2.0727,
+      "step": 3075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980605506064403e-05,
+      "loss": 2.0695,
+      "step": 3076
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9805915759527675e-05,
+      "loss": 2.0708,
+      "step": 3077
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9805776408892928e-05,
+      "loss": 2.0573,
+      "step": 3078
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9805637008740496e-05,
+      "loss": 2.0858,
+      "step": 3079
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980549755907108e-05,
+      "loss": 2.073,
+      "step": 3080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980535805988539e-05,
+      "loss": 2.0841,
+      "step": 3081
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980521851118412e-05,
+      "loss": 2.0839,
+      "step": 3082
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9805078912967982e-05,
+      "loss": 2.075,
+      "step": 3083
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9804939265237682e-05,
+      "loss": 2.093,
+      "step": 3084
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980479956799392e-05,
+      "loss": 2.0867,
+      "step": 3085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9804659821237404e-05,
+      "loss": 2.0766,
+      "step": 3086
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980452002496884e-05,
+      "loss": 2.0748,
+      "step": 3087
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9804380179188934e-05,
+      "loss": 2.0799,
+      "step": 3088
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980424028389839e-05,
+      "loss": 2.0866,
+      "step": 3089
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980410033909792e-05,
+      "loss": 2.0654,
+      "step": 3090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9803960344788223e-05,
+      "loss": 2.0758,
+      "step": 3091
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9803820300970017e-05,
+      "loss": 2.0685,
+      "step": 3092
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9803680207643995e-05,
+      "loss": 2.0759,
+      "step": 3093
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9803540064810876e-05,
+      "loss": 2.0779,
+      "step": 3094
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980339987247136e-05,
+      "loss": 2.0805,
+      "step": 3095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980325963062616e-05,
+      "loss": 2.0768,
+      "step": 3096
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9803119339275983e-05,
+      "loss": 2.0649,
+      "step": 3097
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980297899842154e-05,
+      "loss": 2.0913,
+      "step": 3098
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980283860806353e-05,
+      "loss": 2.0689,
+      "step": 3099
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9802698168202675e-05,
+      "loss": 2.0845,
+      "step": 3100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9802557678839673e-05,
+      "loss": 2.0876,
+      "step": 3101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9802417139975242e-05,
+      "loss": 2.0816,
+      "step": 3102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9802276551610084e-05,
+      "loss": 2.0694,
+      "step": 3103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9802135913744917e-05,
+      "loss": 2.0953,
+      "step": 3104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9801995226380445e-05,
+      "loss": 2.0633,
+      "step": 3105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980185448951738e-05,
+      "loss": 2.0702,
+      "step": 3106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9801713703156435e-05,
+      "loss": 2.0769,
+      "step": 3107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980157286729832e-05,
+      "loss": 2.0541,
+      "step": 3108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980143198194374e-05,
+      "loss": 2.0938,
+      "step": 3109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9801291047093417e-05,
+      "loss": 2.0831,
+      "step": 3110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9801150062748054e-05,
+      "loss": 2.0777,
+      "step": 3111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980100902890837e-05,
+      "loss": 2.0807,
+      "step": 3112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980086794557507e-05,
+      "loss": 2.0829,
+      "step": 3113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9800726812748875e-05,
+      "loss": 2.08,
+      "step": 3114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9800585630430492e-05,
+      "loss": 2.0753,
+      "step": 3115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9800444398620632e-05,
+      "loss": 2.0833,
+      "step": 3116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9800303117320012e-05,
+      "loss": 2.088,
+      "step": 3117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9800161786529344e-05,
+      "loss": 2.0614,
+      "step": 3118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9800020406249346e-05,
+      "loss": 2.0738,
+      "step": 3119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9799878976480723e-05,
+      "loss": 2.0862,
+      "step": 3120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9799737497224197e-05,
+      "loss": 2.0956,
+      "step": 3121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979959596848048e-05,
+      "loss": 2.0583,
+      "step": 3122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9799454390250285e-05,
+      "loss": 2.0752,
+      "step": 3123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9799312762534328e-05,
+      "loss": 2.0747,
+      "step": 3124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9799171085333327e-05,
+      "loss": 2.0807,
+      "step": 3125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979902935864799e-05,
+      "loss": 2.0504,
+      "step": 3126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9798887582479044e-05,
+      "loss": 2.0734,
+      "step": 3127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9798745756827195e-05,
+      "loss": 2.051,
+      "step": 3128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9798603881693165e-05,
+      "loss": 2.0656,
+      "step": 3129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979846195707767e-05,
+      "loss": 2.0788,
+      "step": 3130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979831998298142e-05,
+      "loss": 2.0552,
+      "step": 3131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979817795940514e-05,
+      "loss": 2.0786,
+      "step": 3132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9798035886349545e-05,
+      "loss": 2.0793,
+      "step": 3133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979789376381535e-05,
+      "loss": 2.063,
+      "step": 3134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9797751591803273e-05,
+      "loss": 2.0671,
+      "step": 3135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9797609370314036e-05,
+      "loss": 2.091,
+      "step": 3136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9797467099348357e-05,
+      "loss": 2.0672,
+      "step": 3137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9797324778906944e-05,
+      "loss": 2.0507,
+      "step": 3138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979718240899053e-05,
+      "loss": 2.09,
+      "step": 3139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9797039989599828e-05,
+      "loss": 2.0752,
+      "step": 3140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9796897520735553e-05,
+      "loss": 2.0826,
+      "step": 3141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9796755002398428e-05,
+      "loss": 2.0709,
+      "step": 3142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9796612434589175e-05,
+      "loss": 2.0715,
+      "step": 3143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9796469817308512e-05,
+      "loss": 2.0797,
+      "step": 3144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9796327150557158e-05,
+      "loss": 2.0907,
+      "step": 3145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9796184434335837e-05,
+      "loss": 2.1033,
+      "step": 3146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9796041668645264e-05,
+      "loss": 2.0573,
+      "step": 3147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9795898853486165e-05,
+      "loss": 2.0821,
+      "step": 3148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9795755988859258e-05,
+      "loss": 2.0848,
+      "step": 3149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9795613074765267e-05,
+      "loss": 2.0809,
+      "step": 3150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979547011120491e-05,
+      "loss": 2.076,
+      "step": 3151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9795327098178915e-05,
+      "loss": 2.0834,
+      "step": 3152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9795184035688e-05,
+      "loss": 2.0852,
+      "step": 3153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9795040923732887e-05,
+      "loss": 2.0679,
+      "step": 3154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97948977623143e-05,
+      "loss": 2.079,
+      "step": 3155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979475455143296e-05,
+      "loss": 2.0866,
+      "step": 3156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9794611291089593e-05,
+      "loss": 2.0628,
+      "step": 3157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9794467981284923e-05,
+      "loss": 2.0728,
+      "step": 3158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9794324622019668e-05,
+      "loss": 2.0841,
+      "step": 3159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979418121329456e-05,
+      "loss": 2.0709,
+      "step": 3160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9794037755110318e-05,
+      "loss": 2.0673,
+      "step": 3161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9793894247467666e-05,
+      "loss": 2.0577,
+      "step": 3162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9793750690367336e-05,
+      "loss": 2.0638,
+      "step": 3163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979360708381004e-05,
+      "loss": 2.0671,
+      "step": 3164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9793463427796516e-05,
+      "loss": 2.06,
+      "step": 3165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979331972232748e-05,
+      "loss": 2.0622,
+      "step": 3166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979317596740366e-05,
+      "loss": 2.0935,
+      "step": 3167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9793032163025788e-05,
+      "loss": 2.065,
+      "step": 3168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792888309194587e-05,
+      "loss": 2.0715,
+      "step": 3169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792744405910777e-05,
+      "loss": 2.0867,
+      "step": 3170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792600453175093e-05,
+      "loss": 2.0559,
+      "step": 3171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792456450988256e-05,
+      "loss": 2.0713,
+      "step": 3172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792312399351e-05,
+      "loss": 2.0918,
+      "step": 3173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792168298264044e-05,
+      "loss": 2.0533,
+      "step": 3174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792024147728125e-05,
+      "loss": 2.0872,
+      "step": 3175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979187994774396e-05,
+      "loss": 2.0661,
+      "step": 3176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979173569831229e-05,
+      "loss": 2.0575,
+      "step": 3177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979159139943383e-05,
+      "loss": 2.0797,
+      "step": 3178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979144705110932e-05,
+      "loss": 2.0639,
+      "step": 3179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9791302653339484e-05,
+      "loss": 2.0886,
+      "step": 3180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979115820612505e-05,
+      "loss": 2.077,
+      "step": 3181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979101370946675e-05,
+      "loss": 2.0668,
+      "step": 3182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979086916336531e-05,
+      "loss": 2.0967,
+      "step": 3183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790724567821464e-05,
+      "loss": 2.0748,
+      "step": 3184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979057992283594e-05,
+      "loss": 2.0757,
+      "step": 3185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979043522840947e-05,
+      "loss": 2.0725,
+      "step": 3186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790290484542784e-05,
+      "loss": 2.0834,
+      "step": 3187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790145691236614e-05,
+      "loss": 2.0867,
+      "step": 3188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979000084849169e-05,
+      "loss": 2.0537,
+      "step": 3189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789855956308738e-05,
+      "loss": 2.0796,
+      "step": 3190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789711014688498e-05,
+      "loss": 2.0905,
+      "step": 3191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789566023631702e-05,
+      "loss": 2.0562,
+      "step": 3192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978942098313908e-05,
+      "loss": 2.0727,
+      "step": 3193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789275893211356e-05,
+      "loss": 2.0777,
+      "step": 3194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789130753849276e-05,
+      "loss": 2.0694,
+      "step": 3195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788985565053567e-05,
+      "loss": 2.0731,
+      "step": 3196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788840326824964e-05,
+      "loss": 2.0675,
+      "step": 3197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788695039164193e-05,
+      "loss": 2.0758,
+      "step": 3198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788549702072e-05,
+      "loss": 2.0621,
+      "step": 3199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788404315549107e-05,
+      "loss": 2.0679,
+      "step": 3200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788258879596255e-05,
+      "loss": 2.0681,
+      "step": 3201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788113394214182e-05,
+      "loss": 2.0703,
+      "step": 3202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9787967859403613e-05,
+      "loss": 2.0603,
+      "step": 3203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9787822275165288e-05,
+      "loss": 2.0696,
+      "step": 3204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9787676641499943e-05,
+      "loss": 2.048,
+      "step": 3205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9787530958408312e-05,
+      "loss": 2.0791,
+      "step": 3206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978738522589113e-05,
+      "loss": 2.0839,
+      "step": 3207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9787239443949134e-05,
+      "loss": 2.0686,
+      "step": 3208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978709361258306e-05,
+      "loss": 2.0808,
+      "step": 3209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786947731793644e-05,
+      "loss": 2.0771,
+      "step": 3210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786801801581625e-05,
+      "loss": 2.0673,
+      "step": 3211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786655821947736e-05,
+      "loss": 2.064,
+      "step": 3212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786509792892718e-05,
+      "loss": 2.0678,
+      "step": 3213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786363714417308e-05,
+      "loss": 2.0675,
+      "step": 3214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978621758652224e-05,
+      "loss": 2.0817,
+      "step": 3215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786071409208255e-05,
+      "loss": 2.0855,
+      "step": 3216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978592518247609e-05,
+      "loss": 2.0876,
+      "step": 3217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785778906326483e-05,
+      "loss": 2.0592,
+      "step": 3218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785632580760175e-05,
+      "loss": 2.0869,
+      "step": 3219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97854862057779e-05,
+      "loss": 2.0738,
+      "step": 3220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785339781380405e-05,
+      "loss": 2.0835,
+      "step": 3221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785193307568423e-05,
+      "loss": 2.0785,
+      "step": 3222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785046784342697e-05,
+      "loss": 2.0459,
+      "step": 3223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9784900211703967e-05,
+      "loss": 2.0499,
+      "step": 3224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978475358965297e-05,
+      "loss": 2.0743,
+      "step": 3225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9784606918190447e-05,
+      "loss": 2.0703,
+      "step": 3226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9784460197317142e-05,
+      "loss": 2.0945,
+      "step": 3227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9784313427033793e-05,
+      "loss": 2.0626,
+      "step": 3228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978416660734114e-05,
+      "loss": 2.0775,
+      "step": 3229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978401973823993e-05,
+      "loss": 2.0782,
+      "step": 3230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97838728197309e-05,
+      "loss": 2.073,
+      "step": 3231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783725851814794e-05,
+      "loss": 2.0738,
+      "step": 3232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783578834492348e-05,
+      "loss": 2.0696,
+      "step": 3233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783431767764313e-05,
+      "loss": 2.0717,
+      "step": 3234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978328465163143e-05,
+      "loss": 2.0686,
+      "step": 3235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783137486094436e-05,
+      "loss": 2.0695,
+      "step": 3236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978299027115408e-05,
+      "loss": 2.0836,
+      "step": 3237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9782843006811108e-05,
+      "loss": 2.0625,
+      "step": 3238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9782695693066256e-05,
+      "loss": 2.0721,
+      "step": 3239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978254832992027e-05,
+      "loss": 2.0798,
+      "step": 3240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9782400917373896e-05,
+      "loss": 2.07,
+      "step": 3241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978225345542788e-05,
+      "loss": 2.073,
+      "step": 3242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9782105944082962e-05,
+      "loss": 2.0654,
+      "step": 3243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978195838333989e-05,
+      "loss": 2.0589,
+      "step": 3244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978181077319941e-05,
+      "loss": 2.088,
+      "step": 3245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9781663113662266e-05,
+      "loss": 2.0382,
+      "step": 3246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9781515404729203e-05,
+      "loss": 2.0586,
+      "step": 3247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9781367646400966e-05,
+      "loss": 2.0772,
+      "step": 3248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9781219838678304e-05,
+      "loss": 2.0565,
+      "step": 3249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9781071981561962e-05,
+      "loss": 2.0555,
+      "step": 3250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780924075052687e-05,
+      "loss": 2.052,
+      "step": 3251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780776119151227e-05,
+      "loss": 2.0673,
+      "step": 3252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780628113858323e-05,
+      "loss": 2.0783,
+      "step": 3253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780480059174735e-05,
+      "loss": 2.0646,
+      "step": 3254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780331955101198e-05,
+      "loss": 2.0679,
+      "step": 3255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780183801638464e-05,
+      "loss": 2.0953,
+      "step": 3256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780035598787285e-05,
+      "loss": 2.0742,
+      "step": 3257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779887346548407e-05,
+      "loss": 2.079,
+      "step": 3258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977973904492257e-05,
+      "loss": 2.0751,
+      "step": 3259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977959069391054e-05,
+      "loss": 2.0895,
+      "step": 3260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779442293513055e-05,
+      "loss": 2.0667,
+      "step": 3261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779293843730864e-05,
+      "loss": 2.0561,
+      "step": 3262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779145344564724e-05,
+      "loss": 2.0556,
+      "step": 3263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778996796015374e-05,
+      "loss": 2.0644,
+      "step": 3264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778848198083573e-05,
+      "loss": 2.0808,
+      "step": 3265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977869955077007e-05,
+      "loss": 2.0592,
+      "step": 3266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977855085407561e-05,
+      "loss": 2.0629,
+      "step": 3267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778402108000952e-05,
+      "loss": 2.0548,
+      "step": 3268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977825331254684e-05,
+      "loss": 2.0822,
+      "step": 3269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778104467714033e-05,
+      "loss": 2.0527,
+      "step": 3270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777955573503274e-05,
+      "loss": 2.0655,
+      "step": 3271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977780662991532e-05,
+      "loss": 2.0731,
+      "step": 3272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777657636950925e-05,
+      "loss": 2.0773,
+      "step": 3273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777508594610837e-05,
+      "loss": 2.074,
+      "step": 3274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977735950289581e-05,
+      "loss": 2.0563,
+      "step": 3275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777210361806598e-05,
+      "loss": 2.0552,
+      "step": 3276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777061171343954e-05,
+      "loss": 2.0534,
+      "step": 3277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977691193150863e-05,
+      "loss": 2.0576,
+      "step": 3278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977676264230138e-05,
+      "loss": 2.0661,
+      "step": 3279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776613303722957e-05,
+      "loss": 2.0864,
+      "step": 3280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977646391577412e-05,
+      "loss": 2.0749,
+      "step": 3281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776314478455617e-05,
+      "loss": 2.0695,
+      "step": 3282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776164991768208e-05,
+      "loss": 2.0579,
+      "step": 3283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776015455712643e-05,
+      "loss": 2.0642,
+      "step": 3284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775865870289683e-05,
+      "loss": 2.0732,
+      "step": 3285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775716235500074e-05,
+      "loss": 2.0792,
+      "step": 3286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775566551344582e-05,
+      "loss": 2.0686,
+      "step": 3287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775416817823955e-05,
+      "loss": 2.0422,
+      "step": 3288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775267034938957e-05,
+      "loss": 2.0683,
+      "step": 3289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977511720269034e-05,
+      "loss": 2.0566,
+      "step": 3290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774967321078856e-05,
+      "loss": 2.0766,
+      "step": 3291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774817390105268e-05,
+      "loss": 2.0711,
+      "step": 3292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977466740977033e-05,
+      "loss": 2.0603,
+      "step": 3293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774517380074805e-05,
+      "loss": 2.0586,
+      "step": 3294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774367301019444e-05,
+      "loss": 2.0523,
+      "step": 3295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774217172605007e-05,
+      "loss": 2.078,
+      "step": 3296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774066994832256e-05,
+      "loss": 2.0714,
+      "step": 3297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773916767701943e-05,
+      "loss": 2.0742,
+      "step": 3298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773766491214827e-05,
+      "loss": 2.1062,
+      "step": 3299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773616165371674e-05,
+      "loss": 2.0586,
+      "step": 3300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773465790173235e-05,
+      "loss": 2.0571,
+      "step": 3301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773315365620274e-05,
+      "loss": 2.0736,
+      "step": 3302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773164891713553e-05,
+      "loss": 2.058,
+      "step": 3303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773014368453823e-05,
+      "loss": 2.0684,
+      "step": 3304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772863795841854e-05,
+      "loss": 2.0808,
+      "step": 3305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772713173878402e-05,
+      "loss": 2.0612,
+      "step": 3306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772562502564225e-05,
+      "loss": 2.0523,
+      "step": 3307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772411781900085e-05,
+      "loss": 2.0617,
+      "step": 3308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977226101188675e-05,
+      "loss": 2.0805,
+      "step": 3309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977211019252497e-05,
+      "loss": 2.0776,
+      "step": 3310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771959323815516e-05,
+      "loss": 2.0715,
+      "step": 3311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771808405759144e-05,
+      "loss": 2.0638,
+      "step": 3312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977165743835662e-05,
+      "loss": 2.0655,
+      "step": 3313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771506421608703e-05,
+      "loss": 2.0837,
+      "step": 3314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977135535551616e-05,
+      "loss": 2.0758,
+      "step": 3315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771204240079748e-05,
+      "loss": 2.0878,
+      "step": 3316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771053075300236e-05,
+      "loss": 2.0919,
+      "step": 3317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9770901861178387e-05,
+      "loss": 2.0734,
+      "step": 3318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977075059771496e-05,
+      "loss": 2.0724,
+      "step": 3319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977059928491072e-05,
+      "loss": 2.0593,
+      "step": 3320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9770447922766432e-05,
+      "loss": 2.074,
+      "step": 3321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977029651128286e-05,
+      "loss": 2.0518,
+      "step": 3322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9770145050460774e-05,
+      "loss": 2.06,
+      "step": 3323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976999354030093e-05,
+      "loss": 2.0733,
+      "step": 3324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97698419808041e-05,
+      "loss": 2.0822,
+      "step": 3325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769690371971047e-05,
+      "loss": 2.0443,
+      "step": 3326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769538713802534e-05,
+      "loss": 2.0606,
+      "step": 3327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976938700629933e-05,
+      "loss": 2.0868,
+      "step": 3328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769235249462197e-05,
+      "loss": 2.039,
+      "step": 3329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769083443291908e-05,
+      "loss": 2.0707,
+      "step": 3330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768931587789225e-05,
+      "loss": 2.0552,
+      "step": 3331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768779682954915e-05,
+      "loss": 2.0618,
+      "step": 3332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976862772878975e-05,
+      "loss": 2.0837,
+      "step": 3333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976847572529449e-05,
+      "loss": 2.0466,
+      "step": 3334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768323672469907e-05,
+      "loss": 2.0653,
+      "step": 3335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976817157031677e-05,
+      "loss": 2.067,
+      "step": 3336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768019418835837e-05,
+      "loss": 2.057,
+      "step": 3337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976786721802789e-05,
+      "loss": 2.0701,
+      "step": 3338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9767714967893692e-05,
+      "loss": 2.0374,
+      "step": 3339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9767562668434014e-05,
+      "loss": 2.0643,
+      "step": 3340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9767410319649618e-05,
+      "loss": 2.0675,
+      "step": 3341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976725792154128e-05,
+      "loss": 2.079,
+      "step": 3342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9767105474109768e-05,
+      "loss": 2.0572,
+      "step": 3343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9766952977355852e-05,
+      "loss": 2.0525,
+      "step": 3344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97668004312803e-05,
+      "loss": 2.069,
+      "step": 3345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9766647835883886e-05,
+      "loss": 2.0734,
+      "step": 3346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976649519116738e-05,
+      "loss": 2.0548,
+      "step": 3347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9766342497131547e-05,
+      "loss": 2.0638,
+      "step": 3348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976618975377716e-05,
+      "loss": 2.0625,
+      "step": 3349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9766036961105e-05,
+      "loss": 2.0606,
+      "step": 3350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9765884119115827e-05,
+      "loss": 2.0526,
+      "step": 3351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976573122781042e-05,
+      "loss": 2.0681,
+      "step": 3352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976557828718955e-05,
+      "loss": 2.0534,
+      "step": 3353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9765425297253982e-05,
+      "loss": 2.06,
+      "step": 3354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9765272258004496e-05,
+      "loss": 2.053,
+      "step": 3355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9765119169441863e-05,
+      "loss": 2.0731,
+      "step": 3356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976496603156686e-05,
+      "loss": 2.0557,
+      "step": 3357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9764812844380254e-05,
+      "loss": 2.0674,
+      "step": 3358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976465960788282e-05,
+      "loss": 2.0513,
+      "step": 3359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9764506322075334e-05,
+      "loss": 2.0805,
+      "step": 3360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9764352986958567e-05,
+      "loss": 2.0634,
+      "step": 3361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9764199602533296e-05,
+      "loss": 2.0596,
+      "step": 3362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9764046168800295e-05,
+      "loss": 2.068,
+      "step": 3363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9763892685760338e-05,
+      "loss": 2.0711,
+      "step": 3364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9763739153414203e-05,
+      "loss": 2.0637,
+      "step": 3365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976358557176266e-05,
+      "loss": 2.0599,
+      "step": 3366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976343194080649e-05,
+      "loss": 2.0643,
+      "step": 3367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9763278260546465e-05,
+      "loss": 2.0769,
+      "step": 3368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9763124530983358e-05,
+      "loss": 2.0338,
+      "step": 3369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9762970752117957e-05,
+      "loss": 2.0801,
+      "step": 3370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976281692395103e-05,
+      "loss": 2.0722,
+      "step": 3371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9762663046483354e-05,
+      "loss": 2.0558,
+      "step": 3372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9762509119715704e-05,
+      "loss": 2.0561,
+      "step": 3373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9762355143648862e-05,
+      "loss": 2.0545,
+      "step": 3374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9762201118283606e-05,
+      "loss": 2.0652,
+      "step": 3375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9762047043620713e-05,
+      "loss": 2.032,
+      "step": 3376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9761892919660957e-05,
+      "loss": 2.0846,
+      "step": 3377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976173874640512e-05,
+      "loss": 2.0765,
+      "step": 3378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9761584523853977e-05,
+      "loss": 2.0744,
+      "step": 3379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9761430252008313e-05,
+      "loss": 2.0629,
+      "step": 3380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9761275930868906e-05,
+      "loss": 2.0523,
+      "step": 3381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9761121560436528e-05,
+      "loss": 2.0522,
+      "step": 3382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9760967140711965e-05,
+      "loss": 2.0649,
+      "step": 3383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9760812671695994e-05,
+      "loss": 2.0479,
+      "step": 3384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9760658153389397e-05,
+      "loss": 2.0605,
+      "step": 3385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9760503585792953e-05,
+      "loss": 2.0798,
+      "step": 3386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9760348968907445e-05,
+      "loss": 2.0538,
+      "step": 3387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976019430273365e-05,
+      "loss": 2.0705,
+      "step": 3388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9760039587272354e-05,
+      "loss": 2.0679,
+      "step": 3389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9759884822524333e-05,
+      "loss": 2.0678,
+      "step": 3390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975973000849037e-05,
+      "loss": 2.0674,
+      "step": 3391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9759575145171247e-05,
+      "loss": 2.0497,
+      "step": 3392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9759420232567746e-05,
+      "loss": 2.0615,
+      "step": 3393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9759265270680654e-05,
+      "loss": 2.0666,
+      "step": 3394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975911025951075e-05,
+      "loss": 2.0904,
+      "step": 3395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9758955199058813e-05,
+      "loss": 2.0602,
+      "step": 3396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9758800089325626e-05,
+      "loss": 2.0738,
+      "step": 3397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975864493031198e-05,
+      "loss": 2.0829,
+      "step": 3398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9758489722018654e-05,
+      "loss": 2.069,
+      "step": 3399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975833446444643e-05,
+      "loss": 2.063,
+      "step": 3400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9758179157596093e-05,
+      "loss": 2.0733,
+      "step": 3401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9758023801468432e-05,
+      "loss": 2.05,
+      "step": 3402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9757868396064226e-05,
+      "loss": 2.0618,
+      "step": 3403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975771294138426e-05,
+      "loss": 2.0639,
+      "step": 3404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9757557437429323e-05,
+      "loss": 2.0753,
+      "step": 3405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9757401884200195e-05,
+      "loss": 2.0582,
+      "step": 3406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9757246281697663e-05,
+      "loss": 2.0621,
+      "step": 3407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9757090629922516e-05,
+      "loss": 2.0571,
+      "step": 3408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9756934928875537e-05,
+      "loss": 2.0673,
+      "step": 3409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9756779178557516e-05,
+      "loss": 2.0404,
+      "step": 3410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9756623378969235e-05,
+      "loss": 2.0757,
+      "step": 3411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9756467530111483e-05,
+      "loss": 2.0735,
+      "step": 3412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9756311631985047e-05,
+      "loss": 2.0633,
+      "step": 3413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9756155684590713e-05,
+      "loss": 2.0682,
+      "step": 3414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975599968792927e-05,
+      "loss": 2.0705,
+      "step": 3415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9755843642001503e-05,
+      "loss": 2.0725,
+      "step": 3416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9755687546808203e-05,
+      "loss": 2.0591,
+      "step": 3417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975553140235016e-05,
+      "loss": 2.0461,
+      "step": 3418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9755375208628154e-05,
+      "loss": 2.0763,
+      "step": 3419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9755218965642987e-05,
+      "loss": 2.0511,
+      "step": 3420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9755062673395434e-05,
+      "loss": 2.0799,
+      "step": 3421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9754906331886295e-05,
+      "loss": 2.0751,
+      "step": 3422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9754749941116352e-05,
+      "loss": 2.0618,
+      "step": 3423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9754593501086402e-05,
+      "loss": 2.0858,
+      "step": 3424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9754437011797227e-05,
+      "loss": 2.0681,
+      "step": 3425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9754280473249624e-05,
+      "loss": 2.0641,
+      "step": 3426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9754123885444384e-05,
+      "loss": 2.0817,
+      "step": 3427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9753967248382292e-05,
+      "loss": 2.0673,
+      "step": 3428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975381056206414e-05,
+      "loss": 2.0606,
+      "step": 3429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975365382649072e-05,
+      "loss": 2.0647,
+      "step": 3430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9753497041662825e-05,
+      "loss": 2.0662,
+      "step": 3431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975334020758125e-05,
+      "loss": 2.0514,
+      "step": 3432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975318332424678e-05,
+      "loss": 2.0712,
+      "step": 3433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9753026391660208e-05,
+      "loss": 2.0552,
+      "step": 3434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9752869409822333e-05,
+      "loss": 2.0576,
+      "step": 3435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975271237873394e-05,
+      "loss": 2.0422,
+      "step": 3436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9752555298395828e-05,
+      "loss": 2.0611,
+      "step": 3437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9752398168808786e-05,
+      "loss": 2.0677,
+      "step": 3438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975224098997361e-05,
+      "loss": 2.0651,
+      "step": 3439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9752083761891093e-05,
+      "loss": 2.0654,
+      "step": 3440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975192648456203e-05,
+      "loss": 2.0635,
+      "step": 3441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9751769157987215e-05,
+      "loss": 2.0966,
+      "step": 3442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9751611782167438e-05,
+      "loss": 2.046,
+      "step": 3443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97514543571035e-05,
+      "loss": 2.0635,
+      "step": 3444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975129688279619e-05,
+      "loss": 2.0505,
+      "step": 3445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9751139359246312e-05,
+      "loss": 2.0718,
+      "step": 3446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9750981786454654e-05,
+      "loss": 2.0599,
+      "step": 3447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975082416442201e-05,
+      "loss": 2.0818,
+      "step": 3448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9750666493149183e-05,
+      "loss": 2.0729,
+      "step": 3449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9750508772636965e-05,
+      "loss": 2.0686,
+      "step": 3450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9750351002886155e-05,
+      "loss": 2.0681,
+      "step": 3451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9750193183897545e-05,
+      "loss": 2.0626,
+      "step": 3452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975003531567194e-05,
+      "loss": 2.0705,
+      "step": 3453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9749877398210127e-05,
+      "loss": 2.0576,
+      "step": 3454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974971943151291e-05,
+      "loss": 2.0632,
+      "step": 3455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9749561415581088e-05,
+      "loss": 2.0555,
+      "step": 3456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9749403350415456e-05,
+      "loss": 2.0559,
+      "step": 3457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974924523601681e-05,
+      "loss": 2.0575,
+      "step": 3458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974908707238595e-05,
+      "loss": 2.0622,
+      "step": 3459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974892885952368e-05,
+      "loss": 2.0651,
+      "step": 3460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9748770597430794e-05,
+      "loss": 2.0751,
+      "step": 3461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9748612286108087e-05,
+      "loss": 2.0702,
+      "step": 3462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974845392555637e-05,
+      "loss": 2.0588,
+      "step": 3463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9748295515776436e-05,
+      "loss": 2.0624,
+      "step": 3464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974813705676908e-05,
+      "loss": 2.0688,
+      "step": 3465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974797854853511e-05,
+      "loss": 2.0514,
+      "step": 3466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974781999107532e-05,
+      "loss": 2.0584,
+      "step": 3467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9747661384390518e-05,
+      "loss": 2.0614,
+      "step": 3468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9747502728481503e-05,
+      "loss": 2.0508,
+      "step": 3469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974734402334907e-05,
+      "loss": 2.0786,
+      "step": 3470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9747185268994027e-05,
+      "loss": 2.0575,
+      "step": 3471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9747026465417174e-05,
+      "loss": 2.07,
+      "step": 3472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9746867612619312e-05,
+      "loss": 2.0558,
+      "step": 3473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9746708710601242e-05,
+      "loss": 2.0809,
+      "step": 3474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974654975936377e-05,
+      "loss": 2.0525,
+      "step": 3475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9746390758907698e-05,
+      "loss": 2.0457,
+      "step": 3476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9746231709233825e-05,
+      "loss": 2.0498,
+      "step": 3477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974607261034296e-05,
+      "loss": 2.0675,
+      "step": 3478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97459134622359e-05,
+      "loss": 2.0502,
+      "step": 3479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974575426491345e-05,
+      "loss": 2.0584,
+      "step": 3480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9745595018376423e-05,
+      "loss": 2.0643,
+      "step": 3481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974543572262561e-05,
+      "loss": 2.0595,
+      "step": 3482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9745276377661826e-05,
+      "loss": 2.064,
+      "step": 3483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9745116983485868e-05,
+      "loss": 2.0709,
+      "step": 3484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9744957540098545e-05,
+      "loss": 2.0481,
+      "step": 3485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9744798047500665e-05,
+      "loss": 2.0563,
+      "step": 3486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9744638505693023e-05,
+      "loss": 2.0743,
+      "step": 3487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9744478914676436e-05,
+      "loss": 2.0692,
+      "step": 3488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97443192744517e-05,
+      "loss": 2.0508,
+      "step": 3489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9744159585019633e-05,
+      "loss": 2.0629,
+      "step": 3490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9743999846381033e-05,
+      "loss": 2.0586,
+      "step": 3491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9743840058536707e-05,
+      "loss": 2.0669,
+      "step": 3492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974368022148746e-05,
+      "loss": 2.047,
+      "step": 3493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974352033523411e-05,
+      "loss": 2.0683,
+      "step": 3494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9743360399777452e-05,
+      "loss": 2.06,
+      "step": 3495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9743200415118302e-05,
+      "loss": 2.0583,
+      "step": 3496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9743040381257465e-05,
+      "loss": 2.0507,
+      "step": 3497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9742880298195742e-05,
+      "loss": 2.0607,
+      "step": 3498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9742720165933952e-05,
+      "loss": 2.0527,
+      "step": 3499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9742559984472904e-05,
+      "loss": 2.0811,
+      "step": 3500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9742399753813393e-05,
+      "loss": 2.0589,
+      "step": 3501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9742239473956248e-05,
+      "loss": 2.0526,
+      "step": 3502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974207914490226e-05,
+      "loss": 2.055,
+      "step": 3503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974191876665225e-05,
+      "loss": 2.0646,
+      "step": 3504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9741758339207025e-05,
+      "loss": 2.0491,
+      "step": 3505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9741597862567392e-05,
+      "loss": 2.0585,
+      "step": 3506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9741437336734165e-05,
+      "loss": 2.0658,
+      "step": 3507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9741276761708155e-05,
+      "loss": 2.074,
+      "step": 3508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974111613749017e-05,
+      "loss": 2.0404,
+      "step": 3509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9740955464081027e-05,
+      "loss": 2.0646,
+      "step": 3510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9740794741481532e-05,
+      "loss": 2.0587,
+      "step": 3511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97406339696925e-05,
+      "loss": 2.0656,
+      "step": 3512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974047314871473e-05,
+      "loss": 2.0573,
+      "step": 3513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9740312278549056e-05,
+      "loss": 2.0563,
+      "step": 3514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9740151359196275e-05,
+      "loss": 2.0717,
+      "step": 3515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9739990390657203e-05,
+      "loss": 2.0601,
+      "step": 3516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973982937293265e-05,
+      "loss": 2.0659,
+      "step": 3517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9739668306023442e-05,
+      "loss": 2.0581,
+      "step": 3518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9739507189930376e-05,
+      "loss": 2.0745,
+      "step": 3519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9739346024654275e-05,
+      "loss": 2.0725,
+      "step": 3520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973918481019595e-05,
+      "loss": 2.0459,
+      "step": 3521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9739023546556217e-05,
+      "loss": 2.0485,
+      "step": 3522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973886223373589e-05,
+      "loss": 2.0775,
+      "step": 3523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973870087173578e-05,
+      "loss": 2.0517,
+      "step": 3524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9738539460556703e-05,
+      "loss": 2.0744,
+      "step": 3525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973837800019948e-05,
+      "loss": 2.0569,
+      "step": 3526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973821649066492e-05,
+      "loss": 2.0438,
+      "step": 3527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973805493195384e-05,
+      "loss": 2.0662,
+      "step": 3528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9737893324067054e-05,
+      "loss": 2.0857,
+      "step": 3529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9737731667005385e-05,
+      "loss": 2.0604,
+      "step": 3530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973756996076964e-05,
+      "loss": 2.0654,
+      "step": 3531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9737408205360645e-05,
+      "loss": 2.0598,
+      "step": 3532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973724640077921e-05,
+      "loss": 2.0625,
+      "step": 3533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9737084547026157e-05,
+      "loss": 2.038,
+      "step": 3534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9736922644102296e-05,
+      "loss": 2.0487,
+      "step": 3535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9736760692008453e-05,
+      "loss": 2.0421,
+      "step": 3536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9736598690745443e-05,
+      "loss": 2.0368,
+      "step": 3537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9736436640314077e-05,
+      "loss": 2.0455,
+      "step": 3538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973627454071519e-05,
+      "loss": 2.0657,
+      "step": 3539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973611239194958e-05,
+      "loss": 2.0745,
+      "step": 3540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973595019401808e-05,
+      "loss": 2.0613,
+      "step": 3541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9735787946921502e-05,
+      "loss": 2.0581,
+      "step": 3542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9735625650660674e-05,
+      "loss": 2.0567,
+      "step": 3543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9735463305236406e-05,
+      "loss": 2.0503,
+      "step": 3544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9735300910649523e-05,
+      "loss": 2.0395,
+      "step": 3545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9735138466900845e-05,
+      "loss": 2.056,
+      "step": 3546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9734975973991187e-05,
+      "loss": 2.0601,
+      "step": 3547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9734813431921377e-05,
+      "loss": 2.0532,
+      "step": 3548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9734650840692234e-05,
+      "loss": 2.06,
+      "step": 3549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9734488200304572e-05,
+      "loss": 2.0733,
+      "step": 3550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9734325510759224e-05,
+      "loss": 2.0542,
+      "step": 3551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9734162772057e-05,
+      "loss": 2.0515,
+      "step": 3552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973399998419873e-05,
+      "loss": 2.0625,
+      "step": 3553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9733837147185234e-05,
+      "loss": 2.0519,
+      "step": 3554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9733674261017332e-05,
+      "loss": 2.0679,
+      "step": 3555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9733511325695847e-05,
+      "loss": 2.0753,
+      "step": 3556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9733348341221606e-05,
+      "loss": 2.0502,
+      "step": 3557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973318530759543e-05,
+      "loss": 2.0714,
+      "step": 3558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9733022224818136e-05,
+      "loss": 2.059,
+      "step": 3559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9732859092890556e-05,
+      "loss": 2.0618,
+      "step": 3560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973269591181351e-05,
+      "loss": 2.0674,
+      "step": 3561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973253268158782e-05,
+      "loss": 2.039,
+      "step": 3562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973236940221432e-05,
+      "loss": 2.0722,
+      "step": 3563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973220607369382e-05,
+      "loss": 2.075,
+      "step": 3564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9732042696027158e-05,
+      "loss": 2.0469,
+      "step": 3565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973187926921515e-05,
+      "loss": 2.0711,
+      "step": 3566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9731715793258624e-05,
+      "loss": 2.0542,
+      "step": 3567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9731552268158408e-05,
+      "loss": 2.0617,
+      "step": 3568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9731388693915326e-05,
+      "loss": 2.0593,
+      "step": 3569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97312250705302e-05,
+      "loss": 2.0558,
+      "step": 3570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9731061398003867e-05,
+      "loss": 2.0417,
+      "step": 3571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9730897676337143e-05,
+      "loss": 2.0896,
+      "step": 3572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9730733905530856e-05,
+      "loss": 2.0558,
+      "step": 3573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9730570085585837e-05,
+      "loss": 2.0553,
+      "step": 3574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9730406216502915e-05,
+      "loss": 2.0557,
+      "step": 3575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973024229828291e-05,
+      "loss": 2.0451,
+      "step": 3576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9730078330926658e-05,
+      "loss": 2.0771,
+      "step": 3577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972991431443498e-05,
+      "loss": 2.0591,
+      "step": 3578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9729750248808707e-05,
+      "loss": 2.0678,
+      "step": 3579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972958613404867e-05,
+      "loss": 2.0526,
+      "step": 3580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9729421970155693e-05,
+      "loss": 2.0582,
+      "step": 3581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9729257757130612e-05,
+      "loss": 2.0453,
+      "step": 3582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9729093494974245e-05,
+      "loss": 2.0663,
+      "step": 3583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9728929183687432e-05,
+      "loss": 2.0625,
+      "step": 3584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9728764823271e-05,
+      "loss": 2.0366,
+      "step": 3585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972860041372578e-05,
+      "loss": 2.0422,
+      "step": 3586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9728435955052597e-05,
+      "loss": 2.0526,
+      "step": 3587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9728271447252284e-05,
+      "loss": 2.0553,
+      "step": 3588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972810689032567e-05,
+      "loss": 2.0688,
+      "step": 3589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9727942284273594e-05,
+      "loss": 2.0436,
+      "step": 3590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972777762909688e-05,
+      "loss": 2.0747,
+      "step": 3591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972761292479636e-05,
+      "loss": 2.0603,
+      "step": 3592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972744817137287e-05,
+      "loss": 2.051,
+      "step": 3593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9727283368827236e-05,
+      "loss": 2.0512,
+      "step": 3594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972711851716029e-05,
+      "loss": 2.0528,
+      "step": 3595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9726953616372876e-05,
+      "loss": 2.0679,
+      "step": 3596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9726788666465814e-05,
+      "loss": 2.0366,
+      "step": 3597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972662366743994e-05,
+      "loss": 2.0506,
+      "step": 3598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972645861929609e-05,
+      "loss": 2.0548,
+      "step": 3599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9726293522035095e-05,
+      "loss": 2.0532,
+      "step": 3600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9726128375657795e-05,
+      "loss": 2.067,
+      "step": 3601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9725963180165012e-05,
+      "loss": 2.0613,
+      "step": 3602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9725797935557592e-05,
+      "loss": 2.058,
+      "step": 3603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972563264183636e-05,
+      "loss": 2.0631,
+      "step": 3604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9725467299002155e-05,
+      "loss": 2.0583,
+      "step": 3605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9725301907055814e-05,
+      "loss": 2.0421,
+      "step": 3606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972513646599817e-05,
+      "loss": 2.075,
+      "step": 3607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972497097583006e-05,
+      "loss": 2.0532,
+      "step": 3608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9724805436552316e-05,
+      "loss": 2.073,
+      "step": 3609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972463984816578e-05,
+      "loss": 2.0833,
+      "step": 3610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972447421067128e-05,
+      "loss": 2.0791,
+      "step": 3611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9724308524069658e-05,
+      "loss": 2.0621,
+      "step": 3612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9724142788361755e-05,
+      "loss": 2.0513,
+      "step": 3613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9723977003548397e-05,
+      "loss": 2.0692,
+      "step": 3614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9723811169630426e-05,
+      "loss": 2.0627,
+      "step": 3615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9723645286608683e-05,
+      "loss": 2.036,
+      "step": 3616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9723479354484e-05,
+      "loss": 2.0541,
+      "step": 3617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9723313373257222e-05,
+      "loss": 2.0619,
+      "step": 3618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9723147342929183e-05,
+      "loss": 2.0662,
+      "step": 3619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9722981263500717e-05,
+      "loss": 2.0654,
+      "step": 3620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972281513497267e-05,
+      "loss": 2.0731,
+      "step": 3621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9722648957345874e-05,
+      "loss": 2.0699,
+      "step": 3622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972248273062118e-05,
+      "loss": 2.0726,
+      "step": 3623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9722316454799416e-05,
+      "loss": 2.0733,
+      "step": 3624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9722150129881422e-05,
+      "loss": 2.0629,
+      "step": 3625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721983755868047e-05,
+      "loss": 2.0464,
+      "step": 3626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972181733276012e-05,
+      "loss": 2.0544,
+      "step": 3627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972165086055849e-05,
+      "loss": 2.0581,
+      "step": 3628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721484339263994e-05,
+      "loss": 2.062,
+      "step": 3629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972131776887747e-05,
+      "loss": 2.0421,
+      "step": 3630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972115114939977e-05,
+      "loss": 2.0558,
+      "step": 3631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720984480831723e-05,
+      "loss": 2.0416,
+      "step": 3632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720817763174177e-05,
+      "loss": 2.0643,
+      "step": 3633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972065099642797e-05,
+      "loss": 2.065,
+      "step": 3634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720484180593948e-05,
+      "loss": 2.0519,
+      "step": 3635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720317315672952e-05,
+      "loss": 2.047,
+      "step": 3636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720150401665827e-05,
+      "loss": 2.047,
+      "step": 3637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971998343857341e-05,
+      "loss": 2.0467,
+      "step": 3638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971981642639655e-05,
+      "loss": 2.0635,
+      "step": 3639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719649365136087e-05,
+      "loss": 2.0382,
+      "step": 3640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719482254792864e-05,
+      "loss": 2.0679,
+      "step": 3641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971931509536773e-05,
+      "loss": 2.0545,
+      "step": 3642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719147886861525e-05,
+      "loss": 2.0622,
+      "step": 3643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718980629275092e-05,
+      "loss": 2.0533,
+      "step": 3644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971881332260928e-05,
+      "loss": 2.0699,
+      "step": 3645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718645966864932e-05,
+      "loss": 2.0648,
+      "step": 3646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718478562042892e-05,
+      "loss": 2.0452,
+      "step": 3647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718311108144004e-05,
+      "loss": 2.0825,
+      "step": 3648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718143605169118e-05,
+      "loss": 2.063,
+      "step": 3649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9717976053119077e-05,
+      "loss": 2.0618,
+      "step": 3650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9717808451994726e-05,
+      "loss": 2.0527,
+      "step": 3651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9717640801796917e-05,
+      "loss": 2.061,
+      "step": 3652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971747310252649e-05,
+      "loss": 2.0567,
+      "step": 3653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971730535418429e-05,
+      "loss": 2.0489,
+      "step": 3654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9717137556771178e-05,
+      "loss": 2.0427,
+      "step": 3655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971696971028799e-05,
+      "loss": 2.0556,
+      "step": 3656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9716801814735568e-05,
+      "loss": 2.0596,
+      "step": 3657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9716633870114775e-05,
+      "loss": 2.0508,
+      "step": 3658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971646587642645e-05,
+      "loss": 2.0827,
+      "step": 3659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971629783367144e-05,
+      "loss": 2.073,
+      "step": 3660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9716129741850597e-05,
+      "loss": 2.0622,
+      "step": 3661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715961600964772e-05,
+      "loss": 2.0508,
+      "step": 3662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971579341101481e-05,
+      "loss": 2.0479,
+      "step": 3663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715625172001563e-05,
+      "loss": 2.0401,
+      "step": 3664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715456883925878e-05,
+      "loss": 2.058,
+      "step": 3665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715288546788605e-05,
+      "loss": 2.0608,
+      "step": 3666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715120160590594e-05,
+      "loss": 2.0616,
+      "step": 3667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97149517253327e-05,
+      "loss": 2.0623,
+      "step": 3668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714783241015768e-05,
+      "loss": 2.0574,
+      "step": 3669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714614707640653e-05,
+      "loss": 2.0455,
+      "step": 3670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97144461252082e-05,
+      "loss": 2.0326,
+      "step": 3671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714277493719268e-05,
+      "loss": 2.0638,
+      "step": 3672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714108813174703e-05,
+      "loss": 2.0471,
+      "step": 3673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713940083575362e-05,
+      "loss": 2.0447,
+      "step": 3674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971377130492209e-05,
+      "loss": 2.0724,
+      "step": 3675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713602477215746e-05,
+      "loss": 2.0381,
+      "step": 3676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713433600457177e-05,
+      "loss": 2.0581,
+      "step": 3677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971326467464724e-05,
+      "loss": 2.0401,
+      "step": 3678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713095699786785e-05,
+      "loss": 2.0493,
+      "step": 3679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9712926675876666e-05,
+      "loss": 2.0551,
+      "step": 3680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971275760291774e-05,
+      "loss": 2.0606,
+      "step": 3681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9712588480910856e-05,
+      "loss": 2.0446,
+      "step": 3682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971241930985687e-05,
+      "loss": 2.051,
+      "step": 3683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971225008975664e-05,
+      "loss": 2.0534,
+      "step": 3684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9712080820611014e-05,
+      "loss": 2.0701,
+      "step": 3685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971191150242085e-05,
+      "loss": 2.066,
+      "step": 3686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9711742135187002e-05,
+      "loss": 2.071,
+      "step": 3687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9711572718910326e-05,
+      "loss": 2.0553,
+      "step": 3688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971140325359168e-05,
+      "loss": 2.0606,
+      "step": 3689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9711233739231913e-05,
+      "loss": 2.0459,
+      "step": 3690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971106417583189e-05,
+      "loss": 2.0433,
+      "step": 3691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9710894563392457e-05,
+      "loss": 2.063,
+      "step": 3692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971072490191448e-05,
+      "loss": 2.0443,
+      "step": 3693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971055519139881e-05,
+      "loss": 2.0426,
+      "step": 3694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971038543184631e-05,
+      "loss": 2.0612,
+      "step": 3695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971021562325783e-05,
+      "loss": 2.0386,
+      "step": 3696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9710045765634232e-05,
+      "loss": 2.0393,
+      "step": 3697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9709875858976373e-05,
+      "loss": 2.0583,
+      "step": 3698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970970590328511e-05,
+      "loss": 2.0555,
+      "step": 3699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97095358985613e-05,
+      "loss": 2.0516,
+      "step": 3700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9709365844805802e-05,
+      "loss": 2.0458,
+      "step": 3701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9709195742019477e-05,
+      "loss": 2.042,
+      "step": 3702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970902559020319e-05,
+      "loss": 2.063,
+      "step": 3703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708855389357783e-05,
+      "loss": 2.0525,
+      "step": 3704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708685139484133e-05,
+      "loss": 2.0417,
+      "step": 3705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708514840583087e-05,
+      "loss": 2.0442,
+      "step": 3706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970834449265551e-05,
+      "loss": 2.0614,
+      "step": 3707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708174095702266e-05,
+      "loss": 2.0561,
+      "step": 3708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708003649724215e-05,
+      "loss": 2.054,
+      "step": 3709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9707833154722208e-05,
+      "loss": 2.0542,
+      "step": 3710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9707662610697115e-05,
+      "loss": 2.0725,
+      "step": 3711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9707492017649797e-05,
+      "loss": 2.0527,
+      "step": 3712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970732137558111e-05,
+      "loss": 2.0556,
+      "step": 3713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9707150684491926e-05,
+      "loss": 2.0419,
+      "step": 3714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9706979944383097e-05,
+      "loss": 2.0468,
+      "step": 3715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9706809155255485e-05,
+      "loss": 2.0729,
+      "step": 3716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9706638317109958e-05,
+      "loss": 2.0618,
+      "step": 3717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9706467429947376e-05,
+      "loss": 2.0506,
+      "step": 3718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97062964937686e-05,
+      "loss": 2.0656,
+      "step": 3719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97061255085745e-05,
+      "loss": 2.0663,
+      "step": 3720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9705954474365937e-05,
+      "loss": 2.0669,
+      "step": 3721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970578339114377e-05,
+      "loss": 2.0344,
+      "step": 3722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9705612258908864e-05,
+      "loss": 2.054,
+      "step": 3723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9705441077662088e-05,
+      "loss": 2.051,
+      "step": 3724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97052698474043e-05,
+      "loss": 2.0907,
+      "step": 3725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970509856813637e-05,
+      "loss": 2.0338,
+      "step": 3726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970492723985916e-05,
+      "loss": 2.069,
+      "step": 3727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970475586257354e-05,
+      "loss": 2.0542,
+      "step": 3728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9704584436280366e-05,
+      "loss": 2.0557,
+      "step": 3729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9704412960980512e-05,
+      "loss": 2.0608,
+      "step": 3730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970424143667484e-05,
+      "loss": 2.071,
+      "step": 3731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9704069863364222e-05,
+      "loss": 2.0569,
+      "step": 3732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703898241049514e-05,
+      "loss": 2.0393,
+      "step": 3733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703726569731593e-05,
+      "loss": 2.0442,
+      "step": 3734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970355484941132e-05,
+      "loss": 2.0645,
+      "step": 3735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703383080089566e-05,
+      "loss": 2.0511,
+      "step": 3736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703211261767192e-05,
+      "loss": 2.0445,
+      "step": 3737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970303939444507e-05,
+      "loss": 2.0334,
+      "step": 3738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702867478124068e-05,
+      "loss": 2.057,
+      "step": 3739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702695512805056e-05,
+      "loss": 2.0655,
+      "step": 3740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702523498488897e-05,
+      "loss": 2.0448,
+      "step": 3741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702351435176467e-05,
+      "loss": 2.0574,
+      "step": 3742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702179322868626e-05,
+      "loss": 2.0567,
+      "step": 3743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702007161566247e-05,
+      "loss": 2.0456,
+      "step": 3744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9701834951270205e-05,
+      "loss": 2.044,
+      "step": 3745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970166269198136e-05,
+      "loss": 2.0759,
+      "step": 3746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970149038370059e-05,
+      "loss": 2.064,
+      "step": 3747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9701318026428763e-05,
+      "loss": 2.0437,
+      "step": 3748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9701145620166743e-05,
+      "loss": 2.0545,
+      "step": 3749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970097316491541e-05,
+      "loss": 2.0597,
+      "step": 3750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970080066067563e-05,
+      "loss": 2.0593,
+      "step": 3751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9700628107448273e-05,
+      "loss": 2.0601,
+      "step": 3752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9700455505234213e-05,
+      "loss": 2.0485,
+      "step": 3753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9700282854034324e-05,
+      "loss": 2.0464,
+      "step": 3754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9700110153849473e-05,
+      "loss": 2.081,
+      "step": 3755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969993740468053e-05,
+      "loss": 2.062,
+      "step": 3756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9699764606528376e-05,
+      "loss": 2.059,
+      "step": 3757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9699591759393876e-05,
+      "loss": 2.0626,
+      "step": 3758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9699418863277907e-05,
+      "loss": 2.0608,
+      "step": 3759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969924591818134e-05,
+      "loss": 2.052,
+      "step": 3760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969907292410505e-05,
+      "loss": 2.0562,
+      "step": 3761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698899881049905e-05,
+      "loss": 2.044,
+      "step": 3762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969872678901679e-05,
+      "loss": 2.0512,
+      "step": 3763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698553648006567e-05,
+      "loss": 2.0691,
+      "step": 3764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698380458020123e-05,
+      "loss": 2.0516,
+      "step": 3765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698207219058322e-05,
+      "loss": 2.0631,
+      "step": 3766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698033931122042e-05,
+      "loss": 2.0593,
+      "step": 3767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9697860594212156e-05,
+      "loss": 2.0614,
+      "step": 3768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969768720832954e-05,
+      "loss": 2.045,
+      "step": 3769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9697513773475076e-05,
+      "loss": 2.0596,
+      "step": 3770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9697340289649636e-05,
+      "loss": 2.0587,
+      "step": 3771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969716675685409e-05,
+      "loss": 2.0455,
+      "step": 3772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696993175089326e-05,
+      "loss": 2.0695,
+      "step": 3773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969681954435621e-05,
+      "loss": 2.0435,
+      "step": 3774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696645864655624e-05,
+      "loss": 2.0531,
+      "step": 3775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696472135988443e-05,
+      "loss": 2.0341,
+      "step": 3776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696298358355545e-05,
+      "loss": 2.0399,
+      "step": 3777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696124531757806e-05,
+      "loss": 2.0516,
+      "step": 3778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969595065619611e-05,
+      "loss": 2.0479,
+      "step": 3779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969577673167133e-05,
+      "loss": 2.0589,
+      "step": 3780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9695602758184345e-05,
+      "loss": 2.0333,
+      "step": 3781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969542873573603e-05,
+      "loss": 2.0552,
+      "step": 3782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9695254664327268e-05,
+      "loss": 2.0634,
+      "step": 3783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969508054395894e-05,
+      "loss": 2.0418,
+      "step": 3784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9694906374631922e-05,
+      "loss": 2.0663,
+      "step": 3785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969473215634709e-05,
+      "loss": 2.0586,
+      "step": 3786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9694557889105333e-05,
+      "loss": 2.0425,
+      "step": 3787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9694383572907525e-05,
+      "loss": 2.0599,
+      "step": 3788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9694209207754546e-05,
+      "loss": 2.0646,
+      "step": 3789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9694034793647277e-05,
+      "loss": 2.0554,
+      "step": 3790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96938603305866e-05,
+      "loss": 2.0394,
+      "step": 3791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9693685818573396e-05,
+      "loss": 2.0527,
+      "step": 3792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9693511257608544e-05,
+      "loss": 2.0765,
+      "step": 3793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969333664769293e-05,
+      "loss": 2.0447,
+      "step": 3794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9693161988827428e-05,
+      "loss": 2.0275,
+      "step": 3795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969298728101293e-05,
+      "loss": 2.0553,
+      "step": 3796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9692812524250312e-05,
+      "loss": 2.0835,
+      "step": 3797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9692637718540454e-05,
+      "loss": 2.0655,
+      "step": 3798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969246286388425e-05,
+      "loss": 2.0639,
+      "step": 3799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969228796028257e-05,
+      "loss": 2.0507,
+      "step": 3800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96921130077363e-05,
+      "loss": 2.0308,
+      "step": 3801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969193800624633e-05,
+      "loss": 2.052,
+      "step": 3802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9691762955813543e-05,
+      "loss": 2.0698,
+      "step": 3803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9691587856438815e-05,
+      "loss": 2.0471,
+      "step": 3804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969141270812304e-05,
+      "loss": 2.0351,
+      "step": 3805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9691237510867092e-05,
+      "loss": 2.0532,
+      "step": 3806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9691062264671865e-05,
+      "loss": 2.0429,
+      "step": 3807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9690886969538238e-05,
+      "loss": 2.0524,
+      "step": 3808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9690711625467097e-05,
+      "loss": 2.0375,
+      "step": 3809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9690536232459335e-05,
+      "loss": 2.0632,
+      "step": 3810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9690360790515827e-05,
+      "loss": 2.0656,
+      "step": 3811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9690185299637464e-05,
+      "loss": 2.0501,
+      "step": 3812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969000975982513e-05,
+      "loss": 2.0511,
+      "step": 3813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9689834171079715e-05,
+      "loss": 2.0548,
+      "step": 3814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9689658533402106e-05,
+      "loss": 2.0377,
+      "step": 3815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9689482846793185e-05,
+      "loss": 2.0522,
+      "step": 3816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9689307111253845e-05,
+      "loss": 2.0543,
+      "step": 3817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9689131326784968e-05,
+      "loss": 2.0476,
+      "step": 3818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9688955493387447e-05,
+      "loss": 2.0576,
+      "step": 3819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9688779611062165e-05,
+      "loss": 2.0416,
+      "step": 3820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9688603679810014e-05,
+      "loss": 2.0484,
+      "step": 3821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9688427699631877e-05,
+      "loss": 2.0347,
+      "step": 3822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9688251670528647e-05,
+      "loss": 2.053,
+      "step": 3823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9688075592501214e-05,
+      "loss": 2.0357,
+      "step": 3824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9687899465550466e-05,
+      "loss": 2.0455,
+      "step": 3825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9687723289677293e-05,
+      "loss": 2.0515,
+      "step": 3826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968754706488258e-05,
+      "loss": 2.0515,
+      "step": 3827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9687370791167223e-05,
+      "loss": 2.0605,
+      "step": 3828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9687194468532108e-05,
+      "loss": 2.0454,
+      "step": 3829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968701809697813e-05,
+      "loss": 2.0405,
+      "step": 3830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9686841676506172e-05,
+      "loss": 2.0639,
+      "step": 3831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9686665207117133e-05,
+      "loss": 2.0555,
+      "step": 3832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96864886888119e-05,
+      "loss": 2.075,
+      "step": 3833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9686312121591364e-05,
+      "loss": 2.0325,
+      "step": 3834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9686135505456422e-05,
+      "loss": 2.0421,
+      "step": 3835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9685958840407956e-05,
+      "loss": 2.0657,
+      "step": 3836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9685782126446865e-05,
+      "loss": 2.0449,
+      "step": 3837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9685605363574043e-05,
+      "loss": 2.0511,
+      "step": 3838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9685428551790377e-05,
+      "loss": 2.0347,
+      "step": 3839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9685251691096765e-05,
+      "loss": 2.0666,
+      "step": 3840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96850747814941e-05,
+      "loss": 2.0608,
+      "step": 3841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9684897822983266e-05,
+      "loss": 2.0408,
+      "step": 3842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9684720815565168e-05,
+      "loss": 2.0403,
+      "step": 3843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9684543759240695e-05,
+      "loss": 2.0519,
+      "step": 3844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968436665401074e-05,
+      "loss": 2.0549,
+      "step": 3845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96841894998762e-05,
+      "loss": 2.0433,
+      "step": 3846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968401229683797e-05,
+      "loss": 2.0677,
+      "step": 3847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9683835044896943e-05,
+      "loss": 2.0421,
+      "step": 3848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9683657744054017e-05,
+      "loss": 2.043,
+      "step": 3849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968348039431008e-05,
+      "loss": 2.0464,
+      "step": 3850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9683302995666037e-05,
+      "loss": 2.0368,
+      "step": 3851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9683125548122775e-05,
+      "loss": 2.0435,
+      "step": 3852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9682948051681202e-05,
+      "loss": 2.0377,
+      "step": 3853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96827705063422e-05,
+      "loss": 2.0344,
+      "step": 3854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9682592912106675e-05,
+      "loss": 2.0482,
+      "step": 3855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968241526897552e-05,
+      "loss": 2.0471,
+      "step": 3856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968223757694963e-05,
+      "loss": 2.0416,
+      "step": 3857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968205983602991e-05,
+      "loss": 2.0632,
+      "step": 3858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9681882046217254e-05,
+      "loss": 2.0551,
+      "step": 3859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9681704207512558e-05,
+      "loss": 2.057,
+      "step": 3860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968152631991672e-05,
+      "loss": 2.0494,
+      "step": 3861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968134838343064e-05,
+      "loss": 2.0618,
+      "step": 3862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9681170398055217e-05,
+      "loss": 2.0321,
+      "step": 3863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9680992363791345e-05,
+      "loss": 2.0354,
+      "step": 3864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968081428063993e-05,
+      "loss": 2.0557,
+      "step": 3865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9680636148601866e-05,
+      "loss": 2.0467,
+      "step": 3866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9680457967678054e-05,
+      "loss": 2.0433,
+      "step": 3867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96802797378694e-05,
+      "loss": 2.0673,
+      "step": 3868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968010145917679e-05,
+      "loss": 2.0579,
+      "step": 3869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9679923131601138e-05,
+      "loss": 2.0643,
+      "step": 3870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9679744755143338e-05,
+      "loss": 2.0516,
+      "step": 3871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9679566329804293e-05,
+      "loss": 2.0407,
+      "step": 3872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96793878555849e-05,
+      "loss": 2.0628,
+      "step": 3873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9679209332486067e-05,
+      "loss": 2.0546,
+      "step": 3874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9679030760508692e-05,
+      "loss": 2.0424,
+      "step": 3875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9678852139653674e-05,
+      "loss": 2.0325,
+      "step": 3876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9678673469921917e-05,
+      "loss": 2.0668,
+      "step": 3877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9678494751314323e-05,
+      "loss": 2.0591,
+      "step": 3878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96783159838318e-05,
+      "loss": 2.0379,
+      "step": 3879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967813716747524e-05,
+      "loss": 2.0509,
+      "step": 3880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9677958302245556e-05,
+      "loss": 2.0529,
+      "step": 3881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967777938814365e-05,
+      "loss": 2.0392,
+      "step": 3882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9677600425170417e-05,
+      "loss": 2.0463,
+      "step": 3883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967742141332677e-05,
+      "loss": 2.057,
+      "step": 3884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9677242352613606e-05,
+      "loss": 2.0658,
+      "step": 3885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9677063243031836e-05,
+      "loss": 2.0405,
+      "step": 3886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967688408458236e-05,
+      "loss": 2.031,
+      "step": 3887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9676704877266082e-05,
+      "loss": 2.048,
+      "step": 3888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9676525621083912e-05,
+      "loss": 2.0429,
+      "step": 3889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9676346316036753e-05,
+      "loss": 2.0479,
+      "step": 3890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9676166962125504e-05,
+      "loss": 2.0545,
+      "step": 3891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9675987559351082e-05,
+      "loss": 2.0515,
+      "step": 3892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9675808107714385e-05,
+      "loss": 2.0484,
+      "step": 3893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967562860721632e-05,
+      "loss": 2.0498,
+      "step": 3894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9675449057857797e-05,
+      "loss": 2.0359,
+      "step": 3895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967526945963972e-05,
+      "loss": 2.0471,
+      "step": 3896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9675089812562996e-05,
+      "loss": 2.0754,
+      "step": 3897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9674910116628533e-05,
+      "loss": 2.0544,
+      "step": 3898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9674730371837237e-05,
+      "loss": 2.0484,
+      "step": 3899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967455057819002e-05,
+      "loss": 2.0823,
+      "step": 3900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9674370735687787e-05,
+      "loss": 2.0251,
+      "step": 3901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9674190844331442e-05,
+      "loss": 2.0545,
+      "step": 3902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96740109041219e-05,
+      "loss": 2.0513,
+      "step": 3903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9673830915060067e-05,
+      "loss": 2.0505,
+      "step": 3904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967365087714685e-05,
+      "loss": 2.0559,
+      "step": 3905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9673470790383162e-05,
+      "loss": 2.0825,
+      "step": 3906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967329065476991e-05,
+      "loss": 2.0384,
+      "step": 3907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9673110470308004e-05,
+      "loss": 2.0529,
+      "step": 3908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9672930236998353e-05,
+      "loss": 2.0557,
+      "step": 3909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967274995484187e-05,
+      "loss": 2.0499,
+      "step": 3910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9672569623839465e-05,
+      "loss": 2.0423,
+      "step": 3911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9672389243992047e-05,
+      "loss": 2.0427,
+      "step": 3912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967220881530052e-05,
+      "loss": 2.0701,
+      "step": 3913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9672028337765813e-05,
+      "loss": 2.0534,
+      "step": 3914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9671847811388817e-05,
+      "loss": 2.0318,
+      "step": 3915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967166723617046e-05,
+      "loss": 2.0683,
+      "step": 3916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9671486612111644e-05,
+      "loss": 2.0409,
+      "step": 3917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9671305939213286e-05,
+      "loss": 2.0408,
+      "step": 3918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9671125217476296e-05,
+      "loss": 2.0143,
+      "step": 3919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9670944446901588e-05,
+      "loss": 2.0268,
+      "step": 3920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967076362749007e-05,
+      "loss": 2.0442,
+      "step": 3921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9670582759242663e-05,
+      "loss": 2.0506,
+      "step": 3922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967040184216028e-05,
+      "loss": 2.0484,
+      "step": 3923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9670220876243825e-05,
+      "loss": 2.0775,
+      "step": 3924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9670039861494222e-05,
+      "loss": 2.0554,
+      "step": 3925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966985879791238e-05,
+      "loss": 2.0445,
+      "step": 3926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9669677685499217e-05,
+      "loss": 2.0547,
+      "step": 3927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966949652425564e-05,
+      "loss": 2.0553,
+      "step": 3928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9669315314182575e-05,
+      "loss": 2.0548,
+      "step": 3929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9669134055280925e-05,
+      "loss": 2.0354,
+      "step": 3930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9668952747551614e-05,
+      "loss": 2.05,
+      "step": 3931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966877139099556e-05,
+      "loss": 2.044,
+      "step": 3932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9668589985613663e-05,
+      "loss": 2.0584,
+      "step": 3933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9668408531406856e-05,
+      "loss": 2.0381,
+      "step": 3934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9668227028376054e-05,
+      "loss": 2.0559,
+      "step": 3935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966804547652216e-05,
+      "loss": 2.0457,
+      "step": 3936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9667863875846102e-05,
+      "loss": 2.0271,
+      "step": 3937
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9667682226348798e-05,
+      "loss": 2.0563,
+      "step": 3938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9667500528031162e-05,
+      "loss": 2.035,
+      "step": 3939
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9667318780894107e-05,
+      "loss": 2.0358,
+      "step": 3940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9667136984938556e-05,
+      "loss": 2.0449,
+      "step": 3941
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966695514016543e-05,
+      "loss": 2.0525,
+      "step": 3942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966677324657564e-05,
+      "loss": 2.0547,
+      "step": 3943
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9666591304170112e-05,
+      "loss": 2.0369,
+      "step": 3944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9666409312949756e-05,
+      "loss": 2.039,
+      "step": 3945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9666227272915498e-05,
+      "loss": 2.0641,
+      "step": 3946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9666045184068256e-05,
+      "loss": 2.0288,
+      "step": 3947
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966586304640895e-05,
+      "loss": 2.0513,
+      "step": 3948
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9665680859938496e-05,
+      "loss": 2.0531,
+      "step": 3949
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9665498624657818e-05,
+      "loss": 2.0287,
+      "step": 3950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966531634056783e-05,
+      "loss": 2.0472,
+      "step": 3951
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9665134007669465e-05,
+      "loss": 2.0213,
+      "step": 3952
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966495162596363e-05,
+      "loss": 2.0401,
+      "step": 3953
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9664769195451254e-05,
+      "loss": 2.057,
+      "step": 3954
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9664586716133258e-05,
+      "loss": 2.0492,
+      "step": 3955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966440418801056e-05,
+      "loss": 2.0459,
+      "step": 3956
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966422161108408e-05,
+      "loss": 2.0258,
+      "step": 3957
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9664038985354753e-05,
+      "loss": 2.0307,
+      "step": 3958
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9663856310823483e-05,
+      "loss": 2.0468,
+      "step": 3959
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9663673587491202e-05,
+      "loss": 2.0331,
+      "step": 3960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9663490815358836e-05,
+      "loss": 2.0389,
+      "step": 3961
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9663307994427303e-05,
+      "loss": 2.026,
+      "step": 3962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9663125124697527e-05,
+      "loss": 2.0341,
+      "step": 3963
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966294220617043e-05,
+      "loss": 2.05,
+      "step": 3964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966275923884694e-05,
+      "loss": 2.0427,
+      "step": 3965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9662576222727972e-05,
+      "loss": 2.0264,
+      "step": 3966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9662393157814463e-05,
+      "loss": 2.0408,
+      "step": 3967
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9662210044107327e-05,
+      "loss": 2.0609,
+      "step": 3968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9662026881607494e-05,
+      "loss": 2.042,
+      "step": 3969
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966184367031589e-05,
+      "loss": 2.063,
+      "step": 3970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9661660410233434e-05,
+      "loss": 2.0509,
+      "step": 3971
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9661477101361057e-05,
+      "loss": 2.0267,
+      "step": 3972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966129374369968e-05,
+      "loss": 2.0348,
+      "step": 3973
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9661110337250235e-05,
+      "loss": 2.0632,
+      "step": 3974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9660926882013647e-05,
+      "loss": 2.0477,
+      "step": 3975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9660743377990835e-05,
+      "loss": 2.0411,
+      "step": 3976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9660559825182734e-05,
+      "loss": 2.0343,
+      "step": 3977
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9660376223590267e-05,
+      "loss": 2.0432,
+      "step": 3978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9660192573214365e-05,
+      "loss": 2.0276,
+      "step": 3979
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9660008874055948e-05,
+      "loss": 2.0369,
+      "step": 3980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965982512611595e-05,
+      "loss": 2.038,
+      "step": 3981
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9659641329395297e-05,
+      "loss": 2.0419,
+      "step": 3982
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965945748389492e-05,
+      "loss": 2.0326,
+      "step": 3983
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965927358961574e-05,
+      "loss": 2.0301,
+      "step": 3984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9659089646558693e-05,
+      "loss": 2.0314,
+      "step": 3985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9658905654724704e-05,
+      "loss": 2.0388,
+      "step": 3986
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9658721614114705e-05,
+      "loss": 2.0443,
+      "step": 3987
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9658537524729622e-05,
+      "loss": 2.0459,
+      "step": 3988
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9658353386570387e-05,
+      "loss": 2.0389,
+      "step": 3989
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965816919963793e-05,
+      "loss": 2.0284,
+      "step": 3990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9657984963933177e-05,
+      "loss": 2.0489,
+      "step": 3991
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9657800679457064e-05,
+      "loss": 2.0098,
+      "step": 3992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965761634621052e-05,
+      "loss": 2.0385,
+      "step": 3993
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9657431964194474e-05,
+      "loss": 2.0415,
+      "step": 3994
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9657247533409858e-05,
+      "loss": 2.0491,
+      "step": 3995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9657063053857605e-05,
+      "loss": 2.0586,
+      "step": 3996
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9656878525538642e-05,
+      "loss": 2.0398,
+      "step": 3997
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9656693948453903e-05,
+      "loss": 2.044,
+      "step": 3998
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9656509322604328e-05,
+      "loss": 2.0369,
+      "step": 3999
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9656324647990836e-05,
+      "loss": 2.0416,
+      "step": 4000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9656139924614372e-05,
+      "loss": 2.0493,
+      "step": 4001
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9655955152475857e-05,
+      "loss": 2.0432,
+      "step": 4002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965577033157623e-05,
+      "loss": 2.0314,
+      "step": 4003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9655585461916425e-05,
+      "loss": 2.0407,
+      "step": 4004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9655400543497378e-05,
+      "loss": 2.0548,
+      "step": 4005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9655215576320017e-05,
+      "loss": 2.0474,
+      "step": 4006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9655030560385278e-05,
+      "loss": 2.0485,
+      "step": 4007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9654845495694096e-05,
+      "loss": 2.057,
+      "step": 4008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9654660382247406e-05,
+      "loss": 2.0473,
+      "step": 4009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9654475220046144e-05,
+      "loss": 2.0244,
+      "step": 4010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965429000909124e-05,
+      "loss": 2.0593,
+      "step": 4011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9654104749383635e-05,
+      "loss": 2.0254,
+      "step": 4012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965391944092426e-05,
+      "loss": 2.0507,
+      "step": 4013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965373408371406e-05,
+      "loss": 2.0189,
+      "step": 4014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9653548677753954e-05,
+      "loss": 2.0148,
+      "step": 4015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9653363223044893e-05,
+      "loss": 2.0365,
+      "step": 4016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9653177719587808e-05,
+      "loss": 2.0461,
+      "step": 4017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965299216738364e-05,
+      "loss": 2.058,
+      "step": 4018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9652806566433316e-05,
+      "loss": 2.0211,
+      "step": 4019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9652620916737782e-05,
+      "loss": 2.0225,
+      "step": 4020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9652435218297972e-05,
+      "loss": 2.0325,
+      "step": 4021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9652249471114828e-05,
+      "loss": 2.0326,
+      "step": 4022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9652063675189285e-05,
+      "loss": 2.0455,
+      "step": 4023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965187783052228e-05,
+      "loss": 2.0274,
+      "step": 4024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9651691937114754e-05,
+      "loss": 2.0441,
+      "step": 4025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9651505994967645e-05,
+      "loss": 2.0199,
+      "step": 4026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965132000408189e-05,
+      "loss": 2.0429,
+      "step": 4027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9651133964458425e-05,
+      "loss": 2.0576,
+      "step": 4028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96509478760982e-05,
+      "loss": 2.0175,
+      "step": 4029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9650761739002147e-05,
+      "loss": 2.0379,
+      "step": 4030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9650575553171207e-05,
+      "loss": 2.029,
+      "step": 4031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965038931860632e-05,
+      "loss": 2.0442,
+      "step": 4032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965020303530843e-05,
+      "loss": 2.0428,
+      "step": 4033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965001670327848e-05,
+      "loss": 2.0394,
+      "step": 4034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9649830322517398e-05,
+      "loss": 2.0509,
+      "step": 4035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9649643893026135e-05,
+      "loss": 2.0373,
+      "step": 4036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9649457414805635e-05,
+      "loss": 2.0583,
+      "step": 4037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964927088785683e-05,
+      "loss": 2.0295,
+      "step": 4038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964908431218067e-05,
+      "loss": 2.0337,
+      "step": 4039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9648897687778094e-05,
+      "loss": 2.0253,
+      "step": 4040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9648711014650045e-05,
+      "loss": 2.0478,
+      "step": 4041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964852429279747e-05,
+      "loss": 2.0546,
+      "step": 4042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9648337522221302e-05,
+      "loss": 2.0416,
+      "step": 4043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964815070292249e-05,
+      "loss": 2.0335,
+      "step": 4044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964796383490198e-05,
+      "loss": 2.0538,
+      "step": 4045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964777691816071e-05,
+      "loss": 2.0407,
+      "step": 4046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9647589952699628e-05,
+      "loss": 2.0395,
+      "step": 4047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9647402938519678e-05,
+      "loss": 2.0523,
+      "step": 4048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96472158756218e-05,
+      "loss": 2.0278,
+      "step": 4049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9647028764006947e-05,
+      "loss": 2.0174,
+      "step": 4050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9646841603676053e-05,
+      "loss": 2.0602,
+      "step": 4051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9646654394630072e-05,
+      "loss": 2.0428,
+      "step": 4052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9646467136869945e-05,
+      "loss": 2.0408,
+      "step": 4053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9646279830396618e-05,
+      "loss": 2.0306,
+      "step": 4054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9646092475211043e-05,
+      "loss": 2.0328,
+      "step": 4055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9645905071314158e-05,
+      "loss": 2.0331,
+      "step": 4056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964571761870691e-05,
+      "loss": 2.0548,
+      "step": 4057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964553011739025e-05,
+      "loss": 2.02,
+      "step": 4058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9645342567365123e-05,
+      "loss": 2.0478,
+      "step": 4059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9645154968632472e-05,
+      "loss": 2.0386,
+      "step": 4060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9644967321193253e-05,
+      "loss": 2.0499,
+      "step": 4061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964477962504841e-05,
+      "loss": 2.0407,
+      "step": 4062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9644591880198885e-05,
+      "loss": 2.017,
+      "step": 4063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964440408664563e-05,
+      "loss": 2.0362,
+      "step": 4064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9644216244389596e-05,
+      "loss": 2.0348,
+      "step": 4065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964402835343173e-05,
+      "loss": 2.0344,
+      "step": 4066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9643840413772982e-05,
+      "loss": 2.0422,
+      "step": 4067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9643652425414297e-05,
+      "loss": 2.0409,
+      "step": 4068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9643464388356625e-05,
+      "loss": 2.0418,
+      "step": 4069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964327630260092e-05,
+      "loss": 2.0249,
+      "step": 4070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9643088168148128e-05,
+      "loss": 2.0317,
+      "step": 4071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96428999849992e-05,
+      "loss": 2.0499,
+      "step": 4072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642711753155087e-05,
+      "loss": 2.0319,
+      "step": 4073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964252347261674e-05,
+      "loss": 2.0322,
+      "step": 4074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964233514338511e-05,
+      "loss": 2.051,
+      "step": 4075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642146765461146e-05,
+      "loss": 2.0391,
+      "step": 4076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641958338845798e-05,
+      "loss": 2.0293,
+      "step": 4077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641769863540018e-05,
+      "loss": 2.0376,
+      "step": 4078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641581339544764e-05,
+      "loss": 2.0497,
+      "step": 4079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964139276686098e-05,
+      "loss": 2.0218,
+      "step": 4080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641204145489623e-05,
+      "loss": 2.0431,
+      "step": 4081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641015475431646e-05,
+      "loss": 2.0244,
+      "step": 4082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640826756687998e-05,
+      "loss": 2.0522,
+      "step": 4083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640637989259636e-05,
+      "loss": 2.039,
+      "step": 4084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964044917314751e-05,
+      "loss": 2.0452,
+      "step": 4085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640260308352573e-05,
+      "loss": 2.0265,
+      "step": 4086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964007139487578e-05,
+      "loss": 2.0502,
+      "step": 4087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9639882432718086e-05,
+      "loss": 2.0209,
+      "step": 4088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963969342188045e-05,
+      "loss": 2.0575,
+      "step": 4089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9639504362363813e-05,
+      "loss": 2.023,
+      "step": 4090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9639315254169144e-05,
+      "loss": 2.0306,
+      "step": 4091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9639126097297388e-05,
+      "loss": 2.0299,
+      "step": 4092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963893689174951e-05,
+      "loss": 2.0223,
+      "step": 4093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9638747637526453e-05,
+      "loss": 2.009,
+      "step": 4094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9638558334629182e-05,
+      "loss": 2.0194,
+      "step": 4095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9638368983058647e-05,
+      "loss": 2.0558,
+      "step": 4096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9638179582815812e-05,
+      "loss": 2.0348,
+      "step": 4097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9637990133901625e-05,
+      "loss": 2.0199,
+      "step": 4098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963780063631705e-05,
+      "loss": 2.0272,
+      "step": 4099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9637611090063035e-05,
+      "loss": 2.044,
+      "step": 4100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9637421495140546e-05,
+      "loss": 2.0296,
+      "step": 4101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963723185155054e-05,
+      "loss": 2.0184,
+      "step": 4102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9637042159293965e-05,
+      "loss": 2.0305,
+      "step": 4103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9636852418371792e-05,
+      "loss": 2.0325,
+      "step": 4104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9636662628784968e-05,
+      "loss": 2.0113,
+      "step": 4105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963647279053446e-05,
+      "loss": 2.0334,
+      "step": 4106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963628290362122e-05,
+      "loss": 2.049,
+      "step": 4107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963609296804621e-05,
+      "loss": 2.0317,
+      "step": 4108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963590298381039e-05,
+      "loss": 2.0643,
+      "step": 4109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963571295091472e-05,
+      "loss": 2.0306,
+      "step": 4110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9635522869360155e-05,
+      "loss": 2.0583,
+      "step": 4111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9635332739147657e-05,
+      "loss": 2.0411,
+      "step": 4112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9635142560278192e-05,
+      "loss": 2.0167,
+      "step": 4113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634952332752712e-05,
+      "loss": 2.0356,
+      "step": 4114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634762056572182e-05,
+      "loss": 2.0438,
+      "step": 4115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963457173173756e-05,
+      "loss": 2.0315,
+      "step": 4116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634381358249808e-05,
+      "loss": 2.033,
+      "step": 4117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634190936109888e-05,
+      "loss": 2.0465,
+      "step": 4118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634000465318765e-05,
+      "loss": 2.0217,
+      "step": 4119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9633809945877396e-05,
+      "loss": 2.0245,
+      "step": 4120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9633619377786744e-05,
+      "loss": 2.0364,
+      "step": 4121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9633428761047774e-05,
+      "loss": 2.052,
+      "step": 4122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9633238095661446e-05,
+      "loss": 2.0228,
+      "step": 4123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963304738162872e-05,
+      "loss": 2.043,
+      "step": 4124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9632856618950572e-05,
+      "loss": 2.0519,
+      "step": 4125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9632665807627948e-05,
+      "loss": 2.0194,
+      "step": 4126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963247494766182e-05,
+      "loss": 2.0682,
+      "step": 4127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9632284039053157e-05,
+      "loss": 2.0402,
+      "step": 4128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963209308180291e-05,
+      "loss": 2.0392,
+      "step": 4129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963190207591205e-05,
+      "loss": 2.0383,
+      "step": 4130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9631711021381548e-05,
+      "loss": 2.0246,
+      "step": 4131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9631519918212363e-05,
+      "loss": 2.0169,
+      "step": 4132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963132876640546e-05,
+      "loss": 2.026,
+      "step": 4133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96311375659618e-05,
+      "loss": 2.0315,
+      "step": 4134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963094631688236e-05,
+      "loss": 2.0315,
+      "step": 4135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963075501916809e-05,
+      "loss": 2.0222,
+      "step": 4136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963056367281997e-05,
+      "loss": 2.063,
+      "step": 4137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963037227783896e-05,
+      "loss": 2.0455,
+      "step": 4138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963018083422603e-05,
+      "loss": 2.0378,
+      "step": 4139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962998934198214e-05,
+      "loss": 2.0339,
+      "step": 4140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629797801108262e-05,
+      "loss": 2.039,
+      "step": 4141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629606211605363e-05,
+      "loss": 2.0364,
+      "step": 4142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962941457347441e-05,
+      "loss": 2.0381,
+      "step": 4143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962922288671637e-05,
+      "loss": 2.0703,
+      "step": 4144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629031151332217e-05,
+      "loss": 2.0561,
+      "step": 4145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962883936732291e-05,
+      "loss": 2.054,
+      "step": 4146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962864753468942e-05,
+      "loss": 2.0408,
+      "step": 4147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628455653432722e-05,
+      "loss": 2.029,
+      "step": 4148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628263723553777e-05,
+      "loss": 2.027,
+      "step": 4149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628071745053555e-05,
+      "loss": 2.027,
+      "step": 4150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962787971793303e-05,
+      "loss": 2.0468,
+      "step": 4151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962768764219317e-05,
+      "loss": 2.0293,
+      "step": 4152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9627495517834942e-05,
+      "loss": 2.0321,
+      "step": 4153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962730334485932e-05,
+      "loss": 2.0434,
+      "step": 4154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9627111123267276e-05,
+      "loss": 2.0442,
+      "step": 4155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9626918853059776e-05,
+      "loss": 2.0409,
+      "step": 4156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9626726534237792e-05,
+      "loss": 2.0315,
+      "step": 4157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9626534166802294e-05,
+      "loss": 2.0457,
+      "step": 4158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962634175075426e-05,
+      "loss": 2.0351,
+      "step": 4159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9626149286094657e-05,
+      "loss": 2.0336,
+      "step": 4160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625956772824453e-05,
+      "loss": 2.0155,
+      "step": 4161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625764210944627e-05,
+      "loss": 2.0306,
+      "step": 4162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625571600456145e-05,
+      "loss": 2.0332,
+      "step": 4163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962537894135999e-05,
+      "loss": 2.0502,
+      "step": 4164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625186233657123e-05,
+      "loss": 2.0411,
+      "step": 4165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962499347734852e-05,
+      "loss": 2.0323,
+      "step": 4166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962480067243516e-05,
+      "loss": 2.0367,
+      "step": 4167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624607818918014e-05,
+      "loss": 2.0334,
+      "step": 4168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624414916798057e-05,
+      "loss": 2.0253,
+      "step": 4169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624221966076257e-05,
+      "loss": 2.0257,
+      "step": 4170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624028966753594e-05,
+      "loss": 2.0472,
+      "step": 4171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962383591883104e-05,
+      "loss": 2.0439,
+      "step": 4172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962364282230957e-05,
+      "loss": 2.0344,
+      "step": 4173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9623449677190166e-05,
+      "loss": 2.0299,
+      "step": 4174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9623256483473794e-05,
+      "loss": 2.0348,
+      "step": 4175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9623063241161434e-05,
+      "loss": 2.0309,
+      "step": 4176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962286995025406e-05,
+      "loss": 2.0235,
+      "step": 4177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622676610752648e-05,
+      "loss": 2.0418,
+      "step": 4178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622483222658176e-05,
+      "loss": 2.028,
+      "step": 4179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622289785971623e-05,
+      "loss": 2.0371,
+      "step": 4180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962209630069396e-05,
+      "loss": 2.0441,
+      "step": 4181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962190276682617e-05,
+      "loss": 2.0241,
+      "step": 4182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9621709184369223e-05,
+      "loss": 2.0553,
+      "step": 4183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9621515553324103e-05,
+      "loss": 2.0389,
+      "step": 4184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9621321873691787e-05,
+      "loss": 2.0257,
+      "step": 4185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9621128145473247e-05,
+      "loss": 2.0401,
+      "step": 4186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962093436866947e-05,
+      "loss": 2.0464,
+      "step": 4187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962074054328143e-05,
+      "loss": 2.0488,
+      "step": 4188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962054666931011e-05,
+      "loss": 2.0213,
+      "step": 4189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9620352746756476e-05,
+      "loss": 2.0228,
+      "step": 4190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962015877562152e-05,
+      "loss": 2.0388,
+      "step": 4191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9619964755906222e-05,
+      "loss": 2.0399,
+      "step": 4192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9619770687611558e-05,
+      "loss": 2.0202,
+      "step": 4193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9619576570738507e-05,
+      "loss": 2.0418,
+      "step": 4194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9619382405288045e-05,
+      "loss": 2.032,
+      "step": 4195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961918819126116e-05,
+      "loss": 2.0318,
+      "step": 4196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9618993928658833e-05,
+      "loss": 2.0001,
+      "step": 4197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9618799617482045e-05,
+      "loss": 2.0338,
+      "step": 4198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961860525773177e-05,
+      "loss": 2.0259,
+      "step": 4199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9618410849408996e-05,
+      "loss": 2.0263,
+      "step": 4200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96182163925147e-05,
+      "loss": 2.0374,
+      "step": 4201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961802188704987e-05,
+      "loss": 2.0376,
+      "step": 4202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9617827333015483e-05,
+      "loss": 2.0263,
+      "step": 4203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9617632730412527e-05,
+      "loss": 2.0302,
+      "step": 4204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961743807924198e-05,
+      "loss": 2.022,
+      "step": 4205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9617243379504825e-05,
+      "loss": 2.0411,
+      "step": 4206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9617048631202047e-05,
+      "loss": 2.0316,
+      "step": 4207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961685383433463e-05,
+      "loss": 2.0218,
+      "step": 4208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9616658988903553e-05,
+      "loss": 2.0346,
+      "step": 4209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9616464094909807e-05,
+      "loss": 2.0365,
+      "step": 4210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961626915235437e-05,
+      "loss": 2.0372,
+      "step": 4211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961607416123823e-05,
+      "loss": 2.0459,
+      "step": 4212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9615879121562373e-05,
+      "loss": 2.0326,
+      "step": 4213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961568403332778e-05,
+      "loss": 2.0389,
+      "step": 4214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961548889653544e-05,
+      "loss": 2.0384,
+      "step": 4215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9615293711186334e-05,
+      "loss": 2.0337,
+      "step": 4216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961509847728145e-05,
+      "loss": 2.0283,
+      "step": 4217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961490319482177e-05,
+      "loss": 2.021,
+      "step": 4218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614707863808292e-05,
+      "loss": 2.0192,
+      "step": 4219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961451248424199e-05,
+      "loss": 2.0029,
+      "step": 4220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614317056123857e-05,
+      "loss": 2.0411,
+      "step": 4221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614121579454877e-05,
+      "loss": 2.0506,
+      "step": 4222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613926054236038e-05,
+      "loss": 2.0272,
+      "step": 4223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613730480468323e-05,
+      "loss": 2.0207,
+      "step": 4224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613534858152732e-05,
+      "loss": 2.044,
+      "step": 4225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613339187290237e-05,
+      "loss": 2.0501,
+      "step": 4226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613143467881845e-05,
+      "loss": 2.0436,
+      "step": 4227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9612947699928524e-05,
+      "loss": 2.0231,
+      "step": 4228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9612751883431273e-05,
+      "loss": 2.0332,
+      "step": 4229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9612556018391083e-05,
+      "loss": 2.026,
+      "step": 4230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9612360104808936e-05,
+      "loss": 2.0519,
+      "step": 4231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.961216414268583e-05,
+      "loss": 2.0331,
+      "step": 4232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9611968132022747e-05,
+      "loss": 2.02,
+      "step": 4233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9611772072820677e-05,
+      "loss": 2.0243,
+      "step": 4234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9611575965080618e-05,
+      "loss": 2.0233,
+      "step": 4235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.961137980880355e-05,
+      "loss": 2.0114,
+      "step": 4236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9611183603990477e-05,
+      "loss": 2.033,
+      "step": 4237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9610987350642374e-05,
+      "loss": 2.0315,
+      "step": 4238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9610791048760242e-05,
+      "loss": 2.0409,
+      "step": 4239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9610594698345067e-05,
+      "loss": 2.0463,
+      "step": 4240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9610398299397845e-05,
+      "loss": 2.0258,
+      "step": 4241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9610201851919565e-05,
+      "loss": 2.0452,
+      "step": 4242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9610005355911224e-05,
+      "loss": 2.0393,
+      "step": 4243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9609808811373807e-05,
+      "loss": 2.0183,
+      "step": 4244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960961221830831e-05,
+      "loss": 2.0324,
+      "step": 4245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9609415576715728e-05,
+      "loss": 2.0277,
+      "step": 4246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960921888659705e-05,
+      "loss": 2.0312,
+      "step": 4247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960902214795327e-05,
+      "loss": 2.0416,
+      "step": 4248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9608825360785382e-05,
+      "loss": 2.0276,
+      "step": 4249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9608628525094383e-05,
+      "loss": 2.0514,
+      "step": 4250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9608431640881264e-05,
+      "loss": 2.0423,
+      "step": 4251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9608234708147014e-05,
+      "loss": 2.049,
+      "step": 4252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960803772689264e-05,
+      "loss": 2.0633,
+      "step": 4253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960784069711913e-05,
+      "loss": 2.0169,
+      "step": 4254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960764361882747e-05,
+      "loss": 2.056,
+      "step": 4255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960744649201867e-05,
+      "loss": 2.0202,
+      "step": 4256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9607249316693718e-05,
+      "loss": 2.0391,
+      "step": 4257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960705209285361e-05,
+      "loss": 2.0265,
+      "step": 4258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9606854820499347e-05,
+      "loss": 2.0246,
+      "step": 4259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9606657499631917e-05,
+      "loss": 2.0279,
+      "step": 4260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9606460130252322e-05,
+      "loss": 2.0413,
+      "step": 4261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9606262712361554e-05,
+      "loss": 2.0458,
+      "step": 4262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9606065245960617e-05,
+      "loss": 2.0381,
+      "step": 4263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9605867731050502e-05,
+      "loss": 2.0461,
+      "step": 4264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960567016763221e-05,
+      "loss": 2.0556,
+      "step": 4265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960547255570674e-05,
+      "loss": 2.0325,
+      "step": 4266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9605274895275084e-05,
+      "loss": 2.0459,
+      "step": 4267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9605077186338243e-05,
+      "loss": 2.0381,
+      "step": 4268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9604879428897215e-05,
+      "loss": 2.0488,
+      "step": 4269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9604681622953007e-05,
+      "loss": 2.0456,
+      "step": 4270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.96044837685066e-05,
+      "loss": 2.0311,
+      "step": 4271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960428586555901e-05,
+      "loss": 2.017,
+      "step": 4272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960408791411123e-05,
+      "loss": 2.0153,
+      "step": 4273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960388991416426e-05,
+      "loss": 2.0506,
+      "step": 4274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9603691865719096e-05,
+      "loss": 1.998,
+      "step": 4275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9603493768776743e-05,
+      "loss": 2.0234,
+      "step": 4276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.96032956233382e-05,
+      "loss": 2.0354,
+      "step": 4277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9603097429404468e-05,
+      "loss": 2.0386,
+      "step": 4278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9602899186976545e-05,
+      "loss": 2.0439,
+      "step": 4279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9602700896055438e-05,
+      "loss": 2.0273,
+      "step": 4280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960250255664214e-05,
+      "loss": 2.0484,
+      "step": 4281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9602304168737662e-05,
+      "loss": 2.026,
+      "step": 4282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9602105732343e-05,
+      "loss": 2.0305,
+      "step": 4283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9601907247459156e-05,
+      "loss": 2.0386,
+      "step": 4284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9601708714087136e-05,
+      "loss": 2.0223,
+      "step": 4285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9601510132227938e-05,
+      "loss": 2.0499,
+      "step": 4286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9601311501882568e-05,
+      "loss": 2.0377,
+      "step": 4287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960111282305203e-05,
+      "loss": 2.0412,
+      "step": 4288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960091409573732e-05,
+      "loss": 2.0404,
+      "step": 4289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9600715319939453e-05,
+      "loss": 2.0331,
+      "step": 4290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9600516495659422e-05,
+      "loss": 2.033,
+      "step": 4291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9600317622898237e-05,
+      "loss": 2.0463,
+      "step": 4292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.96001187016569e-05,
+      "loss": 2.0328,
+      "step": 4293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959991973193642e-05,
+      "loss": 2.0284,
+      "step": 4294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9599720713737794e-05,
+      "loss": 2.0379,
+      "step": 4295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9599521647062033e-05,
+      "loss": 2.0318,
+      "step": 4296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959932253191014e-05,
+      "loss": 2.0217,
+      "step": 4297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9599123368283122e-05,
+      "loss": 2.0098,
+      "step": 4298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959892415618198e-05,
+      "loss": 2.0569,
+      "step": 4299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959872489560773e-05,
+      "loss": 2.0298,
+      "step": 4300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959852558656137e-05,
+      "loss": 2.0265,
+      "step": 4301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9598326229043905e-05,
+      "loss": 2.0316,
+      "step": 4302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959812682305635e-05,
+      "loss": 2.0468,
+      "step": 4303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597927368599704e-05,
+      "loss": 2.0314,
+      "step": 4304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597727865674978e-05,
+      "loss": 2.0206,
+      "step": 4305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959752831428318e-05,
+      "loss": 2.0449,
+      "step": 4306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597328714425314e-05,
+      "loss": 2.0248,
+      "step": 4307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597129066102392e-05,
+      "loss": 2.0426,
+      "step": 4308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9596929369315423e-05,
+      "loss": 2.0449,
+      "step": 4309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9596729624065413e-05,
+      "loss": 2.0313,
+      "step": 4310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9596529830353365e-05,
+      "loss": 2.0375,
+      "step": 4311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95963299881803e-05,
+      "loss": 2.0404,
+      "step": 4312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9596130097547217e-05,
+      "loss": 2.0145,
+      "step": 4313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959593015845513e-05,
+      "loss": 2.0315,
+      "step": 4314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959573017090505e-05,
+      "loss": 2.052,
+      "step": 4315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959553013489798e-05,
+      "loss": 2.0273,
+      "step": 4316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959533005043494e-05,
+      "loss": 2.0364,
+      "step": 4317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9595129917516935e-05,
+      "loss": 2.0359,
+      "step": 4318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 2.0044,
+      "step": 4319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9594729506320073e-05,
+      "loss": 2.0156,
+      "step": 4320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9594529228043238e-05,
+      "loss": 2.0607,
+      "step": 4321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9594328901315484e-05,
+      "loss": 2.0339,
+      "step": 4322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9594128526137818e-05,
+      "loss": 2.0307,
+      "step": 4323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9593928102511257e-05,
+      "loss": 2.0313,
+      "step": 4324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959372763043681e-05,
+      "loss": 2.038,
+      "step": 4325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9593527109915493e-05,
+      "loss": 2.0348,
+      "step": 4326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959332654094831e-05,
+      "loss": 2.0319,
+      "step": 4327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9593125923536288e-05,
+      "loss": 2.0246,
+      "step": 4328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9592925257680425e-05,
+      "loss": 2.0314,
+      "step": 4329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9592724543381748e-05,
+      "loss": 2.0228,
+      "step": 4330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959252378064126e-05,
+      "loss": 2.0305,
+      "step": 4331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9592322969459978e-05,
+      "loss": 2.0384,
+      "step": 4332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9592122109838918e-05,
+      "loss": 2.0246,
+      "step": 4333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9591921201779093e-05,
+      "loss": 2.0321,
+      "step": 4334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959172024528152e-05,
+      "loss": 2.0415,
+      "step": 4335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9591519240347204e-05,
+      "loss": 2.0283,
+      "step": 4336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9591318186977172e-05,
+      "loss": 2.0111,
+      "step": 4337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959111708517244e-05,
+      "loss": 2.034,
+      "step": 4338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959091593493401e-05,
+      "loss": 2.0177,
+      "step": 4339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9590714736262912e-05,
+      "loss": 2.0499,
+      "step": 4340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959051348916015e-05,
+      "loss": 2.0416,
+      "step": 4341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959031219362675e-05,
+      "loss": 2.0231,
+      "step": 4342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9590110849663725e-05,
+      "loss": 2.0195,
+      "step": 4343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958990945727209e-05,
+      "loss": 2.0201,
+      "step": 4344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9589708016452868e-05,
+      "loss": 2.0489,
+      "step": 4345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9589506527207068e-05,
+      "loss": 2.0374,
+      "step": 4346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958930498953571e-05,
+      "loss": 2.0362,
+      "step": 4347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9589103403439814e-05,
+      "loss": 2.0328,
+      "step": 4348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95889017689204e-05,
+      "loss": 2.0269,
+      "step": 4349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958870008597848e-05,
+      "loss": 2.0419,
+      "step": 4350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958849835461508e-05,
+      "loss": 2.0276,
+      "step": 4351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9588296574831208e-05,
+      "loss": 2.0313,
+      "step": 4352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958809474662789e-05,
+      "loss": 2.0051,
+      "step": 4353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9587892870006152e-05,
+      "loss": 2.0424,
+      "step": 4354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9587690944967003e-05,
+      "loss": 2.0197,
+      "step": 4355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9587488971511465e-05,
+      "loss": 2.0254,
+      "step": 4356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9587286949640556e-05,
+      "loss": 2.0393,
+      "step": 4357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95870848793553e-05,
+      "loss": 2.0402,
+      "step": 4358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958688276065672e-05,
+      "loss": 2.028,
+      "step": 4359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958668059354583e-05,
+      "loss": 2.0406,
+      "step": 4360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9586478378023655e-05,
+      "loss": 2.0315,
+      "step": 4361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958627611409121e-05,
+      "loss": 2.0181,
+      "step": 4362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9586073801749527e-05,
+      "loss": 2.0361,
+      "step": 4363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9585871440999622e-05,
+      "loss": 2.0359,
+      "step": 4364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958566903184252e-05,
+      "loss": 2.0185,
+      "step": 4365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9585466574279232e-05,
+      "loss": 2.0527,
+      "step": 4366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9585264068310793e-05,
+      "loss": 2.0219,
+      "step": 4367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958506151393822e-05,
+      "loss": 2.0314,
+      "step": 4368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9584858911162538e-05,
+      "loss": 2.0479,
+      "step": 4369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9584656259984768e-05,
+      "loss": 2.0205,
+      "step": 4370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9584453560405936e-05,
+      "loss": 2.0186,
+      "step": 4371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9584250812427065e-05,
+      "loss": 2.039,
+      "step": 4372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9584048016049176e-05,
+      "loss": 2.0378,
+      "step": 4373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958384517127329e-05,
+      "loss": 2.0396,
+      "step": 4374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9583642278100444e-05,
+      "loss": 2.0296,
+      "step": 4375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9583439336531655e-05,
+      "loss": 2.0415,
+      "step": 4376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9583236346567943e-05,
+      "loss": 2.0354,
+      "step": 4377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958303330821034e-05,
+      "loss": 2.0347,
+      "step": 4378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958283022145987e-05,
+      "loss": 2.0359,
+      "step": 4379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9582627086317555e-05,
+      "loss": 1.9995,
+      "step": 4380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958242390278443e-05,
+      "loss": 2.0296,
+      "step": 4381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9582220670861507e-05,
+      "loss": 2.0337,
+      "step": 4382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9582017390549822e-05,
+      "loss": 2.0507,
+      "step": 4383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95818140618504e-05,
+      "loss": 2.0382,
+      "step": 4384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958161068476427e-05,
+      "loss": 2.018,
+      "step": 4385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958140725929245e-05,
+      "loss": 2.0257,
+      "step": 4386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958120378543598e-05,
+      "loss": 2.0118,
+      "step": 4387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9581000263195876e-05,
+      "loss": 2.0403,
+      "step": 4388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9580796692573172e-05,
+      "loss": 2.015,
+      "step": 4389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9580593073568895e-05,
+      "loss": 2.0465,
+      "step": 4390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9580389406184072e-05,
+      "loss": 2.014,
+      "step": 4391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9580185690419735e-05,
+      "loss": 2.0306,
+      "step": 4392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9579981926276908e-05,
+      "loss": 2.0425,
+      "step": 4393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9579778113756625e-05,
+      "loss": 2.0328,
+      "step": 4394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957957425285991e-05,
+      "loss": 2.0365,
+      "step": 4395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9579370343587794e-05,
+      "loss": 2.0426,
+      "step": 4396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9579166385941307e-05,
+      "loss": 2.0389,
+      "step": 4397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9578962379921483e-05,
+      "loss": 2.0049,
+      "step": 4398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9578758325529346e-05,
+      "loss": 2.0035,
+      "step": 4399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9578554222765928e-05,
+      "loss": 2.0308,
+      "step": 4400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9578350071632264e-05,
+      "loss": 2.0396,
+      "step": 4401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957814587212938e-05,
+      "loss": 2.0342,
+      "step": 4402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9577941624258308e-05,
+      "loss": 2.036,
+      "step": 4403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9577737328020084e-05,
+      "loss": 2.0264,
+      "step": 4404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9577532983415732e-05,
+      "loss": 2.0175,
+      "step": 4405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957732859044629e-05,
+      "loss": 2.0233,
+      "step": 4406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957712414911279e-05,
+      "loss": 2.0343,
+      "step": 4407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957691965941626e-05,
+      "loss": 2.0255,
+      "step": 4408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9576715121357734e-05,
+      "loss": 2.0221,
+      "step": 4409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957651053493825e-05,
+      "loss": 2.0347,
+      "step": 4410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9576305900158835e-05,
+      "loss": 2.042,
+      "step": 4411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9576101217020524e-05,
+      "loss": 2.0248,
+      "step": 4412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957589648552435e-05,
+      "loss": 2.0351,
+      "step": 4413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957569170567135e-05,
+      "loss": 2.045,
+      "step": 4414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9575486877462553e-05,
+      "loss": 2.0331,
+      "step": 4415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9575282000899e-05,
+      "loss": 2.042,
+      "step": 4416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957507707598172e-05,
+      "loss": 2.0424,
+      "step": 4417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9574872102711752e-05,
+      "loss": 2.0365,
+      "step": 4418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9574667081090127e-05,
+      "loss": 2.032,
+      "step": 4419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9574462011117883e-05,
+      "loss": 2.01,
+      "step": 4420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9574256892796057e-05,
+      "loss": 2.0274,
+      "step": 4421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957405172612568e-05,
+      "loss": 2.022,
+      "step": 4422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957384651110779e-05,
+      "loss": 2.046,
+      "step": 4423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9573641247743426e-05,
+      "loss": 2.0163,
+      "step": 4424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9573435936033623e-05,
+      "loss": 2.0478,
+      "step": 4425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9573230575979415e-05,
+      "loss": 2.0244,
+      "step": 4426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9573025167581842e-05,
+      "loss": 2.0462,
+      "step": 4427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957281971084194e-05,
+      "loss": 2.0181,
+      "step": 4428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957261420576075e-05,
+      "loss": 2.0251,
+      "step": 4429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957240865233931e-05,
+      "loss": 2.0344,
+      "step": 4430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9572203050578648e-05,
+      "loss": 2.0175,
+      "step": 4431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9571997400479814e-05,
+      "loss": 2.0432,
+      "step": 4432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957179170204384e-05,
+      "loss": 2.0403,
+      "step": 4433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9571585955271767e-05,
+      "loss": 2.0399,
+      "step": 4434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9571380160164632e-05,
+      "loss": 2.0332,
+      "step": 4435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9571174316723477e-05,
+      "loss": 2.0266,
+      "step": 4436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957096842494934e-05,
+      "loss": 2.0359,
+      "step": 4437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957076248484326e-05,
+      "loss": 2.0347,
+      "step": 4438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9570556496406283e-05,
+      "loss": 2.0377,
+      "step": 4439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957035045963944e-05,
+      "loss": 2.0315,
+      "step": 4440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9570144374543774e-05,
+      "loss": 2.0267,
+      "step": 4441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9569938241120327e-05,
+      "loss": 2.0156,
+      "step": 4442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9569732059370143e-05,
+      "loss": 2.0334,
+      "step": 4443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956952582929426e-05,
+      "loss": 2.0179,
+      "step": 4444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956931955089372e-05,
+      "loss": 2.035,
+      "step": 4445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956911322416956e-05,
+      "loss": 2.0154,
+      "step": 4446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9568906849122836e-05,
+      "loss": 2.0159,
+      "step": 4447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9568700425754573e-05,
+      "loss": 2.0257,
+      "step": 4448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9568493954065824e-05,
+      "loss": 2.028,
+      "step": 4449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9568287434057627e-05,
+      "loss": 2.0341,
+      "step": 4450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956808086573103e-05,
+      "loss": 2.0139,
+      "step": 4451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956787424908707e-05,
+      "loss": 2.0426,
+      "step": 4452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956766758412679e-05,
+      "loss": 2.0281,
+      "step": 4453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9567460870851242e-05,
+      "loss": 2.0266,
+      "step": 4454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9567254109261462e-05,
+      "loss": 2.015,
+      "step": 4455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9567047299358495e-05,
+      "loss": 2.0231,
+      "step": 4456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9566840441143392e-05,
+      "loss": 2.0297,
+      "step": 4457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9566633534617188e-05,
+      "loss": 2.0282,
+      "step": 4458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9566426579780934e-05,
+      "loss": 2.0136,
+      "step": 4459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9566219576635674e-05,
+      "loss": 2.0376,
+      "step": 4460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9566012525182452e-05,
+      "loss": 2.0385,
+      "step": 4461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9565805425422317e-05,
+      "loss": 2.0459,
+      "step": 4462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956559827735631e-05,
+      "loss": 2.0459,
+      "step": 4463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956539108098548e-05,
+      "loss": 2.0265,
+      "step": 4464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9565183836310876e-05,
+      "loss": 2.0351,
+      "step": 4465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9564976543333537e-05,
+      "loss": 2.0279,
+      "step": 4466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9564769202054514e-05,
+      "loss": 2.0235,
+      "step": 4467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9564561812474852e-05,
+      "loss": 2.0424,
+      "step": 4468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9564354374595607e-05,
+      "loss": 2.0285,
+      "step": 4469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9564146888417815e-05,
+      "loss": 2.0459,
+      "step": 4470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956393935394253e-05,
+      "loss": 2.0388,
+      "step": 4471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95637317711708e-05,
+      "loss": 2.0311,
+      "step": 4472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956352414010367e-05,
+      "loss": 2.052,
+      "step": 4473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9563316460742193e-05,
+      "loss": 2.0285,
+      "step": 4474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9563108733087413e-05,
+      "loss": 2.0496,
+      "step": 4475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956290095714038e-05,
+      "loss": 2.0187,
+      "step": 4476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9562693132902145e-05,
+      "loss": 2.0296,
+      "step": 4477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956248526037376e-05,
+      "loss": 2.0261,
+      "step": 4478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9562277339556272e-05,
+      "loss": 2.0604,
+      "step": 4479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9562069370450726e-05,
+      "loss": 2.021,
+      "step": 4480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956186135305818e-05,
+      "loss": 2.0245,
+      "step": 4481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956165328737968e-05,
+      "loss": 2.0386,
+      "step": 4482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956144517341628e-05,
+      "loss": 2.0473,
+      "step": 4483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9561237011169027e-05,
+      "loss": 2.0177,
+      "step": 4484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9561028800638975e-05,
+      "loss": 2.0382,
+      "step": 4485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9560820541827175e-05,
+      "loss": 2.0384,
+      "step": 4486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9560612234734676e-05,
+      "loss": 2.0214,
+      "step": 4487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9560403879362536e-05,
+      "loss": 2.0431,
+      "step": 4488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9560195475711802e-05,
+      "loss": 2.0433,
+      "step": 4489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9559987023783523e-05,
+      "loss": 2.0358,
+      "step": 4490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9559778523578765e-05,
+      "loss": 2.0241,
+      "step": 4491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9559569975098562e-05,
+      "loss": 2.015,
+      "step": 4492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9559361378343984e-05,
+      "loss": 2.028,
+      "step": 4493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9559152733316077e-05,
+      "loss": 2.0298,
+      "step": 4494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9558944040015893e-05,
+      "loss": 2.0338,
+      "step": 4495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9558735298444492e-05,
+      "loss": 2.0278,
+      "step": 4496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9558526508602922e-05,
+      "loss": 2.024,
+      "step": 4497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9558317670492242e-05,
+      "loss": 2.0337,
+      "step": 4498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95581087841135e-05,
+      "loss": 2.0414,
+      "step": 4499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9557899849467762e-05,
+      "loss": 2.0296,
+      "step": 4500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955769086655607e-05,
+      "loss": 2.0233,
+      "step": 4501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9557481835379487e-05,
+      "loss": 2.0362,
+      "step": 4502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955727275593907e-05,
+      "loss": 2.043,
+      "step": 4503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955706362823587e-05,
+      "loss": 2.0363,
+      "step": 4504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9556854452270945e-05,
+      "loss": 2.0441,
+      "step": 4505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955664522804535e-05,
+      "loss": 2.0298,
+      "step": 4506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9556435955560145e-05,
+      "loss": 2.0183,
+      "step": 4507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9556226634816383e-05,
+      "loss": 2.0377,
+      "step": 4508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9556017265815124e-05,
+      "loss": 2.044,
+      "step": 4509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9555807848557426e-05,
+      "loss": 2.0155,
+      "step": 4510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9555598383044342e-05,
+      "loss": 2.0263,
+      "step": 4511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9555388869276933e-05,
+      "loss": 2.0323,
+      "step": 4512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9555179307256256e-05,
+      "loss": 2.0535,
+      "step": 4513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955496969698337e-05,
+      "loss": 2.0295,
+      "step": 4514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9554760038459336e-05,
+      "loss": 2.0308,
+      "step": 4515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9554550331685204e-05,
+      "loss": 2.0328,
+      "step": 4516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9554340576662043e-05,
+      "loss": 2.0287,
+      "step": 4517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955413077339091e-05,
+      "loss": 2.0333,
+      "step": 4518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553920921872856e-05,
+      "loss": 2.0328,
+      "step": 4519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955371102210895e-05,
+      "loss": 2.0464,
+      "step": 4520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553501074100253e-05,
+      "loss": 2.0146,
+      "step": 4521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553291077847817e-05,
+      "loss": 2.0394,
+      "step": 4522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553081033352707e-05,
+      "loss": 2.0261,
+      "step": 4523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9552870940615985e-05,
+      "loss": 2.0385,
+      "step": 4524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9552660799638713e-05,
+      "loss": 2.0648,
+      "step": 4525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9552450610421943e-05,
+      "loss": 2.0259,
+      "step": 4526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955224037296675e-05,
+      "loss": 2.0381,
+      "step": 4527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9552030087274186e-05,
+      "loss": 2.0256,
+      "step": 4528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9551819753345316e-05,
+      "loss": 2.0322,
+      "step": 4529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95516093711812e-05,
+      "loss": 2.018,
+      "step": 4530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95513989407829e-05,
+      "loss": 2.0346,
+      "step": 4531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955118846215149e-05,
+      "loss": 2.0255,
+      "step": 4532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9550977935288017e-05,
+      "loss": 2.0383,
+      "step": 4533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9550767360193553e-05,
+      "loss": 2.0158,
+      "step": 4534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955055673686916e-05,
+      "loss": 2.0395,
+      "step": 4535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9550346065315897e-05,
+      "loss": 2.0479,
+      "step": 4536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9550135345534836e-05,
+      "loss": 2.0301,
+      "step": 4537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549924577527033e-05,
+      "loss": 2.0297,
+      "step": 4538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549713761293558e-05,
+      "loss": 2.0188,
+      "step": 4539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549502896835474e-05,
+      "loss": 2.0329,
+      "step": 4540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549291984153847e-05,
+      "loss": 2.0293,
+      "step": 4541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954908102324974e-05,
+      "loss": 2.0171,
+      "step": 4542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9548870014124217e-05,
+      "loss": 2.0341,
+      "step": 4543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9548658956778347e-05,
+      "loss": 2.0552,
+      "step": 4544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9548447851213196e-05,
+      "loss": 2.061,
+      "step": 4545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9548236697429828e-05,
+      "loss": 2.0241,
+      "step": 4546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9548025495429308e-05,
+      "loss": 2.0472,
+      "step": 4547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9547814245212705e-05,
+      "loss": 2.0205,
+      "step": 4548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9547602946781086e-05,
+      "loss": 2.0179,
+      "step": 4549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9547391600135515e-05,
+      "loss": 2.0421,
+      "step": 4550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954718020527706e-05,
+      "loss": 2.0378,
+      "step": 4551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9546968762206796e-05,
+      "loss": 2.0436,
+      "step": 4552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954675727092578e-05,
+      "loss": 2.0138,
+      "step": 4553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9546545731435084e-05,
+      "loss": 2.0319,
+      "step": 4554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954633414373578e-05,
+      "loss": 2.0296,
+      "step": 4555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9546122507828933e-05,
+      "loss": 2.032,
+      "step": 4556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954591082371561e-05,
+      "loss": 2.0203,
+      "step": 4557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9545699091396883e-05,
+      "loss": 2.0443,
+      "step": 4558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954548731087382e-05,
+      "loss": 2.0198,
+      "step": 4559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954527548214749e-05,
+      "loss": 2.0215,
+      "step": 4560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9545063605218965e-05,
+      "loss": 2.0364,
+      "step": 4561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9544851680089315e-05,
+      "loss": 2.043,
+      "step": 4562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9544639706759605e-05,
+      "loss": 2.0336,
+      "step": 4563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954442768523091e-05,
+      "loss": 2.0383,
+      "step": 4564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95442156155043e-05,
+      "loss": 2.0179,
+      "step": 4565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9544003497580844e-05,
+      "loss": 2.045,
+      "step": 4566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954379133146162e-05,
+      "loss": 2.056,
+      "step": 4567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9543579117147692e-05,
+      "loss": 2.0259,
+      "step": 4568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954336685464013e-05,
+      "loss": 2.0256,
+      "step": 4569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9543154543940012e-05,
+      "loss": 2.0282,
+      "step": 4570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954294218504841e-05,
+      "loss": 2.025,
+      "step": 4571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954272977796639e-05,
+      "loss": 2.0412,
+      "step": 4572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954251732269503e-05,
+      "loss": 2.0409,
+      "step": 4573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9542304819235405e-05,
+      "loss": 2.0317,
+      "step": 4574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9542092267588583e-05,
+      "loss": 2.0433,
+      "step": 4575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9541879667755638e-05,
+      "loss": 2.0377,
+      "step": 4576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9541667019737646e-05,
+      "loss": 2.045,
+      "step": 4577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954145432353568e-05,
+      "loss": 2.0389,
+      "step": 4578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9541241579150813e-05,
+      "loss": 2.024,
+      "step": 4579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9541028786584117e-05,
+      "loss": 2.0237,
+      "step": 4580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9540815945836673e-05,
+      "loss": 2.0424,
+      "step": 4581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9540603056909554e-05,
+      "loss": 2.041,
+      "step": 4582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954039011980383e-05,
+      "loss": 2.0483,
+      "step": 4583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9540177134520582e-05,
+      "loss": 2.0346,
+      "step": 4584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9539964101060884e-05,
+      "loss": 2.0353,
+      "step": 4585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953975101942581e-05,
+      "loss": 2.0323,
+      "step": 4586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9539537889616437e-05,
+      "loss": 2.0417,
+      "step": 4587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953932471163384e-05,
+      "loss": 2.0351,
+      "step": 4588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95391114854791e-05,
+      "loss": 2.0295,
+      "step": 4589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9538898211153287e-05,
+      "loss": 2.0312,
+      "step": 4590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9538684888657484e-05,
+      "loss": 2.0457,
+      "step": 4591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9538471517992767e-05,
+      "loss": 2.0231,
+      "step": 4592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9538258099160206e-05,
+      "loss": 2.0479,
+      "step": 4593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9538044632160894e-05,
+      "loss": 2.0258,
+      "step": 4594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9537831116995897e-05,
+      "loss": 2.0322,
+      "step": 4595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9537617553666295e-05,
+      "loss": 2.0225,
+      "step": 4596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9537403942173167e-05,
+      "loss": 2.0415,
+      "step": 4597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9537190282517593e-05,
+      "loss": 2.0282,
+      "step": 4598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953697657470065e-05,
+      "loss": 2.0411,
+      "step": 4599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9536762818723424e-05,
+      "loss": 2.0404,
+      "step": 4600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9536549014586987e-05,
+      "loss": 2.0207,
+      "step": 4601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953633516229242e-05,
+      "loss": 2.0433,
+      "step": 4602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9536121261840803e-05,
+      "loss": 2.0204,
+      "step": 4603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9535907313233217e-05,
+      "loss": 2.0244,
+      "step": 4604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9535693316470744e-05,
+      "loss": 2.0335,
+      "step": 4605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9535479271554462e-05,
+      "loss": 2.0429,
+      "step": 4606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953526517848545e-05,
+      "loss": 2.0263,
+      "step": 4607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9535051037264796e-05,
+      "loss": 2.0455,
+      "step": 4608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9534836847893577e-05,
+      "loss": 2.0291,
+      "step": 4609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9534622610372875e-05,
+      "loss": 2.0592,
+      "step": 4610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9534408324703767e-05,
+      "loss": 2.0221,
+      "step": 4611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9534193990887345e-05,
+      "loss": 2.0213,
+      "step": 4612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533979608924683e-05,
+      "loss": 2.0448,
+      "step": 4613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953376517881687e-05,
+      "loss": 2.0456,
+      "step": 4614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533550700564984e-05,
+      "loss": 2.0225,
+      "step": 4615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533336174170113e-05,
+      "loss": 2.0253,
+      "step": 4616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533121599633334e-05,
+      "loss": 2.0084,
+      "step": 4617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9532906976955733e-05,
+      "loss": 2.0335,
+      "step": 4618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9532692306138393e-05,
+      "loss": 2.0313,
+      "step": 4619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9532477587182402e-05,
+      "loss": 2.0259,
+      "step": 4620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953226282008884e-05,
+      "loss": 2.0146,
+      "step": 4621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9532048004858798e-05,
+      "loss": 2.0101,
+      "step": 4622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9531833141493356e-05,
+      "loss": 2.0529,
+      "step": 4623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9531618229993596e-05,
+      "loss": 2.0267,
+      "step": 4624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9531403270360607e-05,
+      "loss": 2.0382,
+      "step": 4625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9531188262595475e-05,
+      "loss": 2.0077,
+      "step": 4626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953097320669928e-05,
+      "loss": 1.9912,
+      "step": 4627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953075810267312e-05,
+      "loss": 2.0242,
+      "step": 4628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953054295051807e-05,
+      "loss": 2.0533,
+      "step": 4629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9530327750235224e-05,
+      "loss": 2.0445,
+      "step": 4630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953011250182566e-05,
+      "loss": 2.0266,
+      "step": 4631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9529897205290472e-05,
+      "loss": 2.0232,
+      "step": 4632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952968186063075e-05,
+      "loss": 2.0528,
+      "step": 4633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952946646784757e-05,
+      "loss": 2.0432,
+      "step": 4634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952925102694203e-05,
+      "loss": 2.0404,
+      "step": 4635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9529035537915213e-05,
+      "loss": 2.0426,
+      "step": 4636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952882000076821e-05,
+      "loss": 2.0231,
+      "step": 4637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952860441550211e-05,
+      "loss": 2.0243,
+      "step": 4638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9528388782117997e-05,
+      "loss": 2.0448,
+      "step": 4639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9528173100616963e-05,
+      "loss": 2.0299,
+      "step": 4640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9527957371000098e-05,
+      "loss": 2.0408,
+      "step": 4641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952774159326849e-05,
+      "loss": 2.0447,
+      "step": 4642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9527525767423228e-05,
+      "loss": 2.0569,
+      "step": 4643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9527309893465403e-05,
+      "loss": 2.0367,
+      "step": 4644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9527093971396106e-05,
+      "loss": 2.023,
+      "step": 4645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9526878001216425e-05,
+      "loss": 2.0228,
+      "step": 4646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952666198292745e-05,
+      "loss": 2.015,
+      "step": 4647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952644591653028e-05,
+      "loss": 2.0396,
+      "step": 4648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9526229802025996e-05,
+      "loss": 2.0447,
+      "step": 4649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952601363941569e-05,
+      "loss": 2.0376,
+      "step": 4650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952579742870046e-05,
+      "loss": 2.048,
+      "step": 4651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9525581169881395e-05,
+      "loss": 2.0157,
+      "step": 4652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952536486295959e-05,
+      "loss": 2.0184,
+      "step": 4653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952514850793613e-05,
+      "loss": 2.0402,
+      "step": 4654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9524932104812112e-05,
+      "loss": 2.0287,
+      "step": 4655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9524715653588627e-05,
+      "loss": 2.028,
+      "step": 4656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9524499154266773e-05,
+      "loss": 2.0321,
+      "step": 4657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9524282606847637e-05,
+      "loss": 2.0333,
+      "step": 4658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9524066011332317e-05,
+      "loss": 2.0669,
+      "step": 4659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9523849367721904e-05,
+      "loss": 2.0225,
+      "step": 4660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9523632676017495e-05,
+      "loss": 2.0073,
+      "step": 4661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9523415936220178e-05,
+      "loss": 2.0216,
+      "step": 4662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9523199148331055e-05,
+      "loss": 2.0278,
+      "step": 4663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9522982312351216e-05,
+      "loss": 2.0539,
+      "step": 4664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952276542828176e-05,
+      "loss": 2.0514,
+      "step": 4665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9522548496123778e-05,
+      "loss": 2.0065,
+      "step": 4666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952233151587837e-05,
+      "loss": 2.0264,
+      "step": 4667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9522114487546624e-05,
+      "loss": 2.024,
+      "step": 4668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9521897411129645e-05,
+      "loss": 2.0468,
+      "step": 4669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952168028662852e-05,
+      "loss": 2.0326,
+      "step": 4670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9521463114044358e-05,
+      "loss": 2.0308,
+      "step": 4671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9521245893378245e-05,
+      "loss": 2.0278,
+      "step": 4672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952102862463128e-05,
+      "loss": 2.0319,
+      "step": 4673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952081130780456e-05,
+      "loss": 2.035,
+      "step": 4674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952059394289919e-05,
+      "loss": 2.0161,
+      "step": 4675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9520376529916258e-05,
+      "loss": 2.0213,
+      "step": 4676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952015906885686e-05,
+      "loss": 2.0373,
+      "step": 4677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9519941559722107e-05,
+      "loss": 2.0245,
+      "step": 4678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9519724002513087e-05,
+      "loss": 2.0311,
+      "step": 4679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.95195063972309e-05,
+      "loss": 2.0178,
+      "step": 4680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9519288743876652e-05,
+      "loss": 2.0386,
+      "step": 4681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9519071042451433e-05,
+      "loss": 2.0225,
+      "step": 4682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9518853292956345e-05,
+      "loss": 2.0337,
+      "step": 4683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951863549539249e-05,
+      "loss": 2.0421,
+      "step": 4684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9518417649760962e-05,
+      "loss": 2.0254,
+      "step": 4685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951819975606287e-05,
+      "loss": 2.0459,
+      "step": 4686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9517981814299308e-05,
+      "loss": 2.0091,
+      "step": 4687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951776382447138e-05,
+      "loss": 2.0405,
+      "step": 4688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9517545786580186e-05,
+      "loss": 2.0409,
+      "step": 4689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9517327700626823e-05,
+      "loss": 2.0357,
+      "step": 4690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9517109566612398e-05,
+      "loss": 2.0374,
+      "step": 4691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9516891384538008e-05,
+      "loss": 2.0368,
+      "step": 4692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9516673154404757e-05,
+      "loss": 2.0222,
+      "step": 4693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951645487621375e-05,
+      "loss": 2.0349,
+      "step": 4694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951623654996608e-05,
+      "loss": 2.0355,
+      "step": 4695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951601817566286e-05,
+      "loss": 2.0392,
+      "step": 4696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951579975330519e-05,
+      "loss": 2.0337,
+      "step": 4697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951558128289417e-05,
+      "loss": 2.0273,
+      "step": 4698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9515362764430903e-05,
+      "loss": 2.0356,
+      "step": 4699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9515144197916495e-05,
+      "loss": 2.02,
+      "step": 4700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951492558335205e-05,
+      "loss": 2.0405,
+      "step": 4701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951470692073867e-05,
+      "loss": 2.0304,
+      "step": 4702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951448821007746e-05,
+      "loss": 2.044,
+      "step": 4703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951426945136952e-05,
+      "loss": 2.0175,
+      "step": 4704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9514050644615966e-05,
+      "loss": 2.022,
+      "step": 4705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9513831789817895e-05,
+      "loss": 2.0461,
+      "step": 4706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951361288697641e-05,
+      "loss": 2.0341,
+      "step": 4707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9513393936092623e-05,
+      "loss": 2.015,
+      "step": 4708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9513174937167634e-05,
+      "loss": 2.0357,
+      "step": 4709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9512955890202555e-05,
+      "loss": 2.0194,
+      "step": 4710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9512736795198487e-05,
+      "loss": 2.0451,
+      "step": 4711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9512517652156534e-05,
+      "loss": 2.0281,
+      "step": 4712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951229846107781e-05,
+      "loss": 2.0463,
+      "step": 4713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9512079221963417e-05,
+      "loss": 2.0221,
+      "step": 4714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9511859934814465e-05,
+      "loss": 2.0694,
+      "step": 4715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951164059963206e-05,
+      "loss": 2.0176,
+      "step": 4716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951142121641731e-05,
+      "loss": 2.024,
+      "step": 4717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951120178517132e-05,
+      "loss": 2.0273,
+      "step": 4718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.95109823058952e-05,
+      "loss": 2.0093,
+      "step": 4719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9510762778590058e-05,
+      "loss": 2.0478,
+      "step": 4720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9510543203257008e-05,
+      "loss": 2.0331,
+      "step": 4721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951032357989715e-05,
+      "loss": 2.0249,
+      "step": 4722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.95101039085116e-05,
+      "loss": 2.0112,
+      "step": 4723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9509884189101465e-05,
+      "loss": 2.0271,
+      "step": 4724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9509664421667853e-05,
+      "loss": 2.0297,
+      "step": 4725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9509444606211875e-05,
+      "loss": 2.0331,
+      "step": 4726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950922474273464e-05,
+      "loss": 2.035,
+      "step": 4727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950900483123726e-05,
+      "loss": 2.0308,
+      "step": 4728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9508784871720842e-05,
+      "loss": 2.0369,
+      "step": 4729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9508564864186503e-05,
+      "loss": 2.027,
+      "step": 4730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950834480863535e-05,
+      "loss": 2.0181,
+      "step": 4731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9508124705068493e-05,
+      "loss": 2.044,
+      "step": 4732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9507904553487046e-05,
+      "loss": 2.0272,
+      "step": 4733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950768435389212e-05,
+      "loss": 2.0444,
+      "step": 4734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950746410628483e-05,
+      "loss": 2.038,
+      "step": 4735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9507243810666282e-05,
+      "loss": 2.0377,
+      "step": 4736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9507023467037593e-05,
+      "loss": 2.0368,
+      "step": 4737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950680307539987e-05,
+      "loss": 2.0499,
+      "step": 4738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9506582635754237e-05,
+      "loss": 2.0388,
+      "step": 4739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9506362148101796e-05,
+      "loss": 2.0317,
+      "step": 4740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9506141612443667e-05,
+      "loss": 2.0493,
+      "step": 4741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950592102878096e-05,
+      "loss": 2.0246,
+      "step": 4742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950570039711479e-05,
+      "loss": 2.0437,
+      "step": 4743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9505479717446275e-05,
+      "loss": 2.0414,
+      "step": 4744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9505258989776525e-05,
+      "loss": 2.0369,
+      "step": 4745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9505038214106654e-05,
+      "loss": 2.0318,
+      "step": 4746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9504817390437777e-05,
+      "loss": 2.0376,
+      "step": 4747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9504596518771012e-05,
+      "loss": 2.0424,
+      "step": 4748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9504375599107475e-05,
+      "loss": 2.0067,
+      "step": 4749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950415463144828e-05,
+      "loss": 2.0393,
+      "step": 4750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9503933615794537e-05,
+      "loss": 2.0427,
+      "step": 4751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9503712552147375e-05,
+      "loss": 2.0304,
+      "step": 4752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.95034914405079e-05,
+      "loss": 2.0262,
+      "step": 4753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9503270280877228e-05,
+      "loss": 2.0366,
+      "step": 4754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950304907325648e-05,
+      "loss": 2.0368,
+      "step": 4755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9502827817646775e-05,
+      "loss": 2.0278,
+      "step": 4756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950260651404923e-05,
+      "loss": 2.0297,
+      "step": 4757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9502385162464955e-05,
+      "loss": 2.0411,
+      "step": 4758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9502163762895077e-05,
+      "loss": 2.0448,
+      "step": 4759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950194231534071e-05,
+      "loss": 2.0128,
+      "step": 4760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950172081980297e-05,
+      "loss": 2.025,
+      "step": 4761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950149927628298e-05,
+      "loss": 2.0284,
+      "step": 4762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9501277684781855e-05,
+      "loss": 2.0242,
+      "step": 4763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9501056045300716e-05,
+      "loss": 2.0285,
+      "step": 4764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9500834357840684e-05,
+      "loss": 2.0333,
+      "step": 4765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9500612622402876e-05,
+      "loss": 2.0342,
+      "step": 4766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9500390838988408e-05,
+      "loss": 2.0164,
+      "step": 4767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950016900759841e-05,
+      "loss": 2.0418,
+      "step": 4768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9499947128233993e-05,
+      "loss": 2.0225,
+      "step": 4769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9499725200896277e-05,
+      "loss": 2.0447,
+      "step": 4770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9499503225586392e-05,
+      "loss": 2.0268,
+      "step": 4771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949928120230545e-05,
+      "loss": 2.0366,
+      "step": 4772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949905913105458e-05,
+      "loss": 2.0259,
+      "step": 4773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9498837011834895e-05,
+      "loss": 2.04,
+      "step": 4774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9498614844647525e-05,
+      "loss": 2.0241,
+      "step": 4775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9498392629493583e-05,
+      "loss": 2.0181,
+      "step": 4776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9498170366374198e-05,
+      "loss": 2.037,
+      "step": 4777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949794805529049e-05,
+      "loss": 2.044,
+      "step": 4778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949772569624358e-05,
+      "loss": 2.0333,
+      "step": 4779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9497503289234595e-05,
+      "loss": 2.0371,
+      "step": 4780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9497280834264653e-05,
+      "loss": 2.0386,
+      "step": 4781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9497058331334883e-05,
+      "loss": 2.0406,
+      "step": 4782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9496835780446405e-05,
+      "loss": 2.0275,
+      "step": 4783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9496613181600345e-05,
+      "loss": 2.0297,
+      "step": 4784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9496390534797824e-05,
+      "loss": 2.0272,
+      "step": 4785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9496167840039966e-05,
+      "loss": 2.0141,
+      "step": 4786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94959450973279e-05,
+      "loss": 2.0314,
+      "step": 4787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9495722306662746e-05,
+      "loss": 2.0447,
+      "step": 4788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9495499468045637e-05,
+      "loss": 2.0296,
+      "step": 4789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9495276581477687e-05,
+      "loss": 2.0469,
+      "step": 4790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949505364696003e-05,
+      "loss": 2.0141,
+      "step": 4791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949483066449379e-05,
+      "loss": 2.0467,
+      "step": 4792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9494607634080086e-05,
+      "loss": 2.0402,
+      "step": 4793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949438455572006e-05,
+      "loss": 2.0214,
+      "step": 4794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9494161429414822e-05,
+      "loss": 2.0295,
+      "step": 4795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9493938255165508e-05,
+      "loss": 2.0378,
+      "step": 4796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949371503297324e-05,
+      "loss": 2.0352,
+      "step": 4797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9493491762839154e-05,
+      "loss": 2.0402,
+      "step": 4798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9493268444764367e-05,
+      "loss": 2.0453,
+      "step": 4799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949304507875001e-05,
+      "loss": 2.0346,
+      "step": 4800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9492821664797215e-05,
+      "loss": 2.0345,
+      "step": 4801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9492598202907107e-05,
+      "loss": 2.0304,
+      "step": 4802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9492374693080815e-05,
+      "loss": 2.0367,
+      "step": 4803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9492151135319465e-05,
+      "loss": 2.0289,
+      "step": 4804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949192752962419e-05,
+      "loss": 2.0328,
+      "step": 4805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9491703875996117e-05,
+      "loss": 2.0729,
+      "step": 4806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9491480174436376e-05,
+      "loss": 2.0348,
+      "step": 4807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9491256424946096e-05,
+      "loss": 2.0209,
+      "step": 4808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9491032627526407e-05,
+      "loss": 2.0095,
+      "step": 4809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949080878217844e-05,
+      "loss": 2.0296,
+      "step": 4810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9490584888903326e-05,
+      "loss": 2.012,
+      "step": 4811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9490360947702196e-05,
+      "loss": 2.0065,
+      "step": 4812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949013695857618e-05,
+      "loss": 2.0378,
+      "step": 4813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9489912921526402e-05,
+      "loss": 2.0365,
+      "step": 4814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948968883655401e-05,
+      "loss": 2.0307,
+      "step": 4815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948946470366012e-05,
+      "loss": 2.0251,
+      "step": 4816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9489240522845868e-05,
+      "loss": 2.0592,
+      "step": 4817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9489016294112387e-05,
+      "loss": 2.035,
+      "step": 4818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9488792017460812e-05,
+      "loss": 2.0386,
+      "step": 4819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9488567692892275e-05,
+      "loss": 2.0389,
+      "step": 4820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9488343320407904e-05,
+      "loss": 2.0448,
+      "step": 4821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9488118900008837e-05,
+      "loss": 2.0312,
+      "step": 4822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9487894431696207e-05,
+      "loss": 2.036,
+      "step": 4823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9487669915471144e-05,
+      "loss": 2.0391,
+      "step": 4824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9487445351334785e-05,
+      "loss": 2.0566,
+      "step": 4825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948722073928826e-05,
+      "loss": 2.0263,
+      "step": 4826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948699607933271e-05,
+      "loss": 2.0289,
+      "step": 4827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9486771371469266e-05,
+      "loss": 2.0337,
+      "step": 4828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9486546615699057e-05,
+      "loss": 2.0422,
+      "step": 4829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9486321812023227e-05,
+      "loss": 2.0216,
+      "step": 4830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9486096960442906e-05,
+      "loss": 2.0298,
+      "step": 4831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9485872060959233e-05,
+      "loss": 2.0167,
+      "step": 4832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948564711357334e-05,
+      "loss": 2.0434,
+      "step": 4833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9485422118286367e-05,
+      "loss": 2.0334,
+      "step": 4834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9485197075099445e-05,
+      "loss": 2.0097,
+      "step": 4835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9484971984013714e-05,
+      "loss": 2.0446,
+      "step": 4836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948474684503031e-05,
+      "loss": 2.0248,
+      "step": 4837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948452165815037e-05,
+      "loss": 2.0483,
+      "step": 4838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9484296423375034e-05,
+      "loss": 2.0289,
+      "step": 4839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9484071140705433e-05,
+      "loss": 2.0359,
+      "step": 4840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948384581014271e-05,
+      "loss": 2.0447,
+      "step": 4841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9483620431687996e-05,
+      "loss": 2.0459,
+      "step": 4842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948339500534244e-05,
+      "loss": 2.0195,
+      "step": 4843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9483169531107173e-05,
+      "loss": 2.0328,
+      "step": 4844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9482944008983336e-05,
+      "loss": 2.0244,
+      "step": 4845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9482718438972065e-05,
+      "loss": 2.0326,
+      "step": 4846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94824928210745e-05,
+      "loss": 2.0404,
+      "step": 4847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9482267155291783e-05,
+      "loss": 2.0497,
+      "step": 4848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9482041441625052e-05,
+      "loss": 2.0311,
+      "step": 4849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9481815680075447e-05,
+      "loss": 2.0426,
+      "step": 4850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9481589870644106e-05,
+      "loss": 2.0345,
+      "step": 4851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9481364013332174e-05,
+      "loss": 2.0376,
+      "step": 4852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948113810814079e-05,
+      "loss": 2.0297,
+      "step": 4853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948091215507109e-05,
+      "loss": 2.0502,
+      "step": 4854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9480686154124217e-05,
+      "loss": 2.0371,
+      "step": 4855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9480460105301318e-05,
+      "loss": 2.042,
+      "step": 4856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9480234008603527e-05,
+      "loss": 2.0285,
+      "step": 4857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9480007864031993e-05,
+      "loss": 2.025,
+      "step": 4858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947978167158785e-05,
+      "loss": 2.0556,
+      "step": 4859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9479555431272245e-05,
+      "loss": 2.0253,
+      "step": 4860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9479329143086322e-05,
+      "loss": 2.0333,
+      "step": 4861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9479102807031217e-05,
+      "loss": 2.0113,
+      "step": 4862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947887642310808e-05,
+      "loss": 2.0225,
+      "step": 4863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9478649991318055e-05,
+      "loss": 2.0262,
+      "step": 4864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947842351166228e-05,
+      "loss": 2.0407,
+      "step": 4865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9478196984141898e-05,
+      "loss": 2.0228,
+      "step": 4866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9477970408758056e-05,
+      "loss": 2.0522,
+      "step": 4867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94777437855119e-05,
+      "loss": 2.0565,
+      "step": 4868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947751711440457e-05,
+      "loss": 2.0331,
+      "step": 4869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9477290395437216e-05,
+      "loss": 2.0337,
+      "step": 4870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947706362861098e-05,
+      "loss": 2.0439,
+      "step": 4871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9476836813927005e-05,
+      "loss": 2.035,
+      "step": 4872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9476609951386438e-05,
+      "loss": 2.047,
+      "step": 4873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9476383040990424e-05,
+      "loss": 2.0468,
+      "step": 4874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9476156082740114e-05,
+      "loss": 2.0244,
+      "step": 4875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9475929076636645e-05,
+      "loss": 2.036,
+      "step": 4876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9475702022681173e-05,
+      "loss": 2.0513,
+      "step": 4877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947547492087484e-05,
+      "loss": 2.0225,
+      "step": 4878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9475247771218787e-05,
+      "loss": 2.0387,
+      "step": 4879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9475020573714172e-05,
+      "loss": 2.0434,
+      "step": 4880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9474793328362133e-05,
+      "loss": 2.046,
+      "step": 4881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9474566035163826e-05,
+      "loss": 2.0529,
+      "step": 4882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947433869412039e-05,
+      "loss": 2.0189,
+      "step": 4883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9474111305232982e-05,
+      "loss": 2.0462,
+      "step": 4884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9473883868502742e-05,
+      "loss": 2.0237,
+      "step": 4885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9473656383930825e-05,
+      "loss": 2.047,
+      "step": 4886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9473428851518375e-05,
+      "loss": 2.0353,
+      "step": 4887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9473201271266548e-05,
+      "loss": 2.0267,
+      "step": 4888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9472973643176482e-05,
+      "loss": 2.0204,
+      "step": 4889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9472745967249332e-05,
+      "loss": 2.0325,
+      "step": 4890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9472518243486254e-05,
+      "loss": 2.0257,
+      "step": 4891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947229047188839e-05,
+      "loss": 2.0353,
+      "step": 4892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947206265245689e-05,
+      "loss": 2.0229,
+      "step": 4893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9471834785192913e-05,
+      "loss": 2.0306,
+      "step": 4894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94716068700976e-05,
+      "loss": 2.0556,
+      "step": 4895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9471378907172108e-05,
+      "loss": 2.0208,
+      "step": 4896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9471150896417584e-05,
+      "loss": 2.0197,
+      "step": 4897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9470922837835183e-05,
+      "loss": 2.0442,
+      "step": 4898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9470694731426052e-05,
+      "loss": 2.0416,
+      "step": 4899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9470466577191347e-05,
+      "loss": 2.0315,
+      "step": 4900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947023837513222e-05,
+      "loss": 2.0331,
+      "step": 4901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9470010125249822e-05,
+      "loss": 2.0346,
+      "step": 4902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946978182754531e-05,
+      "loss": 2.0402,
+      "step": 4903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9469553482019825e-05,
+      "loss": 2.039,
+      "step": 4904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9469325088674536e-05,
+      "loss": 2.0159,
+      "step": 4905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946909664751058e-05,
+      "loss": 2.0212,
+      "step": 4906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9468868158529126e-05,
+      "loss": 2.0395,
+      "step": 4907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946863962173132e-05,
+      "loss": 2.05,
+      "step": 4908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9468411037118317e-05,
+      "loss": 2.0146,
+      "step": 4909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946818240469127e-05,
+      "loss": 2.033,
+      "step": 4910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9467953724451334e-05,
+      "loss": 2.0211,
+      "step": 4911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9467724996399664e-05,
+      "loss": 2.031,
+      "step": 4912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9467496220537417e-05,
+      "loss": 2.0252,
+      "step": 4913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9467267396865746e-05,
+      "loss": 2.0414,
+      "step": 4914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946703852538581e-05,
+      "loss": 2.0282,
+      "step": 4915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946680960609876e-05,
+      "loss": 2.0164,
+      "step": 4916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9466580639005755e-05,
+      "loss": 2.0131,
+      "step": 4917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946635162410795e-05,
+      "loss": 2.028,
+      "step": 4918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9466122561406502e-05,
+      "loss": 2.039,
+      "step": 4919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946589345090257e-05,
+      "loss": 2.0363,
+      "step": 4920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9465664292597307e-05,
+      "loss": 2.0362,
+      "step": 4921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946543508649187e-05,
+      "loss": 2.0395,
+      "step": 4922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946520583258742e-05,
+      "loss": 2.0279,
+      "step": 4923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9464976530885115e-05,
+      "loss": 2.0666,
+      "step": 4924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9464747181386107e-05,
+      "loss": 2.0622,
+      "step": 4925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946451778409156e-05,
+      "loss": 2.0423,
+      "step": 4926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9464288339002633e-05,
+      "loss": 2.0418,
+      "step": 4927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9464058846120477e-05,
+      "loss": 2.0578,
+      "step": 4928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946382930544626e-05,
+      "loss": 2.0316,
+      "step": 4929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9463599716981137e-05,
+      "loss": 2.0294,
+      "step": 4930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9463370080726267e-05,
+      "loss": 2.0427,
+      "step": 4931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9463140396682808e-05,
+      "loss": 2.0283,
+      "step": 4932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9462910664851926e-05,
+      "loss": 2.046,
+      "step": 4933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9462680885234778e-05,
+      "loss": 2.048,
+      "step": 4934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946245105783252e-05,
+      "loss": 2.0297,
+      "step": 4935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9462221182646316e-05,
+      "loss": 2.0384,
+      "step": 4936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946199125967733e-05,
+      "loss": 2.0411,
+      "step": 4937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946176128892672e-05,
+      "loss": 2.0397,
+      "step": 4938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9461531270395646e-05,
+      "loss": 2.0445,
+      "step": 4939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9461301204085273e-05,
+      "loss": 2.0228,
+      "step": 4940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946107108999676e-05,
+      "loss": 2.0258,
+      "step": 4941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9460840928131268e-05,
+      "loss": 2.0444,
+      "step": 4942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9460610718489963e-05,
+      "loss": 2.0408,
+      "step": 4943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9460380461074005e-05,
+      "loss": 2.0353,
+      "step": 4944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9460150155884556e-05,
+      "loss": 2.0304,
+      "step": 4945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9459919802922783e-05,
+      "loss": 2.0242,
+      "step": 4946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9459689402189847e-05,
+      "loss": 2.0458,
+      "step": 4947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945945895368691e-05,
+      "loss": 2.0378,
+      "step": 4948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9459228457415137e-05,
+      "loss": 2.0474,
+      "step": 4949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945899791337569e-05,
+      "loss": 2.0257,
+      "step": 4950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9458767321569737e-05,
+      "loss": 2.0573,
+      "step": 4951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9458536681998442e-05,
+      "loss": 2.0296,
+      "step": 4952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9458305994662963e-05,
+      "loss": 2.0387,
+      "step": 4953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9458075259564476e-05,
+      "loss": 2.0231,
+      "step": 4954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945784447670414e-05,
+      "loss": 2.0477,
+      "step": 4955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945761364608312e-05,
+      "loss": 2.0443,
+      "step": 4956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945738276770258e-05,
+      "loss": 2.0334,
+      "step": 4957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945715184156369e-05,
+      "loss": 2.0137,
+      "step": 4958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9456920867667613e-05,
+      "loss": 2.0103,
+      "step": 4959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9456689846015518e-05,
+      "loss": 2.0368,
+      "step": 4960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945645877660857e-05,
+      "loss": 2.0471,
+      "step": 4961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945622765944794e-05,
+      "loss": 2.0369,
+      "step": 4962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9455996494534783e-05,
+      "loss": 2.0505,
+      "step": 4963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9455765281870283e-05,
+      "loss": 2.0396,
+      "step": 4964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9455534021455595e-05,
+      "loss": 2.0194,
+      "step": 4965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945530271329189e-05,
+      "loss": 2.0066,
+      "step": 4966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9455071357380345e-05,
+      "loss": 2.0198,
+      "step": 4967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9454839953722113e-05,
+      "loss": 2.037,
+      "step": 4968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945460850231837e-05,
+      "loss": 2.0356,
+      "step": 4969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945437700317029e-05,
+      "loss": 2.0461,
+      "step": 4970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9454145456279038e-05,
+      "loss": 2.0223,
+      "step": 4971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9453913861645776e-05,
+      "loss": 2.0243,
+      "step": 4972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9453682219271687e-05,
+      "loss": 2.0521,
+      "step": 4973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9453450529157927e-05,
+      "loss": 2.038,
+      "step": 4974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9453218791305675e-05,
+      "loss": 2.0242,
+      "step": 4975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94529870057161e-05,
+      "loss": 2.0384,
+      "step": 4976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945275517239037e-05,
+      "loss": 2.0466,
+      "step": 4977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9452523291329657e-05,
+      "loss": 2.0405,
+      "step": 4978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945229136253513e-05,
+      "loss": 2.0363,
+      "step": 4979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9452059386007965e-05,
+      "loss": 2.0346,
+      "step": 4980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9451827361749333e-05,
+      "loss": 2.0447,
+      "step": 4981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9451595289760398e-05,
+      "loss": 2.0328,
+      "step": 4982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945136317004234e-05,
+      "loss": 2.0303,
+      "step": 4983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9451131002596328e-05,
+      "loss": 2.0391,
+      "step": 4984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9450898787423534e-05,
+      "loss": 2.0327,
+      "step": 4985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9450666524525133e-05,
+      "loss": 2.0491,
+      "step": 4986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9450434213902293e-05,
+      "loss": 2.0357,
+      "step": 4987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9450201855556194e-05,
+      "loss": 2.0411,
+      "step": 4988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9449969449488005e-05,
+      "loss": 2.0122,
+      "step": 4989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9449736995698898e-05,
+      "loss": 2.0247,
+      "step": 4990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944950449419005e-05,
+      "loss": 2.0187,
+      "step": 4991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944927194496264e-05,
+      "loss": 2.0175,
+      "step": 4992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944903934801783e-05,
+      "loss": 2.0361,
+      "step": 4993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9448806703356805e-05,
+      "loss": 2.0278,
+      "step": 4994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9448574010980733e-05,
+      "loss": 2.0433,
+      "step": 4995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9448341270890796e-05,
+      "loss": 2.033,
+      "step": 4996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9448108483088165e-05,
+      "loss": 2.0418,
+      "step": 4997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9447875647574014e-05,
+      "loss": 2.054,
+      "step": 4998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9447642764349523e-05,
+      "loss": 2.0269,
+      "step": 4999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9447409833415863e-05,
+      "loss": 2.0273,
+      "step": 5000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9447176854774215e-05,
+      "loss": 2.0332,
+      "step": 5001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9446943828425757e-05,
+      "loss": 2.0362,
+      "step": 5002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9446710754371655e-05,
+      "loss": 2.0254,
+      "step": 5003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94464776326131e-05,
+      "loss": 2.0447,
+      "step": 5004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944624446315126e-05,
+      "loss": 2.027,
+      "step": 5005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9446011245987316e-05,
+      "loss": 2.0492,
+      "step": 5006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9445777981122444e-05,
+      "loss": 2.0477,
+      "step": 5007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9445544668557822e-05,
+      "loss": 2.0449,
+      "step": 5008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944531130829463e-05,
+      "loss": 2.0265,
+      "step": 5009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9445077900334046e-05,
+      "loss": 2.0392,
+      "step": 5010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9444844444677248e-05,
+      "loss": 2.0189,
+      "step": 5011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944461094132541e-05,
+      "loss": 2.036,
+      "step": 5012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944437739027972e-05,
+      "loss": 2.0418,
+      "step": 5013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9444143791541356e-05,
+      "loss": 2.0485,
+      "step": 5014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9443910145111492e-05,
+      "loss": 2.0137,
+      "step": 5015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9443676450991315e-05,
+      "loss": 2.0492,
+      "step": 5016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9443442709181997e-05,
+      "loss": 2.0342,
+      "step": 5017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944320891968472e-05,
+      "loss": 2.0448,
+      "step": 5018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9442975082500672e-05,
+      "loss": 2.0415,
+      "step": 5019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9442741197631026e-05,
+      "loss": 2.0361,
+      "step": 5020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944250726507697e-05,
+      "loss": 2.0253,
+      "step": 5021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944227328483968e-05,
+      "loss": 2.0289,
+      "step": 5022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9442039256920337e-05,
+      "loss": 2.044,
+      "step": 5023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944180518132013e-05,
+      "loss": 2.0266,
+      "step": 5024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9441571058040227e-05,
+      "loss": 2.0199,
+      "step": 5025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9441336887081825e-05,
+      "loss": 2.0151,
+      "step": 5026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9441102668446098e-05,
+      "loss": 2.0408,
+      "step": 5027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9440868402134233e-05,
+      "loss": 2.0302,
+      "step": 5028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9440634088147413e-05,
+      "loss": 2.0108,
+      "step": 5029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9440399726486818e-05,
+      "loss": 2.0236,
+      "step": 5030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9440165317153633e-05,
+      "loss": 2.0309,
+      "step": 5031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943993086014904e-05,
+      "loss": 2.0428,
+      "step": 5032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9439696355474227e-05,
+      "loss": 2.0382,
+      "step": 5033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9439461803130375e-05,
+      "loss": 2.0315,
+      "step": 5034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9439227203118673e-05,
+      "loss": 2.0261,
+      "step": 5035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9438992555440296e-05,
+      "loss": 2.0486,
+      "step": 5036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943875786009644e-05,
+      "loss": 2.0266,
+      "step": 5037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9438523117088285e-05,
+      "loss": 2.0323,
+      "step": 5038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9438288326417015e-05,
+      "loss": 2.0339,
+      "step": 5039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9438053488083817e-05,
+      "loss": 2.0258,
+      "step": 5040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943781860208988e-05,
+      "loss": 2.035,
+      "step": 5041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943758366843638e-05,
+      "loss": 2.0053,
+      "step": 5042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943734868712452e-05,
+      "loss": 2.032,
+      "step": 5043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9437113658155472e-05,
+      "loss": 2.0193,
+      "step": 5044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9436878581530433e-05,
+      "loss": 2.0285,
+      "step": 5045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943664345725058e-05,
+      "loss": 2.0242,
+      "step": 5046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9436408285317107e-05,
+      "loss": 2.031,
+      "step": 5047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94361730657312e-05,
+      "loss": 2.0296,
+      "step": 5048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9435937798494052e-05,
+      "loss": 2.0503,
+      "step": 5049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9435702483606844e-05,
+      "loss": 2.0511,
+      "step": 5050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9435467121070763e-05,
+      "loss": 2.0444,
+      "step": 5051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9435231710887005e-05,
+      "loss": 2.0339,
+      "step": 5052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9434996253056753e-05,
+      "loss": 2.0295,
+      "step": 5053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94347607475812e-05,
+      "loss": 2.0336,
+      "step": 5054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943452519446153e-05,
+      "loss": 2.0452,
+      "step": 5055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9434289593698933e-05,
+      "loss": 2.0288,
+      "step": 5056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9434053945294606e-05,
+      "loss": 2.0381,
+      "step": 5057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9433818249249734e-05,
+      "loss": 2.0343,
+      "step": 5058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9433582505565506e-05,
+      "loss": 2.0378,
+      "step": 5059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9433346714243116e-05,
+      "loss": 2.0261,
+      "step": 5060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9433110875283755e-05,
+      "loss": 2.0269,
+      "step": 5061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9432874988688607e-05,
+      "loss": 2.0318,
+      "step": 5062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943263905445887e-05,
+      "loss": 2.0365,
+      "step": 5063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943240307259573e-05,
+      "loss": 2.0275,
+      "step": 5064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9432167043100387e-05,
+      "loss": 2.0352,
+      "step": 5065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9431930965974027e-05,
+      "loss": 2.0209,
+      "step": 5066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943169484121784e-05,
+      "loss": 2.0485,
+      "step": 5067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9431458668833025e-05,
+      "loss": 2.0425,
+      "step": 5068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9431222448820768e-05,
+      "loss": 2.0485,
+      "step": 5069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9430986181182264e-05,
+      "loss": 2.0267,
+      "step": 5070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9430749865918713e-05,
+      "loss": 2.0135,
+      "step": 5071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9430513503031297e-05,
+      "loss": 2.0398,
+      "step": 5072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943027709252122e-05,
+      "loss": 2.0366,
+      "step": 5073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9430040634389667e-05,
+      "loss": 2.0341,
+      "step": 5074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9429804128637836e-05,
+      "loss": 2.0595,
+      "step": 5075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9429567575266923e-05,
+      "loss": 2.0374,
+      "step": 5076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942933097427812e-05,
+      "loss": 2.0444,
+      "step": 5077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9429094325672625e-05,
+      "loss": 2.0328,
+      "step": 5078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942885762945163e-05,
+      "loss": 2.0339,
+      "step": 5079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942862088561633e-05,
+      "loss": 2.0597,
+      "step": 5080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9428384094167917e-05,
+      "loss": 2.041,
+      "step": 5081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9428147255107594e-05,
+      "loss": 2.0292,
+      "step": 5082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9427910368436555e-05,
+      "loss": 2.0327,
+      "step": 5083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9427673434155997e-05,
+      "loss": 2.0462,
+      "step": 5084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9427436452267115e-05,
+      "loss": 2.0248,
+      "step": 5085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9427199422771105e-05,
+      "loss": 2.0245,
+      "step": 5086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9426962345669163e-05,
+      "loss": 2.029,
+      "step": 5087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942672522096249e-05,
+      "loss": 2.0314,
+      "step": 5088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942648804865228e-05,
+      "loss": 2.0175,
+      "step": 5089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9426250828739733e-05,
+      "loss": 2.039,
+      "step": 5090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9426013561226046e-05,
+      "loss": 2.0198,
+      "step": 5091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9425776246112417e-05,
+      "loss": 2.0406,
+      "step": 5092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9425538883400042e-05,
+      "loss": 2.0453,
+      "step": 5093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9425301473090125e-05,
+      "loss": 2.0293,
+      "step": 5094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942506401518386e-05,
+      "loss": 2.0275,
+      "step": 5095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942482650968245e-05,
+      "loss": 2.0356,
+      "step": 5096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9424588956587092e-05,
+      "loss": 2.0348,
+      "step": 5097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9424351355898988e-05,
+      "loss": 2.0369,
+      "step": 5098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9424113707619332e-05,
+      "loss": 2.0277,
+      "step": 5099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942387601174933e-05,
+      "loss": 2.0284,
+      "step": 5100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942363826829018e-05,
+      "loss": 2.0394,
+      "step": 5101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9423400477243083e-05,
+      "loss": 2.0361,
+      "step": 5102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9423162638609242e-05,
+      "loss": 2.0225,
+      "step": 5103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9422924752389852e-05,
+      "loss": 2.0345,
+      "step": 5104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942268681858612e-05,
+      "loss": 2.0368,
+      "step": 5105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9422448837199247e-05,
+      "loss": 2.0312,
+      "step": 5106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942221080823043e-05,
+      "loss": 2.0395,
+      "step": 5107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9421972731680878e-05,
+      "loss": 2.031,
+      "step": 5108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9421734607551787e-05,
+      "loss": 2.0382,
+      "step": 5109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9421496435844363e-05,
+      "loss": 2.0357,
+      "step": 5110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942125821655981e-05,
+      "loss": 2.0312,
+      "step": 5111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9421019949699325e-05,
+      "loss": 2.0202,
+      "step": 5112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942078163526412e-05,
+      "loss": 2.0177,
+      "step": 5113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942054327325539e-05,
+      "loss": 2.0327,
+      "step": 5114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942030486367434e-05,
+      "loss": 2.0498,
+      "step": 5115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942006640652218e-05,
+      "loss": 2.0536,
+      "step": 5116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941982790180011e-05,
+      "loss": 2.0115,
+      "step": 5117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9419589349509334e-05,
+      "loss": 2.0263,
+      "step": 5118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941935074965106e-05,
+      "loss": 2.0456,
+      "step": 5119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9419112102226487e-05,
+      "loss": 2.0251,
+      "step": 5120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9418873407236823e-05,
+      "loss": 2.0403,
+      "step": 5121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9418634664683275e-05,
+      "loss": 2.0303,
+      "step": 5122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9418395874567047e-05,
+      "loss": 2.0553,
+      "step": 5123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9418157036889344e-05,
+      "loss": 2.0342,
+      "step": 5124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9417918151651375e-05,
+      "loss": 2.0272,
+      "step": 5125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9417679218854345e-05,
+      "loss": 2.0476,
+      "step": 5126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941744023849946e-05,
+      "loss": 2.0361,
+      "step": 5127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9417201210587926e-05,
+      "loss": 2.0374,
+      "step": 5128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9416962135120952e-05,
+      "loss": 2.0437,
+      "step": 5129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9416723012099747e-05,
+      "loss": 2.0307,
+      "step": 5130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941648384152551e-05,
+      "loss": 2.0147,
+      "step": 5131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941624462339946e-05,
+      "loss": 2.0263,
+      "step": 5132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94160053577228e-05,
+      "loss": 2.0331,
+      "step": 5133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9415766044496733e-05,
+      "loss": 2.0361,
+      "step": 5134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9415526683722476e-05,
+      "loss": 2.0442,
+      "step": 5135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9415287275401234e-05,
+      "loss": 2.0366,
+      "step": 5136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9415047819534214e-05,
+      "loss": 2.0442,
+      "step": 5137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941480831612263e-05,
+      "loss": 2.0343,
+      "step": 5138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941456876516769e-05,
+      "loss": 2.0498,
+      "step": 5139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94143291666706e-05,
+      "loss": 2.0461,
+      "step": 5140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9414089520632573e-05,
+      "loss": 2.0348,
+      "step": 5141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941384982705482e-05,
+      "loss": 2.0669,
+      "step": 5142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9413610085938547e-05,
+      "loss": 2.0325,
+      "step": 5143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941337029728497e-05,
+      "loss": 2.0338,
+      "step": 5144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9413130461095298e-05,
+      "loss": 2.0383,
+      "step": 5145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941289057737074e-05,
+      "loss": 2.0148,
+      "step": 5146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9412650646112512e-05,
+      "loss": 2.0369,
+      "step": 5147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941241066732182e-05,
+      "loss": 2.0274,
+      "step": 5148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9412170640999882e-05,
+      "loss": 2.0371,
+      "step": 5149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94119305671479e-05,
+      "loss": 2.048,
+      "step": 5150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94116904457671e-05,
+      "loss": 2.0238,
+      "step": 5151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9411450276858684e-05,
+      "loss": 2.0196,
+      "step": 5152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941121006042387e-05,
+      "loss": 2.0396,
+      "step": 5153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941096979646387e-05,
+      "loss": 2.0167,
+      "step": 5154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9410729484979894e-05,
+      "loss": 2.0322,
+      "step": 5155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9410489125973162e-05,
+      "loss": 2.0161,
+      "step": 5156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9410248719444883e-05,
+      "loss": 2.0507,
+      "step": 5157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941000826539627e-05,
+      "loss": 2.0377,
+      "step": 5158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9409767763828543e-05,
+      "loss": 2.0259,
+      "step": 5159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9409527214742908e-05,
+      "loss": 2.0411,
+      "step": 5160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940928661814059e-05,
+      "loss": 2.0346,
+      "step": 5161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9409045974022796e-05,
+      "loss": 2.0386,
+      "step": 5162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9408805282390742e-05,
+      "loss": 2.0297,
+      "step": 5163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940856454324565e-05,
+      "loss": 2.0308,
+      "step": 5164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940832375658873e-05,
+      "loss": 2.0302,
+      "step": 5165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9408082922421196e-05,
+      "loss": 2.0223,
+      "step": 5166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9407842040744266e-05,
+      "loss": 2.0392,
+      "step": 5167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9407601111559162e-05,
+      "loss": 2.0328,
+      "step": 5168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9407360134867094e-05,
+      "loss": 2.0391,
+      "step": 5169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940711911066928e-05,
+      "loss": 2.0401,
+      "step": 5170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940687803896694e-05,
+      "loss": 2.0525,
+      "step": 5171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9406636919761287e-05,
+      "loss": 2.0363,
+      "step": 5172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9406395753053545e-05,
+      "loss": 2.0345,
+      "step": 5173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9406154538844927e-05,
+      "loss": 2.0416,
+      "step": 5174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9405913277136653e-05,
+      "loss": 2.0455,
+      "step": 5175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9405671967929936e-05,
+      "loss": 2.0444,
+      "step": 5176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9405430611226003e-05,
+      "loss": 2.006,
+      "step": 5177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940518920702607e-05,
+      "loss": 2.0423,
+      "step": 5178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940494775533135e-05,
+      "loss": 2.0195,
+      "step": 5179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940470625614307e-05,
+      "loss": 2.0344,
+      "step": 5180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9404464709462445e-05,
+      "loss": 2.0222,
+      "step": 5181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94042231152907e-05,
+      "loss": 2.043,
+      "step": 5182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9403981473629047e-05,
+      "loss": 2.0268,
+      "step": 5183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9403739784478716e-05,
+      "loss": 2.0272,
+      "step": 5184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9403498047840918e-05,
+      "loss": 2.0268,
+      "step": 5185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9403256263716882e-05,
+      "loss": 2.0472,
+      "step": 5186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9403014432107822e-05,
+      "loss": 2.0241,
+      "step": 5187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940277255301496e-05,
+      "loss": 2.0146,
+      "step": 5188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9402530626439528e-05,
+      "loss": 2.0352,
+      "step": 5189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940228865238273e-05,
+      "loss": 2.0486,
+      "step": 5190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9402046630845803e-05,
+      "loss": 2.0333,
+      "step": 5191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9401804561829963e-05,
+      "loss": 2.0398,
+      "step": 5192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940156244533643e-05,
+      "loss": 2.0328,
+      "step": 5193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940132028136643e-05,
+      "loss": 2.0277,
+      "step": 5194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940107806992119e-05,
+      "loss": 2.039,
+      "step": 5195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9400835811001922e-05,
+      "loss": 1.9934,
+      "step": 5196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940059350460986e-05,
+      "loss": 2.0543,
+      "step": 5197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940035115074622e-05,
+      "loss": 2.0453,
+      "step": 5198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9400108749412233e-05,
+      "loss": 2.0533,
+      "step": 5199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9399866300609117e-05,
+      "loss": 2.0204,
+      "step": 5200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9399623804338105e-05,
+      "loss": 2.0202,
+      "step": 5201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9399381260600407e-05,
+      "loss": 2.0244,
+      "step": 5202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9399138669397263e-05,
+      "loss": 2.0294,
+      "step": 5203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9398896030729885e-05,
+      "loss": 2.0267,
+      "step": 5204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9398653344599508e-05,
+      "loss": 2.0577,
+      "step": 5205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9398410611007353e-05,
+      "loss": 2.0353,
+      "step": 5206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9398167829954645e-05,
+      "loss": 2.0553,
+      "step": 5207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9397925001442613e-05,
+      "loss": 2.0472,
+      "step": 5208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9397682125472488e-05,
+      "loss": 2.0399,
+      "step": 5209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939743920204548e-05,
+      "loss": 2.0336,
+      "step": 5210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9397196231162837e-05,
+      "loss": 2.032,
+      "step": 5211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9396953212825768e-05,
+      "loss": 2.028,
+      "step": 5212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939671014703551e-05,
+      "loss": 2.0287,
+      "step": 5213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939646703379329e-05,
+      "loss": 2.044,
+      "step": 5214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939622387310033e-05,
+      "loss": 2.0512,
+      "step": 5215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939598066495786e-05,
+      "loss": 2.033,
+      "step": 5216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9395737409367116e-05,
+      "loss": 2.0449,
+      "step": 5217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9395494106329314e-05,
+      "loss": 2.0397,
+      "step": 5218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939525075584569e-05,
+      "loss": 2.0468,
+      "step": 5219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9395007357917474e-05,
+      "loss": 2.0379,
+      "step": 5220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9394763912545892e-05,
+      "loss": 2.0275,
+      "step": 5221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939452041973217e-05,
+      "loss": 2.0276,
+      "step": 5222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9394276879477547e-05,
+      "loss": 2.0609,
+      "step": 5223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9394033291783248e-05,
+      "loss": 2.0406,
+      "step": 5224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93937896566505e-05,
+      "loss": 2.0381,
+      "step": 5225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9393545974080537e-05,
+      "loss": 2.033,
+      "step": 5226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9393302244074585e-05,
+      "loss": 2.0228,
+      "step": 5227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9393058466633884e-05,
+      "loss": 2.0531,
+      "step": 5228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9392814641759656e-05,
+      "loss": 2.023,
+      "step": 5229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9392570769453134e-05,
+      "loss": 2.0269,
+      "step": 5230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9392326849715553e-05,
+      "loss": 2.0339,
+      "step": 5231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9392082882548142e-05,
+      "loss": 2.055,
+      "step": 5232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9391838867952137e-05,
+      "loss": 2.0294,
+      "step": 5233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9391594805928765e-05,
+      "loss": 2.034,
+      "step": 5234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939135069647926e-05,
+      "loss": 2.0201,
+      "step": 5235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9391106539604854e-05,
+      "loss": 2.0252,
+      "step": 5236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9390862335306785e-05,
+      "loss": 2.0432,
+      "step": 5237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9390618083586278e-05,
+      "loss": 2.0268,
+      "step": 5238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9390373784444578e-05,
+      "loss": 2.0366,
+      "step": 5239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9390129437882907e-05,
+      "loss": 2.0372,
+      "step": 5240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93898850439025e-05,
+      "loss": 2.0545,
+      "step": 5241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93896406025046e-05,
+      "loss": 2.0351,
+      "step": 5242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9389396113690436e-05,
+      "loss": 2.0456,
+      "step": 5243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938915157746124e-05,
+      "loss": 2.034,
+      "step": 5244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9388906993818248e-05,
+      "loss": 2.0266,
+      "step": 5245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93886623627627e-05,
+      "loss": 2.0256,
+      "step": 5246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938841768429583e-05,
+      "loss": 2.0257,
+      "step": 5247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9388172958418866e-05,
+      "loss": 2.0267,
+      "step": 5248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938792818513305e-05,
+      "loss": 2.043,
+      "step": 5249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938768336443962e-05,
+      "loss": 2.0315,
+      "step": 5250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938743849633981e-05,
+      "loss": 2.0401,
+      "step": 5251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9387193580834856e-05,
+      "loss": 2.0286,
+      "step": 5252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9386948617925992e-05,
+      "loss": 2.0256,
+      "step": 5253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9386703607614463e-05,
+      "loss": 2.049,
+      "step": 5254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93864585499015e-05,
+      "loss": 2.0276,
+      "step": 5255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938621344478834e-05,
+      "loss": 2.0368,
+      "step": 5256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9385968292276223e-05,
+      "loss": 2.0253,
+      "step": 5257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9385723092366388e-05,
+      "loss": 2.0372,
+      "step": 5258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9385477845060072e-05,
+      "loss": 2.0517,
+      "step": 5259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9385232550358512e-05,
+      "loss": 2.0266,
+      "step": 5260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9384987208262948e-05,
+      "loss": 2.0362,
+      "step": 5261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938474181877462e-05,
+      "loss": 2.043,
+      "step": 5262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9384496381894765e-05,
+      "loss": 2.0182,
+      "step": 5263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9384250897624626e-05,
+      "loss": 2.0429,
+      "step": 5264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9384005365965436e-05,
+      "loss": 2.0205,
+      "step": 5265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9383759786918443e-05,
+      "loss": 2.0516,
+      "step": 5266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9383514160484883e-05,
+      "loss": 2.0186,
+      "step": 5267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9383268486665992e-05,
+      "loss": 2.0394,
+      "step": 5268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938302276546302e-05,
+      "loss": 2.0296,
+      "step": 5269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93827769968772e-05,
+      "loss": 2.0104,
+      "step": 5270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938253118090978e-05,
+      "loss": 2.0564,
+      "step": 5271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9382285317561994e-05,
+      "loss": 2.0322,
+      "step": 5272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938203940683509e-05,
+      "loss": 2.0362,
+      "step": 5273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9381793448730305e-05,
+      "loss": 2.0493,
+      "step": 5274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9381547443248885e-05,
+      "loss": 2.0276,
+      "step": 5275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9381301390392067e-05,
+      "loss": 2.0435,
+      "step": 5276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93810552901611e-05,
+      "loss": 2.0214,
+      "step": 5277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938080914255722e-05,
+      "loss": 2.0404,
+      "step": 5278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938056294758168e-05,
+      "loss": 2.0302,
+      "step": 5279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938031670523571e-05,
+      "loss": 2.043,
+      "step": 5280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9380070415520563e-05,
+      "loss": 2.0237,
+      "step": 5281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9379824078437476e-05,
+      "loss": 2.056,
+      "step": 5282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9379577693987704e-05,
+      "loss": 2.0447,
+      "step": 5283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937933126217248e-05,
+      "loss": 2.0546,
+      "step": 5284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9379084782993054e-05,
+      "loss": 2.0313,
+      "step": 5285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937883825645067e-05,
+      "loss": 2.0338,
+      "step": 5286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937859168254657e-05,
+      "loss": 2.0271,
+      "step": 5287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9378345061282e-05,
+      "loss": 2.0383,
+      "step": 5288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937809839265821e-05,
+      "loss": 2.0376,
+      "step": 5289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9377851676676443e-05,
+      "loss": 2.0459,
+      "step": 5290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937760491333794e-05,
+      "loss": 2.0494,
+      "step": 5291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9377358102643958e-05,
+      "loss": 2.0204,
+      "step": 5292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9377111244595734e-05,
+      "loss": 2.038,
+      "step": 5293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9376864339194515e-05,
+      "loss": 2.0128,
+      "step": 5294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9376617386441553e-05,
+      "loss": 2.0333,
+      "step": 5295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9376370386338088e-05,
+      "loss": 2.0502,
+      "step": 5296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9376123338885375e-05,
+      "loss": 2.0428,
+      "step": 5297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937587624408466e-05,
+      "loss": 2.023,
+      "step": 5298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9375629101937187e-05,
+      "loss": 2.0231,
+      "step": 5299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9375381912444205e-05,
+      "loss": 2.0464,
+      "step": 5300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9375134675606963e-05,
+      "loss": 2.0351,
+      "step": 5301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9374887391426713e-05,
+      "loss": 2.0377,
+      "step": 5302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9374640059904696e-05,
+      "loss": 2.0134,
+      "step": 5303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9374392681042167e-05,
+      "loss": 2.0152,
+      "step": 5304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9374145254840377e-05,
+      "loss": 2.0378,
+      "step": 5305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937389778130057e-05,
+      "loss": 2.0357,
+      "step": 5306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9373650260423997e-05,
+      "loss": 2.0403,
+      "step": 5307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9373402692211908e-05,
+      "loss": 2.0325,
+      "step": 5308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9373155076665555e-05,
+      "loss": 2.0277,
+      "step": 5309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937290741378619e-05,
+      "loss": 2.028,
+      "step": 5310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937265970357506e-05,
+      "loss": 2.0361,
+      "step": 5311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9372411946033414e-05,
+      "loss": 2.0442,
+      "step": 5312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937216414116251e-05,
+      "loss": 2.0421,
+      "step": 5313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9371916288963593e-05,
+      "loss": 2.0282,
+      "step": 5314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9371668389437918e-05,
+      "loss": 2.0303,
+      "step": 5315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9371420442586736e-05,
+      "loss": 2.0536,
+      "step": 5316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9371172448411296e-05,
+      "loss": 2.0188,
+      "step": 5317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9370924406912856e-05,
+      "loss": 2.0281,
+      "step": 5318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9370676318092668e-05,
+      "loss": 2.0312,
+      "step": 5319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937042818195198e-05,
+      "loss": 2.033,
+      "step": 5320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937017999849205e-05,
+      "loss": 2.0329,
+      "step": 5321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9369931767714126e-05,
+      "loss": 2.031,
+      "step": 5322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936968348961947e-05,
+      "loss": 2.0358,
+      "step": 5323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9369435164209324e-05,
+      "loss": 2.0302,
+      "step": 5324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9369186791484953e-05,
+      "loss": 2.0265,
+      "step": 5325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9368938371447604e-05,
+      "loss": 2.0257,
+      "step": 5326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9368689904098536e-05,
+      "loss": 2.0342,
+      "step": 5327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9368441389439e-05,
+      "loss": 2.0582,
+      "step": 5328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9368192827470254e-05,
+      "loss": 2.0468,
+      "step": 5329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9367944218193554e-05,
+      "loss": 2.0382,
+      "step": 5330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9367695561610153e-05,
+      "loss": 2.0473,
+      "step": 5331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9367446857721303e-05,
+      "loss": 2.0113,
+      "step": 5332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9367198106528265e-05,
+      "loss": 2.0204,
+      "step": 5333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93669493080323e-05,
+      "loss": 2.0214,
+      "step": 5334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936670046223465e-05,
+      "loss": 2.0396,
+      "step": 5335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9366451569136587e-05,
+      "loss": 2.0443,
+      "step": 5336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936620262873936e-05,
+      "loss": 2.0359,
+      "step": 5337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9365953641044227e-05,
+      "loss": 2.0326,
+      "step": 5338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9365704606052444e-05,
+      "loss": 2.0433,
+      "step": 5339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9365455523765272e-05,
+      "loss": 2.0281,
+      "step": 5340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9365206394183965e-05,
+      "loss": 2.0442,
+      "step": 5341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9364957217309783e-05,
+      "loss": 2.0366,
+      "step": 5342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9364707993143986e-05,
+      "loss": 2.0401,
+      "step": 5343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9364458721687832e-05,
+      "loss": 2.0174,
+      "step": 5344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9364209402942577e-05,
+      "loss": 2.0448,
+      "step": 5345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9363960036909478e-05,
+      "loss": 2.0302,
+      "step": 5346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9363710623589806e-05,
+      "loss": 2.0314,
+      "step": 5347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9363461162984808e-05,
+      "loss": 2.0467,
+      "step": 5348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9363211655095747e-05,
+      "loss": 2.0381,
+      "step": 5349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9362962099923884e-05,
+      "loss": 2.0291,
+      "step": 5350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9362712497470483e-05,
+      "loss": 2.0443,
+      "step": 5351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93624628477368e-05,
+      "loss": 2.0399,
+      "step": 5352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9362213150724093e-05,
+      "loss": 2.0345,
+      "step": 5353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9361963406433632e-05,
+      "loss": 2.0484,
+      "step": 5354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9361713614866668e-05,
+      "loss": 2.0318,
+      "step": 5355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936146377602447e-05,
+      "loss": 2.0454,
+      "step": 5356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9361213889908294e-05,
+      "loss": 2.0419,
+      "step": 5357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9360963956519408e-05,
+      "loss": 2.0467,
+      "step": 5358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936071397585907e-05,
+      "loss": 2.0291,
+      "step": 5359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936046394792854e-05,
+      "loss": 2.0115,
+      "step": 5360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9360213872729088e-05,
+      "loss": 2.0365,
+      "step": 5361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9359963750261967e-05,
+      "loss": 2.0344,
+      "step": 5362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9359713580528453e-05,
+      "loss": 2.0439,
+      "step": 5363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9359463363529795e-05,
+      "loss": 2.0493,
+      "step": 5364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9359213099267265e-05,
+      "loss": 2.0339,
+      "step": 5365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935896278774213e-05,
+      "loss": 2.0345,
+      "step": 5366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9358712428955646e-05,
+      "loss": 2.0499,
+      "step": 5367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935846202290908e-05,
+      "loss": 2.0396,
+      "step": 5368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93582115696037e-05,
+      "loss": 2.0379,
+      "step": 5369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935796106904076e-05,
+      "loss": 2.0496,
+      "step": 5370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935771052122154e-05,
+      "loss": 2.0521,
+      "step": 5371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93574599261473e-05,
+      "loss": 2.031,
+      "step": 5372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93572092838193e-05,
+      "loss": 1.9966,
+      "step": 5373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935695859423881e-05,
+      "loss": 2.0252,
+      "step": 5374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9356707857407095e-05,
+      "loss": 2.0403,
+      "step": 5375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9356457073325422e-05,
+      "loss": 2.0114,
+      "step": 5376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9356206241995056e-05,
+      "loss": 2.0538,
+      "step": 5377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9355955363417263e-05,
+      "loss": 2.0439,
+      "step": 5378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9355704437593313e-05,
+      "loss": 2.0276,
+      "step": 5379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9355453464524472e-05,
+      "loss": 2.041,
+      "step": 5380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9355202444212005e-05,
+      "loss": 2.0261,
+      "step": 5381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9354951376657184e-05,
+      "loss": 2.0395,
+      "step": 5382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9354700261861272e-05,
+      "loss": 2.0133,
+      "step": 5383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935444909982554e-05,
+      "loss": 2.0287,
+      "step": 5384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9354197890551257e-05,
+      "loss": 2.0288,
+      "step": 5385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935394663403969e-05,
+      "loss": 2.0371,
+      "step": 5386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935369533029211e-05,
+      "loss": 2.0236,
+      "step": 5387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935344397930978e-05,
+      "loss": 2.0343,
+      "step": 5388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9353192581093975e-05,
+      "loss": 2.0371,
+      "step": 5389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9352941135645965e-05,
+      "loss": 2.0224,
+      "step": 5390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9352689642967015e-05,
+      "loss": 2.0345,
+      "step": 5391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9352438103058402e-05,
+      "loss": 2.0387,
+      "step": 5392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9352186515921387e-05,
+      "loss": 2.0114,
+      "step": 5393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9351934881557248e-05,
+      "loss": 2.0424,
+      "step": 5394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9351683199967253e-05,
+      "loss": 2.024,
+      "step": 5395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9351431471152672e-05,
+      "loss": 2.0266,
+      "step": 5396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935117969511478e-05,
+      "loss": 2.0375,
+      "step": 5397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9350927871854846e-05,
+      "loss": 2.0332,
+      "step": 5398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9350676001374137e-05,
+      "loss": 2.0535,
+      "step": 5399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9350424083673932e-05,
+      "loss": 2.0311,
+      "step": 5400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93501721187555e-05,
+      "loss": 2.0423,
+      "step": 5401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9349920106620117e-05,
+      "loss": 2.0196,
+      "step": 5402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934966804726905e-05,
+      "loss": 2.0227,
+      "step": 5403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9349415940703572e-05,
+      "loss": 2.0405,
+      "step": 5404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934916378692496e-05,
+      "loss": 2.0501,
+      "step": 5405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934891158593449e-05,
+      "loss": 2.0324,
+      "step": 5406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9348659337733426e-05,
+      "loss": 2.0318,
+      "step": 5407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934840704232305e-05,
+      "loss": 2.0401,
+      "step": 5408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934815469970463e-05,
+      "loss": 2.0425,
+      "step": 5409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934790230987945e-05,
+      "loss": 2.021,
+      "step": 5410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934764987284877e-05,
+      "loss": 2.0256,
+      "step": 5411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934739738861388e-05,
+      "loss": 2.0243,
+      "step": 5412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9347144857176045e-05,
+      "loss": 2.0193,
+      "step": 5413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9346892278536543e-05,
+      "loss": 2.0249,
+      "step": 5414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934663965269665e-05,
+      "loss": 2.0284,
+      "step": 5415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9346386979657643e-05,
+      "loss": 2.0588,
+      "step": 5416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934613425942079e-05,
+      "loss": 2.0303,
+      "step": 5417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9345881491987377e-05,
+      "loss": 2.0221,
+      "step": 5418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9345628677358678e-05,
+      "loss": 2.0473,
+      "step": 5419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934537581553597e-05,
+      "loss": 2.0324,
+      "step": 5420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9345122906520528e-05,
+      "loss": 2.0273,
+      "step": 5421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9344869950313626e-05,
+      "loss": 2.031,
+      "step": 5422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9344616946916548e-05,
+      "loss": 2.044,
+      "step": 5423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9344363896330566e-05,
+      "loss": 2.0193,
+      "step": 5424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9344110798556964e-05,
+      "loss": 2.0437,
+      "step": 5425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9343857653597014e-05,
+      "loss": 2.0379,
+      "step": 5426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9343604461452e-05,
+      "loss": 2.0244,
+      "step": 5427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9343351222123194e-05,
+      "loss": 2.0342,
+      "step": 5428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934309793561188e-05,
+      "loss": 2.0359,
+      "step": 5429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9342844601919337e-05,
+      "loss": 2.0376,
+      "step": 5430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934259122104684e-05,
+      "loss": 2.0336,
+      "step": 5431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934233779299567e-05,
+      "loss": 2.027,
+      "step": 5432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9342084317767112e-05,
+      "loss": 2.0341,
+      "step": 5433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934183079536244e-05,
+      "loss": 2.0423,
+      "step": 5434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934157722578294e-05,
+      "loss": 2.0358,
+      "step": 5435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9341323609029886e-05,
+      "loss": 2.0261,
+      "step": 5436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9341069945104562e-05,
+      "loss": 2.0431,
+      "step": 5437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9340816234008247e-05,
+      "loss": 2.0321,
+      "step": 5438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9340562475742223e-05,
+      "loss": 2.0369,
+      "step": 5439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9340308670307773e-05,
+      "loss": 2.0343,
+      "step": 5440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934005481770618e-05,
+      "loss": 2.0285,
+      "step": 5441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9339800917938723e-05,
+      "loss": 2.0387,
+      "step": 5442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9339546971006684e-05,
+      "loss": 2.0477,
+      "step": 5443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9339292976911346e-05,
+      "loss": 2.0084,
+      "step": 5444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9339038935653993e-05,
+      "loss": 2.0565,
+      "step": 5445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9338784847235905e-05,
+      "loss": 2.0319,
+      "step": 5446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9338530711658366e-05,
+      "loss": 2.0336,
+      "step": 5447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9338276528922665e-05,
+      "loss": 2.0368,
+      "step": 5448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9338022299030075e-05,
+      "loss": 2.0226,
+      "step": 5449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933776802198189e-05,
+      "loss": 2.0351,
+      "step": 5450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9337513697779387e-05,
+      "loss": 2.0246,
+      "step": 5451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9337259326423855e-05,
+      "loss": 2.0502,
+      "step": 5452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9337004907916577e-05,
+      "loss": 2.0325,
+      "step": 5453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9336750442258837e-05,
+      "loss": 2.034,
+      "step": 5454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9336495929451916e-05,
+      "loss": 2.0307,
+      "step": 5455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933624136949711e-05,
+      "loss": 2.0332,
+      "step": 5456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933598676239569e-05,
+      "loss": 2.0467,
+      "step": 5457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933573210814896e-05,
+      "loss": 2.0452,
+      "step": 5458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933547740675819e-05,
+      "loss": 2.0363,
+      "step": 5459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9335222658224673e-05,
+      "loss": 2.031,
+      "step": 5460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9334967862549693e-05,
+      "loss": 2.0103,
+      "step": 5461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9334713019734538e-05,
+      "loss": 2.0539,
+      "step": 5462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93344581297805e-05,
+      "loss": 2.0425,
+      "step": 5463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9334203192688856e-05,
+      "loss": 2.0274,
+      "step": 5464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93339482084609e-05,
+      "loss": 2.0461,
+      "step": 5465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9333693177097918e-05,
+      "loss": 2.0274,
+      "step": 5466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93334380986012e-05,
+      "loss": 2.0488,
+      "step": 5467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933318297297203e-05,
+      "loss": 2.0211,
+      "step": 5468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93329278002117e-05,
+      "loss": 2.0277,
+      "step": 5469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9332672580321492e-05,
+      "loss": 2.0497,
+      "step": 5470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9332417313302706e-05,
+      "loss": 2.0357,
+      "step": 5471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9332161999156623e-05,
+      "loss": 2.0344,
+      "step": 5472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933190663788453e-05,
+      "loss": 2.0406,
+      "step": 5473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9331651229487727e-05,
+      "loss": 2.044,
+      "step": 5474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9331395773967495e-05,
+      "loss": 2.0146,
+      "step": 5475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9331140271325126e-05,
+      "loss": 2.0399,
+      "step": 5476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9330884721561914e-05,
+      "loss": 2.0172,
+      "step": 5477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9330629124679144e-05,
+      "loss": 2.0321,
+      "step": 5478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933037348067811e-05,
+      "loss": 2.0391,
+      "step": 5479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93301177895601e-05,
+      "loss": 2.0275,
+      "step": 5480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9329862051326407e-05,
+      "loss": 2.0079,
+      "step": 5481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9329606265978325e-05,
+      "loss": 2.0347,
+      "step": 5482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9329350433517146e-05,
+      "loss": 2.0285,
+      "step": 5483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9329094553944153e-05,
+      "loss": 2.0255,
+      "step": 5484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9328838627260648e-05,
+      "loss": 2.0518,
+      "step": 5485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932858265346792e-05,
+      "loss": 2.0321,
+      "step": 5486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9328326632567258e-05,
+      "loss": 2.0541,
+      "step": 5487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9328070564559958e-05,
+      "loss": 2.047,
+      "step": 5488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9327814449447316e-05,
+      "loss": 2.033,
+      "step": 5489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9327558287230627e-05,
+      "loss": 2.0369,
+      "step": 5490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9327302077911174e-05,
+      "loss": 2.0214,
+      "step": 5491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932704582149026e-05,
+      "loss": 2.0326,
+      "step": 5492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9326789517969174e-05,
+      "loss": 2.0385,
+      "step": 5493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9326533167349217e-05,
+      "loss": 2.0454,
+      "step": 5494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9326276769631676e-05,
+      "loss": 2.0203,
+      "step": 5495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9326020324817846e-05,
+      "loss": 2.0437,
+      "step": 5496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9325763832909028e-05,
+      "loss": 2.0368,
+      "step": 5497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9325507293906517e-05,
+      "loss": 2.0506,
+      "step": 5498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93252507078116e-05,
+      "loss": 2.0375,
+      "step": 5499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932499407462558e-05,
+      "loss": 2.0299,
+      "step": 5500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932473739434975e-05,
+      "loss": 2.0467,
+      "step": 5501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932448066698541e-05,
+      "loss": 2.058,
+      "step": 5502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9324223892533853e-05,
+      "loss": 2.0469,
+      "step": 5503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323967070996377e-05,
+      "loss": 2.0367,
+      "step": 5504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323710202374277e-05,
+      "loss": 2.0143,
+      "step": 5505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323453286668852e-05,
+      "loss": 2.038,
+      "step": 5506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323196323881398e-05,
+      "loss": 2.0421,
+      "step": 5507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9322939314013217e-05,
+      "loss": 2.0327,
+      "step": 5508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93226822570656e-05,
+      "loss": 2.0405,
+      "step": 5509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932242515303985e-05,
+      "loss": 2.0312,
+      "step": 5510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9322168001937265e-05,
+      "loss": 2.0142,
+      "step": 5511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932191080375914e-05,
+      "loss": 2.0513,
+      "step": 5512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932165355850678e-05,
+      "loss": 2.0327,
+      "step": 5513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9321396266181476e-05,
+      "loss": 2.0355,
+      "step": 5514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9321138926784532e-05,
+      "loss": 2.033,
+      "step": 5515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932088154031725e-05,
+      "loss": 2.0259,
+      "step": 5516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9320624106780922e-05,
+      "loss": 2.0331,
+      "step": 5517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9320366626176857e-05,
+      "loss": 2.0485,
+      "step": 5518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932010909850635e-05,
+      "loss": 2.0288,
+      "step": 5519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9319851523770704e-05,
+      "loss": 2.0427,
+      "step": 5520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9319593901971218e-05,
+      "loss": 2.0499,
+      "step": 5521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931933623310919e-05,
+      "loss": 2.0324,
+      "step": 5522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931907851718593e-05,
+      "loss": 2.0297,
+      "step": 5523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931882075420273e-05,
+      "loss": 2.0096,
+      "step": 5524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93185629441609e-05,
+      "loss": 2.0445,
+      "step": 5525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931830508706173e-05,
+      "loss": 2.0317,
+      "step": 5526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9318047182906535e-05,
+      "loss": 2.0537,
+      "step": 5527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931778923169661e-05,
+      "loss": 2.0446,
+      "step": 5528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931753123343326e-05,
+      "loss": 2.0503,
+      "step": 5529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9317273188117783e-05,
+      "loss": 2.0368,
+      "step": 5530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9317015095751494e-05,
+      "loss": 2.0427,
+      "step": 5531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9316756956335686e-05,
+      "loss": 2.0276,
+      "step": 5532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9316498769871663e-05,
+      "loss": 2.0218,
+      "step": 5533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9316240536360733e-05,
+      "loss": 2.039,
+      "step": 5534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93159822558042e-05,
+      "loss": 2.0296,
+      "step": 5535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9315723928203364e-05,
+      "loss": 2.038,
+      "step": 5536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9315465553559534e-05,
+      "loss": 2.0228,
+      "step": 5537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931520713187401e-05,
+      "loss": 2.0328,
+      "step": 5538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93149486631481e-05,
+      "loss": 2.0142,
+      "step": 5539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9314690147383112e-05,
+      "loss": 2.0409,
+      "step": 5540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9314431584580345e-05,
+      "loss": 2.0325,
+      "step": 5541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9314172974741113e-05,
+      "loss": 2.0193,
+      "step": 5542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9313914317866715e-05,
+      "loss": 2.012,
+      "step": 5543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931365561395846e-05,
+      "loss": 2.051,
+      "step": 5544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931339686301765e-05,
+      "loss": 2.0433,
+      "step": 5545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9313138065045602e-05,
+      "loss": 2.0443,
+      "step": 5546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9312879220043607e-05,
+      "loss": 2.0358,
+      "step": 5547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9312620328012987e-05,
+      "loss": 2.0598,
+      "step": 5548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9312361388955044e-05,
+      "loss": 2.0433,
+      "step": 5549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9312102402871085e-05,
+      "loss": 2.0385,
+      "step": 5550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931184336976242e-05,
+      "loss": 2.0391,
+      "step": 5551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931158428963035e-05,
+      "loss": 2.0295,
+      "step": 5552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9311325162476194e-05,
+      "loss": 2.0317,
+      "step": 5553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931106598830125e-05,
+      "loss": 2.0443,
+      "step": 5554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9310806767106835e-05,
+      "loss": 2.038,
+      "step": 5555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9310547498894254e-05,
+      "loss": 2.0252,
+      "step": 5556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9310288183664818e-05,
+      "loss": 2.0338,
+      "step": 5557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9310028821419836e-05,
+      "loss": 2.049,
+      "step": 5558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9309769412160616e-05,
+      "loss": 2.0243,
+      "step": 5559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930950995588847e-05,
+      "loss": 2.0485,
+      "step": 5560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9309250452604708e-05,
+      "loss": 2.0322,
+      "step": 5561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9308990902310643e-05,
+      "loss": 2.057,
+      "step": 5562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930873130500758e-05,
+      "loss": 2.0414,
+      "step": 5563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9308471660696833e-05,
+      "loss": 2.0345,
+      "step": 5564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9308211969379714e-05,
+      "loss": 2.0323,
+      "step": 5565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930795223105753e-05,
+      "loss": 2.0263,
+      "step": 5566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.93076924457316e-05,
+      "loss": 2.0412,
+      "step": 5567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9307432613403227e-05,
+      "loss": 2.0284,
+      "step": 5568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930717273407373e-05,
+      "loss": 2.0338,
+      "step": 5569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930691280774442e-05,
+      "loss": 2.0355,
+      "step": 5570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930665283441661e-05,
+      "loss": 2.0631,
+      "step": 5571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9306392814091612e-05,
+      "loss": 2.0325,
+      "step": 5572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9306132746770737e-05,
+      "loss": 2.0186,
+      "step": 5573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.93058726324553e-05,
+      "loss": 2.0234,
+      "step": 5574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9305612471146613e-05,
+      "loss": 2.0175,
+      "step": 5575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9305352262845994e-05,
+      "loss": 2.0395,
+      "step": 5576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9305092007554752e-05,
+      "loss": 2.0346,
+      "step": 5577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9304831705274208e-05,
+      "loss": 2.0487,
+      "step": 5578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930457135600567e-05,
+      "loss": 2.0298,
+      "step": 5579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9304310959750456e-05,
+      "loss": 2.0229,
+      "step": 5580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9304050516509877e-05,
+      "loss": 2.0315,
+      "step": 5581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930379002628525e-05,
+      "loss": 2.0445,
+      "step": 5582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9303529489077896e-05,
+      "loss": 2.0378,
+      "step": 5583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930326890488912e-05,
+      "loss": 2.0343,
+      "step": 5584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930300827372025e-05,
+      "loss": 2.0266,
+      "step": 5585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930274759557259e-05,
+      "loss": 2.0099,
+      "step": 5586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930248687044747e-05,
+      "loss": 2.0374,
+      "step": 5587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9302226098346194e-05,
+      "loss": 2.0343,
+      "step": 5588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9301965279270087e-05,
+      "loss": 2.0541,
+      "step": 5589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9301704413220462e-05,
+      "loss": 2.038,
+      "step": 5590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9301443500198636e-05,
+      "loss": 2.0186,
+      "step": 5591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930118254020593e-05,
+      "loss": 2.0364,
+      "step": 5592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930092153324366e-05,
+      "loss": 2.0213,
+      "step": 5593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9300660479313143e-05,
+      "loss": 2.0458,
+      "step": 5594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9300399378415696e-05,
+      "loss": 2.038,
+      "step": 5595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9300138230552643e-05,
+      "loss": 2.0209,
+      "step": 5596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9299877035725297e-05,
+      "loss": 2.0437,
+      "step": 5597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929961579393498e-05,
+      "loss": 2.0362,
+      "step": 5598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9299354505183014e-05,
+      "loss": 2.0399,
+      "step": 5599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9299093169470714e-05,
+      "loss": 2.0516,
+      "step": 5600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9298831786799398e-05,
+      "loss": 2.0355,
+      "step": 5601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929857035717039e-05,
+      "loss": 2.0003,
+      "step": 5602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929830888058501e-05,
+      "loss": 2.0238,
+      "step": 5603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9298047357044576e-05,
+      "loss": 2.0275,
+      "step": 5604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9297785786550413e-05,
+      "loss": 2.039,
+      "step": 5605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9297524169103835e-05,
+      "loss": 2.0405,
+      "step": 5606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929726250470617e-05,
+      "loss": 2.0195,
+      "step": 5607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9297000793358734e-05,
+      "loss": 2.0536,
+      "step": 5608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9296739035062852e-05,
+      "loss": 2.0382,
+      "step": 5609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9296477229819843e-05,
+      "loss": 2.0494,
+      "step": 5610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9296215377631036e-05,
+      "loss": 2.0461,
+      "step": 5611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9295953478497746e-05,
+      "loss": 2.0514,
+      "step": 5612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9295691532421293e-05,
+      "loss": 2.0567,
+      "step": 5613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929542953940301e-05,
+      "loss": 2.0436,
+      "step": 5614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9295167499444213e-05,
+      "loss": 2.044,
+      "step": 5615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929490541254623e-05,
+      "loss": 2.0298,
+      "step": 5616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9294643278710378e-05,
+      "loss": 2.0183,
+      "step": 5617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9294381097937985e-05,
+      "loss": 2.0455,
+      "step": 5618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9294118870230372e-05,
+      "loss": 2.0271,
+      "step": 5619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9293856595588867e-05,
+      "loss": 2.0169,
+      "step": 5620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9293594274014794e-05,
+      "loss": 2.0239,
+      "step": 5621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9293331905509475e-05,
+      "loss": 2.0386,
+      "step": 5622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9293069490074237e-05,
+      "loss": 2.0227,
+      "step": 5623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9292807027710404e-05,
+      "loss": 2.0359,
+      "step": 5624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92925445184193e-05,
+      "loss": 2.0306,
+      "step": 5625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9292281962202258e-05,
+      "loss": 2.055,
+      "step": 5626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9292019359060593e-05,
+      "loss": 2.0402,
+      "step": 5627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929175670899564e-05,
+      "loss": 2.0563,
+      "step": 5628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929149401200872e-05,
+      "loss": 2.0467,
+      "step": 5629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929123126810116e-05,
+      "loss": 2.0168,
+      "step": 5630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929096847727429e-05,
+      "loss": 2.0231,
+      "step": 5631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929070563952944e-05,
+      "loss": 2.03,
+      "step": 5632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9290442754867925e-05,
+      "loss": 2.0556,
+      "step": 5633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9290179823291084e-05,
+      "loss": 2.0506,
+      "step": 5634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9289916844800243e-05,
+      "loss": 2.0474,
+      "step": 5635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9289653819396723e-05,
+      "loss": 2.0497,
+      "step": 5636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928939074708186e-05,
+      "loss": 2.0133,
+      "step": 5637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9289127627856977e-05,
+      "loss": 2.0515,
+      "step": 5638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928886446172341e-05,
+      "loss": 2.032,
+      "step": 5639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928860124868248e-05,
+      "loss": 2.0232,
+      "step": 5640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928833798873552e-05,
+      "loss": 2.0411,
+      "step": 5641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928807468188386e-05,
+      "loss": 2.0451,
+      "step": 5642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9287811328128826e-05,
+      "loss": 2.0541,
+      "step": 5643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9287547927471753e-05,
+      "loss": 2.0253,
+      "step": 5644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9287284479913967e-05,
+      "loss": 2.0337,
+      "step": 5645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92870209854568e-05,
+      "loss": 2.0319,
+      "step": 5646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9286757444101582e-05,
+      "loss": 2.0348,
+      "step": 5647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9286493855849645e-05,
+      "loss": 2.0336,
+      "step": 5648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928623022070232e-05,
+      "loss": 2.0441,
+      "step": 5649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9285966538660938e-05,
+      "loss": 2.0364,
+      "step": 5650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928570280972683e-05,
+      "loss": 2.0449,
+      "step": 5651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9285439033901328e-05,
+      "loss": 2.0418,
+      "step": 5652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9285175211185766e-05,
+      "loss": 2.0334,
+      "step": 5653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928491134158147e-05,
+      "loss": 2.0175,
+      "step": 5654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9284647425089783e-05,
+      "loss": 2.0279,
+      "step": 5655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928438346171203e-05,
+      "loss": 2.0475,
+      "step": 5656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9284119451449543e-05,
+      "loss": 2.0353,
+      "step": 5657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928385539430366e-05,
+      "loss": 2.0514,
+      "step": 5658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928359129027571e-05,
+      "loss": 2.042,
+      "step": 5659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928332713936703e-05,
+      "loss": 2.0381,
+      "step": 5660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9283062941578952e-05,
+      "loss": 2.017,
+      "step": 5661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9282798696912812e-05,
+      "loss": 2.0306,
+      "step": 5662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9282534405369945e-05,
+      "loss": 2.0339,
+      "step": 5663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9282270066951684e-05,
+      "loss": 2.0181,
+      "step": 5664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9282005681659363e-05,
+      "loss": 2.0333,
+      "step": 5665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9281741249494318e-05,
+      "loss": 2.031,
+      "step": 5666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928147677045788e-05,
+      "loss": 2.0249,
+      "step": 5667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9281212244551396e-05,
+      "loss": 2.0487,
+      "step": 5668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928094767177619e-05,
+      "loss": 2.0289,
+      "step": 5669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9280683052133607e-05,
+      "loss": 2.0416,
+      "step": 5670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928041838562498e-05,
+      "loss": 2.0426,
+      "step": 5671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928015367225164e-05,
+      "loss": 2.0454,
+      "step": 5672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927988891201493e-05,
+      "loss": 2.0403,
+      "step": 5673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9279624104916187e-05,
+      "loss": 2.0356,
+      "step": 5674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9279359250956745e-05,
+      "loss": 2.0247,
+      "step": 5675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927909435013794e-05,
+      "loss": 2.0198,
+      "step": 5676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927882940246112e-05,
+      "loss": 2.036,
+      "step": 5677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9278564407927612e-05,
+      "loss": 2.0426,
+      "step": 5678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9278299366538758e-05,
+      "loss": 2.0396,
+      "step": 5679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9278034278295897e-05,
+      "loss": 2.0249,
+      "step": 5680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9277769143200365e-05,
+      "loss": 2.0441,
+      "step": 5681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9277503961253505e-05,
+      "loss": 2.024,
+      "step": 5682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9277238732456657e-05,
+      "loss": 2.036,
+      "step": 5683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927697345681115e-05,
+      "loss": 2.0303,
+      "step": 5684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9276708134318338e-05,
+      "loss": 2.0371,
+      "step": 5685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927644276497955e-05,
+      "loss": 2.0217,
+      "step": 5686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9276177348796134e-05,
+      "loss": 2.034,
+      "step": 5687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9275911885769424e-05,
+      "loss": 2.0339,
+      "step": 5688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927564637590076e-05,
+      "loss": 2.0358,
+      "step": 5689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9275380819191486e-05,
+      "loss": 2.0268,
+      "step": 5690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9275115215642946e-05,
+      "loss": 2.0232,
+      "step": 5691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9274849565256475e-05,
+      "loss": 2.0228,
+      "step": 5692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9274583868033422e-05,
+      "loss": 2.0135,
+      "step": 5693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927431812397512e-05,
+      "loss": 2.0432,
+      "step": 5694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9274052333082914e-05,
+      "loss": 2.054,
+      "step": 5695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9273786495358152e-05,
+      "loss": 2.0413,
+      "step": 5696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927352061080217e-05,
+      "loss": 2.02,
+      "step": 5697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9273254679416313e-05,
+      "loss": 2.053,
+      "step": 5698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927298870120192e-05,
+      "loss": 2.0401,
+      "step": 5699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927272267616034e-05,
+      "loss": 2.0467,
+      "step": 5700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927245660429291e-05,
+      "loss": 2.0428,
+      "step": 5701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927219048560098e-05,
+      "loss": 2.0319,
+      "step": 5702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9271924320085894e-05,
+      "loss": 2.0284,
+      "step": 5703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927165810774899e-05,
+      "loss": 2.0222,
+      "step": 5704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927139184859162e-05,
+      "loss": 2.029,
+      "step": 5705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927112554261512e-05,
+      "loss": 2.0466,
+      "step": 5706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927085918982084e-05,
+      "loss": 2.0358,
+      "step": 5707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9270592790210124e-05,
+      "loss": 2.0458,
+      "step": 5708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927032634378432e-05,
+      "loss": 2.0283,
+      "step": 5709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9270059850544768e-05,
+      "loss": 2.0231,
+      "step": 5710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9269793310492816e-05,
+      "loss": 2.0533,
+      "step": 5711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9269526723629815e-05,
+      "loss": 2.054,
+      "step": 5712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9269260089957104e-05,
+      "loss": 2.0253,
+      "step": 5713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9268993409476034e-05,
+      "loss": 2.0241,
+      "step": 5714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926872668218795e-05,
+      "loss": 2.0219,
+      "step": 5715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92684599080942e-05,
+      "loss": 2.032,
+      "step": 5716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9268193087196127e-05,
+      "loss": 2.0351,
+      "step": 5717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926792621949508e-05,
+      "loss": 2.0475,
+      "step": 5718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9267659304992413e-05,
+      "loss": 2.0291,
+      "step": 5719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9267392343689466e-05,
+      "loss": 2.0437,
+      "step": 5720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926712533558759e-05,
+      "loss": 2.0366,
+      "step": 5721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9266858280688137e-05,
+      "loss": 2.0226,
+      "step": 5722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926659117899245e-05,
+      "loss": 2.0228,
+      "step": 5723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926632403050188e-05,
+      "loss": 2.0219,
+      "step": 5724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9266056835217778e-05,
+      "loss": 2.035,
+      "step": 5725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9265789593141487e-05,
+      "loss": 2.043,
+      "step": 5726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9265522304274364e-05,
+      "loss": 2.0424,
+      "step": 5727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9265254968617758e-05,
+      "loss": 2.0452,
+      "step": 5728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926498758617301e-05,
+      "loss": 2.0381,
+      "step": 5729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9264720156941483e-05,
+      "loss": 2.0261,
+      "step": 5730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9264452680924518e-05,
+      "loss": 2.0465,
+      "step": 5731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926418515812347e-05,
+      "loss": 2.0307,
+      "step": 5732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9263917588539688e-05,
+      "loss": 2.0477,
+      "step": 5733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9263649972174527e-05,
+      "loss": 2.0367,
+      "step": 5734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9263382309029335e-05,
+      "loss": 2.0274,
+      "step": 5735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9263114599105462e-05,
+      "loss": 2.0397,
+      "step": 5736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9262846842404265e-05,
+      "loss": 2.0333,
+      "step": 5737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9262579038927093e-05,
+      "loss": 2.0377,
+      "step": 5738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9262311188675295e-05,
+      "loss": 2.0488,
+      "step": 5739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9262043291650232e-05,
+      "loss": 2.0464,
+      "step": 5740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9261775347853247e-05,
+      "loss": 2.0281,
+      "step": 5741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92615073572857e-05,
+      "loss": 2.0335,
+      "step": 5742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9261239319948947e-05,
+      "loss": 2.0394,
+      "step": 5743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9260971235844335e-05,
+      "loss": 2.0347,
+      "step": 5744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9260703104973215e-05,
+      "loss": 2.0458,
+      "step": 5745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9260434927336952e-05,
+      "loss": 2.0473,
+      "step": 5746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926016670293689e-05,
+      "loss": 2.0302,
+      "step": 5747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925989843177439e-05,
+      "loss": 2.0189,
+      "step": 5748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9259630113850806e-05,
+      "loss": 2.0146,
+      "step": 5749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9259361749167486e-05,
+      "loss": 2.0475,
+      "step": 5750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9259093337725793e-05,
+      "loss": 2.0479,
+      "step": 5751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925882487952708e-05,
+      "loss": 2.0234,
+      "step": 5752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9258556374572706e-05,
+      "loss": 2.0412,
+      "step": 5753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9258287822864017e-05,
+      "loss": 2.0424,
+      "step": 5754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925801922440238e-05,
+      "loss": 2.0147,
+      "step": 5755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9257750579189146e-05,
+      "loss": 2.0385,
+      "step": 5756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9257481887225675e-05,
+      "loss": 2.0264,
+      "step": 5757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9257213148513317e-05,
+      "loss": 2.0192,
+      "step": 5758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9256944363053435e-05,
+      "loss": 2.0275,
+      "step": 5759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925667553084739e-05,
+      "loss": 2.03,
+      "step": 5760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925640665189653e-05,
+      "loss": 2.0223,
+      "step": 5761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9256137726202215e-05,
+      "loss": 2.0285,
+      "step": 5762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925586875376581e-05,
+      "loss": 2.0399,
+      "step": 5763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925559973458867e-05,
+      "loss": 2.0341,
+      "step": 5764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9255330668672146e-05,
+      "loss": 2.0127,
+      "step": 5765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9255061556017605e-05,
+      "loss": 2.0465,
+      "step": 5766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9254792396626405e-05,
+      "loss": 2.0299,
+      "step": 5767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92545231904999e-05,
+      "loss": 2.0305,
+      "step": 5768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9254253937639463e-05,
+      "loss": 2.0432,
+      "step": 5769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9253984638046438e-05,
+      "loss": 2.0314,
+      "step": 5770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925371529172219e-05,
+      "loss": 2.0452,
+      "step": 5771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9253445898668083e-05,
+      "loss": 2.0307,
+      "step": 5772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925317645888547e-05,
+      "loss": 2.0477,
+      "step": 5773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9252906972375724e-05,
+      "loss": 2.0349,
+      "step": 5774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925263743914019e-05,
+      "loss": 2.0492,
+      "step": 5775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9252367859180242e-05,
+      "loss": 2.0515,
+      "step": 5776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9252098232497235e-05,
+      "loss": 2.0393,
+      "step": 5777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9251828559092534e-05,
+      "loss": 2.0533,
+      "step": 5778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92515588389675e-05,
+      "loss": 2.0467,
+      "step": 5779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925128907212349e-05,
+      "loss": 2.0417,
+      "step": 5780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925101925856187e-05,
+      "loss": 2.0595,
+      "step": 5781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925074939828401e-05,
+      "loss": 2.0441,
+      "step": 5782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925047949129126e-05,
+      "loss": 2.015,
+      "step": 5783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9250209537584986e-05,
+      "loss": 2.027,
+      "step": 5784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9249939537166557e-05,
+      "loss": 2.0401,
+      "step": 5785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9249669490037334e-05,
+      "loss": 2.0428,
+      "step": 5786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924939939619868e-05,
+      "loss": 2.0328,
+      "step": 5787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9249129255651956e-05,
+      "loss": 2.0336,
+      "step": 5788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9248859068398533e-05,
+      "loss": 2.0361,
+      "step": 5789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924858883443977e-05,
+      "loss": 2.0234,
+      "step": 5790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924831855377703e-05,
+      "loss": 2.0307,
+      "step": 5791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9248048226411687e-05,
+      "loss": 2.0283,
+      "step": 5792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9247777852345097e-05,
+      "loss": 2.0267,
+      "step": 5793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924750743157863e-05,
+      "loss": 2.0418,
+      "step": 5794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9247236964113646e-05,
+      "loss": 2.0613,
+      "step": 5795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9246966449951516e-05,
+      "loss": 2.0345,
+      "step": 5796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9246695889093608e-05,
+      "loss": 2.0264,
+      "step": 5797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924642528154128e-05,
+      "loss": 2.0286,
+      "step": 5798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924615462729591e-05,
+      "loss": 2.041,
+      "step": 5799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9245883926358855e-05,
+      "loss": 2.0521,
+      "step": 5800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9245613178731487e-05,
+      "loss": 2.0264,
+      "step": 5801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9245342384415168e-05,
+      "loss": 2.0432,
+      "step": 5802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9245071543411274e-05,
+      "loss": 2.0413,
+      "step": 5803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9244800655721167e-05,
+      "loss": 2.0355,
+      "step": 5804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9244529721346212e-05,
+      "loss": 2.0415,
+      "step": 5805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9244258740287785e-05,
+      "loss": 2.0136,
+      "step": 5806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9243987712547248e-05,
+      "loss": 2.0601,
+      "step": 5807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9243716638125973e-05,
+      "loss": 2.0407,
+      "step": 5808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924344551702533e-05,
+      "loss": 2.0492,
+      "step": 5809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924317434924668e-05,
+      "loss": 2.0495,
+      "step": 5810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9242903134791405e-05,
+      "loss": 2.0491,
+      "step": 5811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9242631873660867e-05,
+      "loss": 2.036,
+      "step": 5812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9242360565856435e-05,
+      "loss": 2.0474,
+      "step": 5813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924208921137948e-05,
+      "loss": 2.0238,
+      "step": 5814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9241817810231374e-05,
+      "loss": 2.0426,
+      "step": 5815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9241546362413487e-05,
+      "loss": 2.0294,
+      "step": 5816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924127486792719e-05,
+      "loss": 2.0412,
+      "step": 5817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9241003326773855e-05,
+      "loss": 2.0194,
+      "step": 5818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924073173895485e-05,
+      "loss": 2.0293,
+      "step": 5819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9240460104471547e-05,
+      "loss": 2.0499,
+      "step": 5820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9240188423325322e-05,
+      "loss": 2.0294,
+      "step": 5821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9239916695517543e-05,
+      "loss": 2.0281,
+      "step": 5822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923964492104958e-05,
+      "loss": 2.0282,
+      "step": 5823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923937309992281e-05,
+      "loss": 2.0452,
+      "step": 5824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9239101232138605e-05,
+      "loss": 2.0401,
+      "step": 5825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9238829317698337e-05,
+      "loss": 2.0188,
+      "step": 5826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9238557356603383e-05,
+      "loss": 2.0531,
+      "step": 5827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9238285348855104e-05,
+      "loss": 2.0522,
+      "step": 5828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923801329445489e-05,
+      "loss": 2.0271,
+      "step": 5829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9237741193404106e-05,
+      "loss": 2.0631,
+      "step": 5830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923746904570412e-05,
+      "loss": 2.0249,
+      "step": 5831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923719685135632e-05,
+      "loss": 2.0429,
+      "step": 5832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9236924610362073e-05,
+      "loss": 2.0272,
+      "step": 5833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9236652322722755e-05,
+      "loss": 2.0333,
+      "step": 5834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9236379988439738e-05,
+      "loss": 2.0267,
+      "step": 5835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92361076075144e-05,
+      "loss": 2.0285,
+      "step": 5836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923583517994812e-05,
+      "loss": 2.0261,
+      "step": 5837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9235562705742267e-05,
+      "loss": 2.0126,
+      "step": 5838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9235290184898224e-05,
+      "loss": 2.0248,
+      "step": 5839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923501761741736e-05,
+      "loss": 2.0263,
+      "step": 5840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9234745003301054e-05,
+      "loss": 2.0493,
+      "step": 5841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9234472342550685e-05,
+      "loss": 2.0284,
+      "step": 5842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923419963516763e-05,
+      "loss": 2.0367,
+      "step": 5843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923392688115326e-05,
+      "loss": 2.0375,
+      "step": 5844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923365408050896e-05,
+      "loss": 2.0224,
+      "step": 5845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92333812332361e-05,
+      "loss": 2.019,
+      "step": 5846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9233108339336065e-05,
+      "loss": 2.0465,
+      "step": 5847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9232835398810233e-05,
+      "loss": 2.0385,
+      "step": 5848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923256241165998e-05,
+      "loss": 2.0412,
+      "step": 5849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923228937788668e-05,
+      "loss": 2.0452,
+      "step": 5850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9232016297491715e-05,
+      "loss": 2.026,
+      "step": 5851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923174317047647e-05,
+      "loss": 2.0427,
+      "step": 5852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9231469996842317e-05,
+      "loss": 2.0584,
+      "step": 5853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9231196776590633e-05,
+      "loss": 2.0339,
+      "step": 5854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9230923509722807e-05,
+      "loss": 2.065,
+      "step": 5855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9230650196240216e-05,
+      "loss": 2.0389,
+      "step": 5856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9230376836144233e-05,
+      "loss": 2.0427,
+      "step": 5857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923010342943625e-05,
+      "loss": 2.0316,
+      "step": 5858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9229829976117637e-05,
+      "loss": 2.0395,
+      "step": 5859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9229556476189783e-05,
+      "loss": 2.0352,
+      "step": 5860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9229282929654062e-05,
+      "loss": 2.0409,
+      "step": 5861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922900933651186e-05,
+      "loss": 2.0471,
+      "step": 5862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922873569676456e-05,
+      "loss": 2.0344,
+      "step": 5863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9228462010413538e-05,
+      "loss": 2.0503,
+      "step": 5864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922818827746018e-05,
+      "loss": 2.0224,
+      "step": 5865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9227914497905873e-05,
+      "loss": 2.0251,
+      "step": 5866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922764067175199e-05,
+      "loss": 2.0231,
+      "step": 5867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9227366798999916e-05,
+      "loss": 2.0415,
+      "step": 5868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922709287965104e-05,
+      "loss": 2.0115,
+      "step": 5869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9226818913706742e-05,
+      "loss": 2.0231,
+      "step": 5870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9226544901168402e-05,
+      "loss": 2.0364,
+      "step": 5871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922627084203741e-05,
+      "loss": 2.0388,
+      "step": 5872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9225996736315144e-05,
+      "loss": 2.0272,
+      "step": 5873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922572258400299e-05,
+      "loss": 2.0156,
+      "step": 5874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9225448385102335e-05,
+      "loss": 2.0363,
+      "step": 5875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922517413961456e-05,
+      "loss": 2.0196,
+      "step": 5876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9224899847541055e-05,
+      "loss": 2.0443,
+      "step": 5877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92246255088832e-05,
+      "loss": 2.0303,
+      "step": 5878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9224351123642383e-05,
+      "loss": 2.0248,
+      "step": 5879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9224076691819986e-05,
+      "loss": 2.0125,
+      "step": 5880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92238022134174e-05,
+      "loss": 2.0386,
+      "step": 5881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922352768843601e-05,
+      "loss": 2.0507,
+      "step": 5882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9223253116877202e-05,
+      "loss": 2.061,
+      "step": 5883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9222978498742356e-05,
+      "loss": 2.0236,
+      "step": 5884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9222703834032868e-05,
+      "loss": 2.0335,
+      "step": 5885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922242912275012e-05,
+      "loss": 2.0328,
+      "step": 5886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9222154364895503e-05,
+      "loss": 2.0376,
+      "step": 5887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92218795604704e-05,
+      "loss": 2.0461,
+      "step": 5888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9221604709476206e-05,
+      "loss": 2.0327,
+      "step": 5889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9221329811914298e-05,
+      "loss": 2.038,
+      "step": 5890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922105486778607e-05,
+      "loss": 2.0344,
+      "step": 5891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9220779877092913e-05,
+      "loss": 2.0663,
+      "step": 5892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9220504839836214e-05,
+      "loss": 2.0385,
+      "step": 5893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922022975601736e-05,
+      "loss": 2.0222,
+      "step": 5894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921995462563774e-05,
+      "loss": 2.0535,
+      "step": 5895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9219679448698744e-05,
+      "loss": 2.0324,
+      "step": 5896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9219404225201766e-05,
+      "loss": 2.0325,
+      "step": 5897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921912895514819e-05,
+      "loss": 2.0229,
+      "step": 5898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9218853638539404e-05,
+      "loss": 2.0462,
+      "step": 5899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9218578275376807e-05,
+      "loss": 2.0459,
+      "step": 5900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9218302865661784e-05,
+      "loss": 2.0399,
+      "step": 5901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9218027409395725e-05,
+      "loss": 2.0261,
+      "step": 5902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9217751906580024e-05,
+      "loss": 2.0288,
+      "step": 5903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9217476357216072e-05,
+      "loss": 2.0493,
+      "step": 5904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9217200761305257e-05,
+      "loss": 2.0329,
+      "step": 5905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9216925118848976e-05,
+      "loss": 2.0377,
+      "step": 5906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9216649429848616e-05,
+      "loss": 2.0498,
+      "step": 5907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921637369430557e-05,
+      "loss": 2.0221,
+      "step": 5908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9216097912221235e-05,
+      "loss": 2.0392,
+      "step": 5909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9215822083597e-05,
+      "loss": 2.0288,
+      "step": 5910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9215546208434258e-05,
+      "loss": 2.0403,
+      "step": 5911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92152702867344e-05,
+      "loss": 2.0337,
+      "step": 5912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921499431849882e-05,
+      "loss": 2.0068,
+      "step": 5913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921471830372892e-05,
+      "loss": 2.0311,
+      "step": 5914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9214442242426083e-05,
+      "loss": 2.0344,
+      "step": 5915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9214166134591706e-05,
+      "loss": 2.0199,
+      "step": 5916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9213889980227187e-05,
+      "loss": 2.0528,
+      "step": 5917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9213613779333918e-05,
+      "loss": 2.0396,
+      "step": 5918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921333753191329e-05,
+      "loss": 2.0358,
+      "step": 5919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9213061237966707e-05,
+      "loss": 2.0339,
+      "step": 5920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9212784897495558e-05,
+      "loss": 2.0395,
+      "step": 5921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921250851050124e-05,
+      "loss": 2.039,
+      "step": 5922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9212232076985144e-05,
+      "loss": 2.0446,
+      "step": 5923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9211955596948674e-05,
+      "loss": 2.0409,
+      "step": 5924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9211679070393222e-05,
+      "loss": 2.0259,
+      "step": 5925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9211402497320185e-05,
+      "loss": 2.068,
+      "step": 5926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921112587773096e-05,
+      "loss": 2.014,
+      "step": 5927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9210849211626943e-05,
+      "loss": 2.0415,
+      "step": 5928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9210572499009527e-05,
+      "loss": 2.013,
+      "step": 5929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921029573988012e-05,
+      "loss": 2.0141,
+      "step": 5930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9210018934240112e-05,
+      "loss": 2.0194,
+      "step": 5931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92097420820909e-05,
+      "loss": 2.0121,
+      "step": 5932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9209465183433886e-05,
+      "loss": 2.0174,
+      "step": 5933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9209188238270466e-05,
+      "loss": 2.0151,
+      "step": 5934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920891124660204e-05,
+      "loss": 2.0213,
+      "step": 5935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9208634208430005e-05,
+      "loss": 2.0502,
+      "step": 5936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9208357123755762e-05,
+      "loss": 2.0372,
+      "step": 5937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9208079992580708e-05,
+      "loss": 2.0303,
+      "step": 5938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9207802814906246e-05,
+      "loss": 2.0328,
+      "step": 5939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9207525590733768e-05,
+      "loss": 2.0385,
+      "step": 5940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9207248320064683e-05,
+      "loss": 2.0312,
+      "step": 5941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9206971002900388e-05,
+      "loss": 2.0289,
+      "step": 5942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920669363924228e-05,
+      "loss": 2.0308,
+      "step": 5943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9206416229091767e-05,
+      "loss": 2.0362,
+      "step": 5944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9206138772450242e-05,
+      "loss": 2.0344,
+      "step": 5945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920586126931911e-05,
+      "loss": 2.0275,
+      "step": 5946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9205583719699773e-05,
+      "loss": 2.049,
+      "step": 5947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9205306123593634e-05,
+      "loss": 2.0216,
+      "step": 5948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9205028481002086e-05,
+      "loss": 2.0182,
+      "step": 5949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920475079192654e-05,
+      "loss": 2.0312,
+      "step": 5950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9204473056368394e-05,
+      "loss": 2.0422,
+      "step": 5951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9204195274329053e-05,
+      "loss": 2.0278,
+      "step": 5952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920391744580992e-05,
+      "loss": 2.0338,
+      "step": 5953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9203639570812397e-05,
+      "loss": 2.0422,
+      "step": 5954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920336164933789e-05,
+      "loss": 2.0305,
+      "step": 5955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9203083681387797e-05,
+      "loss": 2.0307,
+      "step": 5956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9202805666963523e-05,
+      "loss": 2.0357,
+      "step": 5957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920252760606647e-05,
+      "loss": 2.0377,
+      "step": 5958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920224949869805e-05,
+      "loss": 2.0391,
+      "step": 5959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9201971344859663e-05,
+      "loss": 2.0346,
+      "step": 5960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9201693144552712e-05,
+      "loss": 2.0411,
+      "step": 5961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9201414897778604e-05,
+      "loss": 2.0473,
+      "step": 5962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920113660453874e-05,
+      "loss": 2.0464,
+      "step": 5963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9200858264834534e-05,
+      "loss": 2.031,
+      "step": 5964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920057987866738e-05,
+      "loss": 2.026,
+      "step": 5965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9200301446038696e-05,
+      "loss": 2.0184,
+      "step": 5966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920002296694988e-05,
+      "loss": 2.0596,
+      "step": 5967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919974444140234e-05,
+      "loss": 2.0398,
+      "step": 5968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9199465869397486e-05,
+      "loss": 2.0349,
+      "step": 5969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9199187250936714e-05,
+      "loss": 2.0154,
+      "step": 5970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9198908586021444e-05,
+      "loss": 2.043,
+      "step": 5971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9198629874653078e-05,
+      "loss": 2.0465,
+      "step": 5972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919835111683302e-05,
+      "loss": 2.0113,
+      "step": 5973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9198072312562684e-05,
+      "loss": 2.0273,
+      "step": 5974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9197793461843476e-05,
+      "loss": 2.0315,
+      "step": 5975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91975145646768e-05,
+      "loss": 2.0457,
+      "step": 5976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919723562106407e-05,
+      "loss": 2.0327,
+      "step": 5977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919695663100669e-05,
+      "loss": 2.0512,
+      "step": 5978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9196677594506072e-05,
+      "loss": 2.0365,
+      "step": 5979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9196398511563627e-05,
+      "loss": 2.0322,
+      "step": 5980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9196119382180757e-05,
+      "loss": 2.022,
+      "step": 5981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9195840206358876e-05,
+      "loss": 2.0478,
+      "step": 5982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9195560984099396e-05,
+      "loss": 2.0297,
+      "step": 5983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9195281715403723e-05,
+      "loss": 2.0293,
+      "step": 5984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9195002400273267e-05,
+      "loss": 2.0507,
+      "step": 5985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9194723038709444e-05,
+      "loss": 2.0268,
+      "step": 5986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919444363071366e-05,
+      "loss": 2.0508,
+      "step": 5987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9194164176287328e-05,
+      "loss": 2.0362,
+      "step": 5988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9193884675431857e-05,
+      "loss": 2.0296,
+      "step": 5989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919360512814866e-05,
+      "loss": 2.0305,
+      "step": 5990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919332553443915e-05,
+      "loss": 2.0307,
+      "step": 5991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9193045894304735e-05,
+      "loss": 2.0443,
+      "step": 5992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9192766207746835e-05,
+      "loss": 2.0092,
+      "step": 5993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919248647476685e-05,
+      "loss": 2.0275,
+      "step": 5994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91922066953662e-05,
+      "loss": 2.0581,
+      "step": 5995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9191926869546303e-05,
+      "loss": 2.0316,
+      "step": 5996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9191646997308563e-05,
+      "loss": 2.0321,
+      "step": 5997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9191367078654396e-05,
+      "loss": 2.0318,
+      "step": 5998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919108711358522e-05,
+      "loss": 2.0225,
+      "step": 5999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919080710210244e-05,
+      "loss": 2.0285,
+      "step": 6000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919052704420748e-05,
+      "loss": 2.0378,
+      "step": 6001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9190246939901745e-05,
+      "loss": 2.0507,
+      "step": 6002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9189966789186656e-05,
+      "loss": 2.0358,
+      "step": 6003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9189686592063625e-05,
+      "loss": 2.0474,
+      "step": 6004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9189406348534067e-05,
+      "loss": 2.0398,
+      "step": 6005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91891260585994e-05,
+      "loss": 2.0423,
+      "step": 6006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9188845722261035e-05,
+      "loss": 2.0329,
+      "step": 6007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.918856533952039e-05,
+      "loss": 2.0346,
+      "step": 6008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9188284910378882e-05,
+      "loss": 2.0278,
+      "step": 6009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9188004434837927e-05,
+      "loss": 2.0258,
+      "step": 6010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9187723912898936e-05,
+      "loss": 2.0384,
+      "step": 6011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9187443344563333e-05,
+      "loss": 2.0347,
+      "step": 6012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918716272983253e-05,
+      "loss": 2.0388,
+      "step": 6013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9186882068707943e-05,
+      "loss": 2.0608,
+      "step": 6014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9186601361190996e-05,
+      "loss": 2.0434,
+      "step": 6015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91863206072831e-05,
+      "loss": 2.0303,
+      "step": 6016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918603980698568e-05,
+      "loss": 2.0347,
+      "step": 6017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9185758960300145e-05,
+      "loss": 2.0285,
+      "step": 6018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9185478067227915e-05,
+      "loss": 2.0295,
+      "step": 6019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918519712777042e-05,
+      "loss": 2.0434,
+      "step": 6020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918491614192906e-05,
+      "loss": 2.0304,
+      "step": 6021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918463510970527e-05,
+      "loss": 2.0492,
+      "step": 6022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9184354031100458e-05,
+      "loss": 2.0195,
+      "step": 6023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9184072906116048e-05,
+      "loss": 2.0423,
+      "step": 6024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9183791734753465e-05,
+      "loss": 2.042,
+      "step": 6025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9183510517014123e-05,
+      "loss": 2.0292,
+      "step": 6026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918322925289944e-05,
+      "loss": 2.0327,
+      "step": 6027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9182947942410837e-05,
+      "loss": 2.0463,
+      "step": 6028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918266658554974e-05,
+      "loss": 2.0269,
+      "step": 6029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9182385182317564e-05,
+      "loss": 2.033,
+      "step": 6030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9182103732715736e-05,
+      "loss": 2.0298,
+      "step": 6031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918182223674567e-05,
+      "loss": 2.0203,
+      "step": 6032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9181540694408796e-05,
+      "loss": 2.0195,
+      "step": 6033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9181259105706527e-05,
+      "loss": 2.0282,
+      "step": 6034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9180977470640293e-05,
+      "loss": 2.0259,
+      "step": 6035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9180695789211507e-05,
+      "loss": 2.0362,
+      "step": 6036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9180414061421603e-05,
+      "loss": 2.0412,
+      "step": 6037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9180132287271994e-05,
+      "loss": 2.0094,
+      "step": 6038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91798504667641e-05,
+      "loss": 2.0299,
+      "step": 6039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917956859989936e-05,
+      "loss": 2.0295,
+      "step": 6040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9179286686679185e-05,
+      "loss": 2.0384,
+      "step": 6041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9179004727105003e-05,
+      "loss": 2.0512,
+      "step": 6042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917872272117823e-05,
+      "loss": 2.0088,
+      "step": 6043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91784406689003e-05,
+      "loss": 2.0394,
+      "step": 6044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9178158570272637e-05,
+      "loss": 2.0335,
+      "step": 6045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9177876425296658e-05,
+      "loss": 2.0395,
+      "step": 6046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9177594233973795e-05,
+      "loss": 2.0431,
+      "step": 6047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917731199630547e-05,
+      "loss": 2.0229,
+      "step": 6048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9177029712293108e-05,
+      "loss": 2.0297,
+      "step": 6049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917674738193813e-05,
+      "loss": 2.0252,
+      "step": 6050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9176465005241975e-05,
+      "loss": 2.0425,
+      "step": 6051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9176182582206055e-05,
+      "loss": 2.0326,
+      "step": 6052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9175900112831803e-05,
+      "loss": 2.0102,
+      "step": 6053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9175617597120643e-05,
+      "loss": 2.0443,
+      "step": 6054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9175335035074003e-05,
+      "loss": 2.0424,
+      "step": 6055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9175052426693312e-05,
+      "loss": 2.0214,
+      "step": 6056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9174769771979994e-05,
+      "loss": 2.032,
+      "step": 6057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9174487070935474e-05,
+      "loss": 2.0406,
+      "step": 6058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9174204323561186e-05,
+      "loss": 2.0504,
+      "step": 6059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9173921529858553e-05,
+      "loss": 2.0264,
+      "step": 6060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917363868982901e-05,
+      "loss": 2.0433,
+      "step": 6061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9173355803473975e-05,
+      "loss": 2.0559,
+      "step": 6062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917307287079488e-05,
+      "loss": 2.0607,
+      "step": 6063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9172789891793157e-05,
+      "loss": 2.0207,
+      "step": 6064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9172506866470235e-05,
+      "loss": 2.0321,
+      "step": 6065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917222379482754e-05,
+      "loss": 2.0458,
+      "step": 6066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91719406768665e-05,
+      "loss": 2.0393,
+      "step": 6067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917165751258855e-05,
+      "loss": 2.0501,
+      "step": 6068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917137430199512e-05,
+      "loss": 2.0509,
+      "step": 6069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917109104508764e-05,
+      "loss": 2.0125,
+      "step": 6070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9170807741867534e-05,
+      "loss": 2.012,
+      "step": 6071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9170524392336237e-05,
+      "loss": 2.0258,
+      "step": 6072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917024099649518e-05,
+      "loss": 2.0295,
+      "step": 6073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9169957554345795e-05,
+      "loss": 2.0376,
+      "step": 6074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916967406588951e-05,
+      "loss": 2.0411,
+      "step": 6075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916939053112776e-05,
+      "loss": 2.0213,
+      "step": 6076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9169106950061975e-05,
+      "loss": 2.0596,
+      "step": 6077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9168823322693588e-05,
+      "loss": 2.0391,
+      "step": 6078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916853964902403e-05,
+      "loss": 2.0446,
+      "step": 6079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9168255929054737e-05,
+      "loss": 2.0272,
+      "step": 6080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9167972162787136e-05,
+      "loss": 2.026,
+      "step": 6081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9167688350222665e-05,
+      "loss": 2.0448,
+      "step": 6082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9167404491362755e-05,
+      "loss": 2.0463,
+      "step": 6083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916712058620884e-05,
+      "loss": 2.0341,
+      "step": 6084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9166836634762353e-05,
+      "loss": 2.0254,
+      "step": 6085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916655263702473e-05,
+      "loss": 2.0433,
+      "step": 6086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9166268592997398e-05,
+      "loss": 2.0355,
+      "step": 6087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91659845026818e-05,
+      "loss": 2.0311,
+      "step": 6088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9165700366079368e-05,
+      "loss": 2.047,
+      "step": 6089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9165416183191537e-05,
+      "loss": 2.0349,
+      "step": 6090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9165131954019743e-05,
+      "loss": 2.0345,
+      "step": 6091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9164847678565418e-05,
+      "loss": 2.0432,
+      "step": 6092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9164563356829998e-05,
+      "loss": 2.0585,
+      "step": 6093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916427898881492e-05,
+      "loss": 2.0336,
+      "step": 6094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9163994574521623e-05,
+      "loss": 2.0407,
+      "step": 6095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9163710113951537e-05,
+      "loss": 2.0249,
+      "step": 6096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9163425607106103e-05,
+      "loss": 2.0274,
+      "step": 6097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9163141053986758e-05,
+      "loss": 2.0465,
+      "step": 6098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9162856454594935e-05,
+      "loss": 2.0284,
+      "step": 6099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9162571808932076e-05,
+      "loss": 2.0252,
+      "step": 6100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9162287116999616e-05,
+      "loss": 2.0304,
+      "step": 6101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9162002378798988e-05,
+      "loss": 2.0327,
+      "step": 6102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9161717594331638e-05,
+      "loss": 2.0264,
+      "step": 6103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9161432763599003e-05,
+      "loss": 2.0469,
+      "step": 6104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916114788660252e-05,
+      "loss": 2.0443,
+      "step": 6105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916086296334362e-05,
+      "loss": 2.0322,
+      "step": 6106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916057799382375e-05,
+      "loss": 2.0293,
+      "step": 6107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916029297804435e-05,
+      "loss": 2.0522,
+      "step": 6108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916000791600686e-05,
+      "loss": 2.0399,
+      "step": 6109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915972280771271e-05,
+      "loss": 2.034,
+      "step": 6110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9159437653163348e-05,
+      "loss": 2.0214,
+      "step": 6111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9159152452360215e-05,
+      "loss": 2.0238,
+      "step": 6112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9158867205304744e-05,
+      "loss": 2.0616,
+      "step": 6113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9158581911998383e-05,
+      "loss": 2.0173,
+      "step": 6114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915829657244257e-05,
+      "loss": 2.0216,
+      "step": 6115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915801118663874e-05,
+      "loss": 2.0375,
+      "step": 6116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9157725754588347e-05,
+      "loss": 2.0184,
+      "step": 6117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9157440276292823e-05,
+      "loss": 2.0257,
+      "step": 6118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9157154751753607e-05,
+      "loss": 2.0139,
+      "step": 6119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9156869180972148e-05,
+      "loss": 2.04,
+      "step": 6120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9156583563949887e-05,
+      "loss": 2.027,
+      "step": 6121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9156297900688263e-05,
+      "loss": 2.0314,
+      "step": 6122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915601219118872e-05,
+      "loss": 2.0356,
+      "step": 6123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91557264354527e-05,
+      "loss": 2.0406,
+      "step": 6124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9155440633481653e-05,
+      "loss": 2.0129,
+      "step": 6125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9155154785277012e-05,
+      "loss": 2.0377,
+      "step": 6126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9154868890840225e-05,
+      "loss": 2.0376,
+      "step": 6127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9154582950172736e-05,
+      "loss": 2.0448,
+      "step": 6128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9154296963275988e-05,
+      "loss": 2.0227,
+      "step": 6129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9154010930151427e-05,
+      "loss": 2.0501,
+      "step": 6130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9153724850800494e-05,
+      "loss": 2.0609,
+      "step": 6131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9153438725224638e-05,
+      "loss": 2.0092,
+      "step": 6132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91531525534253e-05,
+      "loss": 2.0462,
+      "step": 6133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9152866335403927e-05,
+      "loss": 2.0174,
+      "step": 6134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9152580071161968e-05,
+      "loss": 2.0176,
+      "step": 6135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9152293760700863e-05,
+      "loss": 2.0297,
+      "step": 6136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9152007404022056e-05,
+      "loss": 2.0269,
+      "step": 6137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9151721001127e-05,
+      "loss": 2.0437,
+      "step": 6138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915143455201714e-05,
+      "loss": 2.0423,
+      "step": 6139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915114805669392e-05,
+      "loss": 2.0457,
+      "step": 6140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9150861515158784e-05,
+      "loss": 2.0255,
+      "step": 6141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915057492741318e-05,
+      "loss": 2.0511,
+      "step": 6142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915028829345856e-05,
+      "loss": 2.0353,
+      "step": 6143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9150001613296375e-05,
+      "loss": 2.0132,
+      "step": 6144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.914971488692806e-05,
+      "loss": 2.0282,
+      "step": 6145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9149428114355072e-05,
+      "loss": 2.0341,
+      "step": 6146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9149141295578854e-05,
+      "loss": 2.0322,
+      "step": 6147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.914885443060086e-05,
+      "loss": 2.0466,
+      "step": 6148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9148567519422533e-05,
+      "loss": 2.0251,
+      "step": 6149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9148280562045327e-05,
+      "loss": 2.0339,
+      "step": 6150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.914799355847069e-05,
+      "loss": 2.0266,
+      "step": 6151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.914770650870007e-05,
+      "loss": 2.0313,
+      "step": 6152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.914741941273491e-05,
+      "loss": 2.0254,
+      "step": 6153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9147132270576675e-05,
+      "loss": 2.0308,
+      "step": 6154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91468450822268e-05,
+      "loss": 2.0233,
+      "step": 6155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9146557847686745e-05,
+      "loss": 2.0308,
+      "step": 6156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9146270566957957e-05,
+      "loss": 2.0247,
+      "step": 6157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9145983240041886e-05,
+      "loss": 2.0488,
+      "step": 6158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9145695866939985e-05,
+      "loss": 2.0542,
+      "step": 6159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9145408447653704e-05,
+      "loss": 2.0309,
+      "step": 6160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9145120982184493e-05,
+      "loss": 2.0369,
+      "step": 6161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9144833470533806e-05,
+      "loss": 2.0459,
+      "step": 6162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9144545912703092e-05,
+      "loss": 2.0368,
+      "step": 6163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9144258308693806e-05,
+      "loss": 2.0307,
+      "step": 6164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9143970658507404e-05,
+      "loss": 2.0109,
+      "step": 6165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9143682962145326e-05,
+      "loss": 2.0253,
+      "step": 6166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.914339521960904e-05,
+      "loss": 2.0486,
+      "step": 6167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9143107430899988e-05,
+      "loss": 2.0242,
+      "step": 6168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9142819596019627e-05,
+      "loss": 2.0321,
+      "step": 6169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.914253171496941e-05,
+      "loss": 2.0454,
+      "step": 6170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9142243787750797e-05,
+      "loss": 2.0346,
+      "step": 6171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9141955814365235e-05,
+      "loss": 1.9997,
+      "step": 6172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9141667794814176e-05,
+      "loss": 2.045,
+      "step": 6173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9141379729099078e-05,
+      "loss": 2.0371,
+      "step": 6174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91410916172214e-05,
+      "loss": 2.0382,
+      "step": 6175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9140803459182588e-05,
+      "loss": 2.0291,
+      "step": 6176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9140515254984105e-05,
+      "loss": 2.0363,
+      "step": 6177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9140227004627406e-05,
+      "loss": 2.0659,
+      "step": 6178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913993870811394e-05,
+      "loss": 2.0405,
+      "step": 6179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9139650365445168e-05,
+      "loss": 2.0397,
+      "step": 6180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9139361976622542e-05,
+      "loss": 2.041,
+      "step": 6181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9139073541647524e-05,
+      "loss": 2.0613,
+      "step": 6182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9138785060521567e-05,
+      "loss": 2.0266,
+      "step": 6183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913849653324613e-05,
+      "loss": 2.031,
+      "step": 6184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9138207959822667e-05,
+      "loss": 2.0278,
+      "step": 6185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9137919340252637e-05,
+      "loss": 2.0109,
+      "step": 6186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91376306745375e-05,
+      "loss": 2.0329,
+      "step": 6187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9137341962678707e-05,
+      "loss": 2.0215,
+      "step": 6188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9137053204677724e-05,
+      "loss": 2.0422,
+      "step": 6189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9136764400536e-05,
+      "loss": 2.0373,
+      "step": 6190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9136475550255002e-05,
+      "loss": 2.0457,
+      "step": 6191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9136186653836187e-05,
+      "loss": 2.0297,
+      "step": 6192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913589771128101e-05,
+      "loss": 2.0233,
+      "step": 6193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913560872259093e-05,
+      "loss": 2.0084,
+      "step": 6194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913531968776741e-05,
+      "loss": 2.0331,
+      "step": 6195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9135030606811907e-05,
+      "loss": 2.0343,
+      "step": 6196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9134741479725885e-05,
+      "loss": 2.0157,
+      "step": 6197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91344523065108e-05,
+      "loss": 2.0178,
+      "step": 6198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9134163087168112e-05,
+      "loss": 2.043,
+      "step": 6199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9133873821699286e-05,
+      "loss": 2.0369,
+      "step": 6200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9133584510105777e-05,
+      "loss": 2.0205,
+      "step": 6201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913329515238905e-05,
+      "loss": 2.0272,
+      "step": 6202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9133005748550563e-05,
+      "loss": 2.0265,
+      "step": 6203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9132716298591776e-05,
+      "loss": 2.0303,
+      "step": 6204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913242680251416e-05,
+      "loss": 2.0503,
+      "step": 6205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9132137260319168e-05,
+      "loss": 2.0193,
+      "step": 6206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9131847672008267e-05,
+      "loss": 2.0426,
+      "step": 6207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9131558037582916e-05,
+      "loss": 2.0489,
+      "step": 6208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913126835704458e-05,
+      "loss": 2.0551,
+      "step": 6209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9130978630394717e-05,
+      "loss": 2.0218,
+      "step": 6210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9130688857634795e-05,
+      "loss": 2.0237,
+      "step": 6211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913039903876628e-05,
+      "loss": 2.0422,
+      "step": 6212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913010917379063e-05,
+      "loss": 2.0384,
+      "step": 6213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9129819262709305e-05,
+      "loss": 2.0249,
+      "step": 6214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912952930552378e-05,
+      "loss": 2.0291,
+      "step": 6215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9129239302235514e-05,
+      "loss": 2.0518,
+      "step": 6216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912894925284597e-05,
+      "loss": 2.0427,
+      "step": 6217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9128659157356614e-05,
+      "loss": 2.0439,
+      "step": 6218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912836901576891e-05,
+      "loss": 2.0346,
+      "step": 6219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9128078828084323e-05,
+      "loss": 2.0389,
+      "step": 6220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912778859430432e-05,
+      "loss": 2.0252,
+      "step": 6221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912749831443037e-05,
+      "loss": 2.0475,
+      "step": 6222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912720798846393e-05,
+      "loss": 2.0267,
+      "step": 6223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9126917616406473e-05,
+      "loss": 2.0417,
+      "step": 6224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9126627198259464e-05,
+      "loss": 2.0215,
+      "step": 6225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9126336734024368e-05,
+      "loss": 2.0324,
+      "step": 6226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9126046223702654e-05,
+      "loss": 2.028,
+      "step": 6227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9125755667295784e-05,
+      "loss": 2.0141,
+      "step": 6228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9125465064805233e-05,
+      "loss": 2.0471,
+      "step": 6229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9125174416232465e-05,
+      "loss": 2.0206,
+      "step": 6230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9124883721578946e-05,
+      "loss": 2.0104,
+      "step": 6231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9124592980846143e-05,
+      "loss": 2.0281,
+      "step": 6232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9124302194035527e-05,
+      "loss": 2.0299,
+      "step": 6233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912401136114857e-05,
+      "loss": 2.0348,
+      "step": 6234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9123720482186732e-05,
+      "loss": 2.0344,
+      "step": 6235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9123429557151486e-05,
+      "loss": 2.0384,
+      "step": 6236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9123138586044302e-05,
+      "loss": 2.024,
+      "step": 6237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912284756886665e-05,
+      "loss": 2.0276,
+      "step": 6238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912255650562e-05,
+      "loss": 2.0318,
+      "step": 6239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912226539630582e-05,
+      "loss": 2.0378,
+      "step": 6240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912197424092558e-05,
+      "loss": 2.0404,
+      "step": 6241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9121683039480748e-05,
+      "loss": 2.0307,
+      "step": 6242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91213917919728e-05,
+      "loss": 2.0302,
+      "step": 6243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91211004984032e-05,
+      "loss": 2.0332,
+      "step": 6244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912080915877343e-05,
+      "loss": 2.0329,
+      "step": 6245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9120517773084946e-05,
+      "loss": 2.0285,
+      "step": 6246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9120226341339233e-05,
+      "loss": 2.0309,
+      "step": 6247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9119934863537756e-05,
+      "loss": 2.0252,
+      "step": 6248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9119643339681987e-05,
+      "loss": 2.0357,
+      "step": 6249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9119351769773403e-05,
+      "loss": 2.0319,
+      "step": 6250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9119060153813467e-05,
+      "loss": 2.0364,
+      "step": 6251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911876849180366e-05,
+      "loss": 2.0205,
+      "step": 6252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9118476783745455e-05,
+      "loss": 2.0447,
+      "step": 6253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9118185029640317e-05,
+      "loss": 2.0256,
+      "step": 6254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911789322948973e-05,
+      "loss": 2.0408,
+      "step": 6255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9117601383295156e-05,
+      "loss": 2.0386,
+      "step": 6256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911730949105808e-05,
+      "loss": 2.0416,
+      "step": 6257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911701755277997e-05,
+      "loss": 2.0579,
+      "step": 6258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91167255684623e-05,
+      "loss": 2.0329,
+      "step": 6259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9116433538106546e-05,
+      "loss": 2.0258,
+      "step": 6260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9116141461714184e-05,
+      "loss": 2.0447,
+      "step": 6261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9115849339286685e-05,
+      "loss": 2.0548,
+      "step": 6262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911555717082553e-05,
+      "loss": 2.0276,
+      "step": 6263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9115264956332186e-05,
+      "loss": 2.0377,
+      "step": 6264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9114972695808136e-05,
+      "loss": 2.0294,
+      "step": 6265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9114680389254852e-05,
+      "loss": 2.0234,
+      "step": 6266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911438803667381e-05,
+      "loss": 2.0399,
+      "step": 6267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9114095638066494e-05,
+      "loss": 2.0196,
+      "step": 6268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911380319343437e-05,
+      "loss": 2.0228,
+      "step": 6269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911351070277892e-05,
+      "loss": 2.0124,
+      "step": 6270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911321816610162e-05,
+      "loss": 2.0398,
+      "step": 6271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9112925583403947e-05,
+      "loss": 2.0362,
+      "step": 6272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911263295468738e-05,
+      "loss": 2.0317,
+      "step": 6273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9112340279953394e-05,
+      "loss": 2.0341,
+      "step": 6274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9112047559203472e-05,
+      "loss": 2.0137,
+      "step": 6275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911175479243909e-05,
+      "loss": 2.0437,
+      "step": 6276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911146197966172e-05,
+      "loss": 2.0374,
+      "step": 6277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9111169120872848e-05,
+      "loss": 2.0202,
+      "step": 6278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911087621607395e-05,
+      "loss": 2.021,
+      "step": 6279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9110583265266507e-05,
+      "loss": 1.9979,
+      "step": 6280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9110290268451997e-05,
+      "loss": 2.038,
+      "step": 6281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91099972256319e-05,
+      "loss": 2.0237,
+      "step": 6282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9109704136807695e-05,
+      "loss": 2.0132,
+      "step": 6283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9109411001980865e-05,
+      "loss": 2.0373,
+      "step": 6284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910911782115289e-05,
+      "loss": 2.0325,
+      "step": 6285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9108824594325243e-05,
+      "loss": 2.0239,
+      "step": 6286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9108531321499412e-05,
+      "loss": 2.0304,
+      "step": 6287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9108238002676877e-05,
+      "loss": 2.0321,
+      "step": 6288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910794463785912e-05,
+      "loss": 2.0397,
+      "step": 6289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910765122704762e-05,
+      "loss": 2.0383,
+      "step": 6290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910735777024386e-05,
+      "loss": 2.0198,
+      "step": 6291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910706426744932e-05,
+      "loss": 2.0249,
+      "step": 6292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9106770718665486e-05,
+      "loss": 2.0323,
+      "step": 6293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9106477123893837e-05,
+      "loss": 2.0301,
+      "step": 6294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9106183483135857e-05,
+      "loss": 2.0234,
+      "step": 6295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9105889796393028e-05,
+      "loss": 2.0282,
+      "step": 6296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9105596063666833e-05,
+      "loss": 2.027,
+      "step": 6297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9105302284958757e-05,
+      "loss": 2.0323,
+      "step": 6298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9105008460270283e-05,
+      "loss": 2.0355,
+      "step": 6299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9104714589602892e-05,
+      "loss": 1.9942,
+      "step": 6300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9104420672958076e-05,
+      "loss": 2.0351,
+      "step": 6301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9104126710337307e-05,
+      "loss": 2.015,
+      "step": 6302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910383270174208e-05,
+      "loss": 2.0244,
+      "step": 6303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9103538647173875e-05,
+      "loss": 2.0408,
+      "step": 6304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9103244546634173e-05,
+      "loss": 2.0388,
+      "step": 6305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910295040012447e-05,
+      "loss": 2.0348,
+      "step": 6306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9102656207646243e-05,
+      "loss": 2.0307,
+      "step": 6307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910236196920098e-05,
+      "loss": 2.0195,
+      "step": 6308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9102067684790164e-05,
+      "loss": 2.0243,
+      "step": 6309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9101773354415285e-05,
+      "loss": 2.0236,
+      "step": 6310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910147897807783e-05,
+      "loss": 2.0439,
+      "step": 6311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910118455577928e-05,
+      "loss": 2.0378,
+      "step": 6312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910089008752113e-05,
+      "loss": 2.0337,
+      "step": 6313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910059557330486e-05,
+      "loss": 2.0264,
+      "step": 6314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9100301013131962e-05,
+      "loss": 2.0189,
+      "step": 6315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910000640700392e-05,
+      "loss": 2.024,
+      "step": 6316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909971175492222e-05,
+      "loss": 2.0451,
+      "step": 6317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9099417056888354e-05,
+      "loss": 2.0292,
+      "step": 6318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909912231290381e-05,
+      "loss": 2.0296,
+      "step": 6319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9098827522970076e-05,
+      "loss": 2.0368,
+      "step": 6320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909853268708864e-05,
+      "loss": 2.0348,
+      "step": 6321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909823780526099e-05,
+      "loss": 2.0266,
+      "step": 6322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9097942877488617e-05,
+      "loss": 2.0143,
+      "step": 6323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9097647903773008e-05,
+      "loss": 2.0152,
+      "step": 6324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9097352884115653e-05,
+      "loss": 2.0426,
+      "step": 6325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9097057818518047e-05,
+      "loss": 2.0406,
+      "step": 6326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909676270698167e-05,
+      "loss": 2.0302,
+      "step": 6327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9096467549508022e-05,
+      "loss": 2.0373,
+      "step": 6328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909617234609859e-05,
+      "loss": 2.0266,
+      "step": 6329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909587709675486e-05,
+      "loss": 2.045,
+      "step": 6330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9095581801478335e-05,
+      "loss": 2.0215,
+      "step": 6331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9095286460270492e-05,
+      "loss": 2.0351,
+      "step": 6332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9094991073132833e-05,
+      "loss": 2.039,
+      "step": 6333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909469564006684e-05,
+      "loss": 2.0104,
+      "step": 6334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9094400161074013e-05,
+      "loss": 2.0449,
+      "step": 6335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9094104636155843e-05,
+      "loss": 2.0506,
+      "step": 6336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909380906531382e-05,
+      "loss": 2.0745,
+      "step": 6337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909351344854944e-05,
+      "loss": 2.0222,
+      "step": 6338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9093217785864187e-05,
+      "loss": 2.0072,
+      "step": 6339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9092922077259567e-05,
+      "loss": 2.0391,
+      "step": 6340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9092626322737065e-05,
+      "loss": 2.0341,
+      "step": 6341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9092330522298173e-05,
+      "loss": 2.0237,
+      "step": 6342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909203467594439e-05,
+      "loss": 2.0255,
+      "step": 6343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909173878367721e-05,
+      "loss": 2.0255,
+      "step": 6344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9091442845498123e-05,
+      "loss": 2.0093,
+      "step": 6345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9091146861408628e-05,
+      "loss": 2.047,
+      "step": 6346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9090850831410214e-05,
+      "loss": 2.0133,
+      "step": 6347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909055475550438e-05,
+      "loss": 2.0327,
+      "step": 6348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909025863369262e-05,
+      "loss": 2.0444,
+      "step": 6349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908996246597643e-05,
+      "loss": 2.0148,
+      "step": 6350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9089666252357307e-05,
+      "loss": 2.0331,
+      "step": 6351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9089369992836742e-05,
+      "loss": 2.0331,
+      "step": 6352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908907368741624e-05,
+      "loss": 2.0284,
+      "step": 6353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9088777336097283e-05,
+      "loss": 2.0396,
+      "step": 6354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9088480938881382e-05,
+      "loss": 2.0394,
+      "step": 6355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9088184495770025e-05,
+      "loss": 2.036,
+      "step": 6356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908788800676471e-05,
+      "loss": 2.0341,
+      "step": 6357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908759147186694e-05,
+      "loss": 2.0517,
+      "step": 6358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9087294891078206e-05,
+      "loss": 2.0284,
+      "step": 6359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908699826440001e-05,
+      "loss": 2.0153,
+      "step": 6360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9086701591833847e-05,
+      "loss": 2.0357,
+      "step": 6361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9086404873381217e-05,
+      "loss": 2.0386,
+      "step": 6362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908610810904362e-05,
+      "loss": 2.0486,
+      "step": 6363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9085811298822547e-05,
+      "loss": 2.0263,
+      "step": 6364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9085514442719504e-05,
+      "loss": 2.0202,
+      "step": 6365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908521754073599e-05,
+      "loss": 2.0416,
+      "step": 6366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084920592873497e-05,
+      "loss": 2.0222,
+      "step": 6367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084623599133536e-05,
+      "loss": 2.0394,
+      "step": 6368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084326559517595e-05,
+      "loss": 2.013,
+      "step": 6369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084029474027184e-05,
+      "loss": 2.026,
+      "step": 6370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90837323426638e-05,
+      "loss": 2.0096,
+      "step": 6371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908343516542894e-05,
+      "loss": 2.0355,
+      "step": 6372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9083137942324108e-05,
+      "loss": 2.0236,
+      "step": 6373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9082840673350806e-05,
+      "loss": 2.0469,
+      "step": 6374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9082543358510527e-05,
+      "loss": 2.0542,
+      "step": 6375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9082245997804786e-05,
+      "loss": 2.0298,
+      "step": 6376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908194859123507e-05,
+      "loss": 2.0239,
+      "step": 6377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9081651138802893e-05,
+      "loss": 2.0361,
+      "step": 6378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908135364050975e-05,
+      "loss": 2.0342,
+      "step": 6379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908105609635715e-05,
+      "loss": 2.0109,
+      "step": 6380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9080758506346586e-05,
+      "loss": 2.01,
+      "step": 6381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9080460870479567e-05,
+      "loss": 2.0295,
+      "step": 6382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9080163188757595e-05,
+      "loss": 2.0402,
+      "step": 6383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9079865461182173e-05,
+      "loss": 2.0356,
+      "step": 6384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9079567687754808e-05,
+      "loss": 2.0456,
+      "step": 6385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9079269868476996e-05,
+      "loss": 2.0146,
+      "step": 6386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9078972003350247e-05,
+      "loss": 2.0228,
+      "step": 6387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907867409237606e-05,
+      "loss": 2.0433,
+      "step": 6388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9078376135555946e-05,
+      "loss": 2.0423,
+      "step": 6389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907807813289141e-05,
+      "loss": 2.0192,
+      "step": 6390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907778008438395e-05,
+      "loss": 2.0379,
+      "step": 6391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9077481990035073e-05,
+      "loss": 2.0347,
+      "step": 6392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9077183849846287e-05,
+      "loss": 2.022,
+      "step": 6393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9076885663819097e-05,
+      "loss": 2.0215,
+      "step": 6394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9076587431955006e-05,
+      "loss": 2.036,
+      "step": 6395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9076289154255524e-05,
+      "loss": 2.0321,
+      "step": 6396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075990830722157e-05,
+      "loss": 2.0228,
+      "step": 6397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075692461356406e-05,
+      "loss": 2.0354,
+      "step": 6398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075394046159786e-05,
+      "loss": 2.0328,
+      "step": 6399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075095585133795e-05,
+      "loss": 2.0509,
+      "step": 6400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9074797078279947e-05,
+      "loss": 2.0279,
+      "step": 6401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9074498525599746e-05,
+      "loss": 2.0321,
+      "step": 6402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9074199927094703e-05,
+      "loss": 2.0211,
+      "step": 6403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9073901282766323e-05,
+      "loss": 2.0385,
+      "step": 6404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9073602592616114e-05,
+      "loss": 2.0343,
+      "step": 6405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9073303856645586e-05,
+      "loss": 2.0146,
+      "step": 6406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9073005074856244e-05,
+      "loss": 2.0232,
+      "step": 6407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9072706247249603e-05,
+      "loss": 2.0485,
+      "step": 6408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9072407373827166e-05,
+      "loss": 2.0365,
+      "step": 6409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907210845459045e-05,
+      "loss": 2.0507,
+      "step": 6410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9071809489540954e-05,
+      "loss": 2.0413,
+      "step": 6411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907151047868019e-05,
+      "loss": 2.0185,
+      "step": 6412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9071211422009678e-05,
+      "loss": 2.0281,
+      "step": 6413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9070912319530918e-05,
+      "loss": 2.0313,
+      "step": 6414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9070613171245426e-05,
+      "loss": 2.0233,
+      "step": 6415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907031397715471e-05,
+      "loss": 2.0415,
+      "step": 6416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907001473726028e-05,
+      "loss": 2.0509,
+      "step": 6417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9069715451563648e-05,
+      "loss": 2.0353,
+      "step": 6418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9069416120066325e-05,
+      "loss": 2.0214,
+      "step": 6419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9069116742769826e-05,
+      "loss": 2.0365,
+      "step": 6420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9068817319675655e-05,
+      "loss": 2.0211,
+      "step": 6421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9068517850785332e-05,
+      "loss": 2.0273,
+      "step": 6422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9068218336100367e-05,
+      "loss": 2.0541,
+      "step": 6423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906791877562227e-05,
+      "loss": 2.0536,
+      "step": 6424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9067619169352558e-05,
+      "loss": 2.0252,
+      "step": 6425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906731951729274e-05,
+      "loss": 2.0401,
+      "step": 6426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906701981944433e-05,
+      "loss": 2.0383,
+      "step": 6427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9066720075808842e-05,
+      "loss": 2.0163,
+      "step": 6428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906642028638779e-05,
+      "loss": 2.0445,
+      "step": 6429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906612045118269e-05,
+      "loss": 2.0254,
+      "step": 6430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906582057019505e-05,
+      "loss": 2.0263,
+      "step": 6431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9065520643426393e-05,
+      "loss": 2.0484,
+      "step": 6432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9065220670878226e-05,
+      "loss": 2.0199,
+      "step": 6433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9064920652552067e-05,
+      "loss": 2.0359,
+      "step": 6434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906462058844943e-05,
+      "loss": 2.0151,
+      "step": 6435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906432047857183e-05,
+      "loss": 2.0125,
+      "step": 6436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9064020322920785e-05,
+      "loss": 2.023,
+      "step": 6437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906372012149781e-05,
+      "loss": 2.0242,
+      "step": 6438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906341987430442e-05,
+      "loss": 2.0416,
+      "step": 6439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906311958134213e-05,
+      "loss": 2.0213,
+      "step": 6440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9062819242612457e-05,
+      "loss": 2.0119,
+      "step": 6441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9062518858116917e-05,
+      "loss": 2.0377,
+      "step": 6442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9062218427857034e-05,
+      "loss": 2.0292,
+      "step": 6443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9061917951834316e-05,
+      "loss": 2.0341,
+      "step": 6444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9061617430050283e-05,
+      "loss": 2.0238,
+      "step": 6445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9061316862506457e-05,
+      "loss": 2.0368,
+      "step": 6446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9061016249204347e-05,
+      "loss": 2.0494,
+      "step": 6447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906071559014548e-05,
+      "loss": 2.0504,
+      "step": 6448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906041488533137e-05,
+      "loss": 2.0358,
+      "step": 6449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9060114134763537e-05,
+      "loss": 2.0455,
+      "step": 6450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9059813338443495e-05,
+      "loss": 2.0385,
+      "step": 6451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905951249637277e-05,
+      "loss": 2.0294,
+      "step": 6452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9059211608552876e-05,
+      "loss": 2.0045,
+      "step": 6453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9058910674985335e-05,
+      "loss": 2.0447,
+      "step": 6454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905860969567167e-05,
+      "loss": 2.0272,
+      "step": 6455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905830867061339e-05,
+      "loss": 2.0327,
+      "step": 6456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9058007599812028e-05,
+      "loss": 2.026,
+      "step": 6457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9057706483269097e-05,
+      "loss": 2.0301,
+      "step": 6458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905740532098612e-05,
+      "loss": 2.0276,
+      "step": 6459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9057104112964615e-05,
+      "loss": 2.0354,
+      "step": 6460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9056802859206104e-05,
+      "loss": 2.0112,
+      "step": 6461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905650155971211e-05,
+      "loss": 2.0112,
+      "step": 6462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9056200214484154e-05,
+      "loss": 2.037,
+      "step": 6463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905589882352376e-05,
+      "loss": 2.0275,
+      "step": 6464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9055597386832445e-05,
+      "loss": 2.032,
+      "step": 6465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9055295904411735e-05,
+      "loss": 1.9945,
+      "step": 6466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9054994376263148e-05,
+      "loss": 2.0483,
+      "step": 6467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9054692802388213e-05,
+      "loss": 2.0278,
+      "step": 6468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905439118278845e-05,
+      "loss": 2.0446,
+      "step": 6469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9054089517465378e-05,
+      "loss": 2.056,
+      "step": 6470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9053787806420528e-05,
+      "loss": 2.0129,
+      "step": 6471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9053486049655417e-05,
+      "loss": 2.0168,
+      "step": 6472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9053184247171575e-05,
+      "loss": 2.027,
+      "step": 6473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9052882398970522e-05,
+      "loss": 2.0308,
+      "step": 6474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9052580505053783e-05,
+      "loss": 2.0307,
+      "step": 6475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905227856542288e-05,
+      "loss": 2.0324,
+      "step": 6476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9051976580079344e-05,
+      "loss": 2.0419,
+      "step": 6477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905167454902469e-05,
+      "loss": 2.0377,
+      "step": 6478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9051372472260454e-05,
+      "loss": 2.0242,
+      "step": 6479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9051070349788157e-05,
+      "loss": 2.0374,
+      "step": 6480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9050768181609325e-05,
+      "loss": 2.0438,
+      "step": 6481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9050465967725484e-05,
+      "loss": 2.0243,
+      "step": 6482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9050163708138156e-05,
+      "loss": 2.0318,
+      "step": 6483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9049861402848872e-05,
+      "loss": 2.0335,
+      "step": 6484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9049559051859158e-05,
+      "loss": 2.0367,
+      "step": 6485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9049256655170542e-05,
+      "loss": 2.0358,
+      "step": 6486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9048954212784547e-05,
+      "loss": 2.0444,
+      "step": 6487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9048651724702706e-05,
+      "loss": 2.039,
+      "step": 6488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904834919092654e-05,
+      "loss": 2.007,
+      "step": 6489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904804661145758e-05,
+      "loss": 2.0383,
+      "step": 6490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904774398629736e-05,
+      "loss": 2.0301,
+      "step": 6491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9047441315447394e-05,
+      "loss": 2.0294,
+      "step": 6492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9047138598909224e-05,
+      "loss": 2.0019,
+      "step": 6493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904683583668437e-05,
+      "loss": 2.0322,
+      "step": 6494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9046533028774368e-05,
+      "loss": 2.0196,
+      "step": 6495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9046230175180738e-05,
+      "loss": 2.0375,
+      "step": 6496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9045927275905018e-05,
+      "loss": 2.0271,
+      "step": 6497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9045624330948734e-05,
+      "loss": 2.054,
+      "step": 6498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904532134031342e-05,
+      "loss": 2.0255,
+      "step": 6499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9045018304000596e-05,
+      "loss": 2.0406,
+      "step": 6500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.90447152220118e-05,
+      "loss": 2.0198,
+      "step": 6501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904441209434856e-05,
+      "loss": 2.0148,
+      "step": 6502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9044108921012413e-05,
+      "loss": 2.019,
+      "step": 6503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904380570200488e-05,
+      "loss": 2.0287,
+      "step": 6504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9043502437327495e-05,
+      "loss": 2.0157,
+      "step": 6505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9043199126981797e-05,
+      "loss": 2.0387,
+      "step": 6506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904289577096931e-05,
+      "loss": 2.0257,
+      "step": 6507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9042592369291567e-05,
+      "loss": 2.0337,
+      "step": 6508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9042288921950104e-05,
+      "loss": 2.0216,
+      "step": 6509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9041985428946447e-05,
+      "loss": 2.0039,
+      "step": 6510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904168189028213e-05,
+      "loss": 2.027,
+      "step": 6511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9041378305958692e-05,
+      "loss": 2.0396,
+      "step": 6512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904107467597766e-05,
+      "loss": 2.0236,
+      "step": 6513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904077100034057e-05,
+      "loss": 2.0315,
+      "step": 6514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9040467279048958e-05,
+      "loss": 2.044,
+      "step": 6515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904016351210435e-05,
+      "loss": 2.034,
+      "step": 6516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9039859699508287e-05,
+      "loss": 2.0488,
+      "step": 6517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9039555841262295e-05,
+      "loss": 2.0027,
+      "step": 6518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9039251937367922e-05,
+      "loss": 2.0323,
+      "step": 6519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903894798782669e-05,
+      "loss": 2.0429,
+      "step": 6520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903864399264014e-05,
+      "loss": 2.0366,
+      "step": 6521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9038339951809805e-05,
+      "loss": 2.0287,
+      "step": 6522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903803586533722e-05,
+      "loss": 2.0466,
+      "step": 6523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9037731733223924e-05,
+      "loss": 2.0366,
+      "step": 6524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9037427555471453e-05,
+      "loss": 2.0224,
+      "step": 6525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9037123332081334e-05,
+      "loss": 2.041,
+      "step": 6526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9036819063055114e-05,
+      "loss": 2.0092,
+      "step": 6527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9036514748394326e-05,
+      "loss": 2.0353,
+      "step": 6528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9036210388100508e-05,
+      "loss": 2.0088,
+      "step": 6529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903590598217519e-05,
+      "loss": 2.0079,
+      "step": 6530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9035601530619914e-05,
+      "loss": 2.04,
+      "step": 6531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903529703343622e-05,
+      "loss": 2.0056,
+      "step": 6532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9034992490625645e-05,
+      "loss": 2.0421,
+      "step": 6533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9034687902189724e-05,
+      "loss": 2.0376,
+      "step": 6534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9034383268129996e-05,
+      "loss": 2.0278,
+      "step": 6535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9034078588448003e-05,
+      "loss": 2.0263,
+      "step": 6536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9033773863145276e-05,
+      "loss": 2.0288,
+      "step": 6537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9033469092223357e-05,
+      "loss": 2.0266,
+      "step": 6538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9033164275683794e-05,
+      "loss": 2.0135,
+      "step": 6539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9032859413528112e-05,
+      "loss": 2.0392,
+      "step": 6540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903255450575786e-05,
+      "loss": 2.0125,
+      "step": 6541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903224955237457e-05,
+      "loss": 2.0206,
+      "step": 6542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903194455337979e-05,
+      "loss": 2.0314,
+      "step": 6543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903163950877506e-05,
+      "loss": 2.0428,
+      "step": 6544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9031334418561912e-05,
+      "loss": 2.0218,
+      "step": 6545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9031029282741896e-05,
+      "loss": 2.0516,
+      "step": 6546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9030724101316545e-05,
+      "loss": 2.0298,
+      "step": 6547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9030418874287407e-05,
+      "loss": 2.0326,
+      "step": 6548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903011360165602e-05,
+      "loss": 2.0256,
+      "step": 6549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9029808283423926e-05,
+      "loss": 2.0241,
+      "step": 6550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9029502919592666e-05,
+      "loss": 2.028,
+      "step": 6551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9029197510163783e-05,
+      "loss": 2.0489,
+      "step": 6552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.902889205513882e-05,
+      "loss": 2.0382,
+      "step": 6553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.902858655451932e-05,
+      "loss": 2.0266,
+      "step": 6554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9028281008306822e-05,
+      "loss": 2.0385,
+      "step": 6555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.902797541650287e-05,
+      "loss": 2.0278,
+      "step": 6556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9027669779109012e-05,
+      "loss": 2.0101,
+      "step": 6557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9027364096126783e-05,
+      "loss": 2.0265,
+      "step": 6558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9027058367557735e-05,
+      "loss": 2.0296,
+      "step": 6559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.902675259340341e-05,
+      "loss": 2.0438,
+      "step": 6560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9026446773665346e-05,
+      "loss": 2.0245,
+      "step": 6561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9026140908345095e-05,
+      "loss": 2.0165,
+      "step": 6562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9025834997444196e-05,
+      "loss": 2.0217,
+      "step": 6563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9025529040964203e-05,
+      "loss": 2.0075,
+      "step": 6564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9025223038906647e-05,
+      "loss": 2.0149,
+      "step": 6565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9024916991273082e-05,
+      "loss": 2.0424,
+      "step": 6566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.902461089806505e-05,
+      "loss": 2.0278,
+      "step": 6567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9024304759284106e-05,
+      "loss": 2.0227,
+      "step": 6568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9023998574931782e-05,
+      "loss": 2.0434,
+      "step": 6569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9023692345009635e-05,
+      "loss": 2.0117,
+      "step": 6570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9023386069519206e-05,
+      "loss": 2.0257,
+      "step": 6571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9023079748462043e-05,
+      "loss": 2.0401,
+      "step": 6572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.902277338183969e-05,
+      "loss": 2.0391,
+      "step": 6573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9022466969653702e-05,
+      "loss": 2.0304,
+      "step": 6574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9022160511905618e-05,
+      "loss": 2.0207,
+      "step": 6575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.902185400859699e-05,
+      "loss": 2.0311,
+      "step": 6576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9021547459729364e-05,
+      "loss": 2.0153,
+      "step": 6577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9021240865304288e-05,
+      "loss": 2.0484,
+      "step": 6578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9020934225323313e-05,
+      "loss": 2.0431,
+      "step": 6579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9020627539787984e-05,
+      "loss": 2.0215,
+      "step": 6580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9020320808699854e-05,
+      "loss": 2.0254,
+      "step": 6581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9020014032060467e-05,
+      "loss": 2.0189,
+      "step": 6582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9019707209871372e-05,
+      "loss": 2.0377,
+      "step": 6583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9019400342134124e-05,
+      "loss": 2.0599,
+      "step": 6584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9019093428850267e-05,
+      "loss": 2.0308,
+      "step": 6585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9018786470021354e-05,
+      "loss": 2.0141,
+      "step": 6586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9018479465648936e-05,
+      "loss": 2.0252,
+      "step": 6587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901817241573456e-05,
+      "loss": 2.0365,
+      "step": 6588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9017865320279777e-05,
+      "loss": 2.0342,
+      "step": 6589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901755817928614e-05,
+      "loss": 2.0328,
+      "step": 6590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.90172509927552e-05,
+      "loss": 2.0311,
+      "step": 6591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9016943760688508e-05,
+      "loss": 2.0392,
+      "step": 6592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901663648308761e-05,
+      "loss": 2.0137,
+      "step": 6593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9016329159954067e-05,
+      "loss": 2.0347,
+      "step": 6594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9016021791289424e-05,
+      "loss": 2.0353,
+      "step": 6595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9015714377095236e-05,
+      "loss": 2.0055,
+      "step": 6596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9015406917373058e-05,
+      "loss": 2.0154,
+      "step": 6597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9015099412124434e-05,
+      "loss": 2.038,
+      "step": 6598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9014791861350926e-05,
+      "loss": 2.0164,
+      "step": 6599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9014484265054077e-05,
+      "loss": 2.0531,
+      "step": 6600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9014176623235454e-05,
+      "loss": 2.0288,
+      "step": 6601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9013868935896604e-05,
+      "loss": 2.0545,
+      "step": 6602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9013561203039077e-05,
+      "loss": 2.0245,
+      "step": 6603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9013253424664428e-05,
+      "loss": 2.0129,
+      "step": 6604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9012945600774217e-05,
+      "loss": 2.019,
+      "step": 6605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901263773136999e-05,
+      "loss": 2.019,
+      "step": 6606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9012329816453313e-05,
+      "loss": 2.0427,
+      "step": 6607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901202185602573e-05,
+      "loss": 2.0005,
+      "step": 6608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9011713850088803e-05,
+      "loss": 2.0262,
+      "step": 6609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9011405798644085e-05,
+      "loss": 2.0514,
+      "step": 6610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901109770169313e-05,
+      "loss": 2.0291,
+      "step": 6611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9010789559237495e-05,
+      "loss": 2.038,
+      "step": 6612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9010481371278737e-05,
+      "loss": 2.0229,
+      "step": 6613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901017313781841e-05,
+      "loss": 2.0347,
+      "step": 6614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9009864858858077e-05,
+      "loss": 2.0136,
+      "step": 6615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9009556534399288e-05,
+      "loss": 2.0344,
+      "step": 6616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9009248164443602e-05,
+      "loss": 2.0304,
+      "step": 6617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9008939748992574e-05,
+      "loss": 2.0328,
+      "step": 6618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9008631288047765e-05,
+      "loss": 2.0308,
+      "step": 6619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9008322781610734e-05,
+      "loss": 2.0475,
+      "step": 6620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9008014229683033e-05,
+      "loss": 2.0275,
+      "step": 6621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9007705632266224e-05,
+      "loss": 2.0114,
+      "step": 6622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9007396989361865e-05,
+      "loss": 1.9989,
+      "step": 6623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9007088300971515e-05,
+      "loss": 2.0349,
+      "step": 6624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.900677956709673e-05,
+      "loss": 2.0303,
+      "step": 6625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9006470787739072e-05,
+      "loss": 2.026,
+      "step": 6626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9006161962900103e-05,
+      "loss": 2.0456,
+      "step": 6627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9005853092581377e-05,
+      "loss": 2.0229,
+      "step": 6628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9005544176784453e-05,
+      "loss": 2.0381,
+      "step": 6629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.90052352155109e-05,
+      "loss": 2.028,
+      "step": 6630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9004926208762267e-05,
+      "loss": 2.0413,
+      "step": 6631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9004617156540123e-05,
+      "loss": 2.0403,
+      "step": 6632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.900430805884602e-05,
+      "loss": 2.0182,
+      "step": 6633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.900399891568153e-05,
+      "loss": 2.0257,
+      "step": 6634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.900368972704821e-05,
+      "loss": 2.0269,
+      "step": 6635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.900338049294761e-05,
+      "loss": 2.0239,
+      "step": 6636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9003071213381306e-05,
+      "loss": 2.0212,
+      "step": 6637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9002761888350856e-05,
+      "loss": 2.022,
+      "step": 6638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9002452517857823e-05,
+      "loss": 2.0321,
+      "step": 6639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9002143101903762e-05,
+      "loss": 2.0117,
+      "step": 6640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9001833640490243e-05,
+      "loss": 2.0309,
+      "step": 6641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9001524133618824e-05,
+      "loss": 2.0243,
+      "step": 6642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9001214581291077e-05,
+      "loss": 2.0385,
+      "step": 6643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9000904983508553e-05,
+      "loss": 2.025,
+      "step": 6644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9000595340272822e-05,
+      "loss": 2.0371,
+      "step": 6645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.900028565158545e-05,
+      "loss": 2.0255,
+      "step": 6646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8999975917447995e-05,
+      "loss": 2.0123,
+      "step": 6647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899966613786202e-05,
+      "loss": 2.0387,
+      "step": 6648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8999356312829098e-05,
+      "loss": 2.0259,
+      "step": 6649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8999046442350788e-05,
+      "loss": 2.0303,
+      "step": 6650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8998736526428656e-05,
+      "loss": 2.0347,
+      "step": 6651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8998426565064267e-05,
+      "loss": 2.0202,
+      "step": 6652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8998116558259182e-05,
+      "loss": 2.0269,
+      "step": 6653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899780650601497e-05,
+      "loss": 2.0169,
+      "step": 6654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8997496408333203e-05,
+      "loss": 2.026,
+      "step": 6655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8997186265215435e-05,
+      "loss": 2.026,
+      "step": 6656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899687607666324e-05,
+      "loss": 2.0135,
+      "step": 6657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8996565842678183e-05,
+      "loss": 2.0361,
+      "step": 6658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899625556326183e-05,
+      "loss": 2.0352,
+      "step": 6659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8995945238415748e-05,
+      "loss": 2.0295,
+      "step": 6660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89956348681415e-05,
+      "loss": 2.0292,
+      "step": 6661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8995324452440663e-05,
+      "loss": 2.0271,
+      "step": 6662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8995013991314797e-05,
+      "loss": 2.0284,
+      "step": 6663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899470348476547e-05,
+      "loss": 2.0303,
+      "step": 6664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8994392932794254e-05,
+      "loss": 2.0251,
+      "step": 6665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8994082335402715e-05,
+      "loss": 2.0334,
+      "step": 6666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899377169259242e-05,
+      "loss": 2.0296,
+      "step": 6667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899346100436494e-05,
+      "loss": 2.0407,
+      "step": 6668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8993150270721843e-05,
+      "loss": 2.0328,
+      "step": 6669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8992839491664698e-05,
+      "loss": 2.021,
+      "step": 6670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8992528667195074e-05,
+      "loss": 2.0122,
+      "step": 6671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899221779731454e-05,
+      "loss": 1.9952,
+      "step": 6672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899190688202467e-05,
+      "loss": 2.0346,
+      "step": 6673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899159592132703e-05,
+      "loss": 2.0364,
+      "step": 6674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899128491522319e-05,
+      "loss": 2.0145,
+      "step": 6675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8990973863714724e-05,
+      "loss": 2.0437,
+      "step": 6676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89906627668032e-05,
+      "loss": 2.0265,
+      "step": 6677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899035162449019e-05,
+      "loss": 2.0329,
+      "step": 6678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8990040436777265e-05,
+      "loss": 2.0396,
+      "step": 6679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8989729203665994e-05,
+      "loss": 2.0241,
+      "step": 6680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8989417925157955e-05,
+      "loss": 2.0147,
+      "step": 6681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8989106601254714e-05,
+      "loss": 2.0287,
+      "step": 6682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8988795231957845e-05,
+      "loss": 2.028,
+      "step": 6683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8988483817268917e-05,
+      "loss": 2.0167,
+      "step": 6684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898817235718951e-05,
+      "loss": 2.0317,
+      "step": 6685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8987860851721192e-05,
+      "loss": 2.0263,
+      "step": 6686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898754930086554e-05,
+      "loss": 2.0245,
+      "step": 6687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898723770462412e-05,
+      "loss": 2.0324,
+      "step": 6688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8986926062998513e-05,
+      "loss": 2.0412,
+      "step": 6689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898661437599029e-05,
+      "loss": 2.0182,
+      "step": 6690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898630264360102e-05,
+      "loss": 2.0442,
+      "step": 6691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8985990865832285e-05,
+      "loss": 2.0188,
+      "step": 6692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8985679042685656e-05,
+      "loss": 2.0393,
+      "step": 6693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8985367174162706e-05,
+      "loss": 2.0308,
+      "step": 6694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8985055260265014e-05,
+      "loss": 2.0141,
+      "step": 6695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898474330099415e-05,
+      "loss": 2.0242,
+      "step": 6696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8984431296351694e-05,
+      "loss": 2.01,
+      "step": 6697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898411924633922e-05,
+      "loss": 2.044,
+      "step": 6698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89838071509583e-05,
+      "loss": 2.038,
+      "step": 6699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8983495010210516e-05,
+      "loss": 2.0313,
+      "step": 6700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8983182824097442e-05,
+      "loss": 2.0456,
+      "step": 6701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8982870592620654e-05,
+      "loss": 2.0413,
+      "step": 6702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8982558315781728e-05,
+      "loss": 2.012,
+      "step": 6703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8982245993582243e-05,
+      "loss": 2.017,
+      "step": 6704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8981933626023775e-05,
+      "loss": 2.0302,
+      "step": 6705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8981621213107905e-05,
+      "loss": 2.0334,
+      "step": 6706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8981308754836205e-05,
+      "loss": 2.0346,
+      "step": 6707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8980996251210252e-05,
+      "loss": 2.028,
+      "step": 6708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898068370223163e-05,
+      "loss": 2.0297,
+      "step": 6709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8980371107901914e-05,
+      "loss": 2.0164,
+      "step": 6710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8980058468222683e-05,
+      "loss": 2.0231,
+      "step": 6711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8979745783195517e-05,
+      "loss": 2.0237,
+      "step": 6712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897943305282199e-05,
+      "loss": 2.0198,
+      "step": 6713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897912027710369e-05,
+      "loss": 2.0351,
+      "step": 6714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8978807456042188e-05,
+      "loss": 2.0282,
+      "step": 6715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897849458963907e-05,
+      "loss": 2.0243,
+      "step": 6716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897818167789591e-05,
+      "loss": 2.0261,
+      "step": 6717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8977868720814293e-05,
+      "loss": 2.0201,
+      "step": 6718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8977555718395798e-05,
+      "loss": 2.0094,
+      "step": 6719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8977242670642005e-05,
+      "loss": 2.037,
+      "step": 6720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8976929577554494e-05,
+      "loss": 2.0309,
+      "step": 6721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8976616439134847e-05,
+      "loss": 2.0239,
+      "step": 6722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8976303255384647e-05,
+      "loss": 2.0233,
+      "step": 6723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8975990026305474e-05,
+      "loss": 2.0353,
+      "step": 6724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897567675189891e-05,
+      "loss": 2.0282,
+      "step": 6725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8975363432166537e-05,
+      "loss": 2.0456,
+      "step": 6726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8975050067109936e-05,
+      "loss": 2.0302,
+      "step": 6727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8974736656730694e-05,
+      "loss": 2.0476,
+      "step": 6728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8974423201030384e-05,
+      "loss": 2.0262,
+      "step": 6729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89741097000106e-05,
+      "loss": 2.0089,
+      "step": 6730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897379615367292e-05,
+      "loss": 2.0128,
+      "step": 6731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8973482562018925e-05,
+      "loss": 2.0366,
+      "step": 6732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8973168925050203e-05,
+      "loss": 2.0229,
+      "step": 6733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8972855242768333e-05,
+      "loss": 2.0437,
+      "step": 6734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8972541515174905e-05,
+      "loss": 2.0112,
+      "step": 6735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8972227742271497e-05,
+      "loss": 2.0151,
+      "step": 6736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89719139240597e-05,
+      "loss": 2.0358,
+      "step": 6737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8971600060541097e-05,
+      "loss": 2.0268,
+      "step": 6738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8971286151717266e-05,
+      "loss": 2.0075,
+      "step": 6739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89709721975898e-05,
+      "loss": 2.0131,
+      "step": 6740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8970658198160285e-05,
+      "loss": 2.0034,
+      "step": 6741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8970344153430302e-05,
+      "loss": 2.0344,
+      "step": 6742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8970030063401438e-05,
+      "loss": 2.0269,
+      "step": 6743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896971592807528e-05,
+      "loss": 2.0292,
+      "step": 6744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8969401747453414e-05,
+      "loss": 2.0192,
+      "step": 6745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896908752153743e-05,
+      "loss": 1.9965,
+      "step": 6746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8968773250328905e-05,
+      "loss": 2.0225,
+      "step": 6747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8968458933829438e-05,
+      "loss": 2.003,
+      "step": 6748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8968144572040607e-05,
+      "loss": 2.0377,
+      "step": 6749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8967830164964004e-05,
+      "loss": 2.0408,
+      "step": 6750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8967515712601218e-05,
+      "loss": 2.0329,
+      "step": 6751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8967201214953834e-05,
+      "loss": 2.0363,
+      "step": 6752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896688667202344e-05,
+      "loss": 2.0174,
+      "step": 6753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8966572083811626e-05,
+      "loss": 2.0181,
+      "step": 6754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8966257450319978e-05,
+      "loss": 2.0061,
+      "step": 6755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896594277155009e-05,
+      "loss": 2.0103,
+      "step": 6756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8965628047503546e-05,
+      "loss": 2.0107,
+      "step": 6757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8965313278181935e-05,
+      "loss": 2.0445,
+      "step": 6758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8964998463586855e-05,
+      "loss": 2.015,
+      "step": 6759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8964683603719885e-05,
+      "loss": 2.0301,
+      "step": 6760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896436869858262e-05,
+      "loss": 2.0191,
+      "step": 6761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8964053748176652e-05,
+      "loss": 2.0169,
+      "step": 6762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896373875250357e-05,
+      "loss": 2.0218,
+      "step": 6763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896342371156496e-05,
+      "loss": 2.0108,
+      "step": 6764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896310862536242e-05,
+      "loss": 2.0159,
+      "step": 6765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8962793493897534e-05,
+      "loss": 2.0229,
+      "step": 6766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89624783171719e-05,
+      "loss": 2.0183,
+      "step": 6767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8962163095187108e-05,
+      "loss": 2.0324,
+      "step": 6768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8961847827944745e-05,
+      "loss": 2.0166,
+      "step": 6769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8961532515446412e-05,
+      "loss": 2.0223,
+      "step": 6770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8961217157693693e-05,
+      "loss": 2.021,
+      "step": 6771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8960901754688185e-05,
+      "loss": 2.017,
+      "step": 6772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8960586306431476e-05,
+      "loss": 2.0177,
+      "step": 6773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8960270812925163e-05,
+      "loss": 2.0323,
+      "step": 6774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895995527417084e-05,
+      "loss": 2.0147,
+      "step": 6775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89596396901701e-05,
+      "loss": 2.0026,
+      "step": 6776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8959324060924534e-05,
+      "loss": 2.0171,
+      "step": 6777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8959008386435737e-05,
+      "loss": 2.0298,
+      "step": 6778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8958692666705305e-05,
+      "loss": 2.0212,
+      "step": 6779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895837690173483e-05,
+      "loss": 2.0275,
+      "step": 6780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895806109152591e-05,
+      "loss": 2.0252,
+      "step": 6781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8957745236080134e-05,
+      "loss": 2.0218,
+      "step": 6782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89574293353991e-05,
+      "loss": 2.0468,
+      "step": 6783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8957113389484405e-05,
+      "loss": 2.0209,
+      "step": 6784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895679739833764e-05,
+      "loss": 2.024,
+      "step": 6785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895648136196041e-05,
+      "loss": 2.02,
+      "step": 6786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89561652803543e-05,
+      "loss": 2.0531,
+      "step": 6787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895584915352091e-05,
+      "loss": 2.0345,
+      "step": 6788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895553298146184e-05,
+      "loss": 2.0299,
+      "step": 6789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8955216764178683e-05,
+      "loss": 2.0176,
+      "step": 6790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8954900501673034e-05,
+      "loss": 2.0198,
+      "step": 6791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89545841939465e-05,
+      "loss": 2.0055,
+      "step": 6792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8954267841000665e-05,
+      "loss": 2.0295,
+      "step": 6793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8953951442837134e-05,
+      "loss": 2.0233,
+      "step": 6794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8953634999457503e-05,
+      "loss": 2.0432,
+      "step": 6795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8953318510863368e-05,
+      "loss": 2.0152,
+      "step": 6796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8953001977056332e-05,
+      "loss": 2.0206,
+      "step": 6797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895268539803799e-05,
+      "loss": 2.0273,
+      "step": 6798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8952368773809945e-05,
+      "loss": 2.0265,
+      "step": 6799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895205210437379e-05,
+      "loss": 2.0258,
+      "step": 6800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8951735389731127e-05,
+      "loss": 2.005,
+      "step": 6801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8951418629883555e-05,
+      "loss": 2.0395,
+      "step": 6802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8951101824832673e-05,
+      "loss": 2.0129,
+      "step": 6803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8950784974580084e-05,
+      "loss": 2.0192,
+      "step": 6804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8950468079127382e-05,
+      "loss": 2.0178,
+      "step": 6805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895015113847617e-05,
+      "loss": 2.0309,
+      "step": 6806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8949834152628055e-05,
+      "loss": 2.023,
+      "step": 6807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894951712158463e-05,
+      "loss": 2.0372,
+      "step": 6808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8949200045347496e-05,
+      "loss": 2.0154,
+      "step": 6809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8948882923918258e-05,
+      "loss": 2.0263,
+      "step": 6810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8948565757298515e-05,
+      "loss": 2.0185,
+      "step": 6811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8948248545489866e-05,
+      "loss": 2.0196,
+      "step": 6812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8947931288493923e-05,
+      "loss": 2.0343,
+      "step": 6813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8947613986312276e-05,
+      "loss": 2.0345,
+      "step": 6814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894729663894653e-05,
+      "loss": 2.0175,
+      "step": 6815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8946979246398296e-05,
+      "loss": 2.0116,
+      "step": 6816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8946661808669168e-05,
+      "loss": 2.0409,
+      "step": 6817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8946344325760752e-05,
+      "loss": 2.0148,
+      "step": 6818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8946026797674652e-05,
+      "loss": 2.0394,
+      "step": 6819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894570922441247e-05,
+      "loss": 2.0284,
+      "step": 6820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8945391605975812e-05,
+      "loss": 2.031,
+      "step": 6821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8945073942366277e-05,
+      "loss": 2.0225,
+      "step": 6822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8944756233585474e-05,
+      "loss": 2.0192,
+      "step": 6823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8944438479635007e-05,
+      "loss": 2.0345,
+      "step": 6824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894412068051648e-05,
+      "loss": 2.0149,
+      "step": 6825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8943802836231494e-05,
+      "loss": 2.0161,
+      "step": 6826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894348494678166e-05,
+      "loss": 2.0063,
+      "step": 6827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894316701216858e-05,
+      "loss": 2.0363,
+      "step": 6828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894284903239386e-05,
+      "loss": 2.0329,
+      "step": 6829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8942531007459103e-05,
+      "loss": 2.0198,
+      "step": 6830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894221293736592e-05,
+      "loss": 2.0291,
+      "step": 6831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8941894822115917e-05,
+      "loss": 2.0075,
+      "step": 6832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8941576661710693e-05,
+      "loss": 2.0122,
+      "step": 6833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8941258456151867e-05,
+      "loss": 2.0273,
+      "step": 6834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8940940205441034e-05,
+      "loss": 2.0407,
+      "step": 6835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8940621909579805e-05,
+      "loss": 2.0102,
+      "step": 6836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8940303568569796e-05,
+      "loss": 2.0116,
+      "step": 6837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89399851824126e-05,
+      "loss": 2.0145,
+      "step": 6838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8939666751109835e-05,
+      "loss": 2.0075,
+      "step": 6839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8939348274663106e-05,
+      "loss": 2.0267,
+      "step": 6840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8939029753074017e-05,
+      "loss": 2.0105,
+      "step": 6841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8938711186344184e-05,
+      "loss": 2.0225,
+      "step": 6842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8938392574475214e-05,
+      "loss": 2.0064,
+      "step": 6843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893807391746871e-05,
+      "loss": 2.0222,
+      "step": 6844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893775521532629e-05,
+      "loss": 2.0224,
+      "step": 6845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8937436468049557e-05,
+      "loss": 2.014,
+      "step": 6846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893711767564012e-05,
+      "loss": 2.0225,
+      "step": 6847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893679883809959e-05,
+      "loss": 2.0331,
+      "step": 6848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8936479955429586e-05,
+      "loss": 2.0139,
+      "step": 6849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893616102763171e-05,
+      "loss": 2.003,
+      "step": 6850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893584205470757e-05,
+      "loss": 2.0247,
+      "step": 6851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8935523036658778e-05,
+      "loss": 2.0282,
+      "step": 6852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8935203973486946e-05,
+      "loss": 2.0111,
+      "step": 6853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893488486519369e-05,
+      "loss": 2.0406,
+      "step": 6854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8934565711780618e-05,
+      "loss": 2.0349,
+      "step": 6855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8934246513249336e-05,
+      "loss": 2.0359,
+      "step": 6856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893392726960147e-05,
+      "loss": 2.0208,
+      "step": 6857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8933607980838617e-05,
+      "loss": 2.0197,
+      "step": 6858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8933288646962394e-05,
+      "loss": 2.002,
+      "step": 6859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8932969267974423e-05,
+      "loss": 2.0014,
+      "step": 6860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8932649843876302e-05,
+      "loss": 2.0147,
+      "step": 6861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8932330374669656e-05,
+      "loss": 2.0244,
+      "step": 6862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893201086035609e-05,
+      "loss": 2.0363,
+      "step": 6863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8931691300937226e-05,
+      "loss": 2.0407,
+      "step": 6864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8931371696414667e-05,
+      "loss": 2.006,
+      "step": 6865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8931052046790033e-05,
+      "loss": 2.0237,
+      "step": 6866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893073235206494e-05,
+      "loss": 2.0145,
+      "step": 6867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8930412612241e-05,
+      "loss": 2.0008,
+      "step": 6868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893009282731983e-05,
+      "loss": 2.0329,
+      "step": 6869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892977299730304e-05,
+      "loss": 2.0168,
+      "step": 6870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8929453122192248e-05,
+      "loss": 2.0099,
+      "step": 6871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892913320198907e-05,
+      "loss": 2.0269,
+      "step": 6872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8928813236695117e-05,
+      "loss": 2.0063,
+      "step": 6873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8928493226312015e-05,
+      "loss": 2.0111,
+      "step": 6874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8928173170841367e-05,
+      "loss": 2.0244,
+      "step": 6875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8927853070284802e-05,
+      "loss": 1.9993,
+      "step": 6876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8927532924643926e-05,
+      "loss": 2.0101,
+      "step": 6877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892721273392036e-05,
+      "loss": 2.0098,
+      "step": 6878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8926892498115724e-05,
+      "loss": 2.0195,
+      "step": 6879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8926572217231628e-05,
+      "loss": 2.0138,
+      "step": 6880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8926251891269695e-05,
+      "loss": 2.0455,
+      "step": 6881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892593152023154e-05,
+      "loss": 2.0183,
+      "step": 6882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892561110411878e-05,
+      "loss": 2.0249,
+      "step": 6883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892529064293304e-05,
+      "loss": 2.0228,
+      "step": 6884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8924970136675932e-05,
+      "loss": 2.0226,
+      "step": 6885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892464958534907e-05,
+      "loss": 2.0226,
+      "step": 6886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8924328988954084e-05,
+      "loss": 2.0304,
+      "step": 6887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8924008347492586e-05,
+      "loss": 2.0351,
+      "step": 6888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8923687660966193e-05,
+      "loss": 1.9987,
+      "step": 6889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8923366929376533e-05,
+      "loss": 2.0155,
+      "step": 6890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8923046152725218e-05,
+      "loss": 2.0092,
+      "step": 6891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8922725331013868e-05,
+      "loss": 2.0284,
+      "step": 6892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892240446424411e-05,
+      "loss": 2.0256,
+      "step": 6893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8922083552417558e-05,
+      "loss": 2.0062,
+      "step": 6894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8921762595535833e-05,
+      "loss": 2.0187,
+      "step": 6895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892144159360056e-05,
+      "loss": 2.0174,
+      "step": 6896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8921120546613355e-05,
+      "loss": 2.0254,
+      "step": 6897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892079945457584e-05,
+      "loss": 2.03,
+      "step": 6898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8920478317489642e-05,
+      "loss": 2.0156,
+      "step": 6899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8920157135356378e-05,
+      "loss": 2.028,
+      "step": 6900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8919835908177668e-05,
+      "loss": 2.023,
+      "step": 6901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8919514635955138e-05,
+      "loss": 1.9928,
+      "step": 6902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8919193318690404e-05,
+      "loss": 2.0137,
+      "step": 6903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.89188719563851e-05,
+      "loss": 2.0003,
+      "step": 6904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8918550549040843e-05,
+      "loss": 2.0275,
+      "step": 6905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891822909665925e-05,
+      "loss": 2.0068,
+      "step": 6906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8917907599241952e-05,
+      "loss": 2.0264,
+      "step": 6907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8917586056790572e-05,
+      "loss": 2.001,
+      "step": 6908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8917264469306733e-05,
+      "loss": 2.0186,
+      "step": 6909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8916942836792055e-05,
+      "loss": 2.0163,
+      "step": 6910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8916621159248166e-05,
+      "loss": 2.0142,
+      "step": 6911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891629943667669e-05,
+      "loss": 2.0222,
+      "step": 6912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891597766907925e-05,
+      "loss": 2.0145,
+      "step": 6913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8915655856457476e-05,
+      "loss": 2.0094,
+      "step": 6914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8915333998812983e-05,
+      "loss": 2.0321,
+      "step": 6915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8915012096147408e-05,
+      "loss": 2.0144,
+      "step": 6916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8914690148462368e-05,
+      "loss": 2.0292,
+      "step": 6917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891436815575949e-05,
+      "loss": 2.0368,
+      "step": 6918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891404611804041e-05,
+      "loss": 2.0157,
+      "step": 6919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891372403530674e-05,
+      "loss": 2.025,
+      "step": 6920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8913401907560116e-05,
+      "loss": 2.0307,
+      "step": 6921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891307973480216e-05,
+      "loss": 2.015,
+      "step": 6922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8912757517034497e-05,
+      "loss": 2.02,
+      "step": 6923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891243525425876e-05,
+      "loss": 2.0346,
+      "step": 6924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8912112946476575e-05,
+      "loss": 2.0109,
+      "step": 6925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8911790593689567e-05,
+      "loss": 2.0247,
+      "step": 6926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8911468195899364e-05,
+      "loss": 2.0268,
+      "step": 6927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.89111457531076e-05,
+      "loss": 2.0289,
+      "step": 6928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8910823265315894e-05,
+      "loss": 2.0222,
+      "step": 6929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891050073252588e-05,
+      "loss": 2.0054,
+      "step": 6930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8910178154739188e-05,
+      "loss": 2.0427,
+      "step": 6931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8909855531957443e-05,
+      "loss": 2.0306,
+      "step": 6932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8909532864182278e-05,
+      "loss": 2.0135,
+      "step": 6933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8909210151415317e-05,
+      "loss": 2.01,
+      "step": 6934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8908887393658192e-05,
+      "loss": 2.0224,
+      "step": 6935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8908564590912538e-05,
+      "loss": 1.9919,
+      "step": 6936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890824174317998e-05,
+      "loss": 2.0279,
+      "step": 6937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890791885046215e-05,
+      "loss": 2.0178,
+      "step": 6938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890759591276068e-05,
+      "loss": 2.0023,
+      "step": 6939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890727293007719e-05,
+      "loss": 2.0021,
+      "step": 6940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890694990241333e-05,
+      "loss": 2.0208,
+      "step": 6941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8906626829770715e-05,
+      "loss": 2.028,
+      "step": 6942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8906303712150987e-05,
+      "loss": 2.01,
+      "step": 6943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890598054955577e-05,
+      "loss": 2.0193,
+      "step": 6944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8905657341986696e-05,
+      "loss": 2.0168,
+      "step": 6945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8905334089445405e-05,
+      "loss": 2.0073,
+      "step": 6946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8905010791933525e-05,
+      "loss": 2.0119,
+      "step": 6947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8904687449452687e-05,
+      "loss": 2.0212,
+      "step": 6948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8904364062004522e-05,
+      "loss": 1.9925,
+      "step": 6949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890404062959067e-05,
+      "loss": 2.0257,
+      "step": 6950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890371715221276e-05,
+      "loss": 2.0164,
+      "step": 6951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8903393629872423e-05,
+      "loss": 2.0052,
+      "step": 6952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.89030700625713e-05,
+      "loss": 2.0031,
+      "step": 6953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890274645031102e-05,
+      "loss": 2.0288,
+      "step": 6954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8902422793093212e-05,
+      "loss": 2.0032,
+      "step": 6955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890209909091952e-05,
+      "loss": 2.0136,
+      "step": 6956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8901775343791575e-05,
+      "loss": 2.0029,
+      "step": 6957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8901451551711015e-05,
+      "loss": 2.0162,
+      "step": 6958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8901127714679467e-05,
+      "loss": 2.0228,
+      "step": 6959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8900803832698573e-05,
+      "loss": 2.0057,
+      "step": 6960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8900479905769965e-05,
+      "loss": 2.018,
+      "step": 6961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8900155933895284e-05,
+      "loss": 2.0221,
+      "step": 6962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.889983191707616e-05,
+      "loss": 2.0249,
+      "step": 6963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8899507855314233e-05,
+      "loss": 2.0164,
+      "step": 6964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8899183748611138e-05,
+      "loss": 2.02,
+      "step": 6965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.889885959696851e-05,
+      "loss": 2.0138,
+      "step": 6966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8898535400387993e-05,
+      "loss": 2.0245,
+      "step": 6967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8898211158871212e-05,
+      "loss": 2.0334,
+      "step": 6968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8897886872419815e-05,
+      "loss": 2.0245,
+      "step": 6969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8897562541035434e-05,
+      "loss": 2.02,
+      "step": 6970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8897238164719712e-05,
+      "loss": 1.9962,
+      "step": 6971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8896913743474283e-05,
+      "loss": 2.0105,
+      "step": 6972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8896589277300787e-05,
+      "loss": 1.9956,
+      "step": 6973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.889626476620086e-05,
+      "loss": 2.0059,
+      "step": 6974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.889594021017614e-05,
+      "loss": 2.0402,
+      "step": 6975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.889561560922827e-05,
+      "loss": 2.0292,
+      "step": 6976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8895290963358887e-05,
+      "loss": 2.0081,
+      "step": 6977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8894966272569634e-05,
+      "loss": 2.028,
+      "step": 6978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8894641536862143e-05,
+      "loss": 2.0288,
+      "step": 6979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8894316756238063e-05,
+      "loss": 2.0043,
+      "step": 6980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8893991930699028e-05,
+      "loss": 2.0082,
+      "step": 6981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8893667060246677e-05,
+      "loss": 2.0085,
+      "step": 6982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8893342144882657e-05,
+      "loss": 2.0099,
+      "step": 6983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8893017184608605e-05,
+      "loss": 2.0146,
+      "step": 6984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8892692179426157e-05,
+      "loss": 2.0001,
+      "step": 6985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8892367129336966e-05,
+      "loss": 2.0173,
+      "step": 6986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8892042034342662e-05,
+      "loss": 2.0101,
+      "step": 6987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8891716894444893e-05,
+      "loss": 2.0302,
+      "step": 6988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88913917096453e-05,
+      "loss": 2.0212,
+      "step": 6989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8891066479945524e-05,
+      "loss": 2.0313,
+      "step": 6990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8890741205347208e-05,
+      "loss": 2.0226,
+      "step": 6991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8890415885851995e-05,
+      "loss": 2.0144,
+      "step": 6992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8890090521461526e-05,
+      "loss": 2.0007,
+      "step": 6993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8889765112177446e-05,
+      "loss": 1.9846,
+      "step": 6994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88894396580014e-05,
+      "loss": 2.0336,
+      "step": 6995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8889114158935026e-05,
+      "loss": 2.0186,
+      "step": 6996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8888788614979972e-05,
+      "loss": 2.0257,
+      "step": 6997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888846302613788e-05,
+      "loss": 2.0269,
+      "step": 6998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8888137392410396e-05,
+      "loss": 1.9869,
+      "step": 6999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8887811713799163e-05,
+      "loss": 2.0056,
+      "step": 7000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8887485990305827e-05,
+      "loss": 2.0059,
+      "step": 7001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888716022193203e-05,
+      "loss": 2.0197,
+      "step": 7002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8886834408679417e-05,
+      "loss": 2.0013,
+      "step": 7003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8886508550549638e-05,
+      "loss": 2.0096,
+      "step": 7004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8886182647544335e-05,
+      "loss": 2.0077,
+      "step": 7005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8885856699665153e-05,
+      "loss": 2.0139,
+      "step": 7006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8885530706913744e-05,
+      "loss": 2.0374,
+      "step": 7007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8885204669291748e-05,
+      "loss": 2.014,
+      "step": 7008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888487858680081e-05,
+      "loss": 2.0173,
+      "step": 7009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888455245944258e-05,
+      "loss": 1.9954,
+      "step": 7010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888422628721871e-05,
+      "loss": 2.0243,
+      "step": 7011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888390007013084e-05,
+      "loss": 1.991,
+      "step": 7012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8883573808180616e-05,
+      "loss": 2.0118,
+      "step": 7013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8883247501369692e-05,
+      "loss": 2.0059,
+      "step": 7014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888292114969971e-05,
+      "loss": 2.0116,
+      "step": 7015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888259475317232e-05,
+      "loss": 2.0086,
+      "step": 7016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8882268311789175e-05,
+      "loss": 2.0157,
+      "step": 7017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8881941825551916e-05,
+      "loss": 2.0158,
+      "step": 7018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8881615294462195e-05,
+      "loss": 1.9922,
+      "step": 7019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8881288718521663e-05,
+      "loss": 2.0053,
+      "step": 7020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8880962097731963e-05,
+      "loss": 2.0093,
+      "step": 7021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888063543209475e-05,
+      "loss": 2.0042,
+      "step": 7022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8880308721611676e-05,
+      "loss": 1.9965,
+      "step": 7023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887998196628438e-05,
+      "loss": 2.0143,
+      "step": 7024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8879655166114527e-05,
+      "loss": 2.0033,
+      "step": 7025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8879328321103757e-05,
+      "loss": 2.0187,
+      "step": 7026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887900143125372e-05,
+      "loss": 2.0072,
+      "step": 7027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887867449656607e-05,
+      "loss": 1.9959,
+      "step": 7028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887834751704246e-05,
+      "loss": 2.0118,
+      "step": 7029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8878020492684536e-05,
+      "loss": 2.0154,
+      "step": 7030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887769342349395e-05,
+      "loss": 2.0217,
+      "step": 7031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887736630947236e-05,
+      "loss": 2.0085,
+      "step": 7032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8877039150621414e-05,
+      "loss": 2.0231,
+      "step": 7033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8876711946942762e-05,
+      "loss": 2.0091,
+      "step": 7034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887638469843806e-05,
+      "loss": 2.0139,
+      "step": 7035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8876057405108956e-05,
+      "loss": 2.019,
+      "step": 7036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887573006695711e-05,
+      "loss": 2.0061,
+      "step": 7037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8875402683984164e-05,
+      "loss": 2.0091,
+      "step": 7038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8875075256191783e-05,
+      "loss": 2.0332,
+      "step": 7039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8874747783581614e-05,
+      "loss": 2.0057,
+      "step": 7040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8874420266155307e-05,
+      "loss": 2.0,
+      "step": 7041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8874092703914528e-05,
+      "loss": 2.0117,
+      "step": 7042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8873765096860918e-05,
+      "loss": 2.0098,
+      "step": 7043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887343744499614e-05,
+      "loss": 2.0017,
+      "step": 7044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8873109748321847e-05,
+      "loss": 2.0144,
+      "step": 7045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887278200683969e-05,
+      "loss": 2.0147,
+      "step": 7046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887245422055133e-05,
+      "loss": 2.0059,
+      "step": 7047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8872126389458417e-05,
+      "loss": 2.018,
+      "step": 7048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8871798513562607e-05,
+      "loss": 1.9943,
+      "step": 7049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8871470592865558e-05,
+      "loss": 1.9991,
+      "step": 7050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8871142627368926e-05,
+      "loss": 2.0125,
+      "step": 7051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8870814617074365e-05,
+      "loss": 2.0226,
+      "step": 7052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8870486561983538e-05,
+      "loss": 1.9972,
+      "step": 7053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887015846209809e-05,
+      "loss": 2.02,
+      "step": 7054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8869830317419687e-05,
+      "loss": 2.0177,
+      "step": 7055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886950212794998e-05,
+      "loss": 2.0144,
+      "step": 7056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8869173893690634e-05,
+      "loss": 2.0107,
+      "step": 7057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88688456146433e-05,
+      "loss": 2.0034,
+      "step": 7058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8868517290809637e-05,
+      "loss": 2.0046,
+      "step": 7059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8868188922191306e-05,
+      "loss": 2.0126,
+      "step": 7060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886786050878996e-05,
+      "loss": 2.0043,
+      "step": 7061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8867532050607264e-05,
+      "loss": 2.0161,
+      "step": 7062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8867203547644872e-05,
+      "loss": 2.019,
+      "step": 7063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8866874999904447e-05,
+      "loss": 2.0003,
+      "step": 7064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886654640738764e-05,
+      "loss": 2.0212,
+      "step": 7065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8866217770096117e-05,
+      "loss": 2.0264,
+      "step": 7066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886588908803154e-05,
+      "loss": 2.0045,
+      "step": 7067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8865560361195562e-05,
+      "loss": 1.9953,
+      "step": 7068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8865231589589844e-05,
+      "loss": 2.0123,
+      "step": 7069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886490277321605e-05,
+      "loss": 2.0171,
+      "step": 7070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886457391207584e-05,
+      "loss": 2.017,
+      "step": 7071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886424500617087e-05,
+      "loss": 2.0252,
+      "step": 7072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8863916055502807e-05,
+      "loss": 2.0028,
+      "step": 7073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886358706007331e-05,
+      "loss": 2.0053,
+      "step": 7074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8863258019884035e-05,
+      "loss": 2.0148,
+      "step": 7075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8862928934936653e-05,
+      "loss": 2.0328,
+      "step": 7076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8862599805232822e-05,
+      "loss": 2.0177,
+      "step": 7077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88622706307742e-05,
+      "loss": 1.9979,
+      "step": 7078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8861941411562452e-05,
+      "loss": 2.014,
+      "step": 7079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8861612147599244e-05,
+      "loss": 2.0093,
+      "step": 7080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8861282838886238e-05,
+      "loss": 2.0259,
+      "step": 7081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8860953485425088e-05,
+      "loss": 2.0037,
+      "step": 7082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8860624087217467e-05,
+      "loss": 2.0161,
+      "step": 7083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8860294644265037e-05,
+      "loss": 1.9996,
+      "step": 7084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8859965156569456e-05,
+      "loss": 2.0199,
+      "step": 7085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8859635624132396e-05,
+      "loss": 2.0323,
+      "step": 7086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8859306046955516e-05,
+      "loss": 2.0203,
+      "step": 7087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8858976425040484e-05,
+      "loss": 2.0103,
+      "step": 7088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8858646758388956e-05,
+      "loss": 2.0153,
+      "step": 7089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8858317047002606e-05,
+      "loss": 2.0231,
+      "step": 7090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8857987290883094e-05,
+      "loss": 2.0148,
+      "step": 7091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8857657490032087e-05,
+      "loss": 2.0179,
+      "step": 7092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8857327644451252e-05,
+      "loss": 1.987,
+      "step": 7093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8856997754142253e-05,
+      "loss": 2.0131,
+      "step": 7094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8856667819106754e-05,
+      "loss": 2.02,
+      "step": 7095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8856337839346424e-05,
+      "loss": 2.0132,
+      "step": 7096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8856007814862927e-05,
+      "loss": 2.0124,
+      "step": 7097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8855677745657933e-05,
+      "loss": 2.0033,
+      "step": 7098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8855347631733104e-05,
+      "loss": 2.0235,
+      "step": 7099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8855017473090108e-05,
+      "loss": 2.0196,
+      "step": 7100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8854687269730617e-05,
+      "loss": 2.0231,
+      "step": 7101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8854357021656297e-05,
+      "loss": 2.0023,
+      "step": 7102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8854026728868812e-05,
+      "loss": 2.0182,
+      "step": 7103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8853696391369833e-05,
+      "loss": 2.0178,
+      "step": 7104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8853366009161026e-05,
+      "loss": 1.9944,
+      "step": 7105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8853035582244064e-05,
+      "loss": 2.0214,
+      "step": 7106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8852705110620606e-05,
+      "loss": 2.0022,
+      "step": 7107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.885237459429233e-05,
+      "loss": 2.0139,
+      "step": 7108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8852044033260902e-05,
+      "loss": 2.0229,
+      "step": 7109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8851713427527995e-05,
+      "loss": 2.0174,
+      "step": 7110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.885138277709527e-05,
+      "loss": 2.0295,
+      "step": 7111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8851052081964402e-05,
+      "loss": 1.9933,
+      "step": 7112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.885072134213706e-05,
+      "loss": 2.0209,
+      "step": 7113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.885039055761492e-05,
+      "loss": 2.0139,
+      "step": 7114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8850059728399643e-05,
+      "loss": 2.0158,
+      "step": 7115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8849728854492902e-05,
+      "loss": 2.0087,
+      "step": 7116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8849397935896374e-05,
+      "loss": 2.014,
+      "step": 7117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8849066972611724e-05,
+      "loss": 1.9986,
+      "step": 7118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8848735964640623e-05,
+      "loss": 2.0366,
+      "step": 7119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8848404911984746e-05,
+      "loss": 2.0034,
+      "step": 7120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8848073814645764e-05,
+      "loss": 2.0231,
+      "step": 7121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8847742672625347e-05,
+      "loss": 1.987,
+      "step": 7122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8847411485925172e-05,
+      "loss": 2.0253,
+      "step": 7123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8847080254546902e-05,
+      "loss": 2.0085,
+      "step": 7124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8846748978492222e-05,
+      "loss": 2.0072,
+      "step": 7125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8846417657762794e-05,
+      "loss": 1.9915,
+      "step": 7126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8846086292360297e-05,
+      "loss": 2.0005,
+      "step": 7127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8845754882286404e-05,
+      "loss": 1.9861,
+      "step": 7128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8845423427542788e-05,
+      "loss": 2.0098,
+      "step": 7129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884509192813112e-05,
+      "loss": 2.0104,
+      "step": 7130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8844760384053077e-05,
+      "loss": 2.031,
+      "step": 7131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8844428795310336e-05,
+      "loss": 2.0146,
+      "step": 7132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8844097161904564e-05,
+      "loss": 2.0176,
+      "step": 7133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884376548383744e-05,
+      "loss": 1.9796,
+      "step": 7134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884343376111064e-05,
+      "loss": 2.0147,
+      "step": 7135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8843101993725842e-05,
+      "loss": 2.0223,
+      "step": 7136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884277018168471e-05,
+      "loss": 1.9886,
+      "step": 7137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8842438324988932e-05,
+      "loss": 2.0013,
+      "step": 7138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8842106423640178e-05,
+      "loss": 2.0216,
+      "step": 7139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884177447764012e-05,
+      "loss": 2.0117,
+      "step": 7140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8841442486990443e-05,
+      "loss": 2.0148,
+      "step": 7141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884111045169282e-05,
+      "loss": 1.9958,
+      "step": 7142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8840778371748928e-05,
+      "loss": 2.013,
+      "step": 7143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8840446247160442e-05,
+      "loss": 1.9789,
+      "step": 7144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8840114077929038e-05,
+      "loss": 2.0117,
+      "step": 7145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88397818640564e-05,
+      "loss": 1.9878,
+      "step": 7146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8839449605544198e-05,
+      "loss": 2.006,
+      "step": 7147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8839117302394117e-05,
+      "loss": 2.0032,
+      "step": 7148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883878495460783e-05,
+      "loss": 1.9973,
+      "step": 7149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8838452562187014e-05,
+      "loss": 2.0074,
+      "step": 7150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8838120125133353e-05,
+      "loss": 2.0157,
+      "step": 7151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883778764344852e-05,
+      "loss": 2.0106,
+      "step": 7152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88374551171342e-05,
+      "loss": 2.0192,
+      "step": 7153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883712254619207e-05,
+      "loss": 1.9875,
+      "step": 7154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8836789930623802e-05,
+      "loss": 2.0123,
+      "step": 7155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8836457270431088e-05,
+      "loss": 2.0059,
+      "step": 7156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88361245656156e-05,
+      "loss": 2.0296,
+      "step": 7157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8835791816179022e-05,
+      "loss": 2.0119,
+      "step": 7158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8835459022123034e-05,
+      "loss": 2.0108,
+      "step": 7159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883512618344931e-05,
+      "loss": 1.9862,
+      "step": 7160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883479330015954e-05,
+      "loss": 2.0143,
+      "step": 7161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88344603722554e-05,
+      "loss": 2.0283,
+      "step": 7162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8834127399738572e-05,
+      "loss": 2.0191,
+      "step": 7163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8833794382610738e-05,
+      "loss": 2.0177,
+      "step": 7164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883346132087358e-05,
+      "loss": 2.0036,
+      "step": 7165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8833128214528774e-05,
+      "loss": 2.0232,
+      "step": 7166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8832795063578012e-05,
+      "loss": 2.025,
+      "step": 7167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8832461868022972e-05,
+      "loss": 2.0246,
+      "step": 7168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8832128627865336e-05,
+      "loss": 2.0007,
+      "step": 7169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8831795343106787e-05,
+      "loss": 1.9915,
+      "step": 7170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883146201374901e-05,
+      "loss": 2.0071,
+      "step": 7171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8831128639793682e-05,
+      "loss": 2.0188,
+      "step": 7172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8830795221242494e-05,
+      "loss": 2.0168,
+      "step": 7173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8830461758097126e-05,
+      "loss": 1.9891,
+      "step": 7174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883012825035926e-05,
+      "loss": 2.0088,
+      "step": 7175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8829794698030586e-05,
+      "loss": 2.0135,
+      "step": 7176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8829461101112786e-05,
+      "loss": 2.0132,
+      "step": 7177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8829127459607537e-05,
+      "loss": 2.0109,
+      "step": 7178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8828793773516534e-05,
+      "loss": 2.0047,
+      "step": 7179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8828460042841464e-05,
+      "loss": 2.0038,
+      "step": 7180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8828126267584005e-05,
+      "loss": 2.0226,
+      "step": 7181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8827792447745842e-05,
+      "loss": 2.0262,
+      "step": 7182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882745858332866e-05,
+      "loss": 1.996,
+      "step": 7183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8827124674334155e-05,
+      "loss": 1.9973,
+      "step": 7184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8826790720764e-05,
+      "loss": 1.9947,
+      "step": 7185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8826456722619894e-05,
+      "loss": 2.0284,
+      "step": 7186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882612267990351e-05,
+      "loss": 2.0002,
+      "step": 7187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8825788592616548e-05,
+      "loss": 2.0358,
+      "step": 7188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8825454460760687e-05,
+      "loss": 2.0198,
+      "step": 7189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882512028433762e-05,
+      "loss": 2.0124,
+      "step": 7190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8824786063349026e-05,
+      "loss": 2.0121,
+      "step": 7191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88244517977966e-05,
+      "loss": 2.023,
+      "step": 7192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882411748768203e-05,
+      "loss": 2.0205,
+      "step": 7193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8823783133007e-05,
+      "loss": 1.9952,
+      "step": 7194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8823448733773202e-05,
+      "loss": 1.9965,
+      "step": 7195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8823114289982323e-05,
+      "loss": 2.0166,
+      "step": 7196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8822779801636052e-05,
+      "loss": 2.0129,
+      "step": 7197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8822445268736077e-05,
+      "loss": 2.0192,
+      "step": 7198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8822110691284088e-05,
+      "loss": 1.9966,
+      "step": 7199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8821776069281777e-05,
+      "loss": 2.0134,
+      "step": 7200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8821441402730832e-05,
+      "loss": 2.0019,
+      "step": 7201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8821106691632944e-05,
+      "loss": 2.0233,
+      "step": 7202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88207719359898e-05,
+      "loss": 2.0233,
+      "step": 7203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8820437135803093e-05,
+      "loss": 2.0024,
+      "step": 7204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8820102291074516e-05,
+      "loss": 2.0135,
+      "step": 7205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8819767401805754e-05,
+      "loss": 2.0224,
+      "step": 7206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8819432467998503e-05,
+      "loss": 1.9987,
+      "step": 7207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8819097489654452e-05,
+      "loss": 2.0171,
+      "step": 7208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8818762466775293e-05,
+      "loss": 2.0222,
+      "step": 7209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881842739936272e-05,
+      "loss": 2.0214,
+      "step": 7210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881809228741842e-05,
+      "loss": 2.0275,
+      "step": 7211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881775713094409e-05,
+      "loss": 2.0167,
+      "step": 7212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8817421929941428e-05,
+      "loss": 2.012,
+      "step": 7213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8817086684412113e-05,
+      "loss": 2.0039,
+      "step": 7214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8816751394357844e-05,
+      "loss": 2.0211,
+      "step": 7215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8816416059780315e-05,
+      "loss": 2.0263,
+      "step": 7216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881608068068122e-05,
+      "loss": 2.0184,
+      "step": 7217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8815745257062255e-05,
+      "loss": 2.0075,
+      "step": 7218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8815409788925105e-05,
+      "loss": 2.0105,
+      "step": 7219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8815074276271472e-05,
+      "loss": 2.0333,
+      "step": 7220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881473871910305e-05,
+      "loss": 2.0321,
+      "step": 7221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881440311742153e-05,
+      "loss": 2.017,
+      "step": 7222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8814067471228606e-05,
+      "loss": 2.0284,
+      "step": 7223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881373178052598e-05,
+      "loss": 2.0332,
+      "step": 7224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881339604531534e-05,
+      "loss": 2.0187,
+      "step": 7225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8813060265598384e-05,
+      "loss": 2.027,
+      "step": 7226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8812724441376806e-05,
+      "loss": 2.0304,
+      "step": 7227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8812388572652304e-05,
+      "loss": 2.0114,
+      "step": 7228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8812052659426575e-05,
+      "loss": 2.0027,
+      "step": 7229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8811716701701313e-05,
+      "loss": 2.0294,
+      "step": 7230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8811380699478215e-05,
+      "loss": 2.0226,
+      "step": 7231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8811044652758975e-05,
+      "loss": 2.0122,
+      "step": 7232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8810708561545298e-05,
+      "loss": 1.9945,
+      "step": 7233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8810372425838874e-05,
+      "loss": 2.0002,
+      "step": 7234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8810036245641404e-05,
+      "loss": 1.9957,
+      "step": 7235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8809700020954583e-05,
+      "loss": 2.0331,
+      "step": 7236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8809363751780113e-05,
+      "loss": 1.9997,
+      "step": 7237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8809027438119684e-05,
+      "loss": 2.025,
+      "step": 7238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8808691079975e-05,
+      "loss": 2.0251,
+      "step": 7239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8808354677347764e-05,
+      "loss": 2.0205,
+      "step": 7240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880801823023967e-05,
+      "loss": 2.0023,
+      "step": 7241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8807681738652413e-05,
+      "loss": 2.0178,
+      "step": 7242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8807345202587694e-05,
+      "loss": 2.0146,
+      "step": 7243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880700862204722e-05,
+      "loss": 2.0158,
+      "step": 7244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8806671997032685e-05,
+      "loss": 1.9908,
+      "step": 7245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880633532754579e-05,
+      "loss": 2.0244,
+      "step": 7246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8805998613588228e-05,
+      "loss": 2.0084,
+      "step": 7247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880566185516171e-05,
+      "loss": 1.9961,
+      "step": 7248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8805325052267933e-05,
+      "loss": 1.9897,
+      "step": 7249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8804988204908597e-05,
+      "loss": 2.0303,
+      "step": 7250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8804651313085403e-05,
+      "loss": 1.9982,
+      "step": 7251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880431437680005e-05,
+      "loss": 1.9989,
+      "step": 7252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8803977396054244e-05,
+      "loss": 2.0072,
+      "step": 7253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8803640370849686e-05,
+      "loss": 2.0132,
+      "step": 7254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8803303301188075e-05,
+      "loss": 2.0124,
+      "step": 7255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8802966187071116e-05,
+      "loss": 2.0078,
+      "step": 7256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8802629028500505e-05,
+      "loss": 2.0093,
+      "step": 7257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8802291825477958e-05,
+      "loss": 2.0087,
+      "step": 7258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880195457800516e-05,
+      "loss": 2.0106,
+      "step": 7259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880161728608383e-05,
+      "loss": 2.0035,
+      "step": 7260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8801279949715663e-05,
+      "loss": 2.0196,
+      "step": 7261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8800942568902364e-05,
+      "loss": 2.027,
+      "step": 7262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880060514364564e-05,
+      "loss": 2.0268,
+      "step": 7263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8800267673947185e-05,
+      "loss": 2.0129,
+      "step": 7264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8799930159808718e-05,
+      "loss": 2.0173,
+      "step": 7265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8799592601231928e-05,
+      "loss": 2.0131,
+      "step": 7266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8799254998218532e-05,
+      "loss": 2.0114,
+      "step": 7267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879891735077023e-05,
+      "loss": 1.9776,
+      "step": 7268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8798579658888724e-05,
+      "loss": 2.0219,
+      "step": 7269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8798241922575728e-05,
+      "loss": 2.0071,
+      "step": 7270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8797904141832935e-05,
+      "loss": 2.0024,
+      "step": 7271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8797566316662063e-05,
+      "loss": 1.9851,
+      "step": 7272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879722844706481e-05,
+      "loss": 2.0281,
+      "step": 7273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8796890533042884e-05,
+      "loss": 1.984,
+      "step": 7274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8796552574597994e-05,
+      "loss": 2.0058,
+      "step": 7275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8796214571731843e-05,
+      "loss": 2.0264,
+      "step": 7276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879587652444614e-05,
+      "loss": 1.999,
+      "step": 7277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8795538432742594e-05,
+      "loss": 2.021,
+      "step": 7278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8795200296622907e-05,
+      "loss": 2.0007,
+      "step": 7279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879486211608879e-05,
+      "loss": 2.005,
+      "step": 7280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8794523891141955e-05,
+      "loss": 2.0108,
+      "step": 7281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.87941856217841e-05,
+      "loss": 2.0005,
+      "step": 7282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879384730801694e-05,
+      "loss": 1.999,
+      "step": 7283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8793508949842185e-05,
+      "loss": 2.021,
+      "step": 7284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8793170547261536e-05,
+      "loss": 1.9994,
+      "step": 7285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879283210027671e-05,
+      "loss": 2.0069,
+      "step": 7286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879249360888941e-05,
+      "loss": 2.0159,
+      "step": 7287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879215507310135e-05,
+      "loss": 2.012,
+      "step": 7288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8791816492914238e-05,
+      "loss": 1.9991,
+      "step": 7289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8791477868329783e-05,
+      "loss": 2.0124,
+      "step": 7290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879113919934969e-05,
+      "loss": 2.0288,
+      "step": 7291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879080048597568e-05,
+      "loss": 2.035,
+      "step": 7292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8790461728209457e-05,
+      "loss": 1.9901,
+      "step": 7293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8790122926052734e-05,
+      "loss": 2.012,
+      "step": 7294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878978407950722e-05,
+      "loss": 2.018,
+      "step": 7295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8789445188574623e-05,
+      "loss": 2.0043,
+      "step": 7296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878910625325666e-05,
+      "loss": 2.005,
+      "step": 7297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8788767273555042e-05,
+      "loss": 2.0085,
+      "step": 7298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878842824947148e-05,
+      "loss": 2.0044,
+      "step": 7299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8788089181007682e-05,
+      "loss": 1.9985,
+      "step": 7300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8787750068165366e-05,
+      "loss": 2.0065,
+      "step": 7301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878741091094624e-05,
+      "loss": 2.0216,
+      "step": 7302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8787071709352023e-05,
+      "loss": 1.9988,
+      "step": 7303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8786732463384422e-05,
+      "loss": 2.0111,
+      "step": 7304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878639317304515e-05,
+      "loss": 1.9908,
+      "step": 7305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8786053838335926e-05,
+      "loss": 2.0126,
+      "step": 7306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8785714459258456e-05,
+      "loss": 1.9987,
+      "step": 7307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878537503581446e-05,
+      "loss": 2.0027,
+      "step": 7308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8785035568005647e-05,
+      "loss": 2.0057,
+      "step": 7309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8784696055833738e-05,
+      "loss": 2.0166,
+      "step": 7310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878435649930044e-05,
+      "loss": 2.0021,
+      "step": 7311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878401689840747e-05,
+      "loss": 2.0008,
+      "step": 7312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8783677253156547e-05,
+      "loss": 2.0125,
+      "step": 7313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8783337563549385e-05,
+      "loss": 1.9976,
+      "step": 7314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8782997829587695e-05,
+      "loss": 2.0212,
+      "step": 7315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8782658051273195e-05,
+      "loss": 2.0238,
+      "step": 7316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.87823182286076e-05,
+      "loss": 2.0185,
+      "step": 7317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878197836159263e-05,
+      "loss": 1.9892,
+      "step": 7318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8781638450229996e-05,
+      "loss": 2.0175,
+      "step": 7319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878129849452142e-05,
+      "loss": 1.9918,
+      "step": 7320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8780958494468612e-05,
+      "loss": 1.9865,
+      "step": 7321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8780618450073295e-05,
+      "loss": 2.0425,
+      "step": 7322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8780278361337185e-05,
+      "loss": 2.0052,
+      "step": 7323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8779938228261993e-05,
+      "loss": 2.0087,
+      "step": 7324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8779598050849445e-05,
+      "loss": 2.0347,
+      "step": 7325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8779257829101254e-05,
+      "loss": 2.0018,
+      "step": 7326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877891756301914e-05,
+      "loss": 2.0316,
+      "step": 7327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8778577252604822e-05,
+      "loss": 2.0176,
+      "step": 7328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8778236897860017e-05,
+      "loss": 2.0208,
+      "step": 7329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8777896498786445e-05,
+      "loss": 2.0196,
+      "step": 7330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8777556055385822e-05,
+      "loss": 2.0074,
+      "step": 7331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8777215567659873e-05,
+      "loss": 2.0145,
+      "step": 7332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877687503561031e-05,
+      "loss": 2.0041,
+      "step": 7333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877653445923886e-05,
+      "loss": 1.9927,
+      "step": 7334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8776193838547235e-05,
+      "loss": 2.0037,
+      "step": 7335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877585317353716e-05,
+      "loss": 2.0017,
+      "step": 7336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877551246421036e-05,
+      "loss": 2.0106,
+      "step": 7337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8775171710568547e-05,
+      "loss": 2.0133,
+      "step": 7338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877483091261344e-05,
+      "loss": 1.9979,
+      "step": 7339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877449007034677e-05,
+      "loss": 2.023,
+      "step": 7340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8774149183770253e-05,
+      "loss": 1.999,
+      "step": 7341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877380825288561e-05,
+      "loss": 2.0026,
+      "step": 7342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8773467277694563e-05,
+      "loss": 1.9917,
+      "step": 7343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8773126258198833e-05,
+      "loss": 2.0089,
+      "step": 7344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8772785194400146e-05,
+      "loss": 2.0228,
+      "step": 7345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.877244408630022e-05,
+      "loss": 1.9965,
+      "step": 7346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.877210293390078e-05,
+      "loss": 2.0112,
+      "step": 7347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8771761737203547e-05,
+      "loss": 1.9866,
+      "step": 7348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8771420496210243e-05,
+      "loss": 2.0132,
+      "step": 7349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8771079210922595e-05,
+      "loss": 2.0206,
+      "step": 7350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8770737881342326e-05,
+      "loss": 1.9894,
+      "step": 7351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8770396507471152e-05,
+      "loss": 2.0183,
+      "step": 7352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.877005508931081e-05,
+      "loss": 2.01,
+      "step": 7353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8769713626863012e-05,
+      "loss": 2.0105,
+      "step": 7354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876937212012949e-05,
+      "loss": 2.0059,
+      "step": 7355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8769030569111966e-05,
+      "loss": 2.0084,
+      "step": 7356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8768688973812162e-05,
+      "loss": 2.0034,
+      "step": 7357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8768347334231808e-05,
+      "loss": 2.0061,
+      "step": 7358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8768005650372626e-05,
+      "loss": 1.9955,
+      "step": 7359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8767663922236346e-05,
+      "loss": 1.9987,
+      "step": 7360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8767322149824688e-05,
+      "loss": 2.0045,
+      "step": 7361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876698033313938e-05,
+      "loss": 2.0009,
+      "step": 7362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8766638472182147e-05,
+      "loss": 2.0291,
+      "step": 7363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876629656695472e-05,
+      "loss": 2.0081,
+      "step": 7364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8765954617458816e-05,
+      "loss": 2.0125,
+      "step": 7365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876561262369617e-05,
+      "loss": 2.0256,
+      "step": 7366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876527058566851e-05,
+      "loss": 2.0205,
+      "step": 7367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8764928503377556e-05,
+      "loss": 2.0198,
+      "step": 7368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876458637682504e-05,
+      "loss": 2.0148,
+      "step": 7369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8764244206012693e-05,
+      "loss": 1.999,
+      "step": 7370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8763901990942237e-05,
+      "loss": 2.0094,
+      "step": 7371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8763559731615403e-05,
+      "loss": 1.995,
+      "step": 7372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8763217428033917e-05,
+      "loss": 1.9996,
+      "step": 7373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8762875080199513e-05,
+      "loss": 2.0205,
+      "step": 7374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8762532688113913e-05,
+      "loss": 2.0174,
+      "step": 7375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876219025177885e-05,
+      "loss": 2.0066,
+      "step": 7376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876184777119605e-05,
+      "loss": 1.9889,
+      "step": 7377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8761505246367248e-05,
+      "loss": 2.0036,
+      "step": 7378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876116267729417e-05,
+      "loss": 2.01,
+      "step": 7379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8760820063978543e-05,
+      "loss": 2.0048,
+      "step": 7380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8760477406422103e-05,
+      "loss": 2.0164,
+      "step": 7381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8760134704626576e-05,
+      "loss": 2.0161,
+      "step": 7382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8759791958593697e-05,
+      "loss": 1.9963,
+      "step": 7383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8759449168325195e-05,
+      "loss": 2.0281,
+      "step": 7384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8759106333822804e-05,
+      "loss": 2.0081,
+      "step": 7385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8758763455088244e-05,
+      "loss": 1.9794,
+      "step": 7386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875842053212326e-05,
+      "loss": 2.0241,
+      "step": 7387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8758077564929572e-05,
+      "loss": 2.0056,
+      "step": 7388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875773455350892e-05,
+      "loss": 2.012,
+      "step": 7389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8757391497863035e-05,
+      "loss": 2.0046,
+      "step": 7390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8757048397993648e-05,
+      "loss": 2.0107,
+      "step": 7391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8756705253902493e-05,
+      "loss": 2.0031,
+      "step": 7392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.87563620655913e-05,
+      "loss": 1.9996,
+      "step": 7393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8756018833061803e-05,
+      "loss": 2.0084,
+      "step": 7394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875567555631574e-05,
+      "loss": 2.0055,
+      "step": 7395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8755332235354837e-05,
+      "loss": 2.0191,
+      "step": 7396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8754988870180836e-05,
+      "loss": 1.9988,
+      "step": 7397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875464546079546e-05,
+      "loss": 2.028,
+      "step": 7398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875430200720045e-05,
+      "loss": 2.0115,
+      "step": 7399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8753958509397546e-05,
+      "loss": 2.0211,
+      "step": 7400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875361496738847e-05,
+      "loss": 2.005,
+      "step": 7401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8753271381174963e-05,
+      "loss": 2.0118,
+      "step": 7402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8752927750758764e-05,
+      "loss": 2.0261,
+      "step": 7403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8752584076141605e-05,
+      "loss": 2.0143,
+      "step": 7404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875224035732522e-05,
+      "loss": 2.0106,
+      "step": 7405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875189659431134e-05,
+      "loss": 2.0246,
+      "step": 7406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8751552787101714e-05,
+      "loss": 1.997,
+      "step": 7407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8751208935698072e-05,
+      "loss": 2.0264,
+      "step": 7408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8750865040102143e-05,
+      "loss": 2.0335,
+      "step": 7409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8750521100315674e-05,
+      "loss": 2.0184,
+      "step": 7410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8750177116340395e-05,
+      "loss": 2.0155,
+      "step": 7411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874983308817805e-05,
+      "loss": 2.0397,
+      "step": 7412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874948901583037e-05,
+      "loss": 2.0077,
+      "step": 7413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874914489929909e-05,
+      "loss": 2.0046,
+      "step": 7414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874880073858596e-05,
+      "loss": 2.002,
+      "step": 7415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8748456533692707e-05,
+      "loss": 2.0217,
+      "step": 7416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8748112284621075e-05,
+      "loss": 2.0076,
+      "step": 7417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8747767991372793e-05,
+      "loss": 2.0229,
+      "step": 7418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8747423653949614e-05,
+      "loss": 2.018,
+      "step": 7419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8747079272353264e-05,
+      "loss": 2.0057,
+      "step": 7420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874673484658549e-05,
+      "loss": 2.0123,
+      "step": 7421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8746390376648033e-05,
+      "loss": 2.0155,
+      "step": 7422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8746045862542625e-05,
+      "loss": 2.0114,
+      "step": 7423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8745701304271008e-05,
+      "loss": 2.0329,
+      "step": 7424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8745356701834923e-05,
+      "loss": 2.0284,
+      "step": 7425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8745012055236107e-05,
+      "loss": 2.0121,
+      "step": 7426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874466736447631e-05,
+      "loss": 2.0126,
+      "step": 7427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8744322629557265e-05,
+      "loss": 2.0255,
+      "step": 7428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874397785048071e-05,
+      "loss": 2.0216,
+      "step": 7429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8743633027248394e-05,
+      "loss": 2.0052,
+      "step": 7430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874328815986205e-05,
+      "loss": 1.9995,
+      "step": 7431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874294324832343e-05,
+      "loss": 2.0121,
+      "step": 7432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8742598292634264e-05,
+      "loss": 2.0164,
+      "step": 7433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8742253292796305e-05,
+      "loss": 2.0205,
+      "step": 7434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8741908248811286e-05,
+      "loss": 1.9975,
+      "step": 7435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8741563160680957e-05,
+      "loss": 2.0106,
+      "step": 7436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8741218028407054e-05,
+      "loss": 2.0192,
+      "step": 7437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8740872851991324e-05,
+      "loss": 2.0052,
+      "step": 7438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874052763143551e-05,
+      "loss": 1.9978,
+      "step": 7439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8740182366741354e-05,
+      "loss": 2.0199,
+      "step": 7440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.87398370579106e-05,
+      "loss": 2.0094,
+      "step": 7441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8739491704944988e-05,
+      "loss": 2.0172,
+      "step": 7442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873914630784627e-05,
+      "loss": 2.0081,
+      "step": 7443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8738800866616184e-05,
+      "loss": 2.0142,
+      "step": 7444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8738455381256476e-05,
+      "loss": 2.0125,
+      "step": 7445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873810985176889e-05,
+      "loss": 2.0069,
+      "step": 7446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873776427815517e-05,
+      "loss": 2.0145,
+      "step": 7447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8737418660417068e-05,
+      "loss": 2.015,
+      "step": 7448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873707299855632e-05,
+      "loss": 1.9936,
+      "step": 7449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873672729257468e-05,
+      "loss": 2.0175,
+      "step": 7450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8736381542473886e-05,
+      "loss": 2.0146,
+      "step": 7451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873603574825569e-05,
+      "loss": 2.0131,
+      "step": 7452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8735689909921834e-05,
+      "loss": 2.0239,
+      "step": 7453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8735344027474066e-05,
+      "loss": 2.02,
+      "step": 7454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873499810091413e-05,
+      "loss": 2.0121,
+      "step": 7455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8734652130243778e-05,
+      "loss": 1.9872,
+      "step": 7456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8734306115464754e-05,
+      "loss": 2.0087,
+      "step": 7457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8733960056578807e-05,
+      "loss": 2.0119,
+      "step": 7458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8733613953587685e-05,
+      "loss": 1.9987,
+      "step": 7459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8733267806493133e-05,
+      "loss": 2.0009,
+      "step": 7460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.87329216152969e-05,
+      "loss": 2.0242,
+      "step": 7461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873257538000073e-05,
+      "loss": 1.9911,
+      "step": 7462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8732229100606385e-05,
+      "loss": 2.0111,
+      "step": 7463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.87318827771156e-05,
+      "loss": 2.0249,
+      "step": 7464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873153640953013e-05,
+      "loss": 1.9952,
+      "step": 7465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8731189997851718e-05,
+      "loss": 2.0201,
+      "step": 7466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873084354208212e-05,
+      "loss": 2.023,
+      "step": 7467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8730497042223086e-05,
+      "loss": 2.0008,
+      "step": 7468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8730150498276363e-05,
+      "loss": 2.0134,
+      "step": 7469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.87298039102437e-05,
+      "loss": 2.0092,
+      "step": 7470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872945727812685e-05,
+      "loss": 2.0036,
+      "step": 7471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872911060192756e-05,
+      "loss": 2.0026,
+      "step": 7472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872876388164758e-05,
+      "loss": 2.0043,
+      "step": 7473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8728417117288667e-05,
+      "loss": 2.0129,
+      "step": 7474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872807030885257e-05,
+      "loss": 2.0233,
+      "step": 7475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8727723456341036e-05,
+      "loss": 2.0174,
+      "step": 7476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872737655975582e-05,
+      "loss": 1.9994,
+      "step": 7477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8727029619098673e-05,
+      "loss": 1.994,
+      "step": 7478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872668263437135e-05,
+      "loss": 2.0149,
+      "step": 7479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8726335605575597e-05,
+      "loss": 2.0079,
+      "step": 7480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872598853271317e-05,
+      "loss": 2.0121,
+      "step": 7481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8725641415785822e-05,
+      "loss": 2.0168,
+      "step": 7482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8725294254795305e-05,
+      "loss": 2.0008,
+      "step": 7483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8724947049743375e-05,
+      "loss": 2.0067,
+      "step": 7484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872459980063178e-05,
+      "loss": 2.0023,
+      "step": 7485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872425250746228e-05,
+      "loss": 1.9907,
+      "step": 7486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8723905170236623e-05,
+      "loss": 2.0132,
+      "step": 7487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8723557788956564e-05,
+      "loss": 2.0146,
+      "step": 7488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872321036362386e-05,
+      "loss": 2.0161,
+      "step": 7489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8722862894240264e-05,
+      "loss": 1.9963,
+      "step": 7490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8722515380807528e-05,
+      "loss": 1.9969,
+      "step": 7491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8722167823327414e-05,
+      "loss": 1.9989,
+      "step": 7492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8721820221801668e-05,
+      "loss": 2.0283,
+      "step": 7493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8721472576232052e-05,
+      "loss": 1.9964,
+      "step": 7494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8721124886620318e-05,
+      "loss": 2.0171,
+      "step": 7495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8720777152968227e-05,
+      "loss": 2.0099,
+      "step": 7496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8720429375277528e-05,
+      "loss": 2.0035,
+      "step": 7497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8720081553549982e-05,
+      "loss": 2.0142,
+      "step": 7498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871973368778734e-05,
+      "loss": 2.0107,
+      "step": 7499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8719385777991368e-05,
+      "loss": 2.0184,
+      "step": 7500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8719037824163816e-05,
+      "loss": 1.991,
+      "step": 7501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871868982630644e-05,
+      "loss": 2.0047,
+      "step": 7502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8718341784421002e-05,
+      "loss": 2.0098,
+      "step": 7503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8717993698509257e-05,
+      "loss": 2.0193,
+      "step": 7504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871764556857296e-05,
+      "loss": 2.0012,
+      "step": 7505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871729739461388e-05,
+      "loss": 1.9907,
+      "step": 7506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871694917663376e-05,
+      "loss": 2.0168,
+      "step": 7507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871660091463437e-05,
+      "loss": 2.0016,
+      "step": 7508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8716252608617463e-05,
+      "loss": 2.014,
+      "step": 7509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.87159042585848e-05,
+      "loss": 2.0154,
+      "step": 7510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871555586453814e-05,
+      "loss": 1.9863,
+      "step": 7511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871520742647924e-05,
+      "loss": 2.0227,
+      "step": 7512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8714858944409865e-05,
+      "loss": 2.0027,
+      "step": 7513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8714510418331767e-05,
+      "loss": 2.0187,
+      "step": 7514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8714161848246713e-05,
+      "loss": 2.0148,
+      "step": 7515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871381323415646e-05,
+      "loss": 2.0093,
+      "step": 7516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871346457606277e-05,
+      "loss": 2.0184,
+      "step": 7517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8713115873967402e-05,
+      "loss": 2.0216,
+      "step": 7518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8712767127872117e-05,
+      "loss": 2.0003,
+      "step": 7519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8712418337778677e-05,
+      "loss": 2.0243,
+      "step": 7520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8712069503688842e-05,
+      "loss": 1.9816,
+      "step": 7521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8711720625604376e-05,
+      "loss": 1.999,
+      "step": 7522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8711371703527035e-05,
+      "loss": 2.0109,
+      "step": 7523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871102273745859e-05,
+      "loss": 2.0159,
+      "step": 7524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.87106737274008e-05,
+      "loss": 2.0234,
+      "step": 7525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871032467335542e-05,
+      "loss": 2.0023,
+      "step": 7526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8709975575324223e-05,
+      "loss": 2.0027,
+      "step": 7527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8709626433308966e-05,
+      "loss": 2.0155,
+      "step": 7528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8709277247311416e-05,
+      "loss": 2.0184,
+      "step": 7529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8708928017333333e-05,
+      "loss": 2.0125,
+      "step": 7530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8708578743376476e-05,
+      "loss": 2.0082,
+      "step": 7531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8708229425442624e-05,
+      "loss": 2.0065,
+      "step": 7532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8707880063533522e-05,
+      "loss": 2.0225,
+      "step": 7533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870753065765095e-05,
+      "loss": 2.0084,
+      "step": 7534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8707181207796664e-05,
+      "loss": 2.0091,
+      "step": 7535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870683171397243e-05,
+      "loss": 2.0068,
+      "step": 7536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8706482176180016e-05,
+      "loss": 2.0067,
+      "step": 7537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870613259442118e-05,
+      "loss": 1.9984,
+      "step": 7538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8705782968697697e-05,
+      "loss": 1.9983,
+      "step": 7539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8705433299011323e-05,
+      "loss": 2.0298,
+      "step": 7540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8705083585363834e-05,
+      "loss": 2.0095,
+      "step": 7541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8704733827756986e-05,
+      "loss": 1.9997,
+      "step": 7542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870438402619255e-05,
+      "loss": 2.0334,
+      "step": 7543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8704034180672296e-05,
+      "loss": 2.0089,
+      "step": 7544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8703684291197982e-05,
+      "loss": 2.0079,
+      "step": 7545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870333435777138e-05,
+      "loss": 2.0132,
+      "step": 7546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8702984380394257e-05,
+      "loss": 2.0018,
+      "step": 7547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8702634359068382e-05,
+      "loss": 2.0117,
+      "step": 7548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870228429379552e-05,
+      "loss": 1.9996,
+      "step": 7549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870193418457744e-05,
+      "loss": 2.0022,
+      "step": 7550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8701584031415903e-05,
+      "loss": 1.993,
+      "step": 7551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870123383431269e-05,
+      "loss": 2.0119,
+      "step": 7552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8700883593269563e-05,
+      "loss": 2.0087,
+      "step": 7553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870053330828829e-05,
+      "loss": 2.0024,
+      "step": 7554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870018297937064e-05,
+      "loss": 2.0041,
+      "step": 7555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8699832606518383e-05,
+      "loss": 2.0176,
+      "step": 7556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8699482189733287e-05,
+      "loss": 2.0155,
+      "step": 7557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8699131729017124e-05,
+      "loss": 1.9941,
+      "step": 7558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869878122437166e-05,
+      "loss": 2.017,
+      "step": 7559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869843067579867e-05,
+      "loss": 2.0242,
+      "step": 7560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869808008329992e-05,
+      "loss": 2.0001,
+      "step": 7561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8697729446877183e-05,
+      "loss": 1.9906,
+      "step": 7562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8697378766532228e-05,
+      "loss": 2.0135,
+      "step": 7563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8697028042266826e-05,
+      "loss": 2.0122,
+      "step": 7564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869667727408275e-05,
+      "loss": 2.0066,
+      "step": 7565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8696326461981765e-05,
+      "loss": 2.0292,
+      "step": 7566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8695975605965656e-05,
+      "loss": 2.0155,
+      "step": 7567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869562470603618e-05,
+      "loss": 2.0025,
+      "step": 7568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869527376219512e-05,
+      "loss": 2.0043,
+      "step": 7569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8694922774444236e-05,
+      "loss": 2.0093,
+      "step": 7570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8694571742785315e-05,
+      "loss": 2.0061,
+      "step": 7571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869422066722012e-05,
+      "loss": 2.0287,
+      "step": 7572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8693869547750426e-05,
+      "loss": 2.0009,
+      "step": 7573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8693518384378008e-05,
+      "loss": 2.0352,
+      "step": 7574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8693167177104637e-05,
+      "loss": 2.0029,
+      "step": 7575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8692815925932087e-05,
+      "loss": 2.0135,
+      "step": 7576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8692464630862132e-05,
+      "loss": 2.0013,
+      "step": 7577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8692113291896548e-05,
+      "loss": 2.0215,
+      "step": 7578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8691761909037104e-05,
+      "loss": 2.0091,
+      "step": 7579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869141048228558e-05,
+      "loss": 2.001,
+      "step": 7580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8691059011643748e-05,
+      "loss": 2.0124,
+      "step": 7581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8690707497113384e-05,
+      "loss": 1.9912,
+      "step": 7582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8690355938696264e-05,
+      "loss": 2.0092,
+      "step": 7583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8690004336394158e-05,
+      "loss": 2.0087,
+      "step": 7584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868965269020885e-05,
+      "loss": 2.0089,
+      "step": 7585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8689301000142105e-05,
+      "loss": 2.003,
+      "step": 7586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868894926619571e-05,
+      "loss": 2.0218,
+      "step": 7587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8688597488371433e-05,
+      "loss": 2.0175,
+      "step": 7588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8688245666671055e-05,
+      "loss": 2.0276,
+      "step": 7589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868789380109635e-05,
+      "loss": 2.0142,
+      "step": 7590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86875418916491e-05,
+      "loss": 2.0078,
+      "step": 7591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8687189938331074e-05,
+      "loss": 2.0118,
+      "step": 7592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8686837941144057e-05,
+      "loss": 1.9926,
+      "step": 7593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868648590008982e-05,
+      "loss": 2.0165,
+      "step": 7594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8686133815170147e-05,
+      "loss": 2.0111,
+      "step": 7595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868578168638681e-05,
+      "loss": 1.9936,
+      "step": 7596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868542951374159e-05,
+      "loss": 2.0007,
+      "step": 7597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8685077297236267e-05,
+      "loss": 2.0223,
+      "step": 7598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868472503687262e-05,
+      "loss": 1.9922,
+      "step": 7599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8684372732652424e-05,
+      "loss": 2.0029,
+      "step": 7600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868402038457746e-05,
+      "loss": 2.019,
+      "step": 7601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8683667992649503e-05,
+      "loss": 2.0202,
+      "step": 7602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868331555687034e-05,
+      "loss": 1.9911,
+      "step": 7603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868296307724175e-05,
+      "loss": 2.0074,
+      "step": 7604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868261055376551e-05,
+      "loss": 2.0274,
+      "step": 7605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86822579864434e-05,
+      "loss": 2.0166,
+      "step": 7606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86819053752772e-05,
+      "loss": 2.0123,
+      "step": 7607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8681552720268693e-05,
+      "loss": 2.0178,
+      "step": 7608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868120002141966e-05,
+      "loss": 2.0176,
+      "step": 7609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8680847278731876e-05,
+      "loss": 1.9999,
+      "step": 7610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8680494492207134e-05,
+      "loss": 1.9965,
+      "step": 7611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8680141661847204e-05,
+      "loss": 1.9985,
+      "step": 7612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867978878765387e-05,
+      "loss": 2.0163,
+      "step": 7613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8679435869628922e-05,
+      "loss": 2.0158,
+      "step": 7614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8679082907774133e-05,
+      "loss": 2.0151,
+      "step": 7615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867872990209129e-05,
+      "loss": 2.0369,
+      "step": 7616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8678376852582173e-05,
+      "loss": 2.0267,
+      "step": 7617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867802375924857e-05,
+      "loss": 2.0098,
+      "step": 7618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8677670622092256e-05,
+      "loss": 2.0211,
+      "step": 7619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867731744111502e-05,
+      "loss": 2.0241,
+      "step": 7620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8676964216318648e-05,
+      "loss": 2.0049,
+      "step": 7621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8676610947704918e-05,
+      "loss": 2.0099,
+      "step": 7622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8676257635275613e-05,
+      "loss": 2.0024,
+      "step": 7623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867590427903252e-05,
+      "loss": 2.0287,
+      "step": 7624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8675550878977424e-05,
+      "loss": 2.0344,
+      "step": 7625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8675197435112113e-05,
+      "loss": 2.0437,
+      "step": 7626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8674843947438363e-05,
+      "loss": 2.0137,
+      "step": 7627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8674490415957963e-05,
+      "loss": 1.9979,
+      "step": 7628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8674136840672702e-05,
+      "loss": 2.0128,
+      "step": 7629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8673783221584365e-05,
+      "loss": 2.0138,
+      "step": 7630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8673429558694732e-05,
+      "loss": 2.0239,
+      "step": 7631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8673075852005594e-05,
+      "loss": 1.9963,
+      "step": 7632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8672722101518737e-05,
+      "loss": 2.0124,
+      "step": 7633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8672368307235943e-05,
+      "loss": 2.0117,
+      "step": 7634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8672014469159e-05,
+      "loss": 2.0153,
+      "step": 7635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86716605872897e-05,
+      "loss": 2.0136,
+      "step": 7636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8671306661629827e-05,
+      "loss": 2.0202,
+      "step": 7637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8670952692181165e-05,
+      "loss": 2.014,
+      "step": 7638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8670598678945505e-05,
+      "loss": 2.012,
+      "step": 7639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8670244621924634e-05,
+      "loss": 2.021,
+      "step": 7640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8669890521120343e-05,
+      "loss": 2.0127,
+      "step": 7641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8669536376534415e-05,
+      "loss": 1.9912,
+      "step": 7642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866918218816864e-05,
+      "loss": 2.0156,
+      "step": 7643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8668827956024804e-05,
+      "loss": 2.0097,
+      "step": 7644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86684736801047e-05,
+      "loss": 1.9982,
+      "step": 7645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8668119360410117e-05,
+      "loss": 2.0252,
+      "step": 7646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8667764996942847e-05,
+      "loss": 2.0084,
+      "step": 7647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866741058970467e-05,
+      "loss": 2.0091,
+      "step": 7648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8667056138697382e-05,
+      "loss": 2.0285,
+      "step": 7649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8666701643922773e-05,
+      "loss": 1.9958,
+      "step": 7650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8666347105382635e-05,
+      "loss": 2.0075,
+      "step": 7651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866599252307875e-05,
+      "loss": 2.0056,
+      "step": 7652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866563789701292e-05,
+      "loss": 2.0354,
+      "step": 7653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8665283227186928e-05,
+      "loss": 2.0083,
+      "step": 7654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8664928513602566e-05,
+      "loss": 2.0032,
+      "step": 7655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8664573756261627e-05,
+      "loss": 2.0256,
+      "step": 7656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86642189551659e-05,
+      "loss": 2.0242,
+      "step": 7657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8663864110317186e-05,
+      "loss": 2.0174,
+      "step": 7658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866350922171726e-05,
+      "loss": 2.0376,
+      "step": 7659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866315428936793e-05,
+      "loss": 2.0093,
+      "step": 7660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866279931327098e-05,
+      "loss": 2.0195,
+      "step": 7661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8662444293428205e-05,
+      "loss": 2.0045,
+      "step": 7662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8662089229841396e-05,
+      "loss": 2.0268,
+      "step": 7663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866173412251235e-05,
+      "loss": 2.0281,
+      "step": 7664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8661378971442853e-05,
+      "loss": 2.0196,
+      "step": 7665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8661023776634707e-05,
+      "loss": 2.0053,
+      "step": 7666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86606685380897e-05,
+      "loss": 2.0018,
+      "step": 7667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866031325580963e-05,
+      "loss": 2.0196,
+      "step": 7668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8659957929796286e-05,
+      "loss": 2.0084,
+      "step": 7669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8659602560051465e-05,
+      "loss": 2.024,
+      "step": 7670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865924714657696e-05,
+      "loss": 2.0098,
+      "step": 7671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8658891689374573e-05,
+      "loss": 2.0104,
+      "step": 7672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865853618844609e-05,
+      "loss": 2.0256,
+      "step": 7673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865818064379331e-05,
+      "loss": 2.0027,
+      "step": 7674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8657825055418026e-05,
+      "loss": 2.0188,
+      "step": 7675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865746942332204e-05,
+      "loss": 2.0115,
+      "step": 7676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8657113747507142e-05,
+      "loss": 2.011,
+      "step": 7677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865675802797513e-05,
+      "loss": 1.9976,
+      "step": 7678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8656402264727797e-05,
+      "loss": 2.0369,
+      "step": 7679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8656046457766947e-05,
+      "loss": 1.9869,
+      "step": 7680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865569060709437e-05,
+      "loss": 2.0207,
+      "step": 7681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8655334712711862e-05,
+      "loss": 2.0301,
+      "step": 7682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8654978774621232e-05,
+      "loss": 2.0127,
+      "step": 7683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865462279282426e-05,
+      "loss": 1.9879,
+      "step": 7684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8654266767322758e-05,
+      "loss": 2.0184,
+      "step": 7685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865391069811852e-05,
+      "loss": 2.0169,
+      "step": 7686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865355458521334e-05,
+      "loss": 2.0038,
+      "step": 7687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865319842860902e-05,
+      "loss": 1.9856,
+      "step": 7688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8652842228307357e-05,
+      "loss": 2.0172,
+      "step": 7689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865248598431015e-05,
+      "loss": 2.0219,
+      "step": 7690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8652129696619198e-05,
+      "loss": 1.9771,
+      "step": 7691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8651773365236303e-05,
+      "loss": 2.0112,
+      "step": 7692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8651416990163263e-05,
+      "loss": 2.0098,
+      "step": 7693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8651060571401872e-05,
+      "loss": 2.0005,
+      "step": 7694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8650704108953938e-05,
+      "loss": 2.0239,
+      "step": 7695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865034760282125e-05,
+      "loss": 1.9939,
+      "step": 7696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8649991053005626e-05,
+      "loss": 2.0037,
+      "step": 7697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8649634459508853e-05,
+      "loss": 1.9964,
+      "step": 7698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8649277822332733e-05,
+      "loss": 2.0061,
+      "step": 7699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8648921141479067e-05,
+      "loss": 2.0112,
+      "step": 7700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8648564416949662e-05,
+      "loss": 2.0096,
+      "step": 7701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8648207648746313e-05,
+      "loss": 2.0131,
+      "step": 7702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8647850836870827e-05,
+      "loss": 2.0205,
+      "step": 7703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8647493981325e-05,
+      "loss": 2.0131,
+      "step": 7704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8647137082110643e-05,
+      "loss": 2.0097,
+      "step": 7705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8646780139229544e-05,
+      "loss": 2.0115,
+      "step": 7706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864642315268352e-05,
+      "loss": 2.0097,
+      "step": 7707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8646066122474365e-05,
+      "loss": 2.0191,
+      "step": 7708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8645709048603887e-05,
+      "loss": 2.0009,
+      "step": 7709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8645351931073885e-05,
+      "loss": 2.0309,
+      "step": 7710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8644994769886168e-05,
+      "loss": 2.0249,
+      "step": 7711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864463756504253e-05,
+      "loss": 2.0096,
+      "step": 7712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8644280316544783e-05,
+      "loss": 1.9971,
+      "step": 7713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8643923024394728e-05,
+      "loss": 2.0051,
+      "step": 7714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864356568859417e-05,
+      "loss": 2.0116,
+      "step": 7715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8643208309144915e-05,
+      "loss": 2.0115,
+      "step": 7716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8642850886048766e-05,
+      "loss": 1.9956,
+      "step": 7717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8642493419307527e-05,
+      "loss": 2.0095,
+      "step": 7718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8642135908923003e-05,
+      "loss": 1.9986,
+      "step": 7719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8641778354897e-05,
+      "loss": 2.0014,
+      "step": 7720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864142075723133e-05,
+      "loss": 1.9904,
+      "step": 7721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8641063115927788e-05,
+      "loss": 2.0188,
+      "step": 7722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8640705430988185e-05,
+      "loss": 2.0142,
+      "step": 7723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864034770241433e-05,
+      "loss": 1.9969,
+      "step": 7724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8639989930208026e-05,
+      "loss": 2.0117,
+      "step": 7725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863963211437108e-05,
+      "loss": 2.0241,
+      "step": 7726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8639274254905297e-05,
+      "loss": 2.027,
+      "step": 7727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863891635181249e-05,
+      "loss": 2.0046,
+      "step": 7728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8638558405094462e-05,
+      "loss": 2.0241,
+      "step": 7729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863820041475302e-05,
+      "loss": 2.0003,
+      "step": 7730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8637842380789975e-05,
+      "loss": 2.0137,
+      "step": 7731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863748430320713e-05,
+      "loss": 1.9958,
+      "step": 7732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86371261820063e-05,
+      "loss": 2.0056,
+      "step": 7733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863676801718929e-05,
+      "loss": 1.9981,
+      "step": 7734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8636409808757908e-05,
+      "loss": 2.002,
+      "step": 7735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8636051556713962e-05,
+      "loss": 2.019,
+      "step": 7736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8635693261059264e-05,
+      "loss": 2.0185,
+      "step": 7737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863533492179562e-05,
+      "loss": 2.0288,
+      "step": 7738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8634976538924846e-05,
+      "loss": 1.9986,
+      "step": 7739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8634618112448745e-05,
+      "loss": 2.0159,
+      "step": 7740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8634259642369126e-05,
+      "loss": 2.0112,
+      "step": 7741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8633901128687805e-05,
+      "loss": 2.03,
+      "step": 7742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8633542571406592e-05,
+      "loss": 2.0029,
+      "step": 7743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8633183970527295e-05,
+      "loss": 1.9972,
+      "step": 7744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8632825326051724e-05,
+      "loss": 1.9952,
+      "step": 7745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863246663798169e-05,
+      "loss": 2.0222,
+      "step": 7746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863210790631901e-05,
+      "loss": 1.9967,
+      "step": 7747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863174913106549e-05,
+      "loss": 1.9878,
+      "step": 7748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8631390312222946e-05,
+      "loss": 2.0361,
+      "step": 7749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8631031449793184e-05,
+      "loss": 1.9972,
+      "step": 7750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8630672543778016e-05,
+      "loss": 1.987,
+      "step": 7751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8630313594179263e-05,
+      "loss": 2.0321,
+      "step": 7752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8629954600998733e-05,
+      "loss": 2.0151,
+      "step": 7753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8629595564238233e-05,
+      "loss": 1.9795,
+      "step": 7754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8629236483899585e-05,
+      "loss": 2.0165,
+      "step": 7755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8628877359984602e-05,
+      "loss": 1.9997,
+      "step": 7756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862851819249509e-05,
+      "loss": 2.0072,
+      "step": 7757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8628158981432865e-05,
+      "loss": 1.9999,
+      "step": 7758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8627799726799746e-05,
+      "loss": 1.9987,
+      "step": 7759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8627440428597546e-05,
+      "loss": 2.0111,
+      "step": 7760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8627081086828075e-05,
+      "loss": 2.0231,
+      "step": 7761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862672170149315e-05,
+      "loss": 2.0163,
+      "step": 7762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8626362272594587e-05,
+      "loss": 2.0105,
+      "step": 7763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8626002800134203e-05,
+      "loss": 2.009,
+      "step": 7764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8625643284113805e-05,
+      "loss": 2.0166,
+      "step": 7765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8625283724535215e-05,
+      "loss": 1.9971,
+      "step": 7766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862492412140025e-05,
+      "loss": 1.9965,
+      "step": 7767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862456447471072e-05,
+      "loss": 2.0218,
+      "step": 7768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8624204784468446e-05,
+      "loss": 2.027,
+      "step": 7769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8623845050675243e-05,
+      "loss": 2.0223,
+      "step": 7770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8623485273332932e-05,
+      "loss": 2.0339,
+      "step": 7771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862312545244332e-05,
+      "loss": 2.0112,
+      "step": 7772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862276558800823e-05,
+      "loss": 2.0099,
+      "step": 7773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8622405680029482e-05,
+      "loss": 2.0022,
+      "step": 7774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862204572850889e-05,
+      "loss": 2.0143,
+      "step": 7775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862168573344827e-05,
+      "loss": 2.0151,
+      "step": 7776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862132569484944e-05,
+      "loss": 2.0141,
+      "step": 7777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8620965612714226e-05,
+      "loss": 2.0238,
+      "step": 7778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8620605487044436e-05,
+      "loss": 2.0174,
+      "step": 7779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862024531784189e-05,
+      "loss": 1.9957,
+      "step": 7780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8619885105108417e-05,
+      "loss": 2.0298,
+      "step": 7781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8619524848845824e-05,
+      "loss": 1.9999,
+      "step": 7782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8619164549055936e-05,
+      "loss": 2.0004,
+      "step": 7783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8618804205740572e-05,
+      "loss": 2.0238,
+      "step": 7784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8618443818901552e-05,
+      "loss": 2.01,
+      "step": 7785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861808338854069e-05,
+      "loss": 2.0048,
+      "step": 7786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861772291465982e-05,
+      "loss": 2.0197,
+      "step": 7787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8617362397260748e-05,
+      "loss": 1.9947,
+      "step": 7788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86170018363453e-05,
+      "loss": 2.0061,
+      "step": 7789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.86166412319153e-05,
+      "loss": 2.0255,
+      "step": 7790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8616280583972563e-05,
+      "loss": 2.0216,
+      "step": 7791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8615919892518912e-05,
+      "loss": 2.0274,
+      "step": 7792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8615559157556168e-05,
+      "loss": 2.0135,
+      "step": 7793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.861519837908616e-05,
+      "loss": 2.0,
+      "step": 7794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.86148375571107e-05,
+      "loss": 1.9964,
+      "step": 7795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8614476691631614e-05,
+      "loss": 2.0006,
+      "step": 7796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8614115782650728e-05,
+      "loss": 2.0181,
+      "step": 7797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8613754830169858e-05,
+      "loss": 2.0324,
+      "step": 7798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8613393834190835e-05,
+      "loss": 2.0098,
+      "step": 7799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.861303279471547e-05,
+      "loss": 2.0169,
+      "step": 7800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8612671711745593e-05,
+      "loss": 1.9992,
+      "step": 7801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8612310585283035e-05,
+      "loss": 2.0086,
+      "step": 7802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8611949415329603e-05,
+      "loss": 2.0004,
+      "step": 7803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8611588201887135e-05,
+      "loss": 2.007,
+      "step": 7804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8611226944957447e-05,
+      "loss": 2.0194,
+      "step": 7805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8610865644542365e-05,
+      "loss": 2.0159,
+      "step": 7806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8610504300643718e-05,
+      "loss": 2.0229,
+      "step": 7807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.861014291326333e-05,
+      "loss": 1.9872,
+      "step": 7808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8609781482403018e-05,
+      "loss": 2.0158,
+      "step": 7809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8609420008064614e-05,
+      "loss": 2.0181,
+      "step": 7810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8609058490249944e-05,
+      "loss": 2.014,
+      "step": 7811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860869692896083e-05,
+      "loss": 2.0155,
+      "step": 7812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8608335324199098e-05,
+      "loss": 2.0242,
+      "step": 7813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8607973675966578e-05,
+      "loss": 2.0107,
+      "step": 7814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8607611984265093e-05,
+      "loss": 2.0216,
+      "step": 7815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860725024909647e-05,
+      "loss": 2.0086,
+      "step": 7816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8606888470462533e-05,
+      "loss": 2.0074,
+      "step": 7817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8606526648365117e-05,
+      "loss": 2.024,
+      "step": 7818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860616478280604e-05,
+      "loss": 2.0036,
+      "step": 7819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8605802873787136e-05,
+      "loss": 2.0046,
+      "step": 7820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8605440921310228e-05,
+      "loss": 2.0261,
+      "step": 7821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8605078925377148e-05,
+      "loss": 2.0198,
+      "step": 7822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860471688598972e-05,
+      "loss": 2.0072,
+      "step": 7823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8604354803149775e-05,
+      "loss": 2.0327,
+      "step": 7824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8603992676859137e-05,
+      "loss": 2.0227,
+      "step": 7825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860363050711964e-05,
+      "loss": 2.0224,
+      "step": 7826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8603268293933112e-05,
+      "loss": 2.013,
+      "step": 7827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8602906037301384e-05,
+      "loss": 2.0123,
+      "step": 7828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8602543737226277e-05,
+      "loss": 2.0285,
+      "step": 7829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860218139370963e-05,
+      "loss": 2.0211,
+      "step": 7830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860181900675327e-05,
+      "loss": 2.0161,
+      "step": 7831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860145657635902e-05,
+      "loss": 2.0191,
+      "step": 7832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8601094102528718e-05,
+      "loss": 2.0054,
+      "step": 7833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8600731585264196e-05,
+      "loss": 2.0291,
+      "step": 7834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8600369024567277e-05,
+      "loss": 2.0158,
+      "step": 7835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8600006420439797e-05,
+      "loss": 2.0055,
+      "step": 7836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859964377288359e-05,
+      "loss": 2.0093,
+      "step": 7837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859928108190048e-05,
+      "loss": 2.0054,
+      "step": 7838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8598918347492304e-05,
+      "loss": 2.0024,
+      "step": 7839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859855556966089e-05,
+      "loss": 2.0112,
+      "step": 7840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859819274840807e-05,
+      "loss": 2.0108,
+      "step": 7841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859782988373568e-05,
+      "loss": 2.0017,
+      "step": 7842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8597466975645552e-05,
+      "loss": 2.011,
+      "step": 7843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8597104024139514e-05,
+      "loss": 2.0249,
+      "step": 7844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8596741029219404e-05,
+      "loss": 2.0126,
+      "step": 7845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8596377990887054e-05,
+      "loss": 2.0041,
+      "step": 7846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8596014909144292e-05,
+      "loss": 2.0209,
+      "step": 7847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859565178399296e-05,
+      "loss": 2.0036,
+      "step": 7848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8595288615434883e-05,
+      "loss": 2.0168,
+      "step": 7849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8594925403471902e-05,
+      "loss": 2.0104,
+      "step": 7850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8594562148105848e-05,
+      "loss": 2.0196,
+      "step": 7851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8594198849338558e-05,
+      "loss": 2.0009,
+      "step": 7852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859383550717186e-05,
+      "loss": 2.0074,
+      "step": 7853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8593472121607593e-05,
+      "loss": 2.0289,
+      "step": 7854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8593108692647598e-05,
+      "loss": 2.0092,
+      "step": 7855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8592745220293702e-05,
+      "loss": 2.0179,
+      "step": 7856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859238170454774e-05,
+      "loss": 2.0348,
+      "step": 7857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8592018145411556e-05,
+      "loss": 2.0125,
+      "step": 7858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8591654542886978e-05,
+      "loss": 2.0175,
+      "step": 7859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8591290896975845e-05,
+      "loss": 2.0084,
+      "step": 7860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8590927207679992e-05,
+      "loss": 2.0118,
+      "step": 7861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859056347500126e-05,
+      "loss": 2.0123,
+      "step": 7862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859019969894148e-05,
+      "loss": 2.016,
+      "step": 7863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8589835879502494e-05,
+      "loss": 2.0335,
+      "step": 7864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8589472016686135e-05,
+      "loss": 2.0034,
+      "step": 7865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858910811049424e-05,
+      "loss": 2.0242,
+      "step": 7866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8588744160928654e-05,
+      "loss": 1.9997,
+      "step": 7867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8588380167991206e-05,
+      "loss": 2.0308,
+      "step": 7868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858801613168374e-05,
+      "loss": 2.0144,
+      "step": 7869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858765205200809e-05,
+      "loss": 1.9993,
+      "step": 7870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8587287928966097e-05,
+      "loss": 2.0097,
+      "step": 7871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8586923762559604e-05,
+      "loss": 2.025,
+      "step": 7872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8586559552790443e-05,
+      "loss": 2.0027,
+      "step": 7873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8586195299660454e-05,
+      "loss": 2.0083,
+      "step": 7874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858583100317148e-05,
+      "loss": 2.0201,
+      "step": 7875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8585466663325356e-05,
+      "loss": 2.0181,
+      "step": 7876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8585102280123928e-05,
+      "loss": 2.0033,
+      "step": 7877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8584737853569033e-05,
+      "loss": 2.0136,
+      "step": 7878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8584373383662507e-05,
+      "loss": 2.0081,
+      "step": 7879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8584008870406197e-05,
+      "loss": 2.0037,
+      "step": 7880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8583644313801942e-05,
+      "loss": 1.9953,
+      "step": 7881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8583279713851583e-05,
+      "loss": 2.0238,
+      "step": 7882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858291507055696e-05,
+      "loss": 2.0076,
+      "step": 7883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8582550383919917e-05,
+      "loss": 2.0031,
+      "step": 7884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858218565394229e-05,
+      "loss": 2.01,
+      "step": 7885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8581820880625922e-05,
+      "loss": 2.0102,
+      "step": 7886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8581456063972663e-05,
+      "loss": 2.0115,
+      "step": 7887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8581091203984347e-05,
+      "loss": 2.0248,
+      "step": 7888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858072630066282e-05,
+      "loss": 2.0029,
+      "step": 7889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858036135400992e-05,
+      "loss": 2.0187,
+      "step": 7890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8579996364027497e-05,
+      "loss": 2.0213,
+      "step": 7891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8579631330717393e-05,
+      "loss": 2.007,
+      "step": 7892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8579266254081445e-05,
+      "loss": 2.0211,
+      "step": 7893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8578901134121503e-05,
+      "loss": 2.0309,
+      "step": 7894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.857853597083941e-05,
+      "loss": 2.0076,
+      "step": 7895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8578170764237007e-05,
+      "loss": 2.0252,
+      "step": 7896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.857780551431614e-05,
+      "loss": 1.9937,
+      "step": 7897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8577440221078654e-05,
+      "loss": 2.0004,
+      "step": 7898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.857707488452639e-05,
+      "loss": 2.0194,
+      "step": 7899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85767095046612e-05,
+      "loss": 1.9974,
+      "step": 7900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8576344081484926e-05,
+      "loss": 2.011,
+      "step": 7901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8575978614999404e-05,
+      "loss": 2.0212,
+      "step": 7902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8575613105206497e-05,
+      "loss": 2.0077,
+      "step": 7903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.857524755210804e-05,
+      "loss": 1.9983,
+      "step": 7904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.857488195570588e-05,
+      "loss": 2.0192,
+      "step": 7905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.857451631600186e-05,
+      "loss": 2.0158,
+      "step": 7906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8574150632997834e-05,
+      "loss": 2.024,
+      "step": 7907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8573784906695645e-05,
+      "loss": 2.0152,
+      "step": 7908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8573419137097137e-05,
+      "loss": 2.0002,
+      "step": 7909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8573053324204166e-05,
+      "loss": 2.0163,
+      "step": 7910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8572687468018565e-05,
+      "loss": 2.0019,
+      "step": 7911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8572321568542195e-05,
+      "loss": 2.0155,
+      "step": 7912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8571955625776897e-05,
+      "loss": 2.0205,
+      "step": 7913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.857158963972452e-05,
+      "loss": 2.0217,
+      "step": 7914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8571223610386916e-05,
+      "loss": 2.0168,
+      "step": 7915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8570857537765925e-05,
+      "loss": 2.0189,
+      "step": 7916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8570491421863404e-05,
+      "loss": 2.0147,
+      "step": 7917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8570125262681196e-05,
+      "loss": 2.0312,
+      "step": 7918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8569759060221157e-05,
+      "loss": 1.9913,
+      "step": 7919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8569392814485124e-05,
+      "loss": 2.0174,
+      "step": 7920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856902652547496e-05,
+      "loss": 2.0174,
+      "step": 7921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8568660193192506e-05,
+      "loss": 1.9999,
+      "step": 7922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8568293817639617e-05,
+      "loss": 2.0054,
+      "step": 7923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856792739881814e-05,
+      "loss": 2.0001,
+      "step": 7924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856756093672993e-05,
+      "loss": 2.0084,
+      "step": 7925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856719443137683e-05,
+      "loss": 2.0095,
+      "step": 7926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8566827882760696e-05,
+      "loss": 2.0111,
+      "step": 7927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8566461290883375e-05,
+      "loss": 2.0083,
+      "step": 7928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8566094655746723e-05,
+      "loss": 2.0298,
+      "step": 7929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856572797735259e-05,
+      "loss": 2.0079,
+      "step": 7930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8565361255702825e-05,
+      "loss": 1.9882,
+      "step": 7931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8564994490799285e-05,
+      "loss": 2.0132,
+      "step": 7932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8564627682643817e-05,
+      "loss": 2.0013,
+      "step": 7933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8564260831238275e-05,
+      "loss": 2.0366,
+      "step": 7934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8563893936584514e-05,
+      "loss": 2.0143,
+      "step": 7935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856352699868438e-05,
+      "loss": 2.01,
+      "step": 7936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8563160017539738e-05,
+      "loss": 2.0386,
+      "step": 7937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856279299315243e-05,
+      "loss": 2.0112,
+      "step": 7938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856242592552431e-05,
+      "loss": 2.0074,
+      "step": 7939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8562058814657242e-05,
+      "loss": 2.012,
+      "step": 7940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8561691660553068e-05,
+      "loss": 2.0157,
+      "step": 7941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856132446321365e-05,
+      "loss": 2.029,
+      "step": 7942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8560957222640838e-05,
+      "loss": 2.0188,
+      "step": 7943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8560589938836485e-05,
+      "loss": 2.0199,
+      "step": 7944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856022261180245e-05,
+      "loss": 2.0071,
+      "step": 7945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8559855241540588e-05,
+      "loss": 2.0309,
+      "step": 7946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855948782805275e-05,
+      "loss": 2.0146,
+      "step": 7947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8559120371340794e-05,
+      "loss": 2.0237,
+      "step": 7948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855875287140658e-05,
+      "loss": 1.9998,
+      "step": 7949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8558385328251956e-05,
+      "loss": 2.0092,
+      "step": 7950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855801774187878e-05,
+      "loss": 1.9983,
+      "step": 7951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855765011228891e-05,
+      "loss": 2.0038,
+      "step": 7952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8557282439484204e-05,
+      "loss": 2.0149,
+      "step": 7953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8556914723466514e-05,
+      "loss": 2.0074,
+      "step": 7954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8556546964237703e-05,
+      "loss": 2.0075,
+      "step": 7955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8556179161799623e-05,
+      "loss": 2.0036,
+      "step": 7956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8555811316154132e-05,
+      "loss": 2.0075,
+      "step": 7957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8555443427303093e-05,
+      "loss": 2.0004,
+      "step": 7958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8555075495248356e-05,
+      "loss": 2.0056,
+      "step": 7959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855470751999178e-05,
+      "loss": 1.9985,
+      "step": 7960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855433950153523e-05,
+      "loss": 2.0294,
+      "step": 7961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8553971439880564e-05,
+      "loss": 2.0139,
+      "step": 7962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8553603335029628e-05,
+      "loss": 2.0112,
+      "step": 7963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8553235186984295e-05,
+      "loss": 2.0147,
+      "step": 7964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8552866995746417e-05,
+      "loss": 2.0121,
+      "step": 7965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8552498761317855e-05,
+      "loss": 2.0242,
+      "step": 7966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855213048370047e-05,
+      "loss": 2.0216,
+      "step": 7967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8551762162896118e-05,
+      "loss": 2.0296,
+      "step": 7968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8551393798906664e-05,
+      "loss": 2.0157,
+      "step": 7969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8551025391733962e-05,
+      "loss": 2.0286,
+      "step": 7970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855065694137988e-05,
+      "loss": 2.013,
+      "step": 7971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8550288447846273e-05,
+      "loss": 2.0382,
+      "step": 7972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8549919911135003e-05,
+      "loss": 2.0182,
+      "step": 7973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8549551331247933e-05,
+      "loss": 1.9989,
+      "step": 7974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8549182708186923e-05,
+      "loss": 1.9986,
+      "step": 7975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854881404195383e-05,
+      "loss": 2.0167,
+      "step": 7976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8548445332550522e-05,
+      "loss": 2.0092,
+      "step": 7977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854807657997886e-05,
+      "loss": 2.0142,
+      "step": 7978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8547707784240705e-05,
+      "loss": 2.0029,
+      "step": 7979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8547338945337916e-05,
+      "loss": 2.0308,
+      "step": 7980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8546970063272363e-05,
+      "loss": 2.0219,
+      "step": 7981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85466011380459e-05,
+      "loss": 2.0154,
+      "step": 7982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85462321696604e-05,
+      "loss": 2.0308,
+      "step": 7983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8545863158117717e-05,
+      "loss": 1.9928,
+      "step": 7984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8545494103419722e-05,
+      "loss": 2.0116,
+      "step": 7985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8545125005568274e-05,
+      "loss": 1.9982,
+      "step": 7986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8544755864565238e-05,
+      "loss": 2.004,
+      "step": 7987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8544386680412476e-05,
+      "loss": 2.0086,
+      "step": 7988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8544017453111856e-05,
+      "loss": 2.0146,
+      "step": 7989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854364818266524e-05,
+      "loss": 2.0035,
+      "step": 7990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8543278869074495e-05,
+      "loss": 2.0113,
+      "step": 7991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8542909512341484e-05,
+      "loss": 2.0242,
+      "step": 7992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8542540112468073e-05,
+      "loss": 2.0009,
+      "step": 7993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854217066945613e-05,
+      "loss": 2.0201,
+      "step": 7994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8541801183307513e-05,
+      "loss": 2.0165,
+      "step": 7995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8541431654024093e-05,
+      "loss": 2.0167,
+      "step": 7996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854106208160774e-05,
+      "loss": 2.0187,
+      "step": 7997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854069246606031e-05,
+      "loss": 2.0092,
+      "step": 7998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8540322807383683e-05,
+      "loss": 2.0186,
+      "step": 7999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8539953105579716e-05,
+      "loss": 2.0105,
+      "step": 8000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8539583360650276e-05,
+      "loss": 1.9963,
+      "step": 8001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8539213572597233e-05,
+      "loss": 2.0391,
+      "step": 8002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8538843741422457e-05,
+      "loss": 2.0314,
+      "step": 8003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.853847386712781e-05,
+      "loss": 2.0214,
+      "step": 8004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8538103949715163e-05,
+      "loss": 2.0247,
+      "step": 8005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.853773398918638e-05,
+      "loss": 2.0092,
+      "step": 8006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8537363985543334e-05,
+      "loss": 2.0123,
+      "step": 8007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8536993938787896e-05,
+      "loss": 2.0449,
+      "step": 8008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8536623848921927e-05,
+      "loss": 2.0066,
+      "step": 8009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85362537159473e-05,
+      "loss": 2.0203,
+      "step": 8010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.853588353986588e-05,
+      "loss": 2.012,
+      "step": 8011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8535513320679546e-05,
+      "loss": 2.0167,
+      "step": 8012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8535143058390158e-05,
+      "loss": 2.0047,
+      "step": 8013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8534772752999592e-05,
+      "loss": 1.9976,
+      "step": 8014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8534402404509714e-05,
+      "loss": 1.997,
+      "step": 8015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8534032012922392e-05,
+      "loss": 2.0211,
+      "step": 8016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8533661578239505e-05,
+      "loss": 2.0289,
+      "step": 8017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.853329110046292e-05,
+      "loss": 2.0142,
+      "step": 8018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85329205795945e-05,
+      "loss": 2.002,
+      "step": 8019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8532550015636126e-05,
+      "loss": 2.0093,
+      "step": 8020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8532179408589667e-05,
+      "loss": 2.0121,
+      "step": 8021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8531808758456987e-05,
+      "loss": 2.0181,
+      "step": 8022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8531438065239973e-05,
+      "loss": 2.0151,
+      "step": 8023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.853106732894048e-05,
+      "loss": 2.0298,
+      "step": 8024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8530696549560395e-05,
+      "loss": 1.9993,
+      "step": 8025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.853032572710158e-05,
+      "loss": 2.0138,
+      "step": 8026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852995486156591e-05,
+      "loss": 2.0109,
+      "step": 8027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852958395295526e-05,
+      "loss": 1.9911,
+      "step": 8028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8529213001271502e-05,
+      "loss": 2.0218,
+      "step": 8029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852884200651651e-05,
+      "loss": 2.0101,
+      "step": 8030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8528470968692156e-05,
+      "loss": 1.9984,
+      "step": 8031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8528099887800315e-05,
+      "loss": 2.0171,
+      "step": 8032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852772876384286e-05,
+      "loss": 2.0137,
+      "step": 8033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852735759682166e-05,
+      "loss": 2.0119,
+      "step": 8034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8526986386738604e-05,
+      "loss": 2.0061,
+      "step": 8035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8526615133595555e-05,
+      "loss": 2.0121,
+      "step": 8036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8526243837394383e-05,
+      "loss": 1.9937,
+      "step": 8037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8525872498136978e-05,
+      "loss": 2.0131,
+      "step": 8038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8525501115825203e-05,
+      "loss": 1.9983,
+      "step": 8039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8525129690460936e-05,
+      "loss": 2.0156,
+      "step": 8040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852475822204606e-05,
+      "loss": 2.0172,
+      "step": 8041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852438671058244e-05,
+      "loss": 2.001,
+      "step": 8042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8524015156071962e-05,
+      "loss": 2.0103,
+      "step": 8043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8523643558516494e-05,
+      "loss": 1.9913,
+      "step": 8044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852327191791792e-05,
+      "loss": 2.0124,
+      "step": 8045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8522900234278104e-05,
+      "loss": 2.0163,
+      "step": 8046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852252850759894e-05,
+      "loss": 2.0037,
+      "step": 8047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8522156737882296e-05,
+      "loss": 2.0118,
+      "step": 8048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8521784925130045e-05,
+      "loss": 2.0195,
+      "step": 8049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8521413069344074e-05,
+      "loss": 2.0192,
+      "step": 8050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852104117052626e-05,
+      "loss": 2.0235,
+      "step": 8051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852066922867847e-05,
+      "loss": 2.0116,
+      "step": 8052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8520297243802596e-05,
+      "loss": 2.0113,
+      "step": 8053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8519925215900506e-05,
+      "loss": 2.0259,
+      "step": 8054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8519553144974086e-05,
+      "loss": 2.0235,
+      "step": 8055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8519181031025213e-05,
+      "loss": 2.0101,
+      "step": 8056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8518808874055763e-05,
+      "loss": 2.0131,
+      "step": 8057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8518436674067617e-05,
+      "loss": 2.0276,
+      "step": 8058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8518064431062656e-05,
+      "loss": 2.0298,
+      "step": 8059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851769214504276e-05,
+      "loss": 2.0162,
+      "step": 8060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8517319816009805e-05,
+      "loss": 2.0351,
+      "step": 8061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8516947443965673e-05,
+      "loss": 2.007,
+      "step": 8062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851657502891225e-05,
+      "loss": 2.0205,
+      "step": 8063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851620257085141e-05,
+      "loss": 2.011,
+      "step": 8064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8515830069785034e-05,
+      "loss": 2.0016,
+      "step": 8065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851545752571501e-05,
+      "loss": 2.0211,
+      "step": 8066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8515084938643207e-05,
+      "loss": 1.9988,
+      "step": 8067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8514712308571518e-05,
+      "loss": 2.0302,
+      "step": 8068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851433963550182e-05,
+      "loss": 2.0103,
+      "step": 8069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8513966919436e-05,
+      "loss": 1.9983,
+      "step": 8070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851359416037593e-05,
+      "loss": 2.0393,
+      "step": 8071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8513221358323497e-05,
+      "loss": 1.9947,
+      "step": 8072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8512848513280585e-05,
+      "loss": 1.9996,
+      "step": 8073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851247562524908e-05,
+      "loss": 2.015,
+      "step": 8074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851210269423086e-05,
+      "loss": 2.0225,
+      "step": 8075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851172972022781e-05,
+      "loss": 2.0174,
+      "step": 8076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8511356703241813e-05,
+      "loss": 2.02,
+      "step": 8077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851098364327475e-05,
+      "loss": 2.0187,
+      "step": 8078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851061054032851e-05,
+      "loss": 2.0087,
+      "step": 8079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8510237394404977e-05,
+      "loss": 2.0202,
+      "step": 8080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8509864205506028e-05,
+      "loss": 2.0175,
+      "step": 8081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8509490973633557e-05,
+      "loss": 2.0112,
+      "step": 8082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8509117698789444e-05,
+      "loss": 2.0042,
+      "step": 8083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850874438097557e-05,
+      "loss": 1.9986,
+      "step": 8084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8508371020193826e-05,
+      "loss": 2.0179,
+      "step": 8085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85079976164461e-05,
+      "loss": 2.01,
+      "step": 8086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8507624169734274e-05,
+      "loss": 2.0078,
+      "step": 8087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850725068006023e-05,
+      "loss": 2.0078,
+      "step": 8088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850687714742586e-05,
+      "loss": 1.9988,
+      "step": 8089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8506503571833042e-05,
+      "loss": 2.0107,
+      "step": 8090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8506129953283674e-05,
+      "loss": 2.0163,
+      "step": 8091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8505756291779636e-05,
+      "loss": 2.016,
+      "step": 8092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8505382587322818e-05,
+      "loss": 2.0258,
+      "step": 8093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85050088399151e-05,
+      "loss": 2.0347,
+      "step": 8094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850463504955838e-05,
+      "loss": 1.9963,
+      "step": 8095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850426121625454e-05,
+      "loss": 2.0144,
+      "step": 8096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8503887340005464e-05,
+      "loss": 1.998,
+      "step": 8097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8503513420813045e-05,
+      "loss": 2.0177,
+      "step": 8098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8503139458679172e-05,
+      "loss": 2.0258,
+      "step": 8099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850276545360573e-05,
+      "loss": 2.0231,
+      "step": 8100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850239140559461e-05,
+      "loss": 2.0159,
+      "step": 8101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85020173146477e-05,
+      "loss": 2.0209,
+      "step": 8102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850164318076689e-05,
+      "loss": 2.0284,
+      "step": 8103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850126900395407e-05,
+      "loss": 2.0242,
+      "step": 8104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8500894784211124e-05,
+      "loss": 2.0187,
+      "step": 8105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850052052153995e-05,
+      "loss": 2.0257,
+      "step": 8106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8500146215942432e-05,
+      "loss": 2.0107,
+      "step": 8107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849977186742046e-05,
+      "loss": 2.0223,
+      "step": 8108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8499397475975927e-05,
+      "loss": 1.9975,
+      "step": 8109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8499023041610727e-05,
+      "loss": 2.0096,
+      "step": 8110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8498648564326744e-05,
+      "loss": 2.0041,
+      "step": 8111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849827404412587e-05,
+      "loss": 1.9889,
+      "step": 8112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8497899481010002e-05,
+      "loss": 2.0255,
+      "step": 8113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8497524874981026e-05,
+      "loss": 1.9875,
+      "step": 8114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8497150226040835e-05,
+      "loss": 2.0169,
+      "step": 8115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8496775534191324e-05,
+      "loss": 1.993,
+      "step": 8116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849640079943438e-05,
+      "loss": 2.0179,
+      "step": 8117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84960260217719e-05,
+      "loss": 1.9955,
+      "step": 8118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849565120120577e-05,
+      "loss": 2.0212,
+      "step": 8119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849527633773789e-05,
+      "loss": 2.0194,
+      "step": 8120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849490143137015e-05,
+      "loss": 2.0224,
+      "step": 8121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8494526482104443e-05,
+      "loss": 2.0137,
+      "step": 8122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8494151489942663e-05,
+      "loss": 1.9841,
+      "step": 8123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8493776454886708e-05,
+      "loss": 2.0147,
+      "step": 8124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849340137693846e-05,
+      "loss": 2.0014,
+      "step": 8125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8493026256099824e-05,
+      "loss": 2.0177,
+      "step": 8126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849265109237269e-05,
+      "loss": 2.0188,
+      "step": 8127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8492275885758956e-05,
+      "loss": 2.0057,
+      "step": 8128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8491900636260513e-05,
+      "loss": 2.0115,
+      "step": 8129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8491525343879255e-05,
+      "loss": 2.0145,
+      "step": 8130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849115000861708e-05,
+      "loss": 2.0209,
+      "step": 8131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8490774630475884e-05,
+      "loss": 1.9984,
+      "step": 8132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8490399209457555e-05,
+      "loss": 1.9971,
+      "step": 8133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8490023745564005e-05,
+      "loss": 2.0165,
+      "step": 8134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8489648238797112e-05,
+      "loss": 2.0092,
+      "step": 8135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8489272689158785e-05,
+      "loss": 2.0048,
+      "step": 8136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848889709665091e-05,
+      "loss": 2.0145,
+      "step": 8137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8488521461275395e-05,
+      "loss": 2.0254,
+      "step": 8138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848814578303413e-05,
+      "loss": 2.0005,
+      "step": 8139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8487770061929012e-05,
+      "loss": 2.0062,
+      "step": 8140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848739429796194e-05,
+      "loss": 2.0221,
+      "step": 8141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8487018491134813e-05,
+      "loss": 2.0076,
+      "step": 8142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8486642641449525e-05,
+      "loss": 2.008,
+      "step": 8143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8486266748907978e-05,
+      "loss": 2.0097,
+      "step": 8144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8485890813512067e-05,
+      "loss": 2.0088,
+      "step": 8145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8485514835263695e-05,
+      "loss": 2.0054,
+      "step": 8146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8485138814164753e-05,
+      "loss": 1.9999,
+      "step": 8147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8484762750217148e-05,
+      "loss": 2.0061,
+      "step": 8148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848438664342277e-05,
+      "loss": 2.0152,
+      "step": 8149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8484010493783527e-05,
+      "loss": 2.0001,
+      "step": 8150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8483634301301315e-05,
+      "loss": 2.0255,
+      "step": 8151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848325806597803e-05,
+      "loss": 2.0256,
+      "step": 8152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848288178781558e-05,
+      "loss": 2.0103,
+      "step": 8153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848250546681586e-05,
+      "loss": 2.0111,
+      "step": 8154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848212910298077e-05,
+      "loss": 2.01,
+      "step": 8155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8481752696312215e-05,
+      "loss": 2.0224,
+      "step": 8156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848137624681209e-05,
+      "loss": 1.998,
+      "step": 8157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8480999754482298e-05,
+      "loss": 2.0165,
+      "step": 8158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8480623219324738e-05,
+      "loss": 2.0183,
+      "step": 8159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848024664134132e-05,
+      "loss": 2.009,
+      "step": 8160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847987002053394e-05,
+      "loss": 2.0021,
+      "step": 8161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8479493356904494e-05,
+      "loss": 2.0014,
+      "step": 8162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8479116650454894e-05,
+      "loss": 1.9877,
+      "step": 8163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8478739901187036e-05,
+      "loss": 2.0102,
+      "step": 8164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8478363109102825e-05,
+      "loss": 2.0182,
+      "step": 8165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8477986274204165e-05,
+      "loss": 2.0118,
+      "step": 8166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8477609396492957e-05,
+      "loss": 2.0145,
+      "step": 8167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8477232475971104e-05,
+      "loss": 1.9895,
+      "step": 8168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8476855512640514e-05,
+      "loss": 2.0104,
+      "step": 8169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8476478506503078e-05,
+      "loss": 1.9921,
+      "step": 8170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8476101457560717e-05,
+      "loss": 1.9936,
+      "step": 8171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847572436581532e-05,
+      "loss": 2.0181,
+      "step": 8172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84753472312688e-05,
+      "loss": 1.9962,
+      "step": 8173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847497005392306e-05,
+      "loss": 2.005,
+      "step": 8174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8474592833780005e-05,
+      "loss": 1.9985,
+      "step": 8175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847421557084154e-05,
+      "loss": 2.0227,
+      "step": 8176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8473838265109566e-05,
+      "loss": 2.015,
+      "step": 8177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847346091658599e-05,
+      "loss": 2.0168,
+      "step": 8178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8473083525272718e-05,
+      "loss": 2.0103,
+      "step": 8179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847270609117166e-05,
+      "loss": 2.0071,
+      "step": 8180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8472328614284718e-05,
+      "loss": 2.0132,
+      "step": 8181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84719510946138e-05,
+      "loss": 2.0176,
+      "step": 8182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8471573532160806e-05,
+      "loss": 2.0389,
+      "step": 8183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847119592692765e-05,
+      "loss": 1.9964,
+      "step": 8184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847081827891624e-05,
+      "loss": 2.0198,
+      "step": 8185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8470440588128475e-05,
+      "loss": 2.0185,
+      "step": 8186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847006285456627e-05,
+      "loss": 2.0111,
+      "step": 8187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846968507823153e-05,
+      "loss": 2.0098,
+      "step": 8188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8469307259126158e-05,
+      "loss": 2.0121,
+      "step": 8189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846892939725207e-05,
+      "loss": 2.0014,
+      "step": 8190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846855149261117e-05,
+      "loss": 1.9927,
+      "step": 8191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8468173545205367e-05,
+      "loss": 2.0366,
+      "step": 8192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8467795555036566e-05,
+      "loss": 2.0094,
+      "step": 8193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8467417522106682e-05,
+      "loss": 2.0166,
+      "step": 8194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846703944641762e-05,
+      "loss": 2.0268,
+      "step": 8195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846666132797129e-05,
+      "loss": 2.0082,
+      "step": 8196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8466283166769603e-05,
+      "loss": 2.0283,
+      "step": 8197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8465904962814467e-05,
+      "loss": 1.993,
+      "step": 8198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8465526716107792e-05,
+      "loss": 2.0195,
+      "step": 8199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8465148426651487e-05,
+      "loss": 2.0062,
+      "step": 8200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8464770094447465e-05,
+      "loss": 2.0074,
+      "step": 8201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8464391719497633e-05,
+      "loss": 1.9986,
+      "step": 8202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8464013301803904e-05,
+      "loss": 2.0125,
+      "step": 8203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846363484136819e-05,
+      "loss": 2.03,
+      "step": 8204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8463256338192405e-05,
+      "loss": 2.0045,
+      "step": 8205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846287779227845e-05,
+      "loss": 1.9822,
+      "step": 8206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8462499203628247e-05,
+      "loss": 2.0128,
+      "step": 8207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84621205722437e-05,
+      "loss": 2.0163,
+      "step": 8208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846174189812673e-05,
+      "loss": 2.0089,
+      "step": 8209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8461363181279238e-05,
+      "loss": 2.0076,
+      "step": 8210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8460984421703147e-05,
+      "loss": 2.031,
+      "step": 8211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8460605619400365e-05,
+      "loss": 2.0053,
+      "step": 8212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8460226774372802e-05,
+      "loss": 1.9998,
+      "step": 8213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8459847886622377e-05,
+      "loss": 1.9997,
+      "step": 8214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8459468956151002e-05,
+      "loss": 2.0226,
+      "step": 8215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8459089982960584e-05,
+      "loss": 2.0364,
+      "step": 8216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8458710967053048e-05,
+      "loss": 2.0296,
+      "step": 8217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8458331908430297e-05,
+      "loss": 2.0328,
+      "step": 8218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845795280709425e-05,
+      "loss": 2.0188,
+      "step": 8219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845757366304682e-05,
+      "loss": 2.0113,
+      "step": 8220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8457194476289927e-05,
+      "loss": 2.009,
+      "step": 8221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8456815246825482e-05,
+      "loss": 2.0367,
+      "step": 8222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84564359746554e-05,
+      "loss": 2.0143,
+      "step": 8223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8456056659781594e-05,
+      "loss": 2.0041,
+      "step": 8224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845567730220598e-05,
+      "loss": 1.9989,
+      "step": 8225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845529790193048e-05,
+      "loss": 2.0051,
+      "step": 8226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8454918458957e-05,
+      "loss": 1.9915,
+      "step": 8227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8454538973287465e-05,
+      "loss": 2.013,
+      "step": 8228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8454159444923786e-05,
+      "loss": 2.0191,
+      "step": 8229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845377987386788e-05,
+      "loss": 2.0131,
+      "step": 8230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845340026012167e-05,
+      "loss": 2.0106,
+      "step": 8231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8453020603687065e-05,
+      "loss": 2.0058,
+      "step": 8232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8452640904565986e-05,
+      "loss": 2.0037,
+      "step": 8233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8452261162760348e-05,
+      "loss": 2.0141,
+      "step": 8234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8451881378272073e-05,
+      "loss": 2.0247,
+      "step": 8235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8451501551103075e-05,
+      "loss": 1.9989,
+      "step": 8236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.845112168125527e-05,
+      "loss": 2.0127,
+      "step": 8237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8450741768730584e-05,
+      "loss": 2.0153,
+      "step": 8238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.845036181353093e-05,
+      "loss": 2.0072,
+      "step": 8239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844998181565823e-05,
+      "loss": 2.0143,
+      "step": 8240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8449601775114395e-05,
+      "loss": 2.0188,
+      "step": 8241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8449221691901354e-05,
+      "loss": 2.0242,
+      "step": 8242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844884156602102e-05,
+      "loss": 2.023,
+      "step": 8243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8448461397475318e-05,
+      "loss": 1.9954,
+      "step": 8244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8448081186266162e-05,
+      "loss": 2.0226,
+      "step": 8245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8447700932395474e-05,
+      "loss": 2.0241,
+      "step": 8246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8447320635865177e-05,
+      "loss": 2.001,
+      "step": 8247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8446940296677187e-05,
+      "loss": 2.0029,
+      "step": 8248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8446559914833427e-05,
+      "loss": 2.0241,
+      "step": 8249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8446179490335822e-05,
+      "loss": 1.9975,
+      "step": 8250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8445799023186283e-05,
+      "loss": 2.018,
+      "step": 8251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8445418513386738e-05,
+      "loss": 2.0085,
+      "step": 8252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844503796093911e-05,
+      "loss": 2.0,
+      "step": 8253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844465736584532e-05,
+      "loss": 2.0219,
+      "step": 8254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8444276728107285e-05,
+      "loss": 2.0047,
+      "step": 8255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8443896047726932e-05,
+      "loss": 1.9982,
+      "step": 8256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844351532470618e-05,
+      "loss": 2.0195,
+      "step": 8257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8443134559046955e-05,
+      "loss": 2.0172,
+      "step": 8258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844275375075118e-05,
+      "loss": 2.0186,
+      "step": 8259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8442372899820774e-05,
+      "loss": 2.0144,
+      "step": 8260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8441992006257665e-05,
+      "loss": 2.01,
+      "step": 8261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844161107006377e-05,
+      "loss": 2.0121,
+      "step": 8262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844123009124102e-05,
+      "loss": 2.0101,
+      "step": 8263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8440849069791333e-05,
+      "loss": 1.9914,
+      "step": 8264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844046800571664e-05,
+      "loss": 1.9874,
+      "step": 8265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8440086899018857e-05,
+      "loss": 1.9962,
+      "step": 8266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8439705749699913e-05,
+      "loss": 2.0105,
+      "step": 8267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8439324557761733e-05,
+      "loss": 2.0136,
+      "step": 8268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8438943323206243e-05,
+      "loss": 1.9991,
+      "step": 8269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8438562046035362e-05,
+      "loss": 1.9822,
+      "step": 8270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8438180726251024e-05,
+      "loss": 2.0036,
+      "step": 8271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8437799363855148e-05,
+      "loss": 2.0102,
+      "step": 8272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8437417958849662e-05,
+      "loss": 2.0097,
+      "step": 8273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8437036511236495e-05,
+      "loss": 2.0078,
+      "step": 8274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843665502101757e-05,
+      "loss": 2.0151,
+      "step": 8275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8436273488194813e-05,
+      "loss": 1.9961,
+      "step": 8276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843589191277015e-05,
+      "loss": 2.0107,
+      "step": 8277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843551029474551e-05,
+      "loss": 2.0084,
+      "step": 8278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843512863412282e-05,
+      "loss": 2.0064,
+      "step": 8279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843474693090401e-05,
+      "loss": 2.0055,
+      "step": 8280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8434365185091e-05,
+      "loss": 2.0195,
+      "step": 8281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8433983396685723e-05,
+      "loss": 2.024,
+      "step": 8282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843360156569011e-05,
+      "loss": 2.0011,
+      "step": 8283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8433219692106085e-05,
+      "loss": 2.023,
+      "step": 8284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8432837775935572e-05,
+      "loss": 2.0144,
+      "step": 8285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8432455817180508e-05,
+      "loss": 2.0081,
+      "step": 8286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8432073815842818e-05,
+      "loss": 2.0019,
+      "step": 8287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8431691771924432e-05,
+      "loss": 2.0384,
+      "step": 8288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8431309685427276e-05,
+      "loss": 2.0174,
+      "step": 8289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8430927556353287e-05,
+      "loss": 1.9972,
+      "step": 8290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8430545384704386e-05,
+      "loss": 2.0333,
+      "step": 8291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8430163170482505e-05,
+      "loss": 2.0001,
+      "step": 8292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8429780913689578e-05,
+      "loss": 2.0136,
+      "step": 8293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8429398614327536e-05,
+      "loss": 2.0223,
+      "step": 8294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8429016272398304e-05,
+      "loss": 2.0308,
+      "step": 8295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8428633887903814e-05,
+      "loss": 1.996,
+      "step": 8296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8428251460846002e-05,
+      "loss": 2.0079,
+      "step": 8297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8427868991226794e-05,
+      "loss": 2.0293,
+      "step": 8298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842748647904812e-05,
+      "loss": 2.0145,
+      "step": 8299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842710392431192e-05,
+      "loss": 2.0064,
+      "step": 8300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8426721327020117e-05,
+      "loss": 2.01,
+      "step": 8301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842633868717465e-05,
+      "loss": 2.0294,
+      "step": 8302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8425956004777444e-05,
+      "loss": 2.034,
+      "step": 8303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8425573279830437e-05,
+      "loss": 2.0136,
+      "step": 8304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8425190512335563e-05,
+      "loss": 2.0212,
+      "step": 8305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842480770229475e-05,
+      "loss": 2.0195,
+      "step": 8306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842442484970993e-05,
+      "loss": 1.9942,
+      "step": 8307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8424041954583042e-05,
+      "loss": 2.0182,
+      "step": 8308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842365901691602e-05,
+      "loss": 2.0198,
+      "step": 8309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8423276036710792e-05,
+      "loss": 2.0122,
+      "step": 8310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8422893013969295e-05,
+      "loss": 2.0057,
+      "step": 8311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8422509948693462e-05,
+      "loss": 2.0276,
+      "step": 8312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842212684088523e-05,
+      "loss": 2.0056,
+      "step": 8313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842174369054653e-05,
+      "loss": 1.9976,
+      "step": 8314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84213604976793e-05,
+      "loss": 1.9923,
+      "step": 8315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8420977262285476e-05,
+      "loss": 2.0146,
+      "step": 8316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842059398436699e-05,
+      "loss": 2.0214,
+      "step": 8317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8420210663925777e-05,
+      "loss": 2.0173,
+      "step": 8318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8419827300963776e-05,
+      "loss": 2.0176,
+      "step": 8319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841944389548292e-05,
+      "loss": 2.0057,
+      "step": 8320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841906044748515e-05,
+      "loss": 1.9997,
+      "step": 8321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8418676956972397e-05,
+      "loss": 2.0256,
+      "step": 8322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8418293423946597e-05,
+      "loss": 2.0212,
+      "step": 8323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841790984840969e-05,
+      "loss": 2.0032,
+      "step": 8324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8417526230363616e-05,
+      "loss": 2.0268,
+      "step": 8325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8417142569810304e-05,
+      "loss": 2.0198,
+      "step": 8326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84167588667517e-05,
+      "loss": 2.0277,
+      "step": 8327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8416375121189737e-05,
+      "loss": 2.0104,
+      "step": 8328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8415991333126352e-05,
+      "loss": 2.0114,
+      "step": 8329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8415607502563485e-05,
+      "loss": 2.0078,
+      "step": 8330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841522362950307e-05,
+      "loss": 2.0165,
+      "step": 8331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8414839713947055e-05,
+      "loss": 2.0304,
+      "step": 8332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841445575589737e-05,
+      "loss": 2.014,
+      "step": 8333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8414071755355957e-05,
+      "loss": 2.0062,
+      "step": 8334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8413687712324755e-05,
+      "loss": 1.9945,
+      "step": 8335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8413303626805706e-05,
+      "loss": 2.0257,
+      "step": 8336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8412919498800745e-05,
+      "loss": 2.0196,
+      "step": 8337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 2.0156,
+      "step": 8338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841215111534085e-05,
+      "loss": 2.0049,
+      "step": 8339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84117668598898e-05,
+      "loss": 2.0283,
+      "step": 8340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84113825619606e-05,
+      "loss": 2.0088,
+      "step": 8341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841099822155519e-05,
+      "loss": 2.002,
+      "step": 8342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8410613838675512e-05,
+      "loss": 2.0026,
+      "step": 8343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8410229413323508e-05,
+      "loss": 2.0359,
+      "step": 8344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8409844945501115e-05,
+      "loss": 2.004,
+      "step": 8345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840946043521028e-05,
+      "loss": 2.002,
+      "step": 8346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8409075882452946e-05,
+      "loss": 1.9875,
+      "step": 8347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8408691287231045e-05,
+      "loss": 1.9985,
+      "step": 8348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840830664954653e-05,
+      "loss": 2.0238,
+      "step": 8349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8407921969401335e-05,
+      "loss": 2.019,
+      "step": 8350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8407537246797412e-05,
+      "loss": 2.0029,
+      "step": 8351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8407152481736695e-05,
+      "loss": 1.9991,
+      "step": 8352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840676767422113e-05,
+      "loss": 2.0086,
+      "step": 8353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840638282425266e-05,
+      "loss": 2.011,
+      "step": 8354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8405997931833232e-05,
+      "loss": 2.0034,
+      "step": 8355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8405612996964787e-05,
+      "loss": 2.0065,
+      "step": 8356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8405228019649265e-05,
+      "loss": 2.002,
+      "step": 8357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8404842999888613e-05,
+      "loss": 2.02,
+      "step": 8358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840445793768478e-05,
+      "loss": 2.005,
+      "step": 8359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8404072833039704e-05,
+      "loss": 2.0128,
+      "step": 8360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8403687685955334e-05,
+      "loss": 2.0326,
+      "step": 8361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840330249643361e-05,
+      "loss": 2.0229,
+      "step": 8362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840291726447648e-05,
+      "loss": 2.0132,
+      "step": 8363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8402531990085896e-05,
+      "loss": 2.0285,
+      "step": 8364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840214667326379e-05,
+      "loss": 2.0048,
+      "step": 8365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840176131401212e-05,
+      "loss": 1.9987,
+      "step": 8366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8401375912332823e-05,
+      "loss": 2.0391,
+      "step": 8367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8400990468227848e-05,
+      "loss": 2.0217,
+      "step": 8368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8400604981699147e-05,
+      "loss": 1.9971,
+      "step": 8369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8400219452748657e-05,
+      "loss": 2.0046,
+      "step": 8370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8399833881378334e-05,
+      "loss": 2.0106,
+      "step": 8371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8399448267590122e-05,
+      "loss": 1.994,
+      "step": 8372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8399062611385966e-05,
+      "loss": 1.9954,
+      "step": 8373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839867691276781e-05,
+      "loss": 2.0247,
+      "step": 8374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8398291171737616e-05,
+      "loss": 2.008,
+      "step": 8375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839790538829732e-05,
+      "loss": 2.0187,
+      "step": 8376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8397519562448867e-05,
+      "loss": 2.0084,
+      "step": 8377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8397133694194215e-05,
+      "loss": 2.0216,
+      "step": 8378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839674778353531e-05,
+      "loss": 2.0064,
+      "step": 8379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83963618304741e-05,
+      "loss": 2.0052,
+      "step": 8380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8395975835012533e-05,
+      "loss": 2.0133,
+      "step": 8381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8395589797152556e-05,
+      "loss": 2.0159,
+      "step": 8382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839520371689612e-05,
+      "loss": 2.0285,
+      "step": 8383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839481759424518e-05,
+      "loss": 2.029,
+      "step": 8384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839443142920168e-05,
+      "loss": 2.0186,
+      "step": 8385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839404522176757e-05,
+      "loss": 2.027,
+      "step": 8386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8393658971944806e-05,
+      "loss": 1.9905,
+      "step": 8387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8393272679735333e-05,
+      "loss": 2.0177,
+      "step": 8388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83928863451411e-05,
+      "loss": 2.0154,
+      "step": 8389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8392499968164067e-05,
+      "loss": 2.0142,
+      "step": 8390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839211354880617e-05,
+      "loss": 2.0112,
+      "step": 8391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839172708706938e-05,
+      "loss": 1.9997,
+      "step": 8392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8391340582955634e-05,
+      "loss": 2.012,
+      "step": 8393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8390954036466887e-05,
+      "loss": 1.9975,
+      "step": 8394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839056744760509e-05,
+      "loss": 2.0184,
+      "step": 8395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8390180816372202e-05,
+      "loss": 1.9987,
+      "step": 8396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838979414277017e-05,
+      "loss": 1.9931,
+      "step": 8397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8389407426800945e-05,
+      "loss": 2.0107,
+      "step": 8398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8389020668466484e-05,
+      "loss": 2.019,
+      "step": 8399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8388633867768736e-05,
+      "loss": 2.0151,
+      "step": 8400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838824702470966e-05,
+      "loss": 1.995,
+      "step": 8401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8387860139291205e-05,
+      "loss": 2.0295,
+      "step": 8402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8387473211515324e-05,
+      "loss": 2.0204,
+      "step": 8403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8387086241383977e-05,
+      "loss": 2.0112,
+      "step": 8404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838669922889911e-05,
+      "loss": 1.9941,
+      "step": 8405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838631217406268e-05,
+      "loss": 1.9975,
+      "step": 8406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8385925076876644e-05,
+      "loss": 2.0157,
+      "step": 8407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8385537937342956e-05,
+      "loss": 2.0085,
+      "step": 8408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838515075546357e-05,
+      "loss": 2.0287,
+      "step": 8409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8384763531240443e-05,
+      "loss": 2.0199,
+      "step": 8410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838437626467553e-05,
+      "loss": 2.0121,
+      "step": 8411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838398895577078e-05,
+      "loss": 2.0154,
+      "step": 8412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838360160452816e-05,
+      "loss": 1.9993,
+      "step": 8413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838321421094962e-05,
+      "loss": 2.0239,
+      "step": 8414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8382826775037116e-05,
+      "loss": 2.0204,
+      "step": 8415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838243929679261e-05,
+      "loss": 2.0175,
+      "step": 8416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8382051776218046e-05,
+      "loss": 2.0061,
+      "step": 8417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8381664213315394e-05,
+      "loss": 1.9925,
+      "step": 8418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8381276608086604e-05,
+      "loss": 2.0265,
+      "step": 8419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8380888960533637e-05,
+      "loss": 2.0299,
+      "step": 8420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8380501270658452e-05,
+      "loss": 2.0072,
+      "step": 8421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8380113538462998e-05,
+      "loss": 2.0166,
+      "step": 8422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8379725763949242e-05,
+      "loss": 2.02,
+      "step": 8423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837933794711914e-05,
+      "loss": 1.9977,
+      "step": 8424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837895008797465e-05,
+      "loss": 2.0082,
+      "step": 8425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837856218651773e-05,
+      "loss": 2.0103,
+      "step": 8426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8378174242750337e-05,
+      "loss": 2.0151,
+      "step": 8427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837778625667443e-05,
+      "loss": 2.0,
+      "step": 8428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8377398228291973e-05,
+      "loss": 2.0068,
+      "step": 8429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8377010157604925e-05,
+      "loss": 2.0338,
+      "step": 8430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837662204461524e-05,
+      "loss": 2.0041,
+      "step": 8431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8376233889324884e-05,
+      "loss": 2.011,
+      "step": 8432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8375845691735813e-05,
+      "loss": 2.0055,
+      "step": 8433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8375457451849993e-05,
+      "loss": 2.0193,
+      "step": 8434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8375069169669374e-05,
+      "loss": 2.0069,
+      "step": 8435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8374680845195927e-05,
+      "loss": 2.0175,
+      "step": 8436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8374292478431604e-05,
+      "loss": 2.011,
+      "step": 8437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837390406937838e-05,
+      "loss": 2.0338,
+      "step": 8438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8373515618038204e-05,
+      "loss": 1.9915,
+      "step": 8439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837312712441304e-05,
+      "loss": 2.0328,
+      "step": 8440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8372738588504853e-05,
+      "loss": 2.0055,
+      "step": 8441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83723500103156e-05,
+      "loss": 2.0199,
+      "step": 8442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837196138984725e-05,
+      "loss": 2.0285,
+      "step": 8443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837157272710176e-05,
+      "loss": 2.0245,
+      "step": 8444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8371184022081096e-05,
+      "loss": 2.0213,
+      "step": 8445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837079527478722e-05,
+      "loss": 1.9913,
+      "step": 8446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8370406485222095e-05,
+      "loss": 2.0141,
+      "step": 8447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837001765338768e-05,
+      "loss": 2.0201,
+      "step": 8448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8369628779285945e-05,
+      "loss": 1.9865,
+      "step": 8449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8369239862918854e-05,
+      "loss": 1.9826,
+      "step": 8450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8368850904288366e-05,
+      "loss": 2.0142,
+      "step": 8451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8368461903396446e-05,
+      "loss": 2.0188,
+      "step": 8452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8368072860245063e-05,
+      "loss": 2.0174,
+      "step": 8453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836768377483618e-05,
+      "loss": 2.0033,
+      "step": 8454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8367294647171755e-05,
+      "loss": 2.0171,
+      "step": 8455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8366905477253764e-05,
+      "loss": 2.0122,
+      "step": 8456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8366516265084163e-05,
+      "loss": 2.0115,
+      "step": 8457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836612701066492e-05,
+      "loss": 2.0047,
+      "step": 8458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8365737713998e-05,
+      "loss": 1.9889,
+      "step": 8459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8365348375085376e-05,
+      "loss": 2.0107,
+      "step": 8460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836495899392901e-05,
+      "loss": 2.0301,
+      "step": 8461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8364569570530865e-05,
+      "loss": 2.0157,
+      "step": 8462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8364180104892906e-05,
+      "loss": 2.0171,
+      "step": 8463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8363790597017103e-05,
+      "loss": 2.0075,
+      "step": 8464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836340104690543e-05,
+      "loss": 2.0054,
+      "step": 8465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8363011454559842e-05,
+      "loss": 1.9954,
+      "step": 8466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8362621819982313e-05,
+      "loss": 1.9944,
+      "step": 8467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8362232143174812e-05,
+      "loss": 2.0155,
+      "step": 8468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83618424241393e-05,
+      "loss": 2.0068,
+      "step": 8469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836145266287775e-05,
+      "loss": 1.9941,
+      "step": 8470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8361062859392133e-05,
+      "loss": 2.0136,
+      "step": 8471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836067301368441e-05,
+      "loss": 2.0014,
+      "step": 8472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8360283125756557e-05,
+      "loss": 2.0143,
+      "step": 8473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8359893195610537e-05,
+      "loss": 1.9898,
+      "step": 8474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8359503223248324e-05,
+      "loss": 2.0018,
+      "step": 8475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8359113208671883e-05,
+      "loss": 2.0212,
+      "step": 8476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8358723151883186e-05,
+      "loss": 2.027,
+      "step": 8477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83583330528842e-05,
+      "loss": 2.0033,
+      "step": 8478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83579429116769e-05,
+      "loss": 2.006,
+      "step": 8479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8357552728263255e-05,
+      "loss": 1.9983,
+      "step": 8480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8357162502645228e-05,
+      "loss": 2.0242,
+      "step": 8481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8356772234824798e-05,
+      "loss": 2.0124,
+      "step": 8482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8356381924803935e-05,
+      "loss": 2.0123,
+      "step": 8483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8355991572584603e-05,
+      "loss": 2.0235,
+      "step": 8484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8355601178168783e-05,
+      "loss": 2.0074,
+      "step": 8485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835521074155844e-05,
+      "loss": 2.0149,
+      "step": 8486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8354820262755547e-05,
+      "loss": 1.9887,
+      "step": 8487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8354429741762074e-05,
+      "loss": 2.0192,
+      "step": 8488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8354039178579998e-05,
+      "loss": 2.032,
+      "step": 8489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835364857321129e-05,
+      "loss": 2.01,
+      "step": 8490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8353257925657918e-05,
+      "loss": 1.9987,
+      "step": 8491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835286723592186e-05,
+      "loss": 2.0064,
+      "step": 8492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8352476504005087e-05,
+      "loss": 2.0093,
+      "step": 8493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8352085729909567e-05,
+      "loss": 2.0259,
+      "step": 8494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8351694913637282e-05,
+      "loss": 2.0079,
+      "step": 8495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8351304055190196e-05,
+      "loss": 2.009,
+      "step": 8496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83509131545703e-05,
+      "loss": 2.0117,
+      "step": 8497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8350522211779545e-05,
+      "loss": 2.0136,
+      "step": 8498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8350131226819923e-05,
+      "loss": 2.0094,
+      "step": 8499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83497401996934e-05,
+      "loss": 2.012,
+      "step": 8500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8349349130401953e-05,
+      "loss": 2.0071,
+      "step": 8501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8348958018947556e-05,
+      "loss": 1.9786,
+      "step": 8502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8348566865332186e-05,
+      "loss": 2.0136,
+      "step": 8503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8348175669557813e-05,
+      "loss": 2.0205,
+      "step": 8504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834778443162642e-05,
+      "loss": 2.0064,
+      "step": 8505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834739315153998e-05,
+      "loss": 2.0083,
+      "step": 8506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8347001829300467e-05,
+      "loss": 2.0076,
+      "step": 8507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8346610464909854e-05,
+      "loss": 2.013,
+      "step": 8508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8346219058370124e-05,
+      "loss": 2.0137,
+      "step": 8509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834582760968325e-05,
+      "loss": 1.9916,
+      "step": 8510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8345436118851215e-05,
+      "loss": 2.0327,
+      "step": 8511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8345044585875988e-05,
+      "loss": 2.0079,
+      "step": 8512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8344653010759543e-05,
+      "loss": 1.9884,
+      "step": 8513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8344261393503867e-05,
+      "loss": 1.9957,
+      "step": 8514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8343869734110934e-05,
+      "loss": 2.0178,
+      "step": 8515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8343478032582723e-05,
+      "loss": 2.0173,
+      "step": 8516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8343086288921213e-05,
+      "loss": 1.9963,
+      "step": 8517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8342694503128375e-05,
+      "loss": 1.9945,
+      "step": 8518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8342302675206193e-05,
+      "loss": 2.0118,
+      "step": 8519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8341910805156647e-05,
+      "loss": 2.0026,
+      "step": 8520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8341518892981713e-05,
+      "loss": 2.0005,
+      "step": 8521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834112693868337e-05,
+      "loss": 2.0169,
+      "step": 8522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83407349422636e-05,
+      "loss": 2.0185,
+      "step": 8523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834034290372438e-05,
+      "loss": 2.0102,
+      "step": 8524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8339950823067692e-05,
+      "loss": 2.0162,
+      "step": 8525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8339558700295512e-05,
+      "loss": 2.007,
+      "step": 8526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8339166535409823e-05,
+      "loss": 2.008,
+      "step": 8527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833877432841261e-05,
+      "loss": 1.9949,
+      "step": 8528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8338382079305843e-05,
+      "loss": 2.0213,
+      "step": 8529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8337989788091512e-05,
+      "loss": 1.9861,
+      "step": 8530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833759745477159e-05,
+      "loss": 2.0085,
+      "step": 8531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833720507934807e-05,
+      "loss": 2.0382,
+      "step": 8532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833681266182292e-05,
+      "loss": 2.0296,
+      "step": 8533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833642020219813e-05,
+      "loss": 2.0076,
+      "step": 8534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8336027700475677e-05,
+      "loss": 2.0012,
+      "step": 8535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8335635156657548e-05,
+      "loss": 2.0141,
+      "step": 8536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8335242570745723e-05,
+      "loss": 2.0225,
+      "step": 8537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8334849942742185e-05,
+      "loss": 1.9964,
+      "step": 8538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8334457272648912e-05,
+      "loss": 2.0063,
+      "step": 8539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8334064560467897e-05,
+      "loss": 2.001,
+      "step": 8540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8333671806201115e-05,
+      "loss": 2.0282,
+      "step": 8541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833327900985055e-05,
+      "loss": 2.0241,
+      "step": 8542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8332886171418188e-05,
+      "loss": 2.0137,
+      "step": 8543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8332493290906014e-05,
+      "loss": 2.0152,
+      "step": 8544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8332100368316008e-05,
+      "loss": 2.0161,
+      "step": 8545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8331707403650158e-05,
+      "loss": 2.0236,
+      "step": 8546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8331314396910444e-05,
+      "loss": 2.0093,
+      "step": 8547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8330921348098857e-05,
+      "loss": 2.0155,
+      "step": 8548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8330528257217374e-05,
+      "loss": 2.022,
+      "step": 8549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833013512426799e-05,
+      "loss": 2.001,
+      "step": 8550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8329741949252677e-05,
+      "loss": 2.0114,
+      "step": 8551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832934873217343e-05,
+      "loss": 2.0171,
+      "step": 8552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832895547303223e-05,
+      "loss": 1.9976,
+      "step": 8553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832856217183107e-05,
+      "loss": 2.004,
+      "step": 8554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832816882857193e-05,
+      "loss": 2.0186,
+      "step": 8555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83277754432568e-05,
+      "loss": 2.0225,
+      "step": 8556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8327382015887662e-05,
+      "loss": 2.0407,
+      "step": 8557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8326988546466507e-05,
+      "loss": 1.9984,
+      "step": 8558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8326595034995317e-05,
+      "loss": 2.0047,
+      "step": 8559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8326201481476086e-05,
+      "loss": 2.0375,
+      "step": 8560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8325807885910793e-05,
+      "loss": 2.016,
+      "step": 8561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8325414248301434e-05,
+      "loss": 2.0131,
+      "step": 8562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8325020568649992e-05,
+      "loss": 2.0253,
+      "step": 8563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8324626846958454e-05,
+      "loss": 1.9991,
+      "step": 8564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8324233083228814e-05,
+      "loss": 1.997,
+      "step": 8565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8323839277463052e-05,
+      "loss": 1.9955,
+      "step": 8566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832344542966317e-05,
+      "loss": 2.0087,
+      "step": 8567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832305153983114e-05,
+      "loss": 2.0043,
+      "step": 8568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8322657607968962e-05,
+      "loss": 2.0093,
+      "step": 8569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8322263634078623e-05,
+      "loss": 2.011,
+      "step": 8570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832186961816211e-05,
+      "loss": 2.0017,
+      "step": 8571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8321475560221417e-05,
+      "loss": 2.0162,
+      "step": 8572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832108146025853e-05,
+      "loss": 1.9826,
+      "step": 8573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8320687318275446e-05,
+      "loss": 2.0109,
+      "step": 8574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832029313427415e-05,
+      "loss": 2.0176,
+      "step": 8575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8319898908256626e-05,
+      "loss": 2.0126,
+      "step": 8576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8319504640224877e-05,
+      "loss": 2.0192,
+      "step": 8577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8319110330180886e-05,
+      "loss": 1.9837,
+      "step": 8578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831871597812665e-05,
+      "loss": 2.0034,
+      "step": 8579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831832158406415e-05,
+      "loss": 2.018,
+      "step": 8580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8317927147995392e-05,
+      "loss": 2.0227,
+      "step": 8581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831753266992236e-05,
+      "loss": 2.0133,
+      "step": 8582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8317138149847045e-05,
+      "loss": 1.994,
+      "step": 8583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831674358777144e-05,
+      "loss": 2.0246,
+      "step": 8584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8316348983697543e-05,
+      "loss": 2.0223,
+      "step": 8585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831595433762734e-05,
+      "loss": 2.0142,
+      "step": 8586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8315559649562827e-05,
+      "loss": 2.0194,
+      "step": 8587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8315164919505993e-05,
+      "loss": 2.0355,
+      "step": 8588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831477014745884e-05,
+      "loss": 2.0103,
+      "step": 8589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831437533342335e-05,
+      "loss": 2.0439,
+      "step": 8590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8313980477401528e-05,
+      "loss": 2.0115,
+      "step": 8591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831358557939536e-05,
+      "loss": 2.0219,
+      "step": 8592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8313190639406844e-05,
+      "loss": 2.0211,
+      "step": 8593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8312795657437973e-05,
+      "loss": 2.0074,
+      "step": 8594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831240063349074e-05,
+      "loss": 1.9863,
+      "step": 8595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8312005567567146e-05,
+      "loss": 2.0119,
+      "step": 8596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8311610459669178e-05,
+      "loss": 2.0006,
+      "step": 8597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8311215309798837e-05,
+      "loss": 2.015,
+      "step": 8598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831082011795812e-05,
+      "loss": 2.02,
+      "step": 8599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8310424884149015e-05,
+      "loss": 2.0154,
+      "step": 8600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8310029608373522e-05,
+      "loss": 2.0138,
+      "step": 8601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8309634290633638e-05,
+      "loss": 2.0233,
+      "step": 8602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830923893093136e-05,
+      "loss": 2.023,
+      "step": 8603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8308843529268677e-05,
+      "loss": 2.0017,
+      "step": 8604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8308448085647598e-05,
+      "loss": 2.0114,
+      "step": 8605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830805260007011e-05,
+      "loss": 1.9979,
+      "step": 8606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8307657072538218e-05,
+      "loss": 2.0125,
+      "step": 8607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830726150305391e-05,
+      "loss": 2.0156,
+      "step": 8608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8306865891619193e-05,
+      "loss": 2.0181,
+      "step": 8609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830647023823606e-05,
+      "loss": 1.9968,
+      "step": 8610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8306074542906506e-05,
+      "loss": 2.0137,
+      "step": 8611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8305678805632535e-05,
+      "loss": 2.0185,
+      "step": 8612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830528302641614e-05,
+      "loss": 2.0056,
+      "step": 8613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830488720525933e-05,
+      "loss": 2.0053,
+      "step": 8614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830449134216409e-05,
+      "loss": 2.0098,
+      "step": 8615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8304095437132426e-05,
+      "loss": 1.9954,
+      "step": 8616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830369949016634e-05,
+      "loss": 2.0101,
+      "step": 8617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8303303501267823e-05,
+      "loss": 2.0165,
+      "step": 8618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8302907470438883e-05,
+      "loss": 2.0277,
+      "step": 8619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8302511397681518e-05,
+      "loss": 2.0135,
+      "step": 8620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830211528299773e-05,
+      "loss": 2.009,
+      "step": 8621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8301719126389513e-05,
+      "loss": 1.9909,
+      "step": 8622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8301322927858868e-05,
+      "loss": 2.0118,
+      "step": 8623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8300926687407803e-05,
+      "loss": 1.9986,
+      "step": 8624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8300530405038312e-05,
+      "loss": 2.0117,
+      "step": 8625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83001340807524e-05,
+      "loss": 2.0272,
+      "step": 8626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8299737714552063e-05,
+      "loss": 2.0258,
+      "step": 8627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829934130643931e-05,
+      "loss": 2.0281,
+      "step": 8628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829894485641614e-05,
+      "loss": 2.033,
+      "step": 8629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8298548364484558e-05,
+      "loss": 2.023,
+      "step": 8630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8298151830646556e-05,
+      "loss": 2.0043,
+      "step": 8631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8297755254904147e-05,
+      "loss": 2.0114,
+      "step": 8632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8297358637259326e-05,
+      "loss": 2.0081,
+      "step": 8633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8296961977714103e-05,
+      "loss": 2.0098,
+      "step": 8634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8296565276270476e-05,
+      "loss": 2.0109,
+      "step": 8635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829616853293045e-05,
+      "loss": 2.0093,
+      "step": 8636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829577174769603e-05,
+      "loss": 2.0132,
+      "step": 8637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8295374920569217e-05,
+      "loss": 2.0095,
+      "step": 8638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8294978051552018e-05,
+      "loss": 2.0373,
+      "step": 8639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8294581140646433e-05,
+      "loss": 2.018,
+      "step": 8640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829418418785447e-05,
+      "loss": 2.0118,
+      "step": 8641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829378719317813e-05,
+      "loss": 2.0125,
+      "step": 8642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829339015661942e-05,
+      "loss": 2.0166,
+      "step": 8643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829299307818035e-05,
+      "loss": 2.0209,
+      "step": 8644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8292595957862917e-05,
+      "loss": 2.0056,
+      "step": 8645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8292198795669125e-05,
+      "loss": 2.0076,
+      "step": 8646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829180159160099e-05,
+      "loss": 1.9918,
+      "step": 8647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8291404345660512e-05,
+      "loss": 2.001,
+      "step": 8648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829100705784969e-05,
+      "loss": 2.0118,
+      "step": 8649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8290609728170544e-05,
+      "loss": 2.0032,
+      "step": 8650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829021235662507e-05,
+      "loss": 2.0131,
+      "step": 8651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828981494321528e-05,
+      "loss": 2.0114,
+      "step": 8652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828941748794318e-05,
+      "loss": 2.0153,
+      "step": 8653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8289019990810775e-05,
+      "loss": 2.0014,
+      "step": 8654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8288622451820073e-05,
+      "loss": 2.025,
+      "step": 8655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828822487097308e-05,
+      "loss": 2.0198,
+      "step": 8656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828782724827181e-05,
+      "loss": 2.0023,
+      "step": 8657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8287429583718264e-05,
+      "loss": 2.0052,
+      "step": 8658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8287031877314457e-05,
+      "loss": 1.9991,
+      "step": 8659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8286634129062388e-05,
+      "loss": 2.0192,
+      "step": 8660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8286236338964072e-05,
+      "loss": 1.9902,
+      "step": 8661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828583850702152e-05,
+      "loss": 1.9854,
+      "step": 8662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8285440633236734e-05,
+      "loss": 1.9811,
+      "step": 8663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828504271761173e-05,
+      "loss": 2.0078,
+      "step": 8664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828464476014851e-05,
+      "loss": 1.9948,
+      "step": 8665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8284246760849093e-05,
+      "loss": 2.0052,
+      "step": 8666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8283848719715476e-05,
+      "loss": 2.0054,
+      "step": 8667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828345063674968e-05,
+      "loss": 2.0117,
+      "step": 8668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8283052511953716e-05,
+      "loss": 2.0012,
+      "step": 8669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8282654345329587e-05,
+      "loss": 2.0059,
+      "step": 8670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8282256136879307e-05,
+      "loss": 2.0138,
+      "step": 8671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828185788660489e-05,
+      "loss": 2.022,
+      "step": 8672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8281459594508342e-05,
+      "loss": 2.0113,
+      "step": 8673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8281061260591675e-05,
+      "loss": 2.0075,
+      "step": 8674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8280662884856904e-05,
+      "loss": 2.0151,
+      "step": 8675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8280264467306037e-05,
+      "loss": 2.0017,
+      "step": 8676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8279866007941088e-05,
+      "loss": 2.0152,
+      "step": 8677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8279467506764067e-05,
+      "loss": 2.0219,
+      "step": 8678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8279068963776993e-05,
+      "loss": 2.0298,
+      "step": 8679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827867037898187e-05,
+      "loss": 1.9989,
+      "step": 8680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8278271752380713e-05,
+      "loss": 2.0321,
+      "step": 8681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827787308397554e-05,
+      "loss": 1.9869,
+      "step": 8682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8277474373768356e-05,
+      "loss": 1.997,
+      "step": 8683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8277075621761187e-05,
+      "loss": 2.0217,
+      "step": 8684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8276676827956032e-05,
+      "loss": 2.019,
+      "step": 8685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8276277992354917e-05,
+      "loss": 1.9934,
+      "step": 8686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8275879114959845e-05,
+      "loss": 2.012,
+      "step": 8687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827548019577284e-05,
+      "loss": 2.0031,
+      "step": 8688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8275081234795912e-05,
+      "loss": 1.9915,
+      "step": 8689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8274682232031073e-05,
+      "loss": 2.0135,
+      "step": 8690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8274283187480347e-05,
+      "loss": 2.0052,
+      "step": 8691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827388410114574e-05,
+      "loss": 2.0217,
+      "step": 8692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8273484973029268e-05,
+      "loss": 2.0239,
+      "step": 8693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8273085803132953e-05,
+      "loss": 2.0133,
+      "step": 8694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8272686591458804e-05,
+      "loss": 1.9976,
+      "step": 8695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8272287338008842e-05,
+      "loss": 2.0029,
+      "step": 8696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827188804278508e-05,
+      "loss": 2.0144,
+      "step": 8697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8271488705789532e-05,
+      "loss": 2.0127,
+      "step": 8698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827108932702422e-05,
+      "loss": 2.0248,
+      "step": 8699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827068990649116e-05,
+      "loss": 2.0166,
+      "step": 8700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8270290444192368e-05,
+      "loss": 1.998,
+      "step": 8701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826989094012986e-05,
+      "loss": 2.0199,
+      "step": 8702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8269491394305655e-05,
+      "loss": 2.0275,
+      "step": 8703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826909180672177e-05,
+      "loss": 2.0016,
+      "step": 8704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826869217738022e-05,
+      "loss": 2.0187,
+      "step": 8705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8268292506283028e-05,
+      "loss": 1.9938,
+      "step": 8706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826789279343221e-05,
+      "loss": 1.9843,
+      "step": 8707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8267493038829788e-05,
+      "loss": 2.0189,
+      "step": 8708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8267093242477775e-05,
+      "loss": 2.0114,
+      "step": 8709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8266693404378188e-05,
+      "loss": 2.0084,
+      "step": 8710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8266293524533056e-05,
+      "loss": 1.9977,
+      "step": 8711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8265893602944393e-05,
+      "loss": 1.9981,
+      "step": 8712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8265493639614217e-05,
+      "loss": 2.01,
+      "step": 8713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8265093634544544e-05,
+      "loss": 1.9941,
+      "step": 8714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8264693587737404e-05,
+      "loss": 2.0152,
+      "step": 8715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8264293499194813e-05,
+      "loss": 2.0176,
+      "step": 8716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8263893368918787e-05,
+      "loss": 2.0034,
+      "step": 8717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8263493196911355e-05,
+      "loss": 2.0073,
+      "step": 8718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826309298317453e-05,
+      "loss": 2.0161,
+      "step": 8719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8262692727710334e-05,
+      "loss": 2.0187,
+      "step": 8720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8262292430520793e-05,
+      "loss": 2.0057,
+      "step": 8721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8261892091607928e-05,
+      "loss": 2.0222,
+      "step": 8722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8261491710973754e-05,
+      "loss": 1.9984,
+      "step": 8723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.82610912886203e-05,
+      "loss": 2.0015,
+      "step": 8724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826069082454958e-05,
+      "loss": 2.0094,
+      "step": 8725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8260290318763626e-05,
+      "loss": 1.9966,
+      "step": 8726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8259889771264455e-05,
+      "loss": 2.0214,
+      "step": 8727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8259489182054093e-05,
+      "loss": 2.0154,
+      "step": 8728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8259088551134557e-05,
+      "loss": 2.0006,
+      "step": 8729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8258687878507878e-05,
+      "loss": 2.0068,
+      "step": 8730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825828716417607e-05,
+      "loss": 2.0005,
+      "step": 8731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8257886408141167e-05,
+      "loss": 2.0098,
+      "step": 8732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8257485610405184e-05,
+      "loss": 2.0233,
+      "step": 8733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825708477097015e-05,
+      "loss": 2.0333,
+      "step": 8734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8256683889838087e-05,
+      "loss": 2.0051,
+      "step": 8735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8256282967011018e-05,
+      "loss": 2.0051,
+      "step": 8736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825588200249097e-05,
+      "loss": 1.9986,
+      "step": 8737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825548099627997e-05,
+      "loss": 2.0201,
+      "step": 8738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825507994838004e-05,
+      "loss": 2.0172,
+      "step": 8739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8254678858793203e-05,
+      "loss": 2.001,
+      "step": 8740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825427772752149e-05,
+      "loss": 2.0171,
+      "step": 8741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8253876554566924e-05,
+      "loss": 2.0044,
+      "step": 8742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825347533993153e-05,
+      "loss": 2.0127,
+      "step": 8743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825307408361733e-05,
+      "loss": 2.0269,
+      "step": 8744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825267278562636e-05,
+      "loss": 1.9948,
+      "step": 8745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825227144596064e-05,
+      "loss": 1.9837,
+      "step": 8746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.82518700646222e-05,
+      "loss": 2.0107,
+      "step": 8747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825146864161306e-05,
+      "loss": 2.0284,
+      "step": 8748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8251067176935254e-05,
+      "loss": 2.0088,
+      "step": 8749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8250665670590813e-05,
+      "loss": 2.0136,
+      "step": 8750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8250264122581756e-05,
+      "loss": 2.0223,
+      "step": 8751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824986253291011e-05,
+      "loss": 2.0084,
+      "step": 8752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8249460901577908e-05,
+      "loss": 1.9986,
+      "step": 8753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824905922858718e-05,
+      "loss": 2.012,
+      "step": 8754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824865751393995e-05,
+      "loss": 2.0107,
+      "step": 8755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824825575763825e-05,
+      "loss": 2.0234,
+      "step": 8756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8247853959684104e-05,
+      "loss": 2.0144,
+      "step": 8757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8247452120079544e-05,
+      "loss": 2.0059,
+      "step": 8758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8247050238826598e-05,
+      "loss": 2.0112,
+      "step": 8759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8246648315927298e-05,
+      "loss": 2.0144,
+      "step": 8760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824624635138367e-05,
+      "loss": 2.0163,
+      "step": 8761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824584434519775e-05,
+      "loss": 2.0075,
+      "step": 8762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824544229737156e-05,
+      "loss": 1.9932,
+      "step": 8763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824504020790714e-05,
+      "loss": 2.0138,
+      "step": 8764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8244638076806507e-05,
+      "loss": 2.0056,
+      "step": 8765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8244235904071702e-05,
+      "loss": 2.0153,
+      "step": 8766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8243833689704756e-05,
+      "loss": 1.9982,
+      "step": 8767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8243431433707697e-05,
+      "loss": 2.006,
+      "step": 8768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824302913608256e-05,
+      "loss": 2.0121,
+      "step": 8769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8242626796831366e-05,
+      "loss": 1.9913,
+      "step": 8770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8242224415956158e-05,
+      "loss": 1.9993,
+      "step": 8771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8241821993458963e-05,
+      "loss": 2.0026,
+      "step": 8772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8241419529341815e-05,
+      "loss": 2.0097,
+      "step": 8773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8241017023606747e-05,
+      "loss": 2.0071,
+      "step": 8774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824061447625579e-05,
+      "loss": 2.0199,
+      "step": 8775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8240211887290975e-05,
+      "loss": 2.0139,
+      "step": 8776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8239809256714337e-05,
+      "loss": 2.0046,
+      "step": 8777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823940658452791e-05,
+      "loss": 2.0185,
+      "step": 8778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8239003870733726e-05,
+      "loss": 2.0214,
+      "step": 8779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823860111533382e-05,
+      "loss": 2.0013,
+      "step": 8780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823819831833023e-05,
+      "loss": 2.0108,
+      "step": 8781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8237795479724977e-05,
+      "loss": 2.0292,
+      "step": 8782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823739259952011e-05,
+      "loss": 2.0037,
+      "step": 8783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8236989677717653e-05,
+      "loss": 2.0206,
+      "step": 8784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8236586714319648e-05,
+      "loss": 2.0231,
+      "step": 8785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8236183709328125e-05,
+      "loss": 2.0313,
+      "step": 8786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823578066274512e-05,
+      "loss": 2.0126,
+      "step": 8787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823537757457267e-05,
+      "loss": 2.018,
+      "step": 8788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823497444481281e-05,
+      "loss": 1.9955,
+      "step": 8789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8234571273467576e-05,
+      "loss": 2.0159,
+      "step": 8790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8234168060539e-05,
+      "loss": 2.0215,
+      "step": 8791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8233764806029125e-05,
+      "loss": 2.003,
+      "step": 8792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823336150993998e-05,
+      "loss": 2.0222,
+      "step": 8793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8232958172273608e-05,
+      "loss": 2.0111,
+      "step": 8794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8232554793032045e-05,
+      "loss": 2.0014,
+      "step": 8795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8232151372217327e-05,
+      "loss": 2.0118,
+      "step": 8796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8231747909831488e-05,
+      "loss": 1.9983,
+      "step": 8797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8231344405876565e-05,
+      "loss": 2.0119,
+      "step": 8798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.82309408603546e-05,
+      "loss": 2.016,
+      "step": 8799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8230537273267634e-05,
+      "loss": 2.0244,
+      "step": 8800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8230133644617694e-05,
+      "loss": 1.9959,
+      "step": 8801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822972997440683e-05,
+      "loss": 2.0185,
+      "step": 8802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8229326262637073e-05,
+      "loss": 2.0014,
+      "step": 8803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8228922509310463e-05,
+      "loss": 2.0048,
+      "step": 8804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822851871442904e-05,
+      "loss": 2.02,
+      "step": 8805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8228114877994845e-05,
+      "loss": 2.0178,
+      "step": 8806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8227711000009912e-05,
+      "loss": 2.0057,
+      "step": 8807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8227307080476285e-05,
+      "loss": 1.9915,
+      "step": 8808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8226903119396004e-05,
+      "loss": 2.0058,
+      "step": 8809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8226499116771105e-05,
+      "loss": 2.0109,
+      "step": 8810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822609507260363e-05,
+      "loss": 2.0327,
+      "step": 8811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8225690986895624e-05,
+      "loss": 1.9994,
+      "step": 8812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8225286859649118e-05,
+      "loss": 2.0128,
+      "step": 8813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822488269086616e-05,
+      "loss": 2.0086,
+      "step": 8814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8224478480548792e-05,
+      "loss": 2.0227,
+      "step": 8815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822407422869905e-05,
+      "loss": 2.0168,
+      "step": 8816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822366993531898e-05,
+      "loss": 1.9982,
+      "step": 8817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822326560041062e-05,
+      "loss": 2.0183,
+      "step": 8818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822286122397601e-05,
+      "loss": 2.0153,
+      "step": 8819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.82224568060172e-05,
+      "loss": 2.0015,
+      "step": 8820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8222052346536227e-05,
+      "loss": 2.017,
+      "step": 8821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8221647845535133e-05,
+      "loss": 2.0219,
+      "step": 8822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822124330301596e-05,
+      "loss": 2.0126,
+      "step": 8823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8220838718980753e-05,
+      "loss": 2.0177,
+      "step": 8824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8220434093431558e-05,
+      "loss": 2.0147,
+      "step": 8825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822002942637041e-05,
+      "loss": 1.9963,
+      "step": 8826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8219624717799362e-05,
+      "loss": 2.0059,
+      "step": 8827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8219219967720453e-05,
+      "loss": 2.0052,
+      "step": 8828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8218815176135724e-05,
+      "loss": 2.013,
+      "step": 8829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8218410343047224e-05,
+      "loss": 2.0025,
+      "step": 8830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8218005468456994e-05,
+      "loss": 1.9881,
+      "step": 8831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.821760055236708e-05,
+      "loss": 2.0069,
+      "step": 8832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8217195594779528e-05,
+      "loss": 2.0032,
+      "step": 8833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8216790595696382e-05,
+      "loss": 2.0039,
+      "step": 8834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8216385555119687e-05,
+      "loss": 1.9979,
+      "step": 8835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.821598047305149e-05,
+      "loss": 2.0154,
+      "step": 8836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8215575349493835e-05,
+      "loss": 2.0101,
+      "step": 8837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8215170184448766e-05,
+      "loss": 2.0092,
+      "step": 8838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.821476497791833e-05,
+      "loss": 2.0126,
+      "step": 8839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8214359729904575e-05,
+      "loss": 2.0255,
+      "step": 8840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8213954440409546e-05,
+      "loss": 2.0106,
+      "step": 8841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8213549109435293e-05,
+      "loss": 2.0423,
+      "step": 8842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.821314373698386e-05,
+      "loss": 1.9885,
+      "step": 8843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8212738323057292e-05,
+      "loss": 2.009,
+      "step": 8844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8212332867657636e-05,
+      "loss": 2.0308,
+      "step": 8845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8211927370786946e-05,
+      "loss": 2.0199,
+      "step": 8846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8211521832447264e-05,
+      "loss": 2.0065,
+      "step": 8847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.821111625264064e-05,
+      "loss": 2.0177,
+      "step": 8848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8210710631369122e-05,
+      "loss": 1.9987,
+      "step": 8849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8210304968634757e-05,
+      "loss": 2.0044,
+      "step": 8850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8209899264439594e-05,
+      "loss": 2.019,
+      "step": 8851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8209493518785684e-05,
+      "loss": 2.0089,
+      "step": 8852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8209087731675072e-05,
+      "loss": 2.0136,
+      "step": 8853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8208681903109814e-05,
+      "loss": 2.012,
+      "step": 8854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8208276033091952e-05,
+      "loss": 2.0098,
+      "step": 8855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8207870121623537e-05,
+      "loss": 2.009,
+      "step": 8856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8207464168706618e-05,
+      "loss": 1.9992,
+      "step": 8857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8207058174343252e-05,
+      "loss": 1.9961,
+      "step": 8858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8206652138535483e-05,
+      "loss": 1.9951,
+      "step": 8859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820624606128536e-05,
+      "loss": 2.0157,
+      "step": 8860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820583994259494e-05,
+      "loss": 2.022,
+      "step": 8861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820543378246627e-05,
+      "loss": 1.9957,
+      "step": 8862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8205027580901395e-05,
+      "loss": 2.0192,
+      "step": 8863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820462133790238e-05,
+      "loss": 2.0184,
+      "step": 8864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8204215053471264e-05,
+      "loss": 2.0281,
+      "step": 8865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820380872761011e-05,
+      "loss": 1.9964,
+      "step": 8866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8203402360320956e-05,
+      "loss": 2.0126,
+      "step": 8867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8202995951605868e-05,
+      "loss": 2.0037,
+      "step": 8868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8202589501466884e-05,
+      "loss": 2.0046,
+      "step": 8869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8202183009906074e-05,
+      "loss": 2.0007,
+      "step": 8870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8201776476925476e-05,
+      "loss": 2.0023,
+      "step": 8871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8201369902527148e-05,
+      "loss": 2.0196,
+      "step": 8872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8200963286713144e-05,
+      "loss": 2.0109,
+      "step": 8873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8200556629485514e-05,
+      "loss": 2.0135,
+      "step": 8874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8200149930846318e-05,
+      "loss": 2.0174,
+      "step": 8875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8199743190797606e-05,
+      "loss": 2.0022,
+      "step": 8876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8199336409341427e-05,
+      "loss": 1.9978,
+      "step": 8877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8198929586479845e-05,
+      "loss": 2.0057,
+      "step": 8878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819852272221491e-05,
+      "loss": 2.0094,
+      "step": 8879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819811581654867e-05,
+      "loss": 2.0033,
+      "step": 8880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8197708869483192e-05,
+      "loss": 2.0142,
+      "step": 8881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8197301881020523e-05,
+      "loss": 1.9838,
+      "step": 8882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819689485116272e-05,
+      "loss": 2.0057,
+      "step": 8883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8196487779911836e-05,
+      "loss": 2.038,
+      "step": 8884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8196080667269932e-05,
+      "loss": 2.0123,
+      "step": 8885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819567351323906e-05,
+      "loss": 1.9932,
+      "step": 8886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819526631782128e-05,
+      "loss": 2.01,
+      "step": 8887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8194859081018643e-05,
+      "loss": 2.0069,
+      "step": 8888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8194451802833205e-05,
+      "loss": 1.9996,
+      "step": 8889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819404448326703e-05,
+      "loss": 1.9945,
+      "step": 8890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819363712232217e-05,
+      "loss": 2.0067,
+      "step": 8891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8193229720000682e-05,
+      "loss": 2.0042,
+      "step": 8892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819282227630462e-05,
+      "loss": 2.0269,
+      "step": 8893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819241479123605e-05,
+      "loss": 2.0167,
+      "step": 8894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8192007264797025e-05,
+      "loss": 2.0108,
+      "step": 8895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81915996969896e-05,
+      "loss": 2.018,
+      "step": 8896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819119208781584e-05,
+      "loss": 2.0108,
+      "step": 8897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8190784437277796e-05,
+      "loss": 2.0178,
+      "step": 8898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8190376745377537e-05,
+      "loss": 2.0201,
+      "step": 8899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8189969012117107e-05,
+      "loss": 1.9991,
+      "step": 8900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8189561237498575e-05,
+      "loss": 2.0188,
+      "step": 8901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8189153421523998e-05,
+      "loss": 1.9994,
+      "step": 8902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818874556419544e-05,
+      "loss": 2.0194,
+      "step": 8903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8188337665514953e-05,
+      "loss": 2.0166,
+      "step": 8904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8187929725484602e-05,
+      "loss": 2.0159,
+      "step": 8905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8187521744106442e-05,
+      "loss": 1.9982,
+      "step": 8906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8187113721382538e-05,
+      "loss": 2.0328,
+      "step": 8907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818670565731495e-05,
+      "loss": 2.0174,
+      "step": 8908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8186297551905734e-05,
+      "loss": 1.9978,
+      "step": 8909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8185889405156958e-05,
+      "loss": 2.0064,
+      "step": 8910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8185481217070677e-05,
+      "loss": 2.0217,
+      "step": 8911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8185072987648954e-05,
+      "loss": 1.9889,
+      "step": 8912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8184664716893854e-05,
+      "loss": 1.999,
+      "step": 8913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8184256404807435e-05,
+      "loss": 2.0236,
+      "step": 8914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818384805139176e-05,
+      "loss": 2.0118,
+      "step": 8915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818343965664889e-05,
+      "loss": 2.017,
+      "step": 8916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8183031220580888e-05,
+      "loss": 2.0166,
+      "step": 8917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8182622743189816e-05,
+      "loss": 2.0233,
+      "step": 8918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818221422447774e-05,
+      "loss": 2.0184,
+      "step": 8919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8181805664446718e-05,
+      "loss": 1.9977,
+      "step": 8920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818139706309882e-05,
+      "loss": 2.0058,
+      "step": 8921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8180988420436097e-05,
+      "loss": 2.0124,
+      "step": 8922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8180579736460625e-05,
+      "loss": 1.9995,
+      "step": 8923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818017101117447e-05,
+      "loss": 1.9867,
+      "step": 8924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817976224457968e-05,
+      "loss": 2.0093,
+      "step": 8925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8179353436678335e-05,
+      "loss": 2.0088,
+      "step": 8926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8178944587472487e-05,
+      "loss": 2.005,
+      "step": 8927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817853569696421e-05,
+      "loss": 2.0106,
+      "step": 8928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8178126765155564e-05,
+      "loss": 1.9906,
+      "step": 8929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8177717792048617e-05,
+      "loss": 2.0107,
+      "step": 8930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8177308777645433e-05,
+      "loss": 2.007,
+      "step": 8931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8176899721948074e-05,
+      "loss": 1.9805,
+      "step": 8932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8176490624958615e-05,
+      "loss": 2.0078,
+      "step": 8933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8176081486679112e-05,
+      "loss": 2.0246,
+      "step": 8934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8175672307111635e-05,
+      "loss": 1.9978,
+      "step": 8935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817526308625825e-05,
+      "loss": 1.9954,
+      "step": 8936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8174853824121027e-05,
+      "loss": 2.0252,
+      "step": 8937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8174444520702022e-05,
+      "loss": 2.0203,
+      "step": 8938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8174035176003318e-05,
+      "loss": 2.006,
+      "step": 8939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8173625790026966e-05,
+      "loss": 2.0202,
+      "step": 8940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8173216362775044e-05,
+      "loss": 2.0385,
+      "step": 8941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817280689424962e-05,
+      "loss": 2.0035,
+      "step": 8942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8172397384452755e-05,
+      "loss": 1.9945,
+      "step": 8943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8171987833386517e-05,
+      "loss": 2.0157,
+      "step": 8944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817157824105298e-05,
+      "loss": 1.9959,
+      "step": 8945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817116860745421e-05,
+      "loss": 2.0119,
+      "step": 8946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8170758932592274e-05,
+      "loss": 2.0142,
+      "step": 8947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8170349216469243e-05,
+      "loss": 2.0302,
+      "step": 8948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8169939459087185e-05,
+      "loss": 1.988,
+      "step": 8949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816952966044817e-05,
+      "loss": 2.0013,
+      "step": 8950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8169119820554268e-05,
+      "loss": 1.9943,
+      "step": 8951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816870993940754e-05,
+      "loss": 2.0245,
+      "step": 8952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816830001701007e-05,
+      "loss": 1.999,
+      "step": 8953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8167890053363918e-05,
+      "loss": 2.0116,
+      "step": 8954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816748004847116e-05,
+      "loss": 2.0078,
+      "step": 8955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8167070002333863e-05,
+      "loss": 2.0002,
+      "step": 8956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8166659914954098e-05,
+      "loss": 1.9989,
+      "step": 8957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8166249786333936e-05,
+      "loss": 2.0139,
+      "step": 8958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8165839616475446e-05,
+      "loss": 2.0191,
+      "step": 8959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8165429405380705e-05,
+      "loss": 2.0046,
+      "step": 8960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816501915305178e-05,
+      "loss": 2.015,
+      "step": 8961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8164608859490746e-05,
+      "loss": 2.0017,
+      "step": 8962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816419852469967e-05,
+      "loss": 1.9914,
+      "step": 8963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8163788148680633e-05,
+      "loss": 1.9946,
+      "step": 8964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8163377731435698e-05,
+      "loss": 2.0074,
+      "step": 8965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8162967272966937e-05,
+      "loss": 2.0138,
+      "step": 8966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816255677327643e-05,
+      "loss": 2.0145,
+      "step": 8967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8162146232366246e-05,
+      "loss": 2.0018,
+      "step": 8968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816173565023846e-05,
+      "loss": 1.9993,
+      "step": 8969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8161325026895143e-05,
+      "loss": 2.0178,
+      "step": 8970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816091436233837e-05,
+      "loss": 2.0017,
+      "step": 8971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8160503656570215e-05,
+      "loss": 2.0342,
+      "step": 8972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816009290959275e-05,
+      "loss": 2.0192,
+      "step": 8973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8159682121408054e-05,
+      "loss": 2.0007,
+      "step": 8974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8159271292018195e-05,
+      "loss": 2.0149,
+      "step": 8975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815886042142525e-05,
+      "loss": 2.0148,
+      "step": 8976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81584495096313e-05,
+      "loss": 2.0185,
+      "step": 8977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815803855663841e-05,
+      "loss": 2.0169,
+      "step": 8978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8157627562448662e-05,
+      "loss": 2.0092,
+      "step": 8979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815721652706413e-05,
+      "loss": 2.0081,
+      "step": 8980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815680545048689e-05,
+      "loss": 2.0188,
+      "step": 8981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8156394332719012e-05,
+      "loss": 2.0266,
+      "step": 8982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8155983173762582e-05,
+      "loss": 1.9932,
+      "step": 8983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815557197361967e-05,
+      "loss": 2.0162,
+      "step": 8984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8155160732292352e-05,
+      "loss": 2.005,
+      "step": 8985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815474944978271e-05,
+      "loss": 2.0272,
+      "step": 8986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8154338126092815e-05,
+      "loss": 1.992,
+      "step": 8987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8153926761224742e-05,
+      "loss": 1.9817,
+      "step": 8988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8153515355180582e-05,
+      "loss": 2.0026,
+      "step": 8989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8153103907962398e-05,
+      "loss": 1.9906,
+      "step": 8990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8152692419572273e-05,
+      "loss": 2.0124,
+      "step": 8991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8152280890012288e-05,
+      "loss": 1.995,
+      "step": 8992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8151869319284516e-05,
+      "loss": 2.0141,
+      "step": 8993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8151457707391038e-05,
+      "loss": 2.0081,
+      "step": 8994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8151046054333933e-05,
+      "loss": 2.0157,
+      "step": 8995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8150634360115277e-05,
+      "loss": 2.0175,
+      "step": 8996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8150222624737154e-05,
+      "loss": 2.0146,
+      "step": 8997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814981084820164e-05,
+      "loss": 2.0029,
+      "step": 8998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814939903051081e-05,
+      "loss": 2.022,
+      "step": 8999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8148987171666754e-05,
+      "loss": 2.0089,
+      "step": 9000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8148575271671546e-05,
+      "loss": 2.0238,
+      "step": 9001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8148163330527264e-05,
+      "loss": 2.0044,
+      "step": 9002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814775134823599e-05,
+      "loss": 2.0262,
+      "step": 9003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8147339324799803e-05,
+      "loss": 2.0046,
+      "step": 9004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814692726022079e-05,
+      "loss": 1.9967,
+      "step": 9005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8146515154501023e-05,
+      "loss": 2.0085,
+      "step": 9006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814610300764259e-05,
+      "loss": 2.014,
+      "step": 9007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814569081964757e-05,
+      "loss": 2.0108,
+      "step": 9008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8145278590518042e-05,
+      "loss": 1.9966,
+      "step": 9009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8144866320256093e-05,
+      "loss": 2.0131,
+      "step": 9010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81444540088638e-05,
+      "loss": 2.0039,
+      "step": 9011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8144041656343245e-05,
+      "loss": 2.0063,
+      "step": 9012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8143629262696514e-05,
+      "loss": 2.0008,
+      "step": 9013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814321682792569e-05,
+      "loss": 2.0115,
+      "step": 9014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814280435203285e-05,
+      "loss": 2.023,
+      "step": 9015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8142391835020082e-05,
+      "loss": 2.0272,
+      "step": 9016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8141979276889466e-05,
+      "loss": 2.0177,
+      "step": 9017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8141566677643085e-05,
+      "loss": 2.0125,
+      "step": 9018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8141154037283032e-05,
+      "loss": 2.0039,
+      "step": 9019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8140741355811378e-05,
+      "loss": 2.0055,
+      "step": 9020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814032863323021e-05,
+      "loss": 1.9981,
+      "step": 9021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8139915869541618e-05,
+      "loss": 2.0156,
+      "step": 9022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8139503064747683e-05,
+      "loss": 2.0123,
+      "step": 9023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8139090218850488e-05,
+      "loss": 2.0084,
+      "step": 9024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813867733185212e-05,
+      "loss": 2.0303,
+      "step": 9025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8138264403754663e-05,
+      "loss": 2.0148,
+      "step": 9026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81378514345602e-05,
+      "loss": 2.027,
+      "step": 9027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813743842427082e-05,
+      "loss": 2.0125,
+      "step": 9028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813702537288861e-05,
+      "loss": 2.0003,
+      "step": 9029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8136612280415652e-05,
+      "loss": 2.0064,
+      "step": 9030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813619914685403e-05,
+      "loss": 1.9767,
+      "step": 9031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8135785972205842e-05,
+      "loss": 2.0024,
+      "step": 9032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813537275647316e-05,
+      "loss": 2.0043,
+      "step": 9033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8134959499658077e-05,
+      "loss": 2.0035,
+      "step": 9034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8134546201762682e-05,
+      "loss": 2.0101,
+      "step": 9035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813413286278906e-05,
+      "loss": 2.0186,
+      "step": 9036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8133719482739294e-05,
+      "loss": 2.0018,
+      "step": 9037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8133306061615478e-05,
+      "loss": 2.0298,
+      "step": 9038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81328925994197e-05,
+      "loss": 2.0028,
+      "step": 9039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813247909615404e-05,
+      "loss": 2.0187,
+      "step": 9040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8132065551820596e-05,
+      "loss": 2.0201,
+      "step": 9041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813165196642145e-05,
+      "loss": 1.9977,
+      "step": 9042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813123833995869e-05,
+      "loss": 2.0037,
+      "step": 9043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813082467243441e-05,
+      "loss": 2.0135,
+      "step": 9044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8130410963850696e-05,
+      "loss": 1.9937,
+      "step": 9045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8129997214209634e-05,
+      "loss": 2.0209,
+      "step": 9046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812958342351332e-05,
+      "loss": 2.0289,
+      "step": 9047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812916959176384e-05,
+      "loss": 2.0008,
+      "step": 9048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8128755718963282e-05,
+      "loss": 2.0112,
+      "step": 9049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8128341805113737e-05,
+      "loss": 2.0032,
+      "step": 9050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81279278502173e-05,
+      "loss": 2.0134,
+      "step": 9051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8127513854276052e-05,
+      "loss": 2.01,
+      "step": 9052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8127099817292094e-05,
+      "loss": 2.0085,
+      "step": 9053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8126685739267506e-05,
+      "loss": 2.0076,
+      "step": 9054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812627162020439e-05,
+      "loss": 1.9957,
+      "step": 9055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8125857460104832e-05,
+      "loss": 1.9873,
+      "step": 9056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8125443258970924e-05,
+      "loss": 1.9947,
+      "step": 9057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8125029016804757e-05,
+      "loss": 2.0072,
+      "step": 9058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812461473360842e-05,
+      "loss": 2.0256,
+      "step": 9059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8124200409384012e-05,
+      "loss": 2.0107,
+      "step": 9060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812378604413362e-05,
+      "loss": 2.0203,
+      "step": 9061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8123371637859337e-05,
+      "loss": 2.0012,
+      "step": 9062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8122957190563255e-05,
+      "loss": 2.0214,
+      "step": 9063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8122542702247474e-05,
+      "loss": 2.0179,
+      "step": 9064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812212817291408e-05,
+      "loss": 2.0116,
+      "step": 9065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8121713602565163e-05,
+      "loss": 2.0171,
+      "step": 9066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8121298991202824e-05,
+      "loss": 2.0002,
+      "step": 9067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8120884338829158e-05,
+      "loss": 2.0247,
+      "step": 9068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8120469645446254e-05,
+      "loss": 2.0253,
+      "step": 9069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8120054911056204e-05,
+      "loss": 2.035,
+      "step": 9070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8119640135661106e-05,
+      "loss": 2.0296,
+      "step": 9071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8119225319263055e-05,
+      "loss": 2.0037,
+      "step": 9072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8118810461864146e-05,
+      "loss": 2.0182,
+      "step": 9073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811839556346647e-05,
+      "loss": 2.0059,
+      "step": 9074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8117980624072127e-05,
+      "loss": 2.0128,
+      "step": 9075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8117565643683215e-05,
+      "loss": 2.0199,
+      "step": 9076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811715062230182e-05,
+      "loss": 2.0133,
+      "step": 9077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8116735559930043e-05,
+      "loss": 2.0105,
+      "step": 9078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8116320456569982e-05,
+      "loss": 2.0005,
+      "step": 9079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8115905312223728e-05,
+      "loss": 2.0185,
+      "step": 9080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811549012689338e-05,
+      "loss": 2.0161,
+      "step": 9081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8115074900581036e-05,
+      "loss": 2.0268,
+      "step": 9082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8114659633288797e-05,
+      "loss": 2.0049,
+      "step": 9083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811424432501875e-05,
+      "loss": 2.0079,
+      "step": 9084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8113828975773e-05,
+      "loss": 2.0046,
+      "step": 9085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811341358555364e-05,
+      "loss": 2.0048,
+      "step": 9086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8112998154362768e-05,
+      "loss": 2.0222,
+      "step": 9087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8112582682202485e-05,
+      "loss": 2.0212,
+      "step": 9088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8112167169074887e-05,
+      "loss": 2.0299,
+      "step": 9089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8111751614982072e-05,
+      "loss": 2.0196,
+      "step": 9090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8111336019926137e-05,
+      "loss": 1.9802,
+      "step": 9091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8110920383909184e-05,
+      "loss": 1.9978,
+      "step": 9092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8110504706933315e-05,
+      "loss": 1.9873,
+      "step": 9093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811008898900062e-05,
+      "loss": 2.0271,
+      "step": 9094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8109673230113204e-05,
+      "loss": 1.9912,
+      "step": 9095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8109257430273163e-05,
+      "loss": 2.008,
+      "step": 9096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8108841589482606e-05,
+      "loss": 2.0184,
+      "step": 9097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8108425707743623e-05,
+      "loss": 1.9968,
+      "step": 9098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8108009785058313e-05,
+      "loss": 2.0195,
+      "step": 9099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8107593821428785e-05,
+      "loss": 2.0235,
+      "step": 9100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8107177816857136e-05,
+      "loss": 2.0231,
+      "step": 9101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8106761771345462e-05,
+      "loss": 2.0038,
+      "step": 9102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8106345684895872e-05,
+      "loss": 2.0078,
+      "step": 9103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8105929557510464e-05,
+      "loss": 2.0238,
+      "step": 9104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.810551338919133e-05,
+      "loss": 1.9917,
+      "step": 9105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8105097179940588e-05,
+      "loss": 1.979,
+      "step": 9106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.810468092976033e-05,
+      "loss": 2.0006,
+      "step": 9107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.810426463865266e-05,
+      "loss": 2.0138,
+      "step": 9108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.810384830661968e-05,
+      "loss": 2.0247,
+      "step": 9109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8103431933663494e-05,
+      "loss": 2.0126,
+      "step": 9110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8103015519786203e-05,
+      "loss": 2.0302,
+      "step": 9111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8102599064989906e-05,
+      "loss": 2.0076,
+      "step": 9112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8102182569276714e-05,
+      "loss": 2.0165,
+      "step": 9113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8101766032648723e-05,
+      "loss": 2.0028,
+      "step": 9114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8101349455108044e-05,
+      "loss": 2.0393,
+      "step": 9115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8100932836656776e-05,
+      "loss": 2.0133,
+      "step": 9116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8100516177297017e-05,
+      "loss": 2.0095,
+      "step": 9117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8100099477030884e-05,
+      "loss": 2.0005,
+      "step": 9118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8099682735860472e-05,
+      "loss": 2.0369,
+      "step": 9119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8099265953787886e-05,
+      "loss": 2.0127,
+      "step": 9120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8098849130815237e-05,
+      "loss": 1.9985,
+      "step": 9121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8098432266944622e-05,
+      "loss": 2.0018,
+      "step": 9122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809801536217815e-05,
+      "loss": 2.0149,
+      "step": 9123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809759841651793e-05,
+      "loss": 2.006,
+      "step": 9124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809718142996606e-05,
+      "loss": 2.0153,
+      "step": 9125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809676440252465e-05,
+      "loss": 1.9947,
+      "step": 9126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8096347334195803e-05,
+      "loss": 1.9967,
+      "step": 9127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809593022498163e-05,
+      "loss": 2.0037,
+      "step": 9128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8095513074884233e-05,
+      "loss": 2.0031,
+      "step": 9129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8095095883905723e-05,
+      "loss": 1.9965,
+      "step": 9130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.80946786520482e-05,
+      "loss": 1.9969,
+      "step": 9131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809426137931378e-05,
+      "loss": 2.0063,
+      "step": 9132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809384406570456e-05,
+      "loss": 1.9867,
+      "step": 9133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8093426711222656e-05,
+      "loss": 2.0114,
+      "step": 9134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809300931587017e-05,
+      "loss": 2.0152,
+      "step": 9135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809259187964921e-05,
+      "loss": 2.0255,
+      "step": 9136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809217440256189e-05,
+      "loss": 2.0024,
+      "step": 9137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809175688461031e-05,
+      "loss": 1.9903,
+      "step": 9138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8091339325796584e-05,
+      "loss": 1.9982,
+      "step": 9139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809092172612282e-05,
+      "loss": 1.9994,
+      "step": 9140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8090504085591124e-05,
+      "loss": 2.028,
+      "step": 9141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8090086404203607e-05,
+      "loss": 2.0102,
+      "step": 9142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808966868196238e-05,
+      "loss": 2.0087,
+      "step": 9143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808925091886955e-05,
+      "loss": 2.0157,
+      "step": 9144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808883311492723e-05,
+      "loss": 2.022,
+      "step": 9145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8088415270137522e-05,
+      "loss": 2.0084,
+      "step": 9146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808799738450254e-05,
+      "loss": 2.0015,
+      "step": 9147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8087579458024395e-05,
+      "loss": 1.9999,
+      "step": 9148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8087161490705202e-05,
+      "loss": 2.0079,
+      "step": 9149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808674348254707e-05,
+      "loss": 2.0156,
+      "step": 9150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.80863254335521e-05,
+      "loss": 2.0041,
+      "step": 9151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8085907343722414e-05,
+      "loss": 2.0225,
+      "step": 9152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8085489213060118e-05,
+      "loss": 2.0346,
+      "step": 9153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8085071041567327e-05,
+      "loss": 2.0151,
+      "step": 9154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808465282924615e-05,
+      "loss": 2.0168,
+      "step": 9155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.80842345760987e-05,
+      "loss": 2.0206,
+      "step": 9156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808381628212709e-05,
+      "loss": 1.9922,
+      "step": 9157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808339794733343e-05,
+      "loss": 1.9961,
+      "step": 9158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8082979571719835e-05,
+      "loss": 1.9861,
+      "step": 9159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8082561155288412e-05,
+      "loss": 2.0087,
+      "step": 9160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8082142698041284e-05,
+      "loss": 2.0101,
+      "step": 9161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8081724199980554e-05,
+      "loss": 2.0306,
+      "step": 9162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8081305661108344e-05,
+      "loss": 2.022,
+      "step": 9163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8080887081426764e-05,
+      "loss": 2.0119,
+      "step": 9164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8080468460937925e-05,
+      "loss": 2.0009,
+      "step": 9165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808004979964394e-05,
+      "loss": 2.0179,
+      "step": 9166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807963109754693e-05,
+      "loss": 1.9845,
+      "step": 9167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8079212354649006e-05,
+      "loss": 2.0105,
+      "step": 9168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8078793570952283e-05,
+      "loss": 2.0055,
+      "step": 9169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8078374746458873e-05,
+      "loss": 2.004,
+      "step": 9170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8077955881170895e-05,
+      "loss": 2.0001,
+      "step": 9171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807753697509046e-05,
+      "loss": 2.0047,
+      "step": 9172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8077118028219688e-05,
+      "loss": 2.0117,
+      "step": 9173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807669904056069e-05,
+      "loss": 2.0123,
+      "step": 9174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8076280012115587e-05,
+      "loss": 1.9835,
+      "step": 9175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807586094288649e-05,
+      "loss": 2.0019,
+      "step": 9176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807544183287552e-05,
+      "loss": 1.9932,
+      "step": 9177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8075022682084786e-05,
+      "loss": 1.9965,
+      "step": 9178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8074603490516413e-05,
+      "loss": 2.0081,
+      "step": 9179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8074184258172515e-05,
+      "loss": 1.9864,
+      "step": 9180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8073764985055205e-05,
+      "loss": 2.0118,
+      "step": 9181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807334567116661e-05,
+      "loss": 2.0021,
+      "step": 9182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8072926316508835e-05,
+      "loss": 2.0013,
+      "step": 9183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8072506921084006e-05,
+      "loss": 2.026,
+      "step": 9184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807208748489424e-05,
+      "loss": 2.0092,
+      "step": 9185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8071668007941653e-05,
+      "loss": 2.0206,
+      "step": 9186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8071248490228367e-05,
+      "loss": 2.0128,
+      "step": 9187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8070828931756496e-05,
+      "loss": 1.9785,
+      "step": 9188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8070409332528157e-05,
+      "loss": 2.0079,
+      "step": 9189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8069989692545476e-05,
+      "loss": 2.0207,
+      "step": 9190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8069570011810565e-05,
+      "loss": 2.0071,
+      "step": 9191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.806915029032555e-05,
+      "loss": 2.0113,
+      "step": 9192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8068730528092548e-05,
+      "loss": 2.021,
+      "step": 9193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8068310725113673e-05,
+      "loss": 2.0173,
+      "step": 9194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8067890881391053e-05,
+      "loss": 1.9998,
+      "step": 9195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8067470996926805e-05,
+      "loss": 2.0036,
+      "step": 9196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8067051071723048e-05,
+      "loss": 2.0013,
+      "step": 9197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8066631105781905e-05,
+      "loss": 2.0216,
+      "step": 9198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.80662110991055e-05,
+      "loss": 2.0075,
+      "step": 9199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8065791051695943e-05,
+      "loss": 2.0076,
+      "step": 9200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8065370963555366e-05,
+      "loss": 1.9969,
+      "step": 9201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8064950834685887e-05,
+      "loss": 2.0078,
+      "step": 9202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8064530665089625e-05,
+      "loss": 2.0055,
+      "step": 9203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8064110454768702e-05,
+      "loss": 2.0096,
+      "step": 9204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8063690203725245e-05,
+      "loss": 2.0066,
+      "step": 9205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.806326991196137e-05,
+      "loss": 1.9854,
+      "step": 9206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8062849579479203e-05,
+      "loss": 2.0111,
+      "step": 9207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8062429206280865e-05,
+      "loss": 2.0064,
+      "step": 9208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8062008792368483e-05,
+      "loss": 2.0004,
+      "step": 9209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8061588337744173e-05,
+      "loss": 2.0125,
+      "step": 9210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8061167842410064e-05,
+      "loss": 1.9983,
+      "step": 9211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.806074730636828e-05,
+      "loss": 1.9911,
+      "step": 9212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8060326729620938e-05,
+      "loss": 1.9936,
+      "step": 9213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8059906112170167e-05,
+      "loss": 1.9833,
+      "step": 9214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8059485454018087e-05,
+      "loss": 1.9982,
+      "step": 9215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805906475516683e-05,
+      "loss": 2.004,
+      "step": 9216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8058644015618512e-05,
+      "loss": 2.0133,
+      "step": 9217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8058223235375266e-05,
+      "loss": 2.0145,
+      "step": 9218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805780241443921e-05,
+      "loss": 1.9874,
+      "step": 9219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805738155281247e-05,
+      "loss": 2.0092,
+      "step": 9220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8056960650497174e-05,
+      "loss": 2.0095,
+      "step": 9221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8056539707495446e-05,
+      "loss": 1.9886,
+      "step": 9222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8056118723809407e-05,
+      "loss": 2.0315,
+      "step": 9223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8055697699441196e-05,
+      "loss": 2.004,
+      "step": 9224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8055276634392926e-05,
+      "loss": 2.0013,
+      "step": 9225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8054855528666726e-05,
+      "loss": 1.9998,
+      "step": 9226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805443438226473e-05,
+      "loss": 1.9888,
+      "step": 9227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8054013195189055e-05,
+      "loss": 2.0021,
+      "step": 9228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8053591967441833e-05,
+      "loss": 2.0185,
+      "step": 9229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805317069902519e-05,
+      "loss": 1.9934,
+      "step": 9230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805274938994126e-05,
+      "loss": 1.9993,
+      "step": 9231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805232804019216e-05,
+      "loss": 1.9992,
+      "step": 9232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805190664978002e-05,
+      "loss": 2.0,
+      "step": 9233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8051485218706974e-05,
+      "loss": 2.01,
+      "step": 9234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8051063746975144e-05,
+      "loss": 2.009,
+      "step": 9235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8050642234586657e-05,
+      "loss": 1.9937,
+      "step": 9236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805022068154365e-05,
+      "loss": 2.0085,
+      "step": 9237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8049799087848245e-05,
+      "loss": 2.0138,
+      "step": 9238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8049377453502576e-05,
+      "loss": 2.0029,
+      "step": 9239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8048955778508764e-05,
+      "loss": 1.9953,
+      "step": 9240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8048534062868945e-05,
+      "loss": 2.006,
+      "step": 9241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804811230658525e-05,
+      "loss": 2.0093,
+      "step": 9242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.80476905096598e-05,
+      "loss": 2.0024,
+      "step": 9243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8047268672094736e-05,
+      "loss": 2.005,
+      "step": 9244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8046846793892184e-05,
+      "loss": 2.0115,
+      "step": 9245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804642487505427e-05,
+      "loss": 2.0093,
+      "step": 9246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8046002915583132e-05,
+      "loss": 2.0059,
+      "step": 9247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8045580915480894e-05,
+      "loss": 1.9995,
+      "step": 9248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804515887474969e-05,
+      "loss": 2.0212,
+      "step": 9249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8044736793391654e-05,
+      "loss": 1.9904,
+      "step": 9250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804431467140891e-05,
+      "loss": 2.0011,
+      "step": 9251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8043892508803598e-05,
+      "loss": 2.0055,
+      "step": 9252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8043470305577844e-05,
+      "loss": 2.0086,
+      "step": 9253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8043048061733785e-05,
+      "loss": 2.0106,
+      "step": 9254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804262577727355e-05,
+      "loss": 2.023,
+      "step": 9255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804220345219927e-05,
+      "loss": 1.9951,
+      "step": 9256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804178108651308e-05,
+      "loss": 1.9893,
+      "step": 9257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8041358680217115e-05,
+      "loss": 1.9911,
+      "step": 9258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8040936233313505e-05,
+      "loss": 1.9969,
+      "step": 9259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8040513745804382e-05,
+      "loss": 1.9849,
+      "step": 9260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804009121769188e-05,
+      "loss": 2.0284,
+      "step": 9261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803966864897814e-05,
+      "loss": 1.9994,
+      "step": 9262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8039246039665287e-05,
+      "loss": 2.0151,
+      "step": 9263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803882338975546e-05,
+      "loss": 2.0041,
+      "step": 9264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8038400699250784e-05,
+      "loss": 2.0103,
+      "step": 9265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803797796815341e-05,
+      "loss": 2.0107,
+      "step": 9266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803755519646546e-05,
+      "loss": 1.9921,
+      "step": 9267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803713238418907e-05,
+      "loss": 1.9912,
+      "step": 9268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803670953132638e-05,
+      "loss": 2.0072,
+      "step": 9269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8036286637879527e-05,
+      "loss": 1.9993,
+      "step": 9270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8035863703850637e-05,
+      "loss": 2.0134,
+      "step": 9271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8035440729241855e-05,
+      "loss": 2.0096,
+      "step": 9272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8035017714055313e-05,
+      "loss": 1.9998,
+      "step": 9273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803459465829315e-05,
+      "loss": 1.9949,
+      "step": 9274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8034171561957495e-05,
+      "loss": 1.9976,
+      "step": 9275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8033748425050492e-05,
+      "loss": 2.0228,
+      "step": 9276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8033325247574277e-05,
+      "loss": 2.0187,
+      "step": 9277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803290202953098e-05,
+      "loss": 2.0245,
+      "step": 9278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8032478770922748e-05,
+      "loss": 2.0161,
+      "step": 9279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8032055471751713e-05,
+      "loss": 1.9988,
+      "step": 9280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803163213202001e-05,
+      "loss": 1.9951,
+      "step": 9281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8031208751729784e-05,
+      "loss": 2.0096,
+      "step": 9282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803078533088317e-05,
+      "loss": 1.998,
+      "step": 9283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8030361869482305e-05,
+      "loss": 1.9976,
+      "step": 9284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8029938367529326e-05,
+      "loss": 1.9981,
+      "step": 9285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8029514825026376e-05,
+      "loss": 2.0044,
+      "step": 9286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802909124197559e-05,
+      "loss": 1.9818,
+      "step": 9287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802866761837911e-05,
+      "loss": 1.9953,
+      "step": 9288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8028243954239067e-05,
+      "loss": 1.9939,
+      "step": 9289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8027820249557616e-05,
+      "loss": 1.9941,
+      "step": 9290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8027396504336883e-05,
+      "loss": 1.9797,
+      "step": 9291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8026972718579014e-05,
+      "loss": 2.0118,
+      "step": 9292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8026548892286146e-05,
+      "loss": 1.9932,
+      "step": 9293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8026125025460423e-05,
+      "loss": 2.0031,
+      "step": 9294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802570111810398e-05,
+      "loss": 2.0098,
+      "step": 9295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802527717021896e-05,
+      "loss": 1.9972,
+      "step": 9296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802485318180751e-05,
+      "loss": 1.9981,
+      "step": 9297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8024429152871767e-05,
+      "loss": 2.0031,
+      "step": 9298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8024005083413866e-05,
+      "loss": 1.9899,
+      "step": 9299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8023580973435955e-05,
+      "loss": 1.9918,
+      "step": 9300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8023156822940176e-05,
+      "loss": 2.0216,
+      "step": 9301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8022732631928666e-05,
+      "loss": 2.0134,
+      "step": 9302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8022308400403578e-05,
+      "loss": 2.021,
+      "step": 9303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8021884128367037e-05,
+      "loss": 1.9769,
+      "step": 9304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8021459815821204e-05,
+      "loss": 1.996,
+      "step": 9305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8021035462768206e-05,
+      "loss": 2.0122,
+      "step": 9306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8020611069210196e-05,
+      "loss": 1.9915,
+      "step": 9307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8020186635149315e-05,
+      "loss": 2.0122,
+      "step": 9308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8019762160587705e-05,
+      "loss": 2.0143,
+      "step": 9309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8019337645527506e-05,
+      "loss": 1.9965,
+      "step": 9310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8018913089970867e-05,
+      "loss": 2.031,
+      "step": 9311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8018488493919935e-05,
+      "loss": 1.996,
+      "step": 9312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8018063857376847e-05,
+      "loss": 1.9961,
+      "step": 9313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801763918034375e-05,
+      "loss": 2.0036,
+      "step": 9314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8017214462822786e-05,
+      "loss": 2.0109,
+      "step": 9315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8016789704816102e-05,
+      "loss": 2.0064,
+      "step": 9316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8016364906325847e-05,
+      "loss": 2.0212,
+      "step": 9317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801594006735416e-05,
+      "loss": 1.9897,
+      "step": 9318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8015515187903192e-05,
+      "loss": 2.0,
+      "step": 9319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801509026797508e-05,
+      "loss": 1.985,
+      "step": 9320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801466530757198e-05,
+      "loss": 2.009,
+      "step": 9321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8014240306696035e-05,
+      "loss": 1.9964,
+      "step": 9322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8013815265349382e-05,
+      "loss": 2.0297,
+      "step": 9323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801339018353418e-05,
+      "loss": 2.0023,
+      "step": 9324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801296506125257e-05,
+      "loss": 1.9852,
+      "step": 9325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8012539898506696e-05,
+      "loss": 1.9988,
+      "step": 9326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8012114695298712e-05,
+      "loss": 2.0102,
+      "step": 9327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801168945163076e-05,
+      "loss": 1.9993,
+      "step": 9328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801126416750499e-05,
+      "loss": 2.0092,
+      "step": 9329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8010838842923544e-05,
+      "loss": 2.0048,
+      "step": 9330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801041347788858e-05,
+      "loss": 2.0112,
+      "step": 9331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8009988072402234e-05,
+      "loss": 2.0093,
+      "step": 9332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8009562626466666e-05,
+      "loss": 2.0061,
+      "step": 9333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8009137140084018e-05,
+      "loss": 1.9988,
+      "step": 9334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8008711613256437e-05,
+      "loss": 2.0024,
+      "step": 9335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.800828604598608e-05,
+      "loss": 1.9951,
+      "step": 9336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8007860438275086e-05,
+      "loss": 1.9957,
+      "step": 9337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8007434790125607e-05,
+      "loss": 2.0108,
+      "step": 9338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8007009101539793e-05,
+      "loss": 1.9945,
+      "step": 9339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8006583372519803e-05,
+      "loss": 2.0015,
+      "step": 9340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8006157603067773e-05,
+      "loss": 2.0013,
+      "step": 9341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8005731793185858e-05,
+      "loss": 1.9957,
+      "step": 9342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.800530594287621e-05,
+      "loss": 1.9983,
+      "step": 9343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8004880052140983e-05,
+      "loss": 2.0104,
+      "step": 9344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8004454120982317e-05,
+      "loss": 2.003,
+      "step": 9345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8004028149402374e-05,
+      "loss": 1.9937,
+      "step": 9346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8003602137403296e-05,
+      "loss": 1.9881,
+      "step": 9347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8003176084987242e-05,
+      "loss": 1.9982,
+      "step": 9348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8002749992156356e-05,
+      "loss": 2.0067,
+      "step": 9349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.80023238589128e-05,
+      "loss": 2.0173,
+      "step": 9350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.800189768525871e-05,
+      "loss": 2.0035,
+      "step": 9351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8001471471196254e-05,
+      "loss": 2.004,
+      "step": 9352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8001045216727578e-05,
+      "loss": 2.0036,
+      "step": 9353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8000618921854835e-05,
+      "loss": 2.0087,
+      "step": 9354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8000192586580178e-05,
+      "loss": 2.0258,
+      "step": 9355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7999766210905756e-05,
+      "loss": 1.9891,
+      "step": 9356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7999339794833724e-05,
+      "loss": 1.9947,
+      "step": 9357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.799891333836624e-05,
+      "loss": 1.9921,
+      "step": 9358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7998486841505456e-05,
+      "loss": 2.0204,
+      "step": 9359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7998060304253523e-05,
+      "loss": 2.0159,
+      "step": 9360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7997633726612593e-05,
+      "loss": 1.9922,
+      "step": 9361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7997207108584826e-05,
+      "loss": 2.0001,
+      "step": 9362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7996780450172372e-05,
+      "loss": 1.994,
+      "step": 9363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7996353751377385e-05,
+      "loss": 1.9948,
+      "step": 9364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7995927012202027e-05,
+      "loss": 1.9982,
+      "step": 9365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7995500232648445e-05,
+      "loss": 2.0048,
+      "step": 9366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7995073412718795e-05,
+      "loss": 1.9941,
+      "step": 9367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7994646552415235e-05,
+      "loss": 1.9906,
+      "step": 9368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7994219651739923e-05,
+      "loss": 2.0065,
+      "step": 9369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7993792710695007e-05,
+      "loss": 1.9974,
+      "step": 9370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.799336572928265e-05,
+      "loss": 1.9966,
+      "step": 9371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7992938707505007e-05,
+      "loss": 1.9986,
+      "step": 9372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7992511645364233e-05,
+      "loss": 1.9913,
+      "step": 9373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7992084542862484e-05,
+      "loss": 1.9799,
+      "step": 9374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7991657400001912e-05,
+      "loss": 2.0008,
+      "step": 9375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7991230216784686e-05,
+      "loss": 1.9896,
+      "step": 9376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7990802993212958e-05,
+      "loss": 1.9912,
+      "step": 9377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7990375729288884e-05,
+      "loss": 2.0071,
+      "step": 9378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7989948425014617e-05,
+      "loss": 1.9779,
+      "step": 9379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7989521080392322e-05,
+      "loss": 1.985,
+      "step": 9380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7989093695424154e-05,
+      "loss": 1.9972,
+      "step": 9381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7988666270112278e-05,
+      "loss": 1.9885,
+      "step": 9382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798823880445884e-05,
+      "loss": 2.0056,
+      "step": 9383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7987811298466005e-05,
+      "loss": 1.9716,
+      "step": 9384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7987383752135933e-05,
+      "loss": 1.9918,
+      "step": 9385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7986956165470785e-05,
+      "loss": 1.9889,
+      "step": 9386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7986528538472713e-05,
+      "loss": 2.001,
+      "step": 9387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7986100871143883e-05,
+      "loss": 1.9849,
+      "step": 9388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798567316348645e-05,
+      "loss": 1.9895,
+      "step": 9389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7985245415502578e-05,
+      "loss": 2.0156,
+      "step": 9390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7984817627194427e-05,
+      "loss": 2.011,
+      "step": 9391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7984389798564153e-05,
+      "loss": 1.9741,
+      "step": 9392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798396192961392e-05,
+      "loss": 2.0035,
+      "step": 9393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7983534020345885e-05,
+      "loss": 1.9891,
+      "step": 9394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7983106070762214e-05,
+      "loss": 1.9893,
+      "step": 9395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7982678080865063e-05,
+      "loss": 1.9902,
+      "step": 9396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7982250050656594e-05,
+      "loss": 1.9997,
+      "step": 9397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7981821980138976e-05,
+      "loss": 2.0228,
+      "step": 9398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7981393869314364e-05,
+      "loss": 2.0186,
+      "step": 9399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7980965718184918e-05,
+      "loss": 1.9887,
+      "step": 9400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7980537526752805e-05,
+      "loss": 1.9803,
+      "step": 9401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798010929502019e-05,
+      "loss": 2.0132,
+      "step": 9402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797968102298922e-05,
+      "loss": 2.0117,
+      "step": 9403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7979252710662077e-05,
+      "loss": 2.0051,
+      "step": 9404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7978824358040912e-05,
+      "loss": 2.0059,
+      "step": 9405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7978395965127894e-05,
+      "loss": 2.0119,
+      "step": 9406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797796753192518e-05,
+      "loss": 1.9903,
+      "step": 9407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7977539058434938e-05,
+      "loss": 1.9843,
+      "step": 9408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7977110544659334e-05,
+      "loss": 1.994,
+      "step": 9409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7976681990600527e-05,
+      "loss": 1.9801,
+      "step": 9410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7976253396260682e-05,
+      "loss": 2.0099,
+      "step": 9411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7975824761641967e-05,
+      "loss": 1.9885,
+      "step": 9412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7975396086746545e-05,
+      "loss": 1.9977,
+      "step": 9413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797496737157658e-05,
+      "loss": 2.0106,
+      "step": 9414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7974538616134235e-05,
+      "loss": 1.9758,
+      "step": 9415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7974109820421675e-05,
+      "loss": 1.9841,
+      "step": 9416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7973680984441067e-05,
+      "loss": 1.9851,
+      "step": 9417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797325210819458e-05,
+      "loss": 2.0009,
+      "step": 9418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7972823191684377e-05,
+      "loss": 1.9926,
+      "step": 9419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797239423491262e-05,
+      "loss": 1.9785,
+      "step": 9420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7971965237881483e-05,
+      "loss": 2.0034,
+      "step": 9421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7971536200593124e-05,
+      "loss": 1.9936,
+      "step": 9422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7971107123049717e-05,
+      "loss": 1.9988,
+      "step": 9423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797067800525342e-05,
+      "loss": 2.024,
+      "step": 9424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7970248847206407e-05,
+      "loss": 2.0026,
+      "step": 9425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796981964891085e-05,
+      "loss": 1.9854,
+      "step": 9426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7969390410368907e-05,
+      "loss": 1.9999,
+      "step": 9427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7968961131582743e-05,
+      "loss": 2.0124,
+      "step": 9428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7968531812554536e-05,
+      "loss": 1.9657,
+      "step": 9429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796810245328645e-05,
+      "loss": 2.0017,
+      "step": 9430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7967673053780653e-05,
+      "loss": 1.9863,
+      "step": 9431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796724361403931e-05,
+      "loss": 2.0012,
+      "step": 9432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796681413406459e-05,
+      "loss": 2.004,
+      "step": 9433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7966384613858672e-05,
+      "loss": 1.999,
+      "step": 9434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796595505342371e-05,
+      "loss": 1.9889,
+      "step": 9435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7965525452761886e-05,
+      "loss": 2.0079,
+      "step": 9436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7965095811875362e-05,
+      "loss": 1.9982,
+      "step": 9437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796466613076631e-05,
+      "loss": 1.9971,
+      "step": 9438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79642364094369e-05,
+      "loss": 1.9992,
+      "step": 9439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79638066478893e-05,
+      "loss": 2.0141,
+      "step": 9440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7963376846125683e-05,
+      "loss": 2.0009,
+      "step": 9441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7962947004148217e-05,
+      "loss": 1.985,
+      "step": 9442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7962517121959073e-05,
+      "loss": 1.999,
+      "step": 9443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7962087199560423e-05,
+      "loss": 1.9956,
+      "step": 9444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7961657236954438e-05,
+      "loss": 1.9699,
+      "step": 9445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796122723414329e-05,
+      "loss": 2.0007,
+      "step": 9446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796079719112915e-05,
+      "loss": 2.0009,
+      "step": 9447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7960367107914188e-05,
+      "loss": 1.9945,
+      "step": 9448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7959936984500576e-05,
+      "loss": 1.989,
+      "step": 9449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7959506820890488e-05,
+      "loss": 2.005,
+      "step": 9450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7959076617086094e-05,
+      "loss": 2.0072,
+      "step": 9451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7958646373089567e-05,
+      "loss": 1.9653,
+      "step": 9452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795821608890308e-05,
+      "loss": 2.0095,
+      "step": 9453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7957785764528813e-05,
+      "loss": 2.0045,
+      "step": 9454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7957355399968926e-05,
+      "loss": 2.0091,
+      "step": 9455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7956924995225596e-05,
+      "loss": 1.9851,
+      "step": 9456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7956494550301006e-05,
+      "loss": 1.9992,
+      "step": 9457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795606406519732e-05,
+      "loss": 2.0077,
+      "step": 9458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7955633539916712e-05,
+      "loss": 1.9926,
+      "step": 9459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7955202974461362e-05,
+      "loss": 2.0012,
+      "step": 9460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795477236883344e-05,
+      "loss": 1.9776,
+      "step": 9461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7954341723035117e-05,
+      "loss": 1.9995,
+      "step": 9462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795391103706858e-05,
+      "loss": 1.9928,
+      "step": 9463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7953480310935992e-05,
+      "loss": 1.9868,
+      "step": 9464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7953049544639532e-05,
+      "loss": 1.9863,
+      "step": 9465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7952618738181374e-05,
+      "loss": 2.0002,
+      "step": 9466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7952187891563697e-05,
+      "loss": 2.0214,
+      "step": 9467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7951757004788676e-05,
+      "loss": 2.0016,
+      "step": 9468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7951326077858482e-05,
+      "loss": 1.9754,
+      "step": 9469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79508951107753e-05,
+      "loss": 2.001,
+      "step": 9470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7950464103541292e-05,
+      "loss": 1.9851,
+      "step": 9471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795003305615865e-05,
+      "loss": 1.9886,
+      "step": 9472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7949601968629546e-05,
+      "loss": 2.001,
+      "step": 9473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7949170840956152e-05,
+      "loss": 1.9952,
+      "step": 9474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794873967314065e-05,
+      "loss": 1.9953,
+      "step": 9475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7948308465185213e-05,
+      "loss": 1.9881,
+      "step": 9476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7947877217092022e-05,
+      "loss": 1.9956,
+      "step": 9477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7947445928863255e-05,
+      "loss": 1.9891,
+      "step": 9478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794701460050109e-05,
+      "loss": 2.0108,
+      "step": 9479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79465832320077e-05,
+      "loss": 1.998,
+      "step": 9480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7946151823385272e-05,
+      "loss": 2.0068,
+      "step": 9481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794572037463598e-05,
+      "loss": 1.9994,
+      "step": 9482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7945288885761997e-05,
+      "loss": 1.9963,
+      "step": 9483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7944857356765514e-05,
+      "loss": 1.9958,
+      "step": 9484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79444257876487e-05,
+      "loss": 1.9872,
+      "step": 9485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7943994178413734e-05,
+      "loss": 1.9871,
+      "step": 9486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7943562529062804e-05,
+      "loss": 2.0028,
+      "step": 9487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7943130839598088e-05,
+      "loss": 1.9722,
+      "step": 9488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7942699110021764e-05,
+      "loss": 1.9783,
+      "step": 9489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7942267340336008e-05,
+      "loss": 1.9982,
+      "step": 9490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7941835530543004e-05,
+      "loss": 1.9946,
+      "step": 9491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7941403680644935e-05,
+      "loss": 1.9913,
+      "step": 9492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7940971790643976e-05,
+      "loss": 2.0115,
+      "step": 9493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7940539860542312e-05,
+      "loss": 1.9827,
+      "step": 9494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7940107890342125e-05,
+      "loss": 1.9872,
+      "step": 9495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7939675880045594e-05,
+      "loss": 2.0031,
+      "step": 9496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79392438296549e-05,
+      "loss": 1.996,
+      "step": 9497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793881173917223e-05,
+      "loss": 2.0175,
+      "step": 9498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793837960859976e-05,
+      "loss": 1.983,
+      "step": 9499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7937947437939674e-05,
+      "loss": 2.0054,
+      "step": 9500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793751522719416e-05,
+      "loss": 1.9947,
+      "step": 9501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7937082976365387e-05,
+      "loss": 2.0064,
+      "step": 9502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793665068545555e-05,
+      "loss": 2.0014,
+      "step": 9503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7936218354466828e-05,
+      "loss": 1.9835,
+      "step": 9504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7935785983401404e-05,
+      "loss": 1.9917,
+      "step": 9505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7935353572261462e-05,
+      "loss": 1.9805,
+      "step": 9506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7934921121049185e-05,
+      "loss": 1.9859,
+      "step": 9507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793448862976676e-05,
+      "loss": 2.0033,
+      "step": 9508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7934056098416366e-05,
+      "loss": 1.9791,
+      "step": 9509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7933623527000186e-05,
+      "loss": 2.0024,
+      "step": 9510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7933190915520415e-05,
+      "loss": 1.9943,
+      "step": 9511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7932758263979225e-05,
+      "loss": 1.986,
+      "step": 9512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793232557237881e-05,
+      "loss": 2.0022,
+      "step": 9513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793189284072135e-05,
+      "loss": 1.9746,
+      "step": 9514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793146006900903e-05,
+      "loss": 1.9891,
+      "step": 9515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793102725724404e-05,
+      "loss": 1.9922,
+      "step": 9516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793059440542856e-05,
+      "loss": 2.0089,
+      "step": 9517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793016151356478e-05,
+      "loss": 1.9844,
+      "step": 9518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7929728581654884e-05,
+      "loss": 1.9986,
+      "step": 9519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7929295609701062e-05,
+      "loss": 2.0063,
+      "step": 9520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7928862597705496e-05,
+      "loss": 2.0012,
+      "step": 9521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792842954567037e-05,
+      "loss": 1.9854,
+      "step": 9522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7927996453597877e-05,
+      "loss": 1.9833,
+      "step": 9523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79275633214902e-05,
+      "loss": 1.996,
+      "step": 9524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792713014934953e-05,
+      "loss": 1.9782,
+      "step": 9525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792669693717805e-05,
+      "loss": 2.0061,
+      "step": 9526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7926263684977954e-05,
+      "loss": 1.9973,
+      "step": 9527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7925830392751423e-05,
+      "loss": 1.9821,
+      "step": 9528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792539706050065e-05,
+      "loss": 2.0,
+      "step": 9529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792496368822782e-05,
+      "loss": 1.9977,
+      "step": 9530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7924530275935124e-05,
+      "loss": 1.97,
+      "step": 9531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7924096823624747e-05,
+      "loss": 1.9631,
+      "step": 9532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7923663331298883e-05,
+      "loss": 1.9925,
+      "step": 9533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792322979895972e-05,
+      "loss": 2.002,
+      "step": 9534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792279622660944e-05,
+      "loss": 1.981,
+      "step": 9535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7922362614250244e-05,
+      "loss": 1.9907,
+      "step": 9536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792192896188431e-05,
+      "loss": 2.0027,
+      "step": 9537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7921495269513836e-05,
+      "loss": 1.9962,
+      "step": 9538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792106153714101e-05,
+      "loss": 1.9997,
+      "step": 9539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7920627764768026e-05,
+      "loss": 1.9809,
+      "step": 9540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7920193952397065e-05,
+      "loss": 1.9858,
+      "step": 9541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7919760100030327e-05,
+      "loss": 1.9891,
+      "step": 9542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7919326207669996e-05,
+      "loss": 2.0004,
+      "step": 9543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7918892275318268e-05,
+      "loss": 1.9946,
+      "step": 9544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7918458302977332e-05,
+      "loss": 1.9914,
+      "step": 9545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7918024290649377e-05,
+      "loss": 2.0064,
+      "step": 9546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7917590238336605e-05,
+      "loss": 1.9968,
+      "step": 9547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7917156146041194e-05,
+      "loss": 1.9985,
+      "step": 9548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7916722013765343e-05,
+      "loss": 1.9982,
+      "step": 9549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791628784151125e-05,
+      "loss": 1.995,
+      "step": 9550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7915853629281096e-05,
+      "loss": 1.9905,
+      "step": 9551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791541937707708e-05,
+      "loss": 1.9859,
+      "step": 9552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791498508490139e-05,
+      "loss": 1.9981,
+      "step": 9553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791455075275623e-05,
+      "loss": 2.0026,
+      "step": 9554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7914116380643788e-05,
+      "loss": 1.985,
+      "step": 9555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7913681968566253e-05,
+      "loss": 1.9987,
+      "step": 9556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791324751652582e-05,
+      "loss": 1.9928,
+      "step": 9557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7912813024524685e-05,
+      "loss": 1.985,
+      "step": 9558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7912378492565042e-05,
+      "loss": 1.9734,
+      "step": 9559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7911943920649087e-05,
+      "loss": 1.996,
+      "step": 9560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791150930877901e-05,
+      "loss": 1.9987,
+      "step": 9561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791107465695701e-05,
+      "loss": 1.9936,
+      "step": 9562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7910639965185283e-05,
+      "loss": 1.9854,
+      "step": 9563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7910205233466015e-05,
+      "loss": 2.0017,
+      "step": 9564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790977046180141e-05,
+      "loss": 1.979,
+      "step": 9565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7909335650193665e-05,
+      "loss": 1.9915,
+      "step": 9566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7908900798644968e-05,
+      "loss": 2.0101,
+      "step": 9567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790846590715752e-05,
+      "loss": 1.9915,
+      "step": 9568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7908030975733516e-05,
+      "loss": 2.0038,
+      "step": 9569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7907596004375152e-05,
+      "loss": 1.9931,
+      "step": 9570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7907160993084627e-05,
+      "loss": 1.981,
+      "step": 9571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7906725941864135e-05,
+      "loss": 1.9683,
+      "step": 9572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790629085071587e-05,
+      "loss": 1.9855,
+      "step": 9573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790585571964204e-05,
+      "loss": 1.9411,
+      "step": 9574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7905420548644827e-05,
+      "loss": 2.0056,
+      "step": 9575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790498533772644e-05,
+      "loss": 1.9934,
+      "step": 9576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7904550086889078e-05,
+      "loss": 1.9953,
+      "step": 9577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790411479613493e-05,
+      "loss": 1.9829,
+      "step": 9578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7903679465466196e-05,
+      "loss": 2.0083,
+      "step": 9579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790324409488508e-05,
+      "loss": 1.9776,
+      "step": 9580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7902808684393778e-05,
+      "loss": 1.9848,
+      "step": 9581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7902373233994488e-05,
+      "loss": 1.9893,
+      "step": 9582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790193774368941e-05,
+      "loss": 1.9806,
+      "step": 9583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790150221348074e-05,
+      "loss": 1.9995,
+      "step": 9584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790106664337068e-05,
+      "loss": 1.9981,
+      "step": 9585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790063103336143e-05,
+      "loss": 1.9856,
+      "step": 9586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7900195383455188e-05,
+      "loss": 1.9901,
+      "step": 9587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7899759693654155e-05,
+      "loss": 2.0013,
+      "step": 9588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7899323963960532e-05,
+      "loss": 1.9934,
+      "step": 9589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7898888194376517e-05,
+      "loss": 1.9754,
+      "step": 9590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7898452384904314e-05,
+      "loss": 1.9896,
+      "step": 9591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7898016535546123e-05,
+      "loss": 1.9942,
+      "step": 9592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789758064630414e-05,
+      "loss": 1.9923,
+      "step": 9593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7897144717180573e-05,
+      "loss": 2.0011,
+      "step": 9594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789670874817762e-05,
+      "loss": 1.9515,
+      "step": 9595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7896272739297482e-05,
+      "loss": 1.9942,
+      "step": 9596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789583669054236e-05,
+      "loss": 1.9866,
+      "step": 9597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789540060191446e-05,
+      "loss": 1.9902,
+      "step": 9598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7894964473415983e-05,
+      "loss": 1.9927,
+      "step": 9599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7894528305049127e-05,
+      "loss": 1.9818,
+      "step": 9600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78940920968161e-05,
+      "loss": 1.9915,
+      "step": 9601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7893655848719103e-05,
+      "loss": 2.0,
+      "step": 9602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7893219560760336e-05,
+      "loss": 1.9918,
+      "step": 9603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789278323294201e-05,
+      "loss": 2.0034,
+      "step": 9604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7892346865266318e-05,
+      "loss": 2.0031,
+      "step": 9605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789191045773547e-05,
+      "loss": 1.9728,
+      "step": 9606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7891474010351673e-05,
+      "loss": 1.9899,
+      "step": 9607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789103752311712e-05,
+      "loss": 1.9916,
+      "step": 9608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7890600996034026e-05,
+      "loss": 1.9798,
+      "step": 9609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789016442910459e-05,
+      "loss": 1.9809,
+      "step": 9610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7889727822331017e-05,
+      "loss": 2.0074,
+      "step": 9611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788929117571551e-05,
+      "loss": 1.9917,
+      "step": 9612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7888854489260283e-05,
+      "loss": 1.9798,
+      "step": 9613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7888417762967532e-05,
+      "loss": 2.0044,
+      "step": 9614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7887980996839466e-05,
+      "loss": 2.0032,
+      "step": 9615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7887544190878287e-05,
+      "loss": 2.0007,
+      "step": 9616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7887107345086203e-05,
+      "loss": 1.9741,
+      "step": 9617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7886670459465427e-05,
+      "loss": 1.991,
+      "step": 9618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788623353401815e-05,
+      "loss": 1.9804,
+      "step": 9619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788579656874659e-05,
+      "loss": 1.9754,
+      "step": 9620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7885359563652956e-05,
+      "loss": 1.991,
+      "step": 9621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7884922518739446e-05,
+      "loss": 1.9922,
+      "step": 9622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788448543400827e-05,
+      "loss": 1.9918,
+      "step": 9623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7884048309461636e-05,
+      "loss": 1.9919,
+      "step": 9624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7883611145101754e-05,
+      "loss": 1.9994,
+      "step": 9625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7883173940930823e-05,
+      "loss": 1.9961,
+      "step": 9626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788273669695106e-05,
+      "loss": 1.9846,
+      "step": 9627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788229941316467e-05,
+      "loss": 1.9945,
+      "step": 9628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788186208957386e-05,
+      "loss": 2.003,
+      "step": 9629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788142472618084e-05,
+      "loss": 2.0036,
+      "step": 9630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788098732298782e-05,
+      "loss": 1.9737,
+      "step": 9631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7880549879997003e-05,
+      "loss": 1.9871,
+      "step": 9632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7880112397210606e-05,
+      "loss": 1.9928,
+      "step": 9633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.787967487463083e-05,
+      "loss": 1.9698,
+      "step": 9634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.787923731225989e-05,
+      "loss": 1.9987,
+      "step": 9635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7878799710099995e-05,
+      "loss": 1.967,
+      "step": 9636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7878362068153353e-05,
+      "loss": 1.9775,
+      "step": 9637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7877924386422173e-05,
+      "loss": 1.9832,
+      "step": 9638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.787748666490867e-05,
+      "loss": 1.9996,
+      "step": 9639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.787704890361505e-05,
+      "loss": 1.9752,
+      "step": 9640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7876611102543527e-05,
+      "loss": 1.9745,
+      "step": 9641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7876173261696307e-05,
+      "loss": 1.9844,
+      "step": 9642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7875735381075605e-05,
+      "loss": 1.9886,
+      "step": 9643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7875297460683632e-05,
+      "loss": 1.9953,
+      "step": 9644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78748595005226e-05,
+      "loss": 1.9998,
+      "step": 9645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7874421500594717e-05,
+      "loss": 1.9927,
+      "step": 9646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78739834609022e-05,
+      "loss": 2.0033,
+      "step": 9647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7873545381447255e-05,
+      "loss": 2.0278,
+      "step": 9648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78731072622321e-05,
+      "loss": 1.9986,
+      "step": 9649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.787266910325894e-05,
+      "loss": 2.0002,
+      "step": 9650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7872230904529998e-05,
+      "loss": 1.9823,
+      "step": 9651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7871792666047477e-05,
+      "loss": 2.009,
+      "step": 9652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78713543878136e-05,
+      "loss": 1.9722,
+      "step": 9653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.787091606983057e-05,
+      "loss": 1.979,
+      "step": 9654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7870477712100607e-05,
+      "loss": 1.9887,
+      "step": 9655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7870039314625923e-05,
+      "loss": 1.9899,
+      "step": 9656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.786960087740873e-05,
+      "loss": 2.0075,
+      "step": 9657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7869162400451246e-05,
+      "loss": 1.9934,
+      "step": 9658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.786872388375568e-05,
+      "loss": 2.0036,
+      "step": 9659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7868285327324253e-05,
+      "loss": 1.9936,
+      "step": 9660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7867846731159173e-05,
+      "loss": 1.9769,
+      "step": 9661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7867408095262658e-05,
+      "loss": 1.988,
+      "step": 9662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7866969419636923e-05,
+      "loss": 1.9934,
+      "step": 9663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7866530704284183e-05,
+      "loss": 1.9929,
+      "step": 9664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7866091949206653e-05,
+      "loss": 1.9741,
+      "step": 9665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.786565315440655e-05,
+      "loss": 1.9831,
+      "step": 9666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7865214319886087e-05,
+      "loss": 1.9918,
+      "step": 9667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7864775445647487e-05,
+      "loss": 1.9782,
+      "step": 9668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7864336531692955e-05,
+      "loss": 1.987,
+      "step": 9669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7863897578024718e-05,
+      "loss": 2.0095,
+      "step": 9670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7863458584644987e-05,
+      "loss": 1.9911,
+      "step": 9671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.786301955155598e-05,
+      "loss": 1.9865,
+      "step": 9672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7862580478759914e-05,
+      "loss": 1.9624,
+      "step": 9673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7862141366259007e-05,
+      "loss": 1.9938,
+      "step": 9674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7861702214055476e-05,
+      "loss": 1.9809,
+      "step": 9675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.786126302215154e-05,
+      "loss": 1.9942,
+      "step": 9676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7860823790549412e-05,
+      "loss": 2.0042,
+      "step": 9677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7860384519251318e-05,
+      "loss": 1.997,
+      "step": 9678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785994520825947e-05,
+      "loss": 1.999,
+      "step": 9679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7859505857576086e-05,
+      "loss": 1.983,
+      "step": 9680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785906646720339e-05,
+      "loss": 1.9938,
+      "step": 9681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7858627037143597e-05,
+      "loss": 1.9957,
+      "step": 9682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7858187567398927e-05,
+      "loss": 2.0021,
+      "step": 9683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7857748057971597e-05,
+      "loss": 1.9972,
+      "step": 9684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785730850886383e-05,
+      "loss": 1.9995,
+      "step": 9685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7856868920077844e-05,
+      "loss": 1.9934,
+      "step": 9686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785642929161586e-05,
+      "loss": 1.9965,
+      "step": 9687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7855989623480098e-05,
+      "loss": 1.9734,
+      "step": 9688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7855549915672773e-05,
+      "loss": 1.9901,
+      "step": 9689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7855110168196114e-05,
+      "loss": 1.9997,
+      "step": 9690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7854670381052335e-05,
+      "loss": 1.9814,
+      "step": 9691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785423055424366e-05,
+      "loss": 1.977,
+      "step": 9692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785379068777231e-05,
+      "loss": 2.0045,
+      "step": 9693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7853350781640507e-05,
+      "loss": 1.9947,
+      "step": 9694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7852910835850466e-05,
+      "loss": 1.9891,
+      "step": 9695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785247085040442e-05,
+      "loss": 2.0014,
+      "step": 9696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7852030825304584e-05,
+      "loss": 1.9957,
+      "step": 9697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7851590760553173e-05,
+      "loss": 2.0102,
+      "step": 9698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7851150656152425e-05,
+      "loss": 1.9915,
+      "step": 9699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7850710512104552e-05,
+      "loss": 2.0052,
+      "step": 9700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785027032841178e-05,
+      "loss": 1.9871,
+      "step": 9701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784983010507633e-05,
+      "loss": 1.9972,
+      "step": 9702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7849389842100422e-05,
+      "loss": 1.9664,
+      "step": 9703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784894953948629e-05,
+      "loss": 1.9872,
+      "step": 9704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7848509197236147e-05,
+      "loss": 1.9773,
+      "step": 9705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7848068815352223e-05,
+      "loss": 2.0005,
+      "step": 9706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7847628393836736e-05,
+      "loss": 1.9874,
+      "step": 9707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7847187932691917e-05,
+      "loss": 1.9985,
+      "step": 9708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784674743191998e-05,
+      "loss": 2.0045,
+      "step": 9709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784630689152316e-05,
+      "loss": 1.986,
+      "step": 9710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784586631150368e-05,
+      "loss": 2.0069,
+      "step": 9711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7845425691863764e-05,
+      "loss": 2.0031,
+      "step": 9712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7844985032605627e-05,
+      "loss": 2.0045,
+      "step": 9713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7844544333731512e-05,
+      "loss": 1.9951,
+      "step": 9714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784410359524363e-05,
+      "loss": 1.9822,
+      "step": 9715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7843662817144213e-05,
+      "loss": 1.9765,
+      "step": 9716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784322199943549e-05,
+      "loss": 1.9852,
+      "step": 9717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7842781142119677e-05,
+      "loss": 1.9971,
+      "step": 9718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784234024519901e-05,
+      "loss": 1.9937,
+      "step": 9719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7841899308675706e-05,
+      "loss": 1.9894,
+      "step": 9720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7841458332552003e-05,
+      "loss": 1.9782,
+      "step": 9721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7841017316830123e-05,
+      "loss": 1.9972,
+      "step": 9722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784057626151229e-05,
+      "loss": 1.9968,
+      "step": 9723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7840135166600733e-05,
+      "loss": 1.9713,
+      "step": 9724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7839694032097678e-05,
+      "loss": 1.984,
+      "step": 9725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7839252858005357e-05,
+      "loss": 1.9819,
+      "step": 9726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7838811644326e-05,
+      "loss": 1.9782,
+      "step": 9727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783837039106182e-05,
+      "loss": 2.017,
+      "step": 9728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7837929098215063e-05,
+      "loss": 1.9936,
+      "step": 9729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783748776578795e-05,
+      "loss": 2.0062,
+      "step": 9730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7837046393782708e-05,
+      "loss": 1.9813,
+      "step": 9731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783660498220157e-05,
+      "loss": 1.9701,
+      "step": 9732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7836163531046763e-05,
+      "loss": 1.9888,
+      "step": 9733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7835722040320513e-05,
+      "loss": 1.9707,
+      "step": 9734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7835280510025054e-05,
+      "loss": 1.9797,
+      "step": 9735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7834838940162615e-05,
+      "loss": 2.0082,
+      "step": 9736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7834397330735424e-05,
+      "loss": 1.9925,
+      "step": 9737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7833955681745715e-05,
+      "loss": 2.0025,
+      "step": 9738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7833513993195713e-05,
+      "loss": 1.9915,
+      "step": 9739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783307226508765e-05,
+      "loss": 1.9675,
+      "step": 9740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783263049742376e-05,
+      "loss": 1.9954,
+      "step": 9741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783218869020627e-05,
+      "loss": 1.9846,
+      "step": 9742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783174684343741e-05,
+      "loss": 2.0011,
+      "step": 9743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7831304957119418e-05,
+      "loss": 2.0064,
+      "step": 9744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783086303125452e-05,
+      "loss": 1.9973,
+      "step": 9745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783042106584495e-05,
+      "loss": 1.98,
+      "step": 9746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7829979060892934e-05,
+      "loss": 2.0113,
+      "step": 9747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7829537016400713e-05,
+      "loss": 1.969,
+      "step": 9748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7829094932370516e-05,
+      "loss": 1.9867,
+      "step": 9749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782865280880457e-05,
+      "loss": 1.9993,
+      "step": 9750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7828210645705115e-05,
+      "loss": 2.0028,
+      "step": 9751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782776844307438e-05,
+      "loss": 1.9796,
+      "step": 9752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7827326200914598e-05,
+      "loss": 1.9781,
+      "step": 9753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782688391922801e-05,
+      "loss": 1.9859,
+      "step": 9754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7826441598016836e-05,
+      "loss": 1.9878,
+      "step": 9755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7825999237283316e-05,
+      "loss": 2.0161,
+      "step": 9756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782555683702969e-05,
+      "loss": 1.991,
+      "step": 9757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782511439725818e-05,
+      "loss": 1.9814,
+      "step": 9758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7824671917971028e-05,
+      "loss": 2.0151,
+      "step": 9759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782422939917047e-05,
+      "loss": 1.9832,
+      "step": 9760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7823786840858733e-05,
+      "loss": 1.9923,
+      "step": 9761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7823344243038062e-05,
+      "loss": 1.9975,
+      "step": 9762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7822901605710683e-05,
+      "loss": 1.9936,
+      "step": 9763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7822458928878837e-05,
+      "loss": 2.0127,
+      "step": 9764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7822016212544758e-05,
+      "loss": 2.0088,
+      "step": 9765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7821573456710677e-05,
+      "loss": 1.991,
+      "step": 9766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7821130661378836e-05,
+      "loss": 1.9765,
+      "step": 9767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782068782655147e-05,
+      "loss": 2.003,
+      "step": 9768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782024495223081e-05,
+      "loss": 1.9648,
+      "step": 9769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78198020384191e-05,
+      "loss": 1.9912,
+      "step": 9770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7819359085118573e-05,
+      "loss": 1.9784,
+      "step": 9771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7818916092331467e-05,
+      "loss": 1.9906,
+      "step": 9772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7818473060060015e-05,
+      "loss": 1.9648,
+      "step": 9773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.781802998830646e-05,
+      "loss": 1.9841,
+      "step": 9774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7817586877073037e-05,
+      "loss": 2.0114,
+      "step": 9775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.781714372636198e-05,
+      "loss": 2.0232,
+      "step": 9776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7816700536175533e-05,
+      "loss": 1.9821,
+      "step": 9777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.781625730651593e-05,
+      "loss": 1.996,
+      "step": 9778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7815814037385413e-05,
+      "loss": 2.0078,
+      "step": 9779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7815370728786216e-05,
+      "loss": 2.001,
+      "step": 9780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.781492738072058e-05,
+      "loss": 1.9985,
+      "step": 9781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7814483993190743e-05,
+      "loss": 1.9889,
+      "step": 9782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7814040566198945e-05,
+      "loss": 2.0026,
+      "step": 9783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7813597099747423e-05,
+      "loss": 1.9994,
+      "step": 9784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.781315359383842e-05,
+      "loss": 1.9838,
+      "step": 9785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7812710048474175e-05,
+      "loss": 2.0021,
+      "step": 9786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7812266463656924e-05,
+      "loss": 2.0187,
+      "step": 9787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7811822839388913e-05,
+      "loss": 1.9927,
+      "step": 9788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7811379175672376e-05,
+      "loss": 1.9896,
+      "step": 9789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7810935472509555e-05,
+      "loss": 2.017,
+      "step": 9790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7810491729902697e-05,
+      "loss": 1.9973,
+      "step": 9791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7810047947854035e-05,
+      "loss": 1.975,
+      "step": 9792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7809604126365814e-05,
+      "loss": 1.9908,
+      "step": 9793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7809160265440274e-05,
+      "loss": 1.9932,
+      "step": 9794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7808716365079655e-05,
+      "loss": 2.004,
+      "step": 9795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78082724252862e-05,
+      "loss": 1.9817,
+      "step": 9796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7807828446062152e-05,
+      "loss": 1.9925,
+      "step": 9797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780738442740975e-05,
+      "loss": 1.9943,
+      "step": 9798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780694036933124e-05,
+      "loss": 1.9804,
+      "step": 9799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7806496271828863e-05,
+      "loss": 1.9911,
+      "step": 9800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7806052134904865e-05,
+      "loss": 2.0029,
+      "step": 9801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7805607958561477e-05,
+      "loss": 2.0089,
+      "step": 9802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7805163742800955e-05,
+      "loss": 1.9782,
+      "step": 9803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7804719487625536e-05,
+      "loss": 1.9891,
+      "step": 9804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7804275193037467e-05,
+      "loss": 1.9939,
+      "step": 9805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7803830859038988e-05,
+      "loss": 1.9802,
+      "step": 9806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7803386485632342e-05,
+      "loss": 2.0072,
+      "step": 9807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780294207281978e-05,
+      "loss": 1.974,
+      "step": 9808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7802497620603537e-05,
+      "loss": 1.9864,
+      "step": 9809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7802053128985864e-05,
+      "loss": 2.0094,
+      "step": 9810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7801608597969002e-05,
+      "loss": 1.9771,
+      "step": 9811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78011640275552e-05,
+      "loss": 1.9687,
+      "step": 9812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7800719417746698e-05,
+      "loss": 1.9861,
+      "step": 9813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7800274768545745e-05,
+      "loss": 1.961,
+      "step": 9814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7799830079954585e-05,
+      "loss": 1.9746,
+      "step": 9815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779938535197546e-05,
+      "loss": 1.991,
+      "step": 9816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7798940584610623e-05,
+      "loss": 1.9779,
+      "step": 9817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7798495777862318e-05,
+      "loss": 2.0086,
+      "step": 9818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7798050931732788e-05,
+      "loss": 1.987,
+      "step": 9819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779760604622428e-05,
+      "loss": 1.9736,
+      "step": 9820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7797161121339038e-05,
+      "loss": 1.9848,
+      "step": 9821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7796716157079318e-05,
+      "loss": 2.0058,
+      "step": 9822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779627115344736e-05,
+      "loss": 1.9898,
+      "step": 9823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779582611044541e-05,
+      "loss": 1.9753,
+      "step": 9824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779538102807572e-05,
+      "loss": 1.9688,
+      "step": 9825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7794935906340534e-05,
+      "loss": 1.9749,
+      "step": 9826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7794490745242104e-05,
+      "loss": 1.9901,
+      "step": 9827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7794045544782676e-05,
+      "loss": 1.977,
+      "step": 9828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7793600304964494e-05,
+      "loss": 1.9955,
+      "step": 9829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7793155025789813e-05,
+      "loss": 1.9885,
+      "step": 9830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7792709707260877e-05,
+      "loss": 1.9937,
+      "step": 9831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7792264349379936e-05,
+      "loss": 1.981,
+      "step": 9832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7791818952149242e-05,
+      "loss": 1.9929,
+      "step": 9833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779137351557104e-05,
+      "loss": 2.0202,
+      "step": 9834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779092803964758e-05,
+      "loss": 1.9767,
+      "step": 9835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7790482524381116e-05,
+      "loss": 1.9853,
+      "step": 9836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779003696977389e-05,
+      "loss": 1.9915,
+      "step": 9837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7789591375828158e-05,
+      "loss": 2.0067,
+      "step": 9838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7789145742546167e-05,
+      "loss": 1.9875,
+      "step": 9839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778870006993017e-05,
+      "loss": 1.9908,
+      "step": 9840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7788254357982417e-05,
+      "loss": 1.9912,
+      "step": 9841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7787808606705158e-05,
+      "loss": 1.9988,
+      "step": 9842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7787362816100644e-05,
+      "loss": 2.0165,
+      "step": 9843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7786916986171124e-05,
+      "loss": 2.0108,
+      "step": 9844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7786471116918855e-05,
+      "loss": 1.9934,
+      "step": 9845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7786025208346084e-05,
+      "loss": 1.9835,
+      "step": 9846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7785579260455066e-05,
+      "loss": 1.9871,
+      "step": 9847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778513327324805e-05,
+      "loss": 1.9908,
+      "step": 9848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7784687246727288e-05,
+      "loss": 1.992,
+      "step": 9849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7784241180895033e-05,
+      "loss": 2.0125,
+      "step": 9850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778379507575354e-05,
+      "loss": 1.9879,
+      "step": 9851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778334893130506e-05,
+      "loss": 1.9955,
+      "step": 9852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7782902747551842e-05,
+      "loss": 1.9939,
+      "step": 9853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7782456524496148e-05,
+      "loss": 1.9905,
+      "step": 9854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7782010262140225e-05,
+      "loss": 2.0039,
+      "step": 9855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7781563960486327e-05,
+      "loss": 1.9771,
+      "step": 9856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778111761953671e-05,
+      "loss": 1.9973,
+      "step": 9857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7780671239293626e-05,
+      "loss": 1.9943,
+      "step": 9858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778022481975933e-05,
+      "loss": 1.9926,
+      "step": 9859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7779778360936072e-05,
+      "loss": 2.0018,
+      "step": 9860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7779331862826116e-05,
+      "loss": 1.9804,
+      "step": 9861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777888532543171e-05,
+      "loss": 1.9819,
+      "step": 9862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7778438748755112e-05,
+      "loss": 2.0101,
+      "step": 9863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777799213279857e-05,
+      "loss": 2.0067,
+      "step": 9864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777754547756435e-05,
+      "loss": 2.0106,
+      "step": 9865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77770987830547e-05,
+      "loss": 1.9813,
+      "step": 9866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777665204927188e-05,
+      "loss": 1.9507,
+      "step": 9867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7776205276218143e-05,
+      "loss": 1.9978,
+      "step": 9868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7775758463895746e-05,
+      "loss": 1.9815,
+      "step": 9869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7775311612306947e-05,
+      "loss": 2.0046,
+      "step": 9870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7774864721454e-05,
+      "loss": 1.986,
+      "step": 9871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7774417791339162e-05,
+      "loss": 1.9989,
+      "step": 9872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7773970821964693e-05,
+      "loss": 2.0067,
+      "step": 9873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777352381333285e-05,
+      "loss": 1.9955,
+      "step": 9874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7773076765445882e-05,
+      "loss": 1.9814,
+      "step": 9875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7772629678306054e-05,
+      "loss": 1.9738,
+      "step": 9876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7772182551915626e-05,
+      "loss": 1.9855,
+      "step": 9877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777173538627685e-05,
+      "loss": 1.9864,
+      "step": 9878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7771288181391988e-05,
+      "loss": 1.9905,
+      "step": 9879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7770840937263293e-05,
+      "loss": 2.0045,
+      "step": 9880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7770393653893033e-05,
+      "loss": 1.9864,
+      "step": 9881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7769946331283456e-05,
+      "loss": 1.9848,
+      "step": 9882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776949896943683e-05,
+      "loss": 1.9987,
+      "step": 9883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776905156835541e-05,
+      "loss": 2.009,
+      "step": 9884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776860412804145e-05,
+      "loss": 1.9877,
+      "step": 9885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776815664849722e-05,
+      "loss": 1.9996,
+      "step": 9886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7767709129724974e-05,
+      "loss": 1.9938,
+      "step": 9887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776726157172697e-05,
+      "loss": 1.9917,
+      "step": 9888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7766813974505473e-05,
+      "loss": 2.0032,
+      "step": 9889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776636633806274e-05,
+      "loss": 1.9939,
+      "step": 9890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776591866240103e-05,
+      "loss": 1.9999,
+      "step": 9891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776547094752261e-05,
+      "loss": 2.0068,
+      "step": 9892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7765023193429738e-05,
+      "loss": 2.0009,
+      "step": 9893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776457540012467e-05,
+      "loss": 1.9989,
+      "step": 9894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7764127567609672e-05,
+      "loss": 1.9897,
+      "step": 9895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7763679695887006e-05,
+      "loss": 1.989,
+      "step": 9896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7763231784958935e-05,
+      "loss": 1.9868,
+      "step": 9897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7762783834827716e-05,
+      "loss": 1.9932,
+      "step": 9898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7762335845495613e-05,
+      "loss": 1.9821,
+      "step": 9899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7761887816964888e-05,
+      "loss": 1.9991,
+      "step": 9900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7761439749237802e-05,
+      "loss": 1.9997,
+      "step": 9901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7760991642316624e-05,
+      "loss": 1.9874,
+      "step": 9902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776054349620361e-05,
+      "loss": 1.9857,
+      "step": 9903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776009531090103e-05,
+      "loss": 1.9945,
+      "step": 9904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775964708641114e-05,
+      "loss": 1.9834,
+      "step": 9905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775919882273621e-05,
+      "loss": 2.0061,
+      "step": 9906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77587505198785e-05,
+      "loss": 1.9924,
+      "step": 9907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775830217784027e-05,
+      "loss": 1.9969,
+      "step": 9908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775785379662379e-05,
+      "loss": 1.9668,
+      "step": 9909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775740537623132e-05,
+      "loss": 2.0122,
+      "step": 9910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775695691666513e-05,
+      "loss": 1.9755,
+      "step": 9911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7756508417927483e-05,
+      "loss": 1.9844,
+      "step": 9912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7756059880020637e-05,
+      "loss": 1.9937,
+      "step": 9913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775561130294687e-05,
+      "loss": 1.9931,
+      "step": 9914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775516268670843e-05,
+      "loss": 1.9898,
+      "step": 9915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77547140313076e-05,
+      "loss": 2.0023,
+      "step": 9916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7754265336746637e-05,
+      "loss": 1.9938,
+      "step": 9917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7753816603027803e-05,
+      "loss": 1.9838,
+      "step": 9918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7753367830153374e-05,
+      "loss": 1.9872,
+      "step": 9919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7752919018125607e-05,
+      "loss": 1.9978,
+      "step": 9920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7752470166946773e-05,
+      "loss": 1.9803,
+      "step": 9921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775202127661914e-05,
+      "loss": 2.0159,
+      "step": 9922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775157234714497e-05,
+      "loss": 1.9947,
+      "step": 9923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7751123378526535e-05,
+      "loss": 1.987,
+      "step": 9924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7750674370766102e-05,
+      "loss": 2.0154,
+      "step": 9925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775022532386593e-05,
+      "loss": 2.0118,
+      "step": 9926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7749776237828297e-05,
+      "loss": 1.9691,
+      "step": 9927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7749327112655467e-05,
+      "loss": 1.9845,
+      "step": 9928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7748877948349708e-05,
+      "loss": 2.0052,
+      "step": 9929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7748428744913287e-05,
+      "loss": 1.9703,
+      "step": 9930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7747979502348474e-05,
+      "loss": 1.9892,
+      "step": 9931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774753022065754e-05,
+      "loss": 1.9923,
+      "step": 9932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7747080899842748e-05,
+      "loss": 1.985,
+      "step": 9933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7746631539906368e-05,
+      "loss": 1.9935,
+      "step": 9934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7746182140850673e-05,
+      "loss": 2.007,
+      "step": 9935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7745732702677933e-05,
+      "loss": 1.9941,
+      "step": 9936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774528322539041e-05,
+      "loss": 2.0024,
+      "step": 9937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7744833708990382e-05,
+      "loss": 1.9826,
+      "step": 9938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7744384153480116e-05,
+      "loss": 1.983,
+      "step": 9939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7743934558861884e-05,
+      "loss": 2.0063,
+      "step": 9940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774348492513795e-05,
+      "loss": 1.9938,
+      "step": 9941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7743035252310595e-05,
+      "loss": 1.9973,
+      "step": 9942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774258554038208e-05,
+      "loss": 1.9921,
+      "step": 9943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774213578935468e-05,
+      "loss": 1.9808,
+      "step": 9944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7741685999230665e-05,
+      "loss": 1.9785,
+      "step": 9945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7741236170012307e-05,
+      "loss": 1.9975,
+      "step": 9946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774078630170188e-05,
+      "loss": 2.0034,
+      "step": 9947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7740336394301652e-05,
+      "loss": 2.0145,
+      "step": 9948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77398864478139e-05,
+      "loss": 1.9735,
+      "step": 9949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7739436462240892e-05,
+      "loss": 1.9936,
+      "step": 9950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7738986437584897e-05,
+      "loss": 2.0044,
+      "step": 9951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7738536373848194e-05,
+      "loss": 1.9944,
+      "step": 9952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7738086271033054e-05,
+      "loss": 1.9884,
+      "step": 9953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773763612914175e-05,
+      "loss": 1.9944,
+      "step": 9954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7737185948176554e-05,
+      "loss": 1.989,
+      "step": 9955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773673572813974e-05,
+      "loss": 1.992,
+      "step": 9956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7736285469033582e-05,
+      "loss": 1.9992,
+      "step": 9957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773583517086035e-05,
+      "loss": 1.9786,
+      "step": 9958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7735384833622324e-05,
+      "loss": 1.9922,
+      "step": 9959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7734934457321777e-05,
+      "loss": 1.9931,
+      "step": 9960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773448404196098e-05,
+      "loss": 1.9967,
+      "step": 9961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773403358754221e-05,
+      "loss": 1.9842,
+      "step": 9962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7733583094067738e-05,
+      "loss": 1.9822,
+      "step": 9963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7733132561539844e-05,
+      "loss": 1.9795,
+      "step": 9964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77326819899608e-05,
+      "loss": 2.0053,
+      "step": 9965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773223137933288e-05,
+      "loss": 1.999,
+      "step": 9966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773178072965837e-05,
+      "loss": 1.9922,
+      "step": 9967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773133004093953e-05,
+      "loss": 1.9882,
+      "step": 9968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7730879313178644e-05,
+      "loss": 2.0024,
+      "step": 9969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7730428546377988e-05,
+      "loss": 2.0049,
+      "step": 9970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772997774053984e-05,
+      "loss": 2.0095,
+      "step": 9971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772952689566647e-05,
+      "loss": 2.0069,
+      "step": 9972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7729076011760165e-05,
+      "loss": 2.0092,
+      "step": 9973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7728625088823192e-05,
+      "loss": 1.9826,
+      "step": 9974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7728174126857834e-05,
+      "loss": 2.0126,
+      "step": 9975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7727723125866364e-05,
+      "loss": 2.0215,
+      "step": 9976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7727272085851064e-05,
+      "loss": 1.9823,
+      "step": 9977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772682100681421e-05,
+      "loss": 2.0038,
+      "step": 9978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7726369888758074e-05,
+      "loss": 1.9845,
+      "step": 9979
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7725918731684944e-05,
+      "loss": 1.9785,
+      "step": 9980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772546753559709e-05,
+      "loss": 1.9949,
+      "step": 9981
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7725016300496797e-05,
+      "loss": 2.0021,
+      "step": 9982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772456502638634e-05,
+      "loss": 1.9993,
+      "step": 9983
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7724113713268e-05,
+      "loss": 1.9815,
+      "step": 9984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7723662361144058e-05,
+      "loss": 1.9912,
+      "step": 9985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7723210970016782e-05,
+      "loss": 1.9798,
+      "step": 9986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7722759539888464e-05,
+      "loss": 2.003,
+      "step": 9987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772230807076138e-05,
+      "loss": 1.993,
+      "step": 9988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7721856562637807e-05,
+      "loss": 1.9976,
+      "step": 9989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772140501552003e-05,
+      "loss": 1.972,
+      "step": 9990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7720953429410322e-05,
+      "loss": 2.0025,
+      "step": 9991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772050180431097e-05,
+      "loss": 2.0023,
+      "step": 9992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772005014022425e-05,
+      "loss": 1.9867,
+      "step": 9993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7719598437152447e-05,
+      "loss": 1.9967,
+      "step": 9994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7719146695097842e-05,
+      "loss": 1.9977,
+      "step": 9995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7718694914062714e-05,
+      "loss": 1.9887,
+      "step": 9996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7718243094049345e-05,
+      "loss": 1.9628,
+      "step": 9997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7717791235060012e-05,
+      "loss": 1.9994,
+      "step": 9998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7717339337097003e-05,
+      "loss": 1.9769,
+      "step": 9999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77168874001626e-05,
+      "loss": 1.9824,
+      "step": 10000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7716435424259085e-05,
+      "loss": 1.9852,
+      "step": 10001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7715983409388738e-05,
+      "loss": 2.004,
+      "step": 10002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7715531355553838e-05,
+      "loss": 1.9771,
+      "step": 10003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771507926275668e-05,
+      "loss": 1.9652,
+      "step": 10004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7714627130999532e-05,
+      "loss": 2.0106,
+      "step": 10005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771417496028469e-05,
+      "loss": 1.9885,
+      "step": 10006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771372275061443e-05,
+      "loss": 1.9771,
+      "step": 10007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7713270501991037e-05,
+      "loss": 2.0158,
+      "step": 10008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7712818214416796e-05,
+      "loss": 2.0082,
+      "step": 10009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771236588789399e-05,
+      "loss": 1.9998,
+      "step": 10010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7711913522424905e-05,
+      "loss": 2.0213,
+      "step": 10011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771146111801182e-05,
+      "loss": 1.9899,
+      "step": 10012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7711008674657025e-05,
+      "loss": 2.0008,
+      "step": 10013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7710556192362808e-05,
+      "loss": 2.0055,
+      "step": 10014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7710103671131445e-05,
+      "loss": 1.9816,
+      "step": 10015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7709651110965223e-05,
+      "loss": 2.0046,
+      "step": 10016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7709198511866433e-05,
+      "loss": 1.9882,
+      "step": 10017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7708745873837358e-05,
+      "loss": 1.9813,
+      "step": 10018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770829319688028e-05,
+      "loss": 1.9931,
+      "step": 10019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770784048099749e-05,
+      "loss": 1.9906,
+      "step": 10020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7707387726191273e-05,
+      "loss": 1.9935,
+      "step": 10021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7706934932463912e-05,
+      "loss": 1.9873,
+      "step": 10022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7706482099817698e-05,
+      "loss": 1.9787,
+      "step": 10023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7706029228254914e-05,
+      "loss": 2.0072,
+      "step": 10024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7705576317777847e-05,
+      "loss": 1.9989,
+      "step": 10025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7705123368388788e-05,
+      "loss": 2.0021,
+      "step": 10026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7704670380090024e-05,
+      "loss": 1.9885,
+      "step": 10027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770421735288384e-05,
+      "loss": 2.0013,
+      "step": 10028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770376428677252e-05,
+      "loss": 2.0024,
+      "step": 10029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7703311181758363e-05,
+      "loss": 2.0239,
+      "step": 10030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7702858037843648e-05,
+      "loss": 1.9962,
+      "step": 10031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7702404855030662e-05,
+      "loss": 1.978,
+      "step": 10032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.77019516333217e-05,
+      "loss": 1.9794,
+      "step": 10033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7701498372719048e-05,
+      "loss": 2.0065,
+      "step": 10034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7701045073224997e-05,
+      "loss": 2.0097,
+      "step": 10035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770059173484183e-05,
+      "loss": 2.0233,
+      "step": 10036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7700138357571842e-05,
+      "loss": 1.9918,
+      "step": 10037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769968494141732e-05,
+      "loss": 1.997,
+      "step": 10038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7699231486380555e-05,
+      "loss": 1.9809,
+      "step": 10039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7698777992463837e-05,
+      "loss": 2.0098,
+      "step": 10040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769832445966945e-05,
+      "loss": 2.0044,
+      "step": 10041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7697870887999696e-05,
+      "loss": 1.9846,
+      "step": 10042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7697417277456857e-05,
+      "loss": 1.9829,
+      "step": 10043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7696963628043223e-05,
+      "loss": 1.9819,
+      "step": 10044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7696509939761093e-05,
+      "loss": 1.9964,
+      "step": 10045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769605621261275e-05,
+      "loss": 1.9914,
+      "step": 10046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7695602446600486e-05,
+      "loss": 2.0133,
+      "step": 10047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.76951486417266e-05,
+      "loss": 1.9818,
+      "step": 10048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769469479799337e-05,
+      "loss": 1.9865,
+      "step": 10049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.76942409154031e-05,
+      "loss": 1.9963,
+      "step": 10050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7693786993958078e-05,
+      "loss": 1.982,
+      "step": 10051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7693333033660597e-05,
+      "loss": 1.9966,
+      "step": 10052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7692879034512947e-05,
+      "loss": 2.0027,
+      "step": 10053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7692424996517423e-05,
+      "loss": 1.9737,
+      "step": 10054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7691970919676316e-05,
+      "loss": 1.9867,
+      "step": 10055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769151680399192e-05,
+      "loss": 1.9756,
+      "step": 10056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7691062649466527e-05,
+      "loss": 1.9802,
+      "step": 10057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7690608456102432e-05,
+      "loss": 1.9968,
+      "step": 10058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769015422390193e-05,
+      "loss": 2.0134,
+      "step": 10059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7689699952867314e-05,
+      "loss": 2.0085,
+      "step": 10060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7689245643000874e-05,
+      "loss": 1.9933,
+      "step": 10061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768879129430491e-05,
+      "loss": 1.9841,
+      "step": 10062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7688336906781713e-05,
+      "loss": 1.9959,
+      "step": 10063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7687882480433572e-05,
+      "loss": 1.9791,
+      "step": 10064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7687428015262798e-05,
+      "loss": 1.9912,
+      "step": 10065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768697351127167e-05,
+      "loss": 2.0008,
+      "step": 10066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7686518968462493e-05,
+      "loss": 1.9967,
+      "step": 10067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7686064386837553e-05,
+      "loss": 2.0143,
+      "step": 10068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7685609766399155e-05,
+      "loss": 2.0037,
+      "step": 10069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7685155107149588e-05,
+      "loss": 2.003,
+      "step": 10070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7684700409091157e-05,
+      "loss": 1.983,
+      "step": 10071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7684245672226142e-05,
+      "loss": 2.0128,
+      "step": 10072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7683790896556857e-05,
+      "loss": 1.999,
+      "step": 10073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768333608208559e-05,
+      "loss": 1.996,
+      "step": 10074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768288122881464e-05,
+      "loss": 1.9915,
+      "step": 10075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7682426336746296e-05,
+      "loss": 2.0016,
+      "step": 10076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768197140588287e-05,
+      "loss": 1.9893,
+      "step": 10077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7681516436226644e-05,
+      "loss": 1.9947,
+      "step": 10078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7681061427779925e-05,
+      "loss": 1.9676,
+      "step": 10079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768060638054501e-05,
+      "loss": 2.003,
+      "step": 10080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7680151294524194e-05,
+      "loss": 1.9897,
+      "step": 10081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7679696169719777e-05,
+      "loss": 1.9927,
+      "step": 10082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7679241006134057e-05,
+      "loss": 1.9878,
+      "step": 10083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7678785803769326e-05,
+      "loss": 2.0013,
+      "step": 10084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7678330562627895e-05,
+      "loss": 1.9971,
+      "step": 10085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7677875282712056e-05,
+      "loss": 2.0043,
+      "step": 10086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.767741996402411e-05,
+      "loss": 1.9994,
+      "step": 10087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.767696460656635e-05,
+      "loss": 1.9767,
+      "step": 10088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7676509210341083e-05,
+      "loss": 2.0001,
+      "step": 10089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.767605377535061e-05,
+      "loss": 1.9902,
+      "step": 10090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7675598301597223e-05,
+      "loss": 2.0069,
+      "step": 10091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.767514278908323e-05,
+      "loss": 1.9721,
+      "step": 10092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7674687237810924e-05,
+      "loss": 1.9896,
+      "step": 10093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7674231647782608e-05,
+      "loss": 2.0089,
+      "step": 10094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7673776019000587e-05,
+      "loss": 2.0051,
+      "step": 10095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7673320351467157e-05,
+      "loss": 1.9721,
+      "step": 10096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7672864645184623e-05,
+      "loss": 1.9914,
+      "step": 10097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.767240890015528e-05,
+      "loss": 1.9899,
+      "step": 10098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7671953116381433e-05,
+      "loss": 2.013,
+      "step": 10099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7671497293865386e-05,
+      "loss": 2.0046,
+      "step": 10100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.767104143260944e-05,
+      "loss": 1.9884,
+      "step": 10101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7670585532615895e-05,
+      "loss": 1.9955,
+      "step": 10102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7670129593887052e-05,
+      "loss": 1.9845,
+      "step": 10103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7669673616425216e-05,
+      "loss": 1.9871,
+      "step": 10104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.766921760023269e-05,
+      "loss": 1.9799,
+      "step": 10105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7668761545311774e-05,
+      "loss": 1.9989,
+      "step": 10106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7668305451664776e-05,
+      "loss": 1.9946,
+      "step": 10107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7667849319293994e-05,
+      "loss": 1.9931,
+      "step": 10108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7667393148201733e-05,
+      "loss": 1.9924,
+      "step": 10109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7666936938390296e-05,
+      "loss": 2.0014,
+      "step": 10110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7666480689861992e-05,
+      "loss": 2.003,
+      "step": 10111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7666024402619116e-05,
+      "loss": 2.0022,
+      "step": 10112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7665568076663978e-05,
+      "loss": 1.9944,
+      "step": 10113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7665111711998885e-05,
+      "loss": 2.0114,
+      "step": 10114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.766465530862613e-05,
+      "loss": 1.999,
+      "step": 10115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7664198866548035e-05,
+      "loss": 1.9964,
+      "step": 10116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.766374238576689e-05,
+      "loss": 1.9908,
+      "step": 10117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.766328586628501e-05,
+      "loss": 1.9789,
+      "step": 10118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7662829308104693e-05,
+      "loss": 2.0141,
+      "step": 10119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7662372711228245e-05,
+      "loss": 1.9742,
+      "step": 10120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.766191607565798e-05,
+      "loss": 1.9936,
+      "step": 10121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7661459401396196e-05,
+      "loss": 1.9999,
+      "step": 10122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7661002688445203e-05,
+      "loss": 1.9872,
+      "step": 10123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7660545936807304e-05,
+      "loss": 1.9901,
+      "step": 10124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7660089146484808e-05,
+      "loss": 1.9979,
+      "step": 10125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.765963231748002e-05,
+      "loss": 1.9908,
+      "step": 10126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.765917544979525e-05,
+      "loss": 1.9724,
+      "step": 10127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7658718543432805e-05,
+      "loss": 1.9942,
+      "step": 10128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7658261598394987e-05,
+      "loss": 1.9931,
+      "step": 10129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7657804614684105e-05,
+      "loss": 1.9937,
+      "step": 10130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7657347592302476e-05,
+      "loss": 1.9717,
+      "step": 10131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7656890531252393e-05,
+      "loss": 1.9824,
+      "step": 10132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7656433431536174e-05,
+      "loss": 1.9869,
+      "step": 10133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7655976293156127e-05,
+      "loss": 1.9959,
+      "step": 10134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7655519116114555e-05,
+      "loss": 1.9734,
+      "step": 10135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7655061900413773e-05,
+      "loss": 1.995,
+      "step": 10136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7654604646056088e-05,
+      "loss": 2.0221,
+      "step": 10137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7654147353043806e-05,
+      "loss": 1.9916,
+      "step": 10138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7653690021379236e-05,
+      "loss": 1.9914,
+      "step": 10139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7653232651064692e-05,
+      "loss": 2.0044,
+      "step": 10140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7652775242102484e-05,
+      "loss": 1.974,
+      "step": 10141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7652317794494912e-05,
+      "loss": 1.9984,
+      "step": 10142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7651860308244302e-05,
+      "loss": 1.9855,
+      "step": 10143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.765140278335295e-05,
+      "loss": 2.005,
+      "step": 10144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7650945219823173e-05,
+      "loss": 1.993,
+      "step": 10145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.765048761765728e-05,
+      "loss": 1.9858,
+      "step": 10146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7650029976857587e-05,
+      "loss": 1.9871,
+      "step": 10147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7649572297426395e-05,
+      "loss": 1.9915,
+      "step": 10148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.764911457936602e-05,
+      "loss": 1.9794,
+      "step": 10149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7648656822678777e-05,
+      "loss": 1.9823,
+      "step": 10150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7648199027366973e-05,
+      "loss": 1.9875,
+      "step": 10151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7647741193432922e-05,
+      "loss": 2.0026,
+      "step": 10152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7647283320878937e-05,
+      "loss": 1.9849,
+      "step": 10153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.764682540970733e-05,
+      "loss": 2.0079,
+      "step": 10154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7646367459920405e-05,
+      "loss": 1.9989,
+      "step": 10155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7645909471520488e-05,
+      "loss": 1.9851,
+      "step": 10156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7645451444509882e-05,
+      "loss": 1.9982,
+      "step": 10157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7644993378890904e-05,
+      "loss": 1.9859,
+      "step": 10158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.764453527466587e-05,
+      "loss": 1.9879,
+      "step": 10159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7644077131837087e-05,
+      "loss": 1.9903,
+      "step": 10160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7643618950406872e-05,
+      "loss": 1.9775,
+      "step": 10161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7643160730377535e-05,
+      "loss": 2.0001,
+      "step": 10162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7642702471751393e-05,
+      "loss": 1.9941,
+      "step": 10163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.764224417453077e-05,
+      "loss": 1.9916,
+      "step": 10164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7641785838717957e-05,
+      "loss": 1.9959,
+      "step": 10165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7641327464315286e-05,
+      "loss": 1.9966,
+      "step": 10166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7640869051325075e-05,
+      "loss": 2.0033,
+      "step": 10167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7640410599749626e-05,
+      "loss": 1.9826,
+      "step": 10168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763995210959126e-05,
+      "loss": 1.9886,
+      "step": 10169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763949358085229e-05,
+      "loss": 1.9738,
+      "step": 10170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763903501353504e-05,
+      "loss": 2.0001,
+      "step": 10171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763857640764181e-05,
+      "loss": 2.0072,
+      "step": 10172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7638117763174937e-05,
+      "loss": 1.9937,
+      "step": 10173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7637659080136717e-05,
+      "loss": 2.0118,
+      "step": 10174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7637200358529474e-05,
+      "loss": 2.0065,
+      "step": 10175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7636741598355528e-05,
+      "loss": 1.9894,
+      "step": 10176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763628279961719e-05,
+      "loss": 1.9849,
+      "step": 10177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7635823962316782e-05,
+      "loss": 1.992,
+      "step": 10178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763536508645662e-05,
+      "loss": 1.9928,
+      "step": 10179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7634906172039018e-05,
+      "loss": 1.9905,
+      "step": 10180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7634447219066295e-05,
+      "loss": 1.9874,
+      "step": 10181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763398822754077e-05,
+      "loss": 1.9972,
+      "step": 10182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763352919746476e-05,
+      "loss": 1.9815,
+      "step": 10183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7633070128840583e-05,
+      "loss": 1.9816,
+      "step": 10184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7632611021670557e-05,
+      "loss": 1.9872,
+      "step": 10185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7632151875956998e-05,
+      "loss": 1.9814,
+      "step": 10186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763169269170223e-05,
+      "loss": 2.0127,
+      "step": 10187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7631233468908568e-05,
+      "loss": 1.9854,
+      "step": 10188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7630774207578332e-05,
+      "loss": 1.9959,
+      "step": 10189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763031490771384e-05,
+      "loss": 1.9994,
+      "step": 10190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7629855569317413e-05,
+      "loss": 2.02,
+      "step": 10191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7629396192391374e-05,
+      "loss": 1.9836,
+      "step": 10192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7628936776938032e-05,
+      "loss": 2.0017,
+      "step": 10193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7628477322959717e-05,
+      "loss": 2.0079,
+      "step": 10194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7628017830458746e-05,
+      "loss": 1.9925,
+      "step": 10195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.762755829943744e-05,
+      "loss": 1.9908,
+      "step": 10196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.762709872989812e-05,
+      "loss": 2.002,
+      "step": 10197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7626639121843103e-05,
+      "loss": 1.9671,
+      "step": 10198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7626179475274713e-05,
+      "loss": 2.0069,
+      "step": 10199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7625719790195273e-05,
+      "loss": 1.9946,
+      "step": 10200
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 44509,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 3.2246058176204933e+24,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}