diff --git "a/checkpoint-1800/trainer_state.json" "b/checkpoint-1800/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1800/trainer_state.json"
@@ -0,0 +1,10866 @@
+{
+  "best_metric": 1.063901662826538,
+  "best_model_checkpoint": "epoch_musical_instruments_identification_2/checkpoint-1800",
+  "epoch": 5.0,
+  "global_step": 1800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.1932,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 2.1933,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 2.2353,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.1787,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.1998,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.196,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 2.1959,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 2.196,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 2.2003,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.2016,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1e-06,
+      "loss": 2.1455,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.221,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3e-06,
+      "loss": 2.2121,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.2041,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5e-06,
+      "loss": 2.1947,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.1899,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.1889,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.2253,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 2.2034,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.2046,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.2111,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2e-06,
+      "loss": 2.2018,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 2.1965,
+      "step": 23
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.2016,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5e-06,
+      "loss": 2.2339,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6e-06,
+      "loss": 2.1855,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 2.1893,
+      "step": 27
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.2081,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9e-06,
+      "loss": 2.1951,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3e-06,
+      "loss": 2.1838,
+      "step": 30
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 2.2006,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.2097,
+      "step": 32
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 2.1979,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.211,
+      "step": 34
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5e-06,
+      "loss": 2.1876,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.186,
+      "step": 36
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7e-06,
+      "loss": 2.1857,
+      "step": 37
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.2117,
+      "step": 38
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 2.2052,
+      "step": 39
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.1862,
+      "step": 40
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.1e-06,
+      "loss": 2.1854,
+      "step": 41
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.165,
+      "step": 42
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3e-06,
+      "loss": 2.2396,
+      "step": 43
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4e-06,
+      "loss": 2.196,
+      "step": 44
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5e-06,
+      "loss": 2.1991,
+      "step": 45
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.1991,
+      "step": 46
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7e-06,
+      "loss": 2.1753,
+      "step": 47
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.1923,
+      "step": 48
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.2002,
+      "step": 49
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 2.1949,
+      "step": 50
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.997142857142858e-06,
+      "loss": 2.1941,
+      "step": 51
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.994285714285715e-06,
+      "loss": 2.1922,
+      "step": 52
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.991428571428572e-06,
+      "loss": 2.2029,
+      "step": 53
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.988571428571429e-06,
+      "loss": 2.183,
+      "step": 54
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9857142857142855e-06,
+      "loss": 2.1831,
+      "step": 55
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.982857142857143e-06,
+      "loss": 2.1746,
+      "step": 56
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 2.1989,
+      "step": 57
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977142857142857e-06,
+      "loss": 2.1753,
+      "step": 58
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.974285714285715e-06,
+      "loss": 2.163,
+      "step": 59
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971428571428572e-06,
+      "loss": 2.1741,
+      "step": 60
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.968571428571429e-06,
+      "loss": 2.1819,
+      "step": 61
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.965714285714286e-06,
+      "loss": 2.1967,
+      "step": 62
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962857142857143e-06,
+      "loss": 2.1903,
+      "step": 63
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 2.1706,
+      "step": 64
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9571428571428575e-06,
+      "loss": 2.1847,
+      "step": 65
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.954285714285715e-06,
+      "loss": 2.1749,
+      "step": 66
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.951428571428572e-06,
+      "loss": 2.1806,
+      "step": 67
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.948571428571429e-06,
+      "loss": 2.1897,
+      "step": 68
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.945714285714286e-06,
+      "loss": 2.1754,
+      "step": 69
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9428571428571435e-06,
+      "loss": 2.1851,
+      "step": 70
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.94e-06,
+      "loss": 2.1981,
+      "step": 71
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.937142857142858e-06,
+      "loss": 2.1983,
+      "step": 72
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.934285714285715e-06,
+      "loss": 2.1977,
+      "step": 73
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.931428571428572e-06,
+      "loss": 2.1701,
+      "step": 74
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.928571428571429e-06,
+      "loss": 2.1536,
+      "step": 75
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.925714285714286e-06,
+      "loss": 2.193,
+      "step": 76
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.922857142857143e-06,
+      "loss": 2.1881,
+      "step": 77
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.92e-06,
+      "loss": 2.1741,
+      "step": 78
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.917142857142858e-06,
+      "loss": 2.1698,
+      "step": 79
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9142857142857145e-06,
+      "loss": 2.1986,
+      "step": 80
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.911428571428572e-06,
+      "loss": 2.1697,
+      "step": 81
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.90857142857143e-06,
+      "loss": 2.1688,
+      "step": 82
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9057142857142855e-06,
+      "loss": 2.1943,
+      "step": 83
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.902857142857143e-06,
+      "loss": 2.1921,
+      "step": 84
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.1796,
+      "step": 85
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.897142857142857e-06,
+      "loss": 2.1817,
+      "step": 86
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.894285714285715e-06,
+      "loss": 2.144,
+      "step": 87
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.891428571428572e-06,
+      "loss": 2.1604,
+      "step": 88
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.888571428571429e-06,
+      "loss": 2.1456,
+      "step": 89
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.885714285714286e-06,
+      "loss": 2.1786,
+      "step": 90
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.882857142857143e-06,
+      "loss": 2.2043,
+      "step": 91
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 2.1779,
+      "step": 92
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.877142857142857e-06,
+      "loss": 2.1919,
+      "step": 93
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.874285714285715e-06,
+      "loss": 2.1818,
+      "step": 94
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.871428571428572e-06,
+      "loss": 2.1581,
+      "step": 95
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.868571428571429e-06,
+      "loss": 2.1711,
+      "step": 96
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.865714285714287e-06,
+      "loss": 2.1767,
+      "step": 97
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.862857142857143e-06,
+      "loss": 2.1986,
+      "step": 98
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86e-06,
+      "loss": 2.1936,
+      "step": 99
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 2.1835,
+      "step": 100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.854285714285715e-06,
+      "loss": 2.1711,
+      "step": 101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.851428571428572e-06,
+      "loss": 2.1341,
+      "step": 102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.848571428571429e-06,
+      "loss": 2.1733,
+      "step": 103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.845714285714286e-06,
+      "loss": 2.1703,
+      "step": 104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.842857142857143e-06,
+      "loss": 2.1671,
+      "step": 105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.84e-06,
+      "loss": 2.1516,
+      "step": 106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.837142857142858e-06,
+      "loss": 2.1397,
+      "step": 107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8342857142857145e-06,
+      "loss": 2.1844,
+      "step": 108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.831428571428572e-06,
+      "loss": 2.1409,
+      "step": 109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8285714285714295e-06,
+      "loss": 2.121,
+      "step": 110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.825714285714286e-06,
+      "loss": 2.1793,
+      "step": 111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.822857142857143e-06,
+      "loss": 2.1918,
+      "step": 112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 2.1678,
+      "step": 113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.817142857142857e-06,
+      "loss": 2.1384,
+      "step": 114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.814285714285715e-06,
+      "loss": 2.1457,
+      "step": 115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.811428571428572e-06,
+      "loss": 2.164,
+      "step": 116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.808571428571429e-06,
+      "loss": 2.1782,
+      "step": 117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.805714285714286e-06,
+      "loss": 2.1442,
+      "step": 118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.802857142857143e-06,
+      "loss": 2.1097,
+      "step": 119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.129,
+      "step": 120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.797142857142857e-06,
+      "loss": 2.1831,
+      "step": 121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.794285714285715e-06,
+      "loss": 2.1444,
+      "step": 122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7914285714285715e-06,
+      "loss": 2.2004,
+      "step": 123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.788571428571429e-06,
+      "loss": 2.1258,
+      "step": 124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.785714285714287e-06,
+      "loss": 2.1294,
+      "step": 125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.782857142857143e-06,
+      "loss": 2.1206,
+      "step": 126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.78e-06,
+      "loss": 2.1452,
+      "step": 127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7771428571428575e-06,
+      "loss": 2.16,
+      "step": 128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.774285714285715e-06,
+      "loss": 2.1479,
+      "step": 129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.771428571428572e-06,
+      "loss": 2.1557,
+      "step": 130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.768571428571429e-06,
+      "loss": 2.1623,
+      "step": 131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.765714285714286e-06,
+      "loss": 2.1186,
+      "step": 132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7628571428571435e-06,
+      "loss": 2.1335,
+      "step": 133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.76e-06,
+      "loss": 2.0984,
+      "step": 134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.757142857142858e-06,
+      "loss": 2.1283,
+      "step": 135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.754285714285714e-06,
+      "loss": 2.1365,
+      "step": 136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.751428571428572e-06,
+      "loss": 2.0958,
+      "step": 137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7485714285714295e-06,
+      "loss": 2.169,
+      "step": 138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.745714285714286e-06,
+      "loss": 2.0877,
+      "step": 139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.742857142857144e-06,
+      "loss": 2.0932,
+      "step": 140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.74e-06,
+      "loss": 2.094,
+      "step": 141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.737142857142857e-06,
+      "loss": 2.1569,
+      "step": 142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.734285714285715e-06,
+      "loss": 2.0613,
+      "step": 143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.731428571428572e-06,
+      "loss": 2.116,
+      "step": 144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.728571428571429e-06,
+      "loss": 2.0285,
+      "step": 145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.725714285714286e-06,
+      "loss": 2.1351,
+      "step": 146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.722857142857144e-06,
+      "loss": 2.174,
+      "step": 147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 2.1555,
+      "step": 148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.717142857142857e-06,
+      "loss": 2.104,
+      "step": 149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.714285714285715e-06,
+      "loss": 2.1726,
+      "step": 150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7114285714285715e-06,
+      "loss": 2.1484,
+      "step": 151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.708571428571429e-06,
+      "loss": 2.1684,
+      "step": 152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7057142857142865e-06,
+      "loss": 2.0468,
+      "step": 153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.702857142857143e-06,
+      "loss": 2.1174,
+      "step": 154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7e-06,
+      "loss": 2.1044,
+      "step": 155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.6971428571428574e-06,
+      "loss": 2.1413,
+      "step": 156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.694285714285715e-06,
+      "loss": 2.1223,
+      "step": 157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.691428571428572e-06,
+      "loss": 2.0807,
+      "step": 158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.688571428571429e-06,
+      "loss": 2.0802,
+      "step": 159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.685714285714286e-06,
+      "loss": 2.1053,
+      "step": 160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.682857142857143e-06,
+      "loss": 2.158,
+      "step": 161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 2.1759,
+      "step": 162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.677142857142858e-06,
+      "loss": 2.1431,
+      "step": 163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.674285714285714e-06,
+      "loss": 2.1355,
+      "step": 164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.671428571428572e-06,
+      "loss": 2.0845,
+      "step": 165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.668571428571429e-06,
+      "loss": 2.1096,
+      "step": 166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.665714285714286e-06,
+      "loss": 2.1554,
+      "step": 167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.662857142857144e-06,
+      "loss": 2.0881,
+      "step": 168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.66e-06,
+      "loss": 2.0361,
+      "step": 169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.657142857142857e-06,
+      "loss": 2.0334,
+      "step": 170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.6542857142857145e-06,
+      "loss": 2.1693,
+      "step": 171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.651428571428572e-06,
+      "loss": 2.0666,
+      "step": 172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.648571428571429e-06,
+      "loss": 2.0814,
+      "step": 173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.645714285714286e-06,
+      "loss": 2.1696,
+      "step": 174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 2.0501,
+      "step": 175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 2.1494,
+      "step": 176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.637142857142857e-06,
+      "loss": 2.1464,
+      "step": 177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.634285714285715e-06,
+      "loss": 2.0821,
+      "step": 178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.631428571428571e-06,
+      "loss": 2.0578,
+      "step": 179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.628571428571429e-06,
+      "loss": 2.0885,
+      "step": 180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6257142857142864e-06,
+      "loss": 2.1092,
+      "step": 181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.622857142857143e-06,
+      "loss": 2.1026,
+      "step": 182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 2.1134,
+      "step": 183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.617142857142857e-06,
+      "loss": 2.1566,
+      "step": 184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.614285714285715e-06,
+      "loss": 2.1304,
+      "step": 185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6114285714285716e-06,
+      "loss": 2.0944,
+      "step": 186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.608571428571429e-06,
+      "loss": 2.0699,
+      "step": 187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.605714285714286e-06,
+      "loss": 2.0218,
+      "step": 188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.602857142857143e-06,
+      "loss": 2.2031,
+      "step": 189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.0347,
+      "step": 190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5971428571428575e-06,
+      "loss": 2.0347,
+      "step": 191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.594285714285714e-06,
+      "loss": 2.1245,
+      "step": 192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.591428571428572e-06,
+      "loss": 2.1068,
+      "step": 193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.588571428571429e-06,
+      "loss": 2.06,
+      "step": 194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.585714285714286e-06,
+      "loss": 2.0105,
+      "step": 195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5828571428571435e-06,
+      "loss": 2.0148,
+      "step": 196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.58e-06,
+      "loss": 2.0729,
+      "step": 197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.577142857142858e-06,
+      "loss": 2.1496,
+      "step": 198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.574285714285714e-06,
+      "loss": 2.1057,
+      "step": 199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 2.094,
+      "step": 200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.568571428571429e-06,
+      "loss": 2.0715,
+      "step": 201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.565714285714286e-06,
+      "loss": 2.0591,
+      "step": 202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.562857142857144e-06,
+      "loss": 2.1628,
+      "step": 203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.56e-06,
+      "loss": 2.1043,
+      "step": 204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.557142857142858e-06,
+      "loss": 2.0607,
+      "step": 205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.554285714285715e-06,
+      "loss": 2.0445,
+      "step": 206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.551428571428571e-06,
+      "loss": 2.1152,
+      "step": 207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.548571428571429e-06,
+      "loss": 2.0502,
+      "step": 208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.545714285714286e-06,
+      "loss": 2.0174,
+      "step": 209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.542857142857143e-06,
+      "loss": 2.0485,
+      "step": 210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 2.0891,
+      "step": 211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.537142857142858e-06,
+      "loss": 2.0484,
+      "step": 212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.534285714285715e-06,
+      "loss": 2.0807,
+      "step": 213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5314285714285715e-06,
+      "loss": 2.1129,
+      "step": 214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.528571428571429e-06,
+      "loss": 2.2036,
+      "step": 215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.525714285714286e-06,
+      "loss": 2.0202,
+      "step": 216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.522857142857143e-06,
+      "loss": 2.097,
+      "step": 217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 2.006,
+      "step": 218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5171428571428575e-06,
+      "loss": 2.1062,
+      "step": 219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.514285714285714e-06,
+      "loss": 2.0305,
+      "step": 220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.511428571428572e-06,
+      "loss": 2.1076,
+      "step": 221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.508571428571429e-06,
+      "loss": 2.1966,
+      "step": 222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.505714285714286e-06,
+      "loss": 2.1069,
+      "step": 223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.5028571428571434e-06,
+      "loss": 2.0831,
+      "step": 224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.5e-06,
+      "loss": 2.1507,
+      "step": 225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.497142857142858e-06,
+      "loss": 2.0783,
+      "step": 226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.494285714285715e-06,
+      "loss": 1.999,
+      "step": 227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.491428571428572e-06,
+      "loss": 1.9835,
+      "step": 228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4885714285714286e-06,
+      "loss": 2.0462,
+      "step": 229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.485714285714286e-06,
+      "loss": 2.0569,
+      "step": 230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.482857142857144e-06,
+      "loss": 2.0587,
+      "step": 231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.48e-06,
+      "loss": 2.055,
+      "step": 232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.477142857142858e-06,
+      "loss": 2.082,
+      "step": 233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4742857142857145e-06,
+      "loss": 2.1492,
+      "step": 234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.471428571428571e-06,
+      "loss": 2.0753,
+      "step": 235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.468571428571429e-06,
+      "loss": 1.9967,
+      "step": 236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.465714285714286e-06,
+      "loss": 2.1059,
+      "step": 237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.462857142857143e-06,
+      "loss": 2.0575,
+      "step": 238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 2.1207,
+      "step": 239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.457142857142858e-06,
+      "loss": 2.0252,
+      "step": 240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.454285714285715e-06,
+      "loss": 2.0658,
+      "step": 241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.451428571428571e-06,
+      "loss": 2.0873,
+      "step": 242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.448571428571429e-06,
+      "loss": 2.121,
+      "step": 243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.445714285714286e-06,
+      "loss": 2.0365,
+      "step": 244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.442857142857143e-06,
+      "loss": 2.1615,
+      "step": 245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 2.0512,
+      "step": 246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.437142857142857e-06,
+      "loss": 2.0244,
+      "step": 247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.434285714285715e-06,
+      "loss": 1.9428,
+      "step": 248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.431428571428572e-06,
+      "loss": 2.0025,
+      "step": 249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.428571428571429e-06,
+      "loss": 2.1209,
+      "step": 250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425714285714286e-06,
+      "loss": 2.0812,
+      "step": 251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422857142857143e-06,
+      "loss": 1.9982,
+      "step": 252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.42e-06,
+      "loss": 2.0075,
+      "step": 253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4171428571428576e-06,
+      "loss": 2.1638,
+      "step": 254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.414285714285715e-06,
+      "loss": 2.0029,
+      "step": 255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.411428571428572e-06,
+      "loss": 2.1706,
+      "step": 256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4085714285714285e-06,
+      "loss": 2.1189,
+      "step": 257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.405714285714286e-06,
+      "loss": 1.9728,
+      "step": 258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4028571428571435e-06,
+      "loss": 2.0224,
+      "step": 259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4e-06,
+      "loss": 2.0377,
+      "step": 260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.397142857142858e-06,
+      "loss": 1.9478,
+      "step": 261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3942857142857144e-06,
+      "loss": 2.051,
+      "step": 262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.391428571428572e-06,
+      "loss": 1.9506,
+      "step": 263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.388571428571429e-06,
+      "loss": 2.1893,
+      "step": 264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.385714285714286e-06,
+      "loss": 1.9695,
+      "step": 265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.382857142857143e-06,
+      "loss": 1.9766,
+      "step": 266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.38e-06,
+      "loss": 2.0487,
+      "step": 267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.377142857142858e-06,
+      "loss": 2.0865,
+      "step": 268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.374285714285715e-06,
+      "loss": 2.0527,
+      "step": 269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.371428571428572e-06,
+      "loss": 1.9204,
+      "step": 270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.368571428571429e-06,
+      "loss": 2.0328,
+      "step": 271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.3657142857142855e-06,
+      "loss": 2.0388,
+      "step": 272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.362857142857143e-06,
+      "loss": 1.983,
+      "step": 273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 2.1196,
+      "step": 274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.357142857142857e-06,
+      "loss": 2.0359,
+      "step": 275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.354285714285715e-06,
+      "loss": 1.968,
+      "step": 276
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.351428571428572e-06,
+      "loss": 2.0986,
+      "step": 277
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.348571428571429e-06,
+      "loss": 2.078,
+      "step": 278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.345714285714286e-06,
+      "loss": 2.0297,
+      "step": 279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.342857142857143e-06,
+      "loss": 1.9645,
+      "step": 280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.34e-06,
+      "loss": 1.8953,
+      "step": 281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3371428571428575e-06,
+      "loss": 1.9604,
+      "step": 282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.334285714285715e-06,
+      "loss": 1.9947,
+      "step": 283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.331428571428572e-06,
+      "loss": 2.0304,
+      "step": 284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.328571428571429e-06,
+      "loss": 2.0579,
+      "step": 285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.325714285714286e-06,
+      "loss": 2.022,
+      "step": 286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.3228571428571435e-06,
+      "loss": 2.0657,
+      "step": 287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.32e-06,
+      "loss": 2.0465,
+      "step": 288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.317142857142858e-06,
+      "loss": 1.9437,
+      "step": 289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.314285714285714e-06,
+      "loss": 1.9647,
+      "step": 290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.311428571428572e-06,
+      "loss": 1.9792,
+      "step": 291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3085714285714294e-06,
+      "loss": 2.0021,
+      "step": 292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.305714285714286e-06,
+      "loss": 1.9907,
+      "step": 293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.302857142857143e-06,
+      "loss": 2.025,
+      "step": 294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3e-06,
+      "loss": 2.0517,
+      "step": 295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.297142857142858e-06,
+      "loss": 2.0641,
+      "step": 296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2942857142857146e-06,
+      "loss": 1.9877,
+      "step": 297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.291428571428572e-06,
+      "loss": 1.9957,
+      "step": 298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.288571428571429e-06,
+      "loss": 2.0505,
+      "step": 299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.9501,
+      "step": 300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.282857142857143e-06,
+      "loss": 1.9726,
+      "step": 301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 1.9893,
+      "step": 302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.277142857142857e-06,
+      "loss": 1.8235,
+      "step": 303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.274285714285715e-06,
+      "loss": 1.9638,
+      "step": 304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.271428571428572e-06,
+      "loss": 1.9909,
+      "step": 305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.268571428571429e-06,
+      "loss": 2.108,
+      "step": 306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.265714285714286e-06,
+      "loss": 2.0446,
+      "step": 307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.262857142857143e-06,
+      "loss": 1.9752,
+      "step": 308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.26e-06,
+      "loss": 1.9344,
+      "step": 309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.257142857142857e-06,
+      "loss": 1.9019,
+      "step": 310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.254285714285715e-06,
+      "loss": 1.9577,
+      "step": 311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.251428571428572e-06,
+      "loss": 1.9521,
+      "step": 312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.248571428571429e-06,
+      "loss": 1.9943,
+      "step": 313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.245714285714286e-06,
+      "loss": 2.0362,
+      "step": 314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.242857142857143e-06,
+      "loss": 2.0162,
+      "step": 315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.24e-06,
+      "loss": 2.0704,
+      "step": 316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.237142857142858e-06,
+      "loss": 2.0576,
+      "step": 317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.234285714285714e-06,
+      "loss": 2.1032,
+      "step": 318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.231428571428572e-06,
+      "loss": 2.005,
+      "step": 319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.228571428571429e-06,
+      "loss": 2.0006,
+      "step": 320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.225714285714286e-06,
+      "loss": 1.9812,
+      "step": 321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.222857142857143e-06,
+      "loss": 2.0278,
+      "step": 322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.22e-06,
+      "loss": 2.0428,
+      "step": 323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.217142857142858e-06,
+      "loss": 2.0551,
+      "step": 324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2142857142857145e-06,
+      "loss": 2.0748,
+      "step": 325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.211428571428572e-06,
+      "loss": 1.9723,
+      "step": 326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.208571428571429e-06,
+      "loss": 1.9586,
+      "step": 327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.205714285714286e-06,
+      "loss": 1.976,
+      "step": 328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.202857142857143e-06,
+      "loss": 2.0041,
+      "step": 329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.0654,
+      "step": 330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.197142857142857e-06,
+      "loss": 2.0142,
+      "step": 331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.194285714285715e-06,
+      "loss": 2.0903,
+      "step": 332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.191428571428572e-06,
+      "loss": 1.8168,
+      "step": 333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.188571428571429e-06,
+      "loss": 1.8949,
+      "step": 334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.185714285714286e-06,
+      "loss": 1.8861,
+      "step": 335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.182857142857143e-06,
+      "loss": 2.0482,
+      "step": 336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.18e-06,
+      "loss": 1.8523,
+      "step": 337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.177142857142857e-06,
+      "loss": 1.8483,
+      "step": 338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.174285714285715e-06,
+      "loss": 2.0046,
+      "step": 339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1714285714285715e-06,
+      "loss": 1.8765,
+      "step": 340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.168571428571429e-06,
+      "loss": 2.0096,
+      "step": 341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.165714285714287e-06,
+      "loss": 1.7553,
+      "step": 342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.162857142857143e-06,
+      "loss": 1.9423,
+      "step": 343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.16e-06,
+      "loss": 1.8945,
+      "step": 344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1571428571428575e-06,
+      "loss": 2.0944,
+      "step": 345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.154285714285714e-06,
+      "loss": 1.9509,
+      "step": 346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.151428571428572e-06,
+      "loss": 1.8314,
+      "step": 347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.148571428571429e-06,
+      "loss": 1.9151,
+      "step": 348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.145714285714286e-06,
+      "loss": 1.9887,
+      "step": 349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1428571428571435e-06,
+      "loss": 1.973,
+      "step": 350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.14e-06,
+      "loss": 2.032,
+      "step": 351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.137142857142858e-06,
+      "loss": 1.8415,
+      "step": 352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.134285714285714e-06,
+      "loss": 2.0056,
+      "step": 353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.131428571428572e-06,
+      "loss": 2.0085,
+      "step": 354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.128571428571429e-06,
+      "loss": 2.0766,
+      "step": 355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.125714285714286e-06,
+      "loss": 1.9257,
+      "step": 356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.122857142857144e-06,
+      "loss": 2.0219,
+      "step": 357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.12e-06,
+      "loss": 2.0715,
+      "step": 358
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.117142857142857e-06,
+      "loss": 1.8464,
+      "step": 359
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.114285714285715e-06,
+      "loss": 1.9872,
+      "step": 360
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.6638888888888889,
+      "eval_loss": 1.8874504566192627,
+      "eval_roc_auc": 0.9247877436364222,
+      "eval_runtime": 14.5356,
+      "eval_samples_per_second": 24.767,
+      "eval_steps_per_second": 24.767,
+      "step": 360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.111428571428572e-06,
+      "loss": 2.0502,
+      "step": 361
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.108571428571429e-06,
+      "loss": 1.9585,
+      "step": 362
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.105714285714286e-06,
+      "loss": 1.9773,
+      "step": 363
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.102857142857143e-06,
+      "loss": 1.9256,
+      "step": 364
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1e-06,
+      "loss": 1.9924,
+      "step": 365
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.097142857142857e-06,
+      "loss": 1.8911,
+      "step": 366
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.094285714285715e-06,
+      "loss": 2.0606,
+      "step": 367
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0914285714285715e-06,
+      "loss": 1.9584,
+      "step": 368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.088571428571429e-06,
+      "loss": 1.9242,
+      "step": 369
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.0857142857142865e-06,
+      "loss": 1.9972,
+      "step": 370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.082857142857143e-06,
+      "loss": 1.6981,
+      "step": 371
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.08e-06,
+      "loss": 1.8308,
+      "step": 372
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.0771428571428574e-06,
+      "loss": 1.8146,
+      "step": 373
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.074285714285714e-06,
+      "loss": 1.9584,
+      "step": 374
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.071428571428572e-06,
+      "loss": 2.0404,
+      "step": 375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.068571428571429e-06,
+      "loss": 1.978,
+      "step": 376
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.065714285714286e-06,
+      "loss": 2.0382,
+      "step": 377
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.062857142857143e-06,
+      "loss": 1.9766,
+      "step": 378
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 1.9053,
+      "step": 379
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.057142857142858e-06,
+      "loss": 1.9484,
+      "step": 380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.054285714285714e-06,
+      "loss": 1.799,
+      "step": 381
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.051428571428572e-06,
+      "loss": 2.0658,
+      "step": 382
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0485714285714285e-06,
+      "loss": 1.7029,
+      "step": 383
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.045714285714286e-06,
+      "loss": 2.0117,
+      "step": 384
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.042857142857144e-06,
+      "loss": 1.9691,
+      "step": 385
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.04e-06,
+      "loss": 1.7773,
+      "step": 386
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.037142857142857e-06,
+      "loss": 1.9018,
+      "step": 387
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0342857142857145e-06,
+      "loss": 1.8628,
+      "step": 388
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.031428571428572e-06,
+      "loss": 1.9908,
+      "step": 389
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.028571428571429e-06,
+      "loss": 2.0124,
+      "step": 390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.025714285714286e-06,
+      "loss": 1.965,
+      "step": 391
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.022857142857143e-06,
+      "loss": 1.9771,
+      "step": 392
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 1.9533,
+      "step": 393
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.017142857142857e-06,
+      "loss": 1.9756,
+      "step": 394
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.014285714285715e-06,
+      "loss": 1.8801,
+      "step": 395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.011428571428571e-06,
+      "loss": 1.9098,
+      "step": 396
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.008571428571429e-06,
+      "loss": 2.0659,
+      "step": 397
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0057142857142864e-06,
+      "loss": 1.8009,
+      "step": 398
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.002857142857143e-06,
+      "loss": 1.9536,
+      "step": 399
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.0196,
+      "step": 400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.997142857142857e-06,
+      "loss": 1.9536,
+      "step": 401
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.994285714285714e-06,
+      "loss": 1.9374,
+      "step": 402
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9914285714285716e-06,
+      "loss": 1.8491,
+      "step": 403
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.988571428571429e-06,
+      "loss": 2.0295,
+      "step": 404
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.985714285714286e-06,
+      "loss": 2.0157,
+      "step": 405
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.982857142857143e-06,
+      "loss": 1.8623,
+      "step": 406
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 1.906,
+      "step": 407
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9771428571428575e-06,
+      "loss": 2.0565,
+      "step": 408
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.974285714285714e-06,
+      "loss": 1.9324,
+      "step": 409
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.971428571428572e-06,
+      "loss": 1.687,
+      "step": 410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9685714285714284e-06,
+      "loss": 1.9473,
+      "step": 411
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.965714285714286e-06,
+      "loss": 1.9125,
+      "step": 412
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9628571428571435e-06,
+      "loss": 1.9577,
+      "step": 413
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.96e-06,
+      "loss": 2.0081,
+      "step": 414
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.957142857142858e-06,
+      "loss": 1.9097,
+      "step": 415
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.954285714285714e-06,
+      "loss": 1.8646,
+      "step": 416
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.951428571428572e-06,
+      "loss": 1.8697,
+      "step": 417
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.948571428571429e-06,
+      "loss": 1.7188,
+      "step": 418
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.945714285714286e-06,
+      "loss": 1.91,
+      "step": 419
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.942857142857143e-06,
+      "loss": 1.7726,
+      "step": 420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.94e-06,
+      "loss": 1.9715,
+      "step": 421
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.937142857142858e-06,
+      "loss": 1.9678,
+      "step": 422
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.934285714285715e-06,
+      "loss": 1.6602,
+      "step": 423
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.931428571428571e-06,
+      "loss": 1.8935,
+      "step": 424
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 1.7851,
+      "step": 425
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.925714285714286e-06,
+      "loss": 2.0748,
+      "step": 426
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.922857142857143e-06,
+      "loss": 1.9957,
+      "step": 427
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 2.0067,
+      "step": 428
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917142857142857e-06,
+      "loss": 1.8203,
+      "step": 429
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.914285714285714e-06,
+      "loss": 1.951,
+      "step": 430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9114285714285715e-06,
+      "loss": 1.9583,
+      "step": 431
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.908571428571429e-06,
+      "loss": 1.8257,
+      "step": 432
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.905714285714286e-06,
+      "loss": 1.9062,
+      "step": 433
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.902857142857143e-06,
+      "loss": 1.9505,
+      "step": 434
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 1.7437,
+      "step": 435
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8971428571428575e-06,
+      "loss": 1.7374,
+      "step": 436
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.894285714285714e-06,
+      "loss": 1.8603,
+      "step": 437
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.891428571428572e-06,
+      "loss": 1.9937,
+      "step": 438
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.888571428571428e-06,
+      "loss": 1.9649,
+      "step": 439
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.885714285714286e-06,
+      "loss": 1.945,
+      "step": 440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8828571428571434e-06,
+      "loss": 2.0189,
+      "step": 441
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.88e-06,
+      "loss": 1.7734,
+      "step": 442
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.877142857142858e-06,
+      "loss": 1.8368,
+      "step": 443
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.874285714285715e-06,
+      "loss": 1.8034,
+      "step": 444
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.871428571428572e-06,
+      "loss": 1.8384,
+      "step": 445
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8685714285714286e-06,
+      "loss": 1.9943,
+      "step": 446
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.865714285714286e-06,
+      "loss": 1.9424,
+      "step": 447
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.862857142857143e-06,
+      "loss": 1.6219,
+      "step": 448
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.86e-06,
+      "loss": 1.8607,
+      "step": 449
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.857142857142858e-06,
+      "loss": 1.9349,
+      "step": 450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8542857142857145e-06,
+      "loss": 1.9662,
+      "step": 451
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.851428571428571e-06,
+      "loss": 1.8681,
+      "step": 452
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.848571428571429e-06,
+      "loss": 1.8158,
+      "step": 453
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.845714285714286e-06,
+      "loss": 1.9716,
+      "step": 454
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.842857142857143e-06,
+      "loss": 1.9319,
+      "step": 455
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 1.9485,
+      "step": 456
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.837142857142857e-06,
+      "loss": 1.9522,
+      "step": 457
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.834285714285715e-06,
+      "loss": 1.7359,
+      "step": 458
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.831428571428571e-06,
+      "loss": 1.7189,
+      "step": 459
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.828571428571429e-06,
+      "loss": 1.7212,
+      "step": 460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.825714285714286e-06,
+      "loss": 1.8984,
+      "step": 461
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.822857142857143e-06,
+      "loss": 1.8072,
+      "step": 462
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 1.7526,
+      "step": 463
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.817142857142857e-06,
+      "loss": 1.867,
+      "step": 464
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.814285714285715e-06,
+      "loss": 1.724,
+      "step": 465
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8114285714285716e-06,
+      "loss": 1.9271,
+      "step": 466
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8085714285714287e-06,
+      "loss": 1.9329,
+      "step": 467
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.805714285714286e-06,
+      "loss": 1.8766,
+      "step": 468
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.802857142857143e-06,
+      "loss": 1.8001,
+      "step": 469
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 1.9626,
+      "step": 470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7971428571428576e-06,
+      "loss": 1.9313,
+      "step": 471
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7942857142857147e-06,
+      "loss": 1.609,
+      "step": 472
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7914285714285722e-06,
+      "loss": 1.7565,
+      "step": 473
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7885714285714285e-06,
+      "loss": 1.8305,
+      "step": 474
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.785714285714286e-06,
+      "loss": 1.7863,
+      "step": 475
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.782857142857143e-06,
+      "loss": 1.8918,
+      "step": 476
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 1.8953,
+      "step": 477
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7771428571428578e-06,
+      "loss": 1.8333,
+      "step": 478
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.774285714285715e-06,
+      "loss": 1.8612,
+      "step": 479
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.771428571428572e-06,
+      "loss": 1.8793,
+      "step": 480
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7685714285714287e-06,
+      "loss": 1.828,
+      "step": 481
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7657142857142858e-06,
+      "loss": 1.9759,
+      "step": 482
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.762857142857143e-06,
+      "loss": 1.8489,
+      "step": 483
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 1.9614,
+      "step": 484
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7571428571428575e-06,
+      "loss": 1.831,
+      "step": 485
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7542857142857146e-06,
+      "loss": 1.743,
+      "step": 486
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.751428571428572e-06,
+      "loss": 1.9234,
+      "step": 487
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7485714285714284e-06,
+      "loss": 1.7721,
+      "step": 488
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.745714285714286e-06,
+      "loss": 2.0322,
+      "step": 489
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.742857142857143e-06,
+      "loss": 1.6842,
+      "step": 490
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.74e-06,
+      "loss": 1.7099,
+      "step": 491
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7371428571428577e-06,
+      "loss": 1.803,
+      "step": 492
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.734285714285715e-06,
+      "loss": 1.8711,
+      "step": 493
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.731428571428572e-06,
+      "loss": 1.9093,
+      "step": 494
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.7285714285714286e-06,
+      "loss": 1.7652,
+      "step": 495
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.7257142857142857e-06,
+      "loss": 1.7999,
+      "step": 496
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.722857142857143e-06,
+      "loss": 1.7845,
+      "step": 497
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 1.7273,
+      "step": 498
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.7171428571428575e-06,
+      "loss": 1.8609,
+      "step": 499
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 1.8652,
+      "step": 500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.711428571428572e-06,
+      "loss": 1.9027,
+      "step": 501
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.7085714285714284e-06,
+      "loss": 1.7647,
+      "step": 502
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.705714285714286e-06,
+      "loss": 1.9296,
+      "step": 503
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.702857142857143e-06,
+      "loss": 1.8013,
+      "step": 504
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.7e-06,
+      "loss": 1.9541,
+      "step": 505
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6971428571428577e-06,
+      "loss": 1.7596,
+      "step": 506
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.694285714285715e-06,
+      "loss": 1.8071,
+      "step": 507
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.691428571428572e-06,
+      "loss": 1.7947,
+      "step": 508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.688571428571429e-06,
+      "loss": 1.8373,
+      "step": 509
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.6857142857142857e-06,
+      "loss": 1.6664,
+      "step": 510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.682857142857143e-06,
+      "loss": 1.815,
+      "step": 511
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 1.8522,
+      "step": 512
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.6771428571428575e-06,
+      "loss": 1.9314,
+      "step": 513
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.6742857142857146e-06,
+      "loss": 1.785,
+      "step": 514
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.671428571428572e-06,
+      "loss": 1.9292,
+      "step": 515
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.668571428571429e-06,
+      "loss": 1.9212,
+      "step": 516
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.665714285714286e-06,
+      "loss": 1.9099,
+      "step": 517
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.662857142857143e-06,
+      "loss": 1.8149,
+      "step": 518
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.66e-06,
+      "loss": 1.7249,
+      "step": 519
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6571428571428576e-06,
+      "loss": 1.7215,
+      "step": 520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6542857142857148e-06,
+      "loss": 1.8079,
+      "step": 521
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.651428571428572e-06,
+      "loss": 1.6177,
+      "step": 522
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.648571428571429e-06,
+      "loss": 1.5941,
+      "step": 523
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6457142857142857e-06,
+      "loss": 1.9819,
+      "step": 524
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.642857142857143e-06,
+      "loss": 1.7426,
+      "step": 525
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 1.6513,
+      "step": 526
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6371428571428574e-06,
+      "loss": 1.8205,
+      "step": 527
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.6342857142857145e-06,
+      "loss": 1.751,
+      "step": 528
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.631428571428572e-06,
+      "loss": 1.8294,
+      "step": 529
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.628571428571429e-06,
+      "loss": 1.6369,
+      "step": 530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.625714285714286e-06,
+      "loss": 1.7827,
+      "step": 531
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.622857142857143e-06,
+      "loss": 1.7378,
+      "step": 532
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.62e-06,
+      "loss": 1.8974,
+      "step": 533
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.6171428571428576e-06,
+      "loss": 1.8167,
+      "step": 534
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6142857142857147e-06,
+      "loss": 1.8567,
+      "step": 535
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.611428571428572e-06,
+      "loss": 1.8288,
+      "step": 536
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.608571428571429e-06,
+      "loss": 1.6032,
+      "step": 537
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6057142857142865e-06,
+      "loss": 1.7409,
+      "step": 538
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.602857142857143e-06,
+      "loss": 1.6674,
+      "step": 539
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 1.8895,
+      "step": 540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5971428571428574e-06,
+      "loss": 1.9083,
+      "step": 541
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5942857142857145e-06,
+      "loss": 1.8025,
+      "step": 542
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.591428571428572e-06,
+      "loss": 1.8331,
+      "step": 543
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.588571428571429e-06,
+      "loss": 1.6825,
+      "step": 544
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5857142857142862e-06,
+      "loss": 1.6106,
+      "step": 545
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.582857142857143e-06,
+      "loss": 1.6528,
+      "step": 546
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.58e-06,
+      "loss": 1.8734,
+      "step": 547
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.5771428571428576e-06,
+      "loss": 1.7637,
+      "step": 548
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.5742857142857147e-06,
+      "loss": 1.7601,
+      "step": 549
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 1.8477,
+      "step": 550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.568571428571429e-06,
+      "loss": 1.6751,
+      "step": 551
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.5657142857142864e-06,
+      "loss": 1.7417,
+      "step": 552
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.562857142857143e-06,
+      "loss": 1.7496,
+      "step": 553
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 1.6978,
+      "step": 554
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.5571428571428573e-06,
+      "loss": 1.7292,
+      "step": 555
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.5542857142857144e-06,
+      "loss": 1.8879,
+      "step": 556
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.551428571428572e-06,
+      "loss": 1.7426,
+      "step": 557
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.548571428571429e-06,
+      "loss": 1.7067,
+      "step": 558
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.545714285714286e-06,
+      "loss": 1.9416,
+      "step": 559
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.542857142857143e-06,
+      "loss": 1.521,
+      "step": 560
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.54e-06,
+      "loss": 1.8943,
+      "step": 561
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5371428571428575e-06,
+      "loss": 1.8069,
+      "step": 562
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5342857142857146e-06,
+      "loss": 1.8196,
+      "step": 563
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5314285714285717e-06,
+      "loss": 1.7068,
+      "step": 564
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.528571428571429e-06,
+      "loss": 1.711,
+      "step": 565
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5257142857142864e-06,
+      "loss": 1.7806,
+      "step": 566
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5228571428571435e-06,
+      "loss": 1.5231,
+      "step": 567
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.52e-06,
+      "loss": 1.82,
+      "step": 568
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.5171428571428573e-06,
+      "loss": 1.7285,
+      "step": 569
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.5142857142857144e-06,
+      "loss": 1.7307,
+      "step": 570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.511428571428572e-06,
+      "loss": 1.7761,
+      "step": 571
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.508571428571429e-06,
+      "loss": 1.8074,
+      "step": 572
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.505714285714286e-06,
+      "loss": 1.8069,
+      "step": 573
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.5028571428571433e-06,
+      "loss": 1.9314,
+      "step": 574
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.5e-06,
+      "loss": 1.7339,
+      "step": 575
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4971428571428575e-06,
+      "loss": 1.7319,
+      "step": 576
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4942857142857146e-06,
+      "loss": 1.7083,
+      "step": 577
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4914285714285717e-06,
+      "loss": 1.8224,
+      "step": 578
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.488571428571429e-06,
+      "loss": 1.6424,
+      "step": 579
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4857142857142863e-06,
+      "loss": 1.6987,
+      "step": 580
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4828571428571435e-06,
+      "loss": 1.6532,
+      "step": 581
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.48e-06,
+      "loss": 1.7992,
+      "step": 582
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.4771428571428572e-06,
+      "loss": 1.8066,
+      "step": 583
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.4742857142857144e-06,
+      "loss": 1.7914,
+      "step": 584
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.471428571428572e-06,
+      "loss": 1.6616,
+      "step": 585
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.468571428571429e-06,
+      "loss": 1.7442,
+      "step": 586
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.465714285714286e-06,
+      "loss": 1.6163,
+      "step": 587
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4628571428571432e-06,
+      "loss": 1.5736,
+      "step": 588
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.46e-06,
+      "loss": 1.701,
+      "step": 589
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.4571428571428574e-06,
+      "loss": 1.5466,
+      "step": 590
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.4542857142857145e-06,
+      "loss": 1.5862,
+      "step": 591
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.4514285714285717e-06,
+      "loss": 1.8687,
+      "step": 592
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4485714285714288e-06,
+      "loss": 1.726,
+      "step": 593
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4457142857142863e-06,
+      "loss": 1.8111,
+      "step": 594
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4428571428571434e-06,
+      "loss": 1.8131,
+      "step": 595
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.44e-06,
+      "loss": 1.4534,
+      "step": 596
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.437142857142857e-06,
+      "loss": 1.8011,
+      "step": 597
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4342857142857143e-06,
+      "loss": 1.6137,
+      "step": 598
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.431428571428572e-06,
+      "loss": 1.9868,
+      "step": 599
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 1.465,
+      "step": 600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.425714285714286e-06,
+      "loss": 1.8132,
+      "step": 601
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.422857142857143e-06,
+      "loss": 1.6087,
+      "step": 602
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 1.6952,
+      "step": 603
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.4171428571428574e-06,
+      "loss": 1.6137,
+      "step": 604
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.4142857142857145e-06,
+      "loss": 1.5146,
+      "step": 605
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.4114285714285716e-06,
+      "loss": 1.4461,
+      "step": 606
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4085714285714287e-06,
+      "loss": 1.777,
+      "step": 607
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4057142857142863e-06,
+      "loss": 1.7292,
+      "step": 608
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4028571428571434e-06,
+      "loss": 1.8803,
+      "step": 609
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.654,
+      "step": 610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.397142857142857e-06,
+      "loss": 1.6622,
+      "step": 611
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3942857142857143e-06,
+      "loss": 1.8187,
+      "step": 612
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.391428571428572e-06,
+      "loss": 1.7113,
+      "step": 613
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.388571428571429e-06,
+      "loss": 1.7879,
+      "step": 614
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.385714285714286e-06,
+      "loss": 1.4493,
+      "step": 615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.382857142857143e-06,
+      "loss": 1.7407,
+      "step": 616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 1.8809,
+      "step": 617
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.3771428571428574e-06,
+      "loss": 1.8549,
+      "step": 618
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.3742857142857145e-06,
+      "loss": 1.5986,
+      "step": 619
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.3714285714285716e-06,
+      "loss": 1.5568,
+      "step": 620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.3685714285714287e-06,
+      "loss": 1.7953,
+      "step": 621
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.3657142857142862e-06,
+      "loss": 1.8026,
+      "step": 622
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.3628571428571433e-06,
+      "loss": 1.8626,
+      "step": 623
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 1.537,
+      "step": 624
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.357142857142857e-06,
+      "loss": 1.8223,
+      "step": 625
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.3542857142857142e-06,
+      "loss": 1.7767,
+      "step": 626
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.3514285714285718e-06,
+      "loss": 1.7917,
+      "step": 627
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.348571428571429e-06,
+      "loss": 1.708,
+      "step": 628
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.345714285714286e-06,
+      "loss": 1.7811,
+      "step": 629
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.342857142857143e-06,
+      "loss": 1.8043,
+      "step": 630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 1.7107,
+      "step": 631
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.3371428571428577e-06,
+      "loss": 1.4523,
+      "step": 632
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.3342857142857144e-06,
+      "loss": 1.7388,
+      "step": 633
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.3314285714285715e-06,
+      "loss": 1.6317,
+      "step": 634
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.3285714285714286e-06,
+      "loss": 1.5422,
+      "step": 635
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.325714285714286e-06,
+      "loss": 1.7646,
+      "step": 636
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3228571428571433e-06,
+      "loss": 1.567,
+      "step": 637
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 1.7483,
+      "step": 638
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3171428571428575e-06,
+      "loss": 1.7621,
+      "step": 639
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.314285714285714e-06,
+      "loss": 1.7892,
+      "step": 640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3114285714285717e-06,
+      "loss": 1.6641,
+      "step": 641
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.308571428571429e-06,
+      "loss": 1.557,
+      "step": 642
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.305714285714286e-06,
+      "loss": 1.6772,
+      "step": 643
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.302857142857143e-06,
+      "loss": 1.7496,
+      "step": 644
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 1.713,
+      "step": 645
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2971428571428577e-06,
+      "loss": 1.7208,
+      "step": 646
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2942857142857144e-06,
+      "loss": 1.6512,
+      "step": 647
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2914285714285715e-06,
+      "loss": 1.8503,
+      "step": 648
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2885714285714286e-06,
+      "loss": 2.0026,
+      "step": 649
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.285714285714286e-06,
+      "loss": 1.7878,
+      "step": 650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2828571428571433e-06,
+      "loss": 1.6831,
+      "step": 651
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 1.7798,
+      "step": 652
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2771428571428575e-06,
+      "loss": 1.5792,
+      "step": 653
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.274285714285714e-06,
+      "loss": 1.7921,
+      "step": 654
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.2714285714285717e-06,
+      "loss": 1.5742,
+      "step": 655
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.268571428571429e-06,
+      "loss": 1.5994,
+      "step": 656
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.265714285714286e-06,
+      "loss": 1.5841,
+      "step": 657
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.262857142857143e-06,
+      "loss": 1.5748,
+      "step": 658
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 1.6029,
+      "step": 659
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.2571428571428577e-06,
+      "loss": 1.6603,
+      "step": 660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2542857142857148e-06,
+      "loss": 1.6852,
+      "step": 661
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2514285714285715e-06,
+      "loss": 1.6603,
+      "step": 662
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2485714285714286e-06,
+      "loss": 1.7836,
+      "step": 663
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.245714285714286e-06,
+      "loss": 1.7572,
+      "step": 664
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.242857142857143e-06,
+      "loss": 1.8096,
+      "step": 665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 1.5988,
+      "step": 666
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.2371428571428574e-06,
+      "loss": 1.6786,
+      "step": 667
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.234285714285715e-06,
+      "loss": 1.6813,
+      "step": 668
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.2314285714285716e-06,
+      "loss": 1.8044,
+      "step": 669
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.2285714285714288e-06,
+      "loss": 1.5841,
+      "step": 670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.225714285714286e-06,
+      "loss": 1.307,
+      "step": 671
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.222857142857143e-06,
+      "loss": 1.8389,
+      "step": 672
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 1.5203,
+      "step": 673
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2171428571428576e-06,
+      "loss": 1.5722,
+      "step": 674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 1.4124,
+      "step": 675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.2114285714285714e-06,
+      "loss": 1.4529,
+      "step": 676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.2085714285714285e-06,
+      "loss": 1.5764,
+      "step": 677
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.205714285714286e-06,
+      "loss": 1.7846,
+      "step": 678
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.202857142857143e-06,
+      "loss": 1.7508,
+      "step": 679
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.6298,
+      "step": 680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1971428571428574e-06,
+      "loss": 1.665,
+      "step": 681
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.194285714285715e-06,
+      "loss": 1.5514,
+      "step": 682
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1914285714285716e-06,
+      "loss": 1.8115,
+      "step": 683
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1885714285714287e-06,
+      "loss": 1.3078,
+      "step": 684
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.185714285714286e-06,
+      "loss": 1.6379,
+      "step": 685
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.182857142857143e-06,
+      "loss": 1.6617,
+      "step": 686
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 1.6468,
+      "step": 687
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1771428571428576e-06,
+      "loss": 1.4932,
+      "step": 688
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1742857142857147e-06,
+      "loss": 1.5922,
+      "step": 689
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.1714285714285714e-06,
+      "loss": 1.6475,
+      "step": 690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.1685714285714285e-06,
+      "loss": 1.7,
+      "step": 691
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.165714285714286e-06,
+      "loss": 1.7761,
+      "step": 692
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.162857142857143e-06,
+      "loss": 1.4641,
+      "step": 693
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 1.6819,
+      "step": 694
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1571428571428573e-06,
+      "loss": 1.6963,
+      "step": 695
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.154285714285715e-06,
+      "loss": 1.3995,
+      "step": 696
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.151428571428572e-06,
+      "loss": 1.5134,
+      "step": 697
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.1485714285714287e-06,
+      "loss": 1.7919,
+      "step": 698
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.1457142857142858e-06,
+      "loss": 1.6602,
+      "step": 699
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.142857142857143e-06,
+      "loss": 1.7187,
+      "step": 700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 1.5335,
+      "step": 701
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1371428571428575e-06,
+      "loss": 1.7295,
+      "step": 702
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1342857142857146e-06,
+      "loss": 1.6874,
+      "step": 703
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.1314285714285718e-06,
+      "loss": 1.4432,
+      "step": 704
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.1285714285714284e-06,
+      "loss": 1.5774,
+      "step": 705
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.125714285714286e-06,
+      "loss": 1.9866,
+      "step": 706
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.122857142857143e-06,
+      "loss": 1.7453,
+      "step": 707
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.12e-06,
+      "loss": 1.4233,
+      "step": 708
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.1171428571428573e-06,
+      "loss": 1.8549,
+      "step": 709
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.114285714285715e-06,
+      "loss": 1.5036,
+      "step": 710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.111428571428572e-06,
+      "loss": 1.8674,
+      "step": 711
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1085714285714286e-06,
+      "loss": 1.7126,
+      "step": 712
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1057142857142857e-06,
+      "loss": 1.3111,
+      "step": 713
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.102857142857143e-06,
+      "loss": 1.7646,
+      "step": 714
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 1.3261,
+      "step": 715
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0971428571428575e-06,
+      "loss": 1.5023,
+      "step": 716
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0942857142857146e-06,
+      "loss": 1.7389,
+      "step": 717
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0914285714285717e-06,
+      "loss": 1.7626,
+      "step": 718
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0885714285714284e-06,
+      "loss": 1.8136,
+      "step": 719
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.085714285714286e-06,
+      "loss": 1.8652,
+      "step": 720
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8,
+      "eval_loss": 1.4792804718017578,
+      "eval_roc_auc": 0.9799176264615048,
+      "eval_runtime": 13.6162,
+      "eval_samples_per_second": 26.439,
+      "eval_steps_per_second": 26.439,
+      "step": 720
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.082857142857143e-06,
+      "loss": 1.6689,
+      "step": 721
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.08e-06,
+      "loss": 1.6841,
+      "step": 722
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0771428571428573e-06,
+      "loss": 1.7421,
+      "step": 723
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.074285714285715e-06,
+      "loss": 1.2222,
+      "step": 724
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.071428571428572e-06,
+      "loss": 1.725,
+      "step": 725
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.068571428571429e-06,
+      "loss": 1.532,
+      "step": 726
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.0657142857142857e-06,
+      "loss": 1.5709,
+      "step": 727
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.062857142857143e-06,
+      "loss": 1.622,
+      "step": 728
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 1.776,
+      "step": 729
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.0571428571428575e-06,
+      "loss": 1.5034,
+      "step": 730
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.0542857142857146e-06,
+      "loss": 1.8443,
+      "step": 731
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.0514285714285717e-06,
+      "loss": 1.7849,
+      "step": 732
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.048571428571429e-06,
+      "loss": 1.7772,
+      "step": 733
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.045714285714286e-06,
+      "loss": 1.7634,
+      "step": 734
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.042857142857143e-06,
+      "loss": 1.5789,
+      "step": 735
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.04e-06,
+      "loss": 1.7546,
+      "step": 736
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0371428571428572e-06,
+      "loss": 1.4528,
+      "step": 737
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0342857142857148e-06,
+      "loss": 1.9031,
+      "step": 738
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.031428571428572e-06,
+      "loss": 1.5219,
+      "step": 739
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.028571428571429e-06,
+      "loss": 1.4481,
+      "step": 740
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0257142857142857e-06,
+      "loss": 1.8673,
+      "step": 741
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0228571428571428e-06,
+      "loss": 1.5892,
+      "step": 742
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 1.6518,
+      "step": 743
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.0171428571428574e-06,
+      "loss": 1.4721,
+      "step": 744
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.0142857142857145e-06,
+      "loss": 1.5267,
+      "step": 745
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.0114285714285716e-06,
+      "loss": 1.5166,
+      "step": 746
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.008571428571429e-06,
+      "loss": 1.7242,
+      "step": 747
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.005714285714286e-06,
+      "loss": 1.62,
+      "step": 748
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.002857142857143e-06,
+      "loss": 1.3975,
+      "step": 749
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3e-06,
+      "loss": 1.7414,
+      "step": 750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.997142857142857e-06,
+      "loss": 1.5971,
+      "step": 751
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9942857142857147e-06,
+      "loss": 1.6538,
+      "step": 752
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.991428571428572e-06,
+      "loss": 1.682,
+      "step": 753
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.988571428571429e-06,
+      "loss": 1.4865,
+      "step": 754
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.985714285714286e-06,
+      "loss": 1.5599,
+      "step": 755
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9828571428571427e-06,
+      "loss": 1.6431,
+      "step": 756
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 1.3486,
+      "step": 757
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9771428571428574e-06,
+      "loss": 1.5415,
+      "step": 758
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9742857142857145e-06,
+      "loss": 1.7118,
+      "step": 759
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9714285714285716e-06,
+      "loss": 1.6731,
+      "step": 760
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.968571428571429e-06,
+      "loss": 1.4124,
+      "step": 761
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9657142857142862e-06,
+      "loss": 1.8559,
+      "step": 762
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.962857142857143e-06,
+      "loss": 1.4425,
+      "step": 763
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.96e-06,
+      "loss": 1.7942,
+      "step": 764
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.957142857142857e-06,
+      "loss": 1.766,
+      "step": 765
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.9542857142857147e-06,
+      "loss": 1.6803,
+      "step": 766
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.9514285714285718e-06,
+      "loss": 1.2501,
+      "step": 767
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.948571428571429e-06,
+      "loss": 1.5156,
+      "step": 768
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.945714285714286e-06,
+      "loss": 1.6828,
+      "step": 769
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9428571428571427e-06,
+      "loss": 1.3066,
+      "step": 770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 1.5085,
+      "step": 771
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9371428571428573e-06,
+      "loss": 1.6893,
+      "step": 772
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.9342857142857144e-06,
+      "loss": 1.6498,
+      "step": 773
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.9314285714285716e-06,
+      "loss": 1.7774,
+      "step": 774
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.928571428571429e-06,
+      "loss": 1.6118,
+      "step": 775
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.925714285714286e-06,
+      "loss": 1.6796,
+      "step": 776
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.922857142857143e-06,
+      "loss": 1.6238,
+      "step": 777
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.92e-06,
+      "loss": 1.6668,
+      "step": 778
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.917142857142857e-06,
+      "loss": 1.8241,
+      "step": 779
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.9142857142857146e-06,
+      "loss": 1.6056,
+      "step": 780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.9114285714285717e-06,
+      "loss": 1.5538,
+      "step": 781
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.908571428571429e-06,
+      "loss": 1.4826,
+      "step": 782
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.905714285714286e-06,
+      "loss": 1.5916,
+      "step": 783
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.9028571428571427e-06,
+      "loss": 1.4505,
+      "step": 784
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.9e-06,
+      "loss": 1.8038,
+      "step": 785
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8971428571428573e-06,
+      "loss": 1.6589,
+      "step": 786
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8942857142857144e-06,
+      "loss": 1.5753,
+      "step": 787
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8914285714285715e-06,
+      "loss": 1.6851,
+      "step": 788
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.888571428571429e-06,
+      "loss": 1.7117,
+      "step": 789
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.885714285714286e-06,
+      "loss": 1.5337,
+      "step": 790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8828571428571433e-06,
+      "loss": 1.6549,
+      "step": 791
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.88e-06,
+      "loss": 1.5171,
+      "step": 792
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.877142857142857e-06,
+      "loss": 1.8204,
+      "step": 793
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8742857142857146e-06,
+      "loss": 1.4409,
+      "step": 794
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8714285714285717e-06,
+      "loss": 1.7335,
+      "step": 795
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.868571428571429e-06,
+      "loss": 1.8171,
+      "step": 796
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.865714285714286e-06,
+      "loss": 1.5588,
+      "step": 797
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8628571428571435e-06,
+      "loss": 1.4991,
+      "step": 798
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.86e-06,
+      "loss": 1.4178,
+      "step": 799
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.635,
+      "step": 800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8542857142857144e-06,
+      "loss": 1.8914,
+      "step": 801
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8514285714285715e-06,
+      "loss": 1.4128,
+      "step": 802
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.848571428571429e-06,
+      "loss": 1.4845,
+      "step": 803
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.845714285714286e-06,
+      "loss": 1.5794,
+      "step": 804
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8428571428571432e-06,
+      "loss": 1.7482,
+      "step": 805
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.84e-06,
+      "loss": 1.3293,
+      "step": 806
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.837142857142857e-06,
+      "loss": 1.3219,
+      "step": 807
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8342857142857146e-06,
+      "loss": 1.4887,
+      "step": 808
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.8314285714285717e-06,
+      "loss": 1.7636,
+      "step": 809
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.8285714285714288e-06,
+      "loss": 1.7226,
+      "step": 810
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.825714285714286e-06,
+      "loss": 1.6487,
+      "step": 811
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8228571428571434e-06,
+      "loss": 1.5583,
+      "step": 812
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.82e-06,
+      "loss": 1.6148,
+      "step": 813
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8171428571428572e-06,
+      "loss": 1.4927,
+      "step": 814
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8142857142857143e-06,
+      "loss": 1.7495,
+      "step": 815
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8114285714285714e-06,
+      "loss": 1.1738,
+      "step": 816
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.808571428571429e-06,
+      "loss": 1.2689,
+      "step": 817
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.805714285714286e-06,
+      "loss": 1.4504,
+      "step": 818
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.802857142857143e-06,
+      "loss": 1.4335,
+      "step": 819
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.5411,
+      "step": 820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.797142857142857e-06,
+      "loss": 1.5447,
+      "step": 821
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7942857142857145e-06,
+      "loss": 1.4374,
+      "step": 822
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7914285714285716e-06,
+      "loss": 1.4279,
+      "step": 823
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7885714285714287e-06,
+      "loss": 1.7728,
+      "step": 824
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.785714285714286e-06,
+      "loss": 1.7025,
+      "step": 825
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7828571428571434e-06,
+      "loss": 1.2551,
+      "step": 826
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 1.4571,
+      "step": 827
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.777142857142857e-06,
+      "loss": 1.7484,
+      "step": 828
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7742857142857143e-06,
+      "loss": 1.6948,
+      "step": 829
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7714285714285714e-06,
+      "loss": 1.5485,
+      "step": 830
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.768571428571429e-06,
+      "loss": 1.4506,
+      "step": 831
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.765714285714286e-06,
+      "loss": 1.744,
+      "step": 832
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.762857142857143e-06,
+      "loss": 1.6615,
+      "step": 833
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 1.7488,
+      "step": 834
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.757142857142857e-06,
+      "loss": 1.6663,
+      "step": 835
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.7542857142857145e-06,
+      "loss": 1.7418,
+      "step": 836
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.7514285714285716e-06,
+      "loss": 1.3478,
+      "step": 837
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.7485714285714287e-06,
+      "loss": 1.7521,
+      "step": 838
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.745714285714286e-06,
+      "loss": 1.5926,
+      "step": 839
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.7428571428571433e-06,
+      "loss": 1.7245,
+      "step": 840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 1.8914,
+      "step": 841
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.737142857142857e-06,
+      "loss": 1.4469,
+      "step": 842
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.7342857142857142e-06,
+      "loss": 1.605,
+      "step": 843
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.7314285714285714e-06,
+      "loss": 1.7493,
+      "step": 844
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.728571428571429e-06,
+      "loss": 1.5556,
+      "step": 845
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.725714285714286e-06,
+      "loss": 1.4871,
+      "step": 846
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.722857142857143e-06,
+      "loss": 1.7342,
+      "step": 847
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 1.6271,
+      "step": 848
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7171428571428577e-06,
+      "loss": 1.5814,
+      "step": 849
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7142857142857144e-06,
+      "loss": 1.468,
+      "step": 850
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7114285714285715e-06,
+      "loss": 1.2324,
+      "step": 851
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7085714285714287e-06,
+      "loss": 1.6486,
+      "step": 852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7057142857142858e-06,
+      "loss": 1.4265,
+      "step": 853
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7028571428571433e-06,
+      "loss": 1.4776,
+      "step": 854
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 1.5793,
+      "step": 855
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6971428571428575e-06,
+      "loss": 1.4445,
+      "step": 856
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.694285714285714e-06,
+      "loss": 1.5134,
+      "step": 857
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6914285714285713e-06,
+      "loss": 1.5164,
+      "step": 858
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.688571428571429e-06,
+      "loss": 1.4598,
+      "step": 859
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.685714285714286e-06,
+      "loss": 1.7106,
+      "step": 860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.682857142857143e-06,
+      "loss": 1.643,
+      "step": 861
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.68e-06,
+      "loss": 1.6752,
+      "step": 862
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6771428571428577e-06,
+      "loss": 1.7136,
+      "step": 863
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6742857142857144e-06,
+      "loss": 1.6802,
+      "step": 864
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6714285714285715e-06,
+      "loss": 1.5568,
+      "step": 865
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6685714285714286e-06,
+      "loss": 1.2284,
+      "step": 866
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6657142857142857e-06,
+      "loss": 1.6611,
+      "step": 867
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6628571428571433e-06,
+      "loss": 1.4115,
+      "step": 868
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 1.4414,
+      "step": 869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.6571428571428575e-06,
+      "loss": 1.4086,
+      "step": 870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.654285714285714e-06,
+      "loss": 1.4835,
+      "step": 871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.6514285714285713e-06,
+      "loss": 1.6848,
+      "step": 872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.648571428571429e-06,
+      "loss": 1.6476,
+      "step": 873
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.645714285714286e-06,
+      "loss": 1.5199,
+      "step": 874
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.642857142857143e-06,
+      "loss": 1.561,
+      "step": 875
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.64e-06,
+      "loss": 1.7153,
+      "step": 876
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.6371428571428577e-06,
+      "loss": 1.6178,
+      "step": 877
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.6342857142857144e-06,
+      "loss": 1.6024,
+      "step": 878
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.6314285714285715e-06,
+      "loss": 1.5242,
+      "step": 879
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.6285714285714286e-06,
+      "loss": 1.5345,
+      "step": 880
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6257142857142857e-06,
+      "loss": 1.5828,
+      "step": 881
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6228571428571432e-06,
+      "loss": 1.4609,
+      "step": 882
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 1.6896,
+      "step": 883
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6171428571428574e-06,
+      "loss": 1.5425,
+      "step": 884
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.614285714285715e-06,
+      "loss": 1.3028,
+      "step": 885
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6114285714285712e-06,
+      "loss": 1.5638,
+      "step": 886
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6085714285714288e-06,
+      "loss": 1.6394,
+      "step": 887
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.605714285714286e-06,
+      "loss": 1.5175,
+      "step": 888
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.602857142857143e-06,
+      "loss": 1.4311,
+      "step": 889
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6e-06,
+      "loss": 1.6864,
+      "step": 890
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5971428571428576e-06,
+      "loss": 1.5064,
+      "step": 891
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5942857142857147e-06,
+      "loss": 1.6242,
+      "step": 892
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5914285714285714e-06,
+      "loss": 1.6449,
+      "step": 893
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5885714285714285e-06,
+      "loss": 1.5679,
+      "step": 894
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5857142857142856e-06,
+      "loss": 1.256,
+      "step": 895
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.582857142857143e-06,
+      "loss": 1.5144,
+      "step": 896
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 1.5909,
+      "step": 897
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5771428571428574e-06,
+      "loss": 1.6504,
+      "step": 898
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.574285714285715e-06,
+      "loss": 1.3408,
+      "step": 899
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 1.4703,
+      "step": 900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5685714285714287e-06,
+      "loss": 1.588,
+      "step": 901
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.565714285714286e-06,
+      "loss": 1.5107,
+      "step": 902
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.562857142857143e-06,
+      "loss": 1.5318,
+      "step": 903
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.56e-06,
+      "loss": 1.6164,
+      "step": 904
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5571428571428576e-06,
+      "loss": 1.5653,
+      "step": 905
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5542857142857147e-06,
+      "loss": 1.5437,
+      "step": 906
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5514285714285714e-06,
+      "loss": 1.7246,
+      "step": 907
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5485714285714285e-06,
+      "loss": 1.589,
+      "step": 908
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5457142857142856e-06,
+      "loss": 1.6464,
+      "step": 909
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.542857142857143e-06,
+      "loss": 1.4733,
+      "step": 910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 1.4139,
+      "step": 911
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5371428571428574e-06,
+      "loss": 1.4221,
+      "step": 912
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.534285714285715e-06,
+      "loss": 1.5861,
+      "step": 913
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.531428571428572e-06,
+      "loss": 1.358,
+      "step": 914
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5285714285714287e-06,
+      "loss": 1.7324,
+      "step": 915
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.525714285714286e-06,
+      "loss": 1.4807,
+      "step": 916
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.522857142857143e-06,
+      "loss": 1.3605,
+      "step": 917
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.52e-06,
+      "loss": 1.6587,
+      "step": 918
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5171428571428575e-06,
+      "loss": 1.3962,
+      "step": 919
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5142857142857147e-06,
+      "loss": 1.4527,
+      "step": 920
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5114285714285718e-06,
+      "loss": 1.719,
+      "step": 921
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5085714285714285e-06,
+      "loss": 1.4732,
+      "step": 922
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5057142857142856e-06,
+      "loss": 1.7571,
+      "step": 923
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.502857142857143e-06,
+      "loss": 1.1767,
+      "step": 924
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5e-06,
+      "loss": 1.2872,
+      "step": 925
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4971428571428573e-06,
+      "loss": 1.6733,
+      "step": 926
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4942857142857144e-06,
+      "loss": 1.3742,
+      "step": 927
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4914285714285715e-06,
+      "loss": 1.3528,
+      "step": 928
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4885714285714286e-06,
+      "loss": 1.4607,
+      "step": 929
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.485714285714286e-06,
+      "loss": 1.5616,
+      "step": 930
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.482857142857143e-06,
+      "loss": 1.6478,
+      "step": 931
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 1.3899,
+      "step": 932
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4771428571428575e-06,
+      "loss": 1.664,
+      "step": 933
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4742857142857146e-06,
+      "loss": 1.4991,
+      "step": 934
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4714285714285717e-06,
+      "loss": 1.3425,
+      "step": 935
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.468571428571429e-06,
+      "loss": 1.6704,
+      "step": 936
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.465714285714286e-06,
+      "loss": 1.3997,
+      "step": 937
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.462857142857143e-06,
+      "loss": 1.4987,
+      "step": 938
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.46e-06,
+      "loss": 1.3269,
+      "step": 939
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4571428571428573e-06,
+      "loss": 1.6327,
+      "step": 940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.454285714285715e-06,
+      "loss": 1.6927,
+      "step": 941
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4514285714285715e-06,
+      "loss": 1.5714,
+      "step": 942
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4485714285714286e-06,
+      "loss": 1.4299,
+      "step": 943
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.445714285714286e-06,
+      "loss": 1.5286,
+      "step": 944
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.442857142857143e-06,
+      "loss": 1.5674,
+      "step": 945
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 1.7289,
+      "step": 946
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4371428571428575e-06,
+      "loss": 1.4824,
+      "step": 947
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4342857142857146e-06,
+      "loss": 1.5711,
+      "step": 948
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4314285714285717e-06,
+      "loss": 1.464,
+      "step": 949
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.428571428571429e-06,
+      "loss": 1.5649,
+      "step": 950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.425714285714286e-06,
+      "loss": 1.482,
+      "step": 951
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.422857142857143e-06,
+      "loss": 1.664,
+      "step": 952
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.42e-06,
+      "loss": 1.3421,
+      "step": 953
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4171428571428572e-06,
+      "loss": 1.7061,
+      "step": 954
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4142857142857148e-06,
+      "loss": 1.2879,
+      "step": 955
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.4114285714285715e-06,
+      "loss": 1.6165,
+      "step": 956
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.4085714285714286e-06,
+      "loss": 1.3925,
+      "step": 957
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.405714285714286e-06,
+      "loss": 1.3673,
+      "step": 958
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.402857142857143e-06,
+      "loss": 1.3176,
+      "step": 959
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.4529,
+      "step": 960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3971428571428574e-06,
+      "loss": 1.6843,
+      "step": 961
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3942857142857145e-06,
+      "loss": 1.4249,
+      "step": 962
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3914285714285716e-06,
+      "loss": 1.5824,
+      "step": 963
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3885714285714288e-06,
+      "loss": 1.5507,
+      "step": 964
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.385714285714286e-06,
+      "loss": 1.2428,
+      "step": 965
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.382857142857143e-06,
+      "loss": 1.5306,
+      "step": 966
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.38e-06,
+      "loss": 1.4414,
+      "step": 967
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.377142857142857e-06,
+      "loss": 1.5333,
+      "step": 968
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3742857142857147e-06,
+      "loss": 1.3987,
+      "step": 969
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.371428571428572e-06,
+      "loss": 1.5593,
+      "step": 970
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3685714285714285e-06,
+      "loss": 1.6914,
+      "step": 971
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.365714285714286e-06,
+      "loss": 1.345,
+      "step": 972
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.362857142857143e-06,
+      "loss": 1.525,
+      "step": 973
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 1.3446,
+      "step": 974
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3571428571428574e-06,
+      "loss": 1.5563,
+      "step": 975
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3542857142857145e-06,
+      "loss": 1.5701,
+      "step": 976
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3514285714285716e-06,
+      "loss": 1.5058,
+      "step": 977
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.3485714285714287e-06,
+      "loss": 1.7296,
+      "step": 978
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.345714285714286e-06,
+      "loss": 1.408,
+      "step": 979
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.342857142857143e-06,
+      "loss": 1.3206,
+      "step": 980
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 1.4441,
+      "step": 981
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.337142857142857e-06,
+      "loss": 1.4775,
+      "step": 982
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3342857142857147e-06,
+      "loss": 1.6177,
+      "step": 983
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.331428571428572e-06,
+      "loss": 1.4726,
+      "step": 984
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.3285714285714285e-06,
+      "loss": 1.2535,
+      "step": 985
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.325714285714286e-06,
+      "loss": 1.4802,
+      "step": 986
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.322857142857143e-06,
+      "loss": 1.6861,
+      "step": 987
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 1.5452,
+      "step": 988
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3171428571428573e-06,
+      "loss": 1.5299,
+      "step": 989
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3142857142857145e-06,
+      "loss": 1.3005,
+      "step": 990
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3114285714285716e-06,
+      "loss": 1.5031,
+      "step": 991
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.3085714285714287e-06,
+      "loss": 1.31,
+      "step": 992
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.3057142857142858e-06,
+      "loss": 1.3873,
+      "step": 993
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.302857142857143e-06,
+      "loss": 1.5716,
+      "step": 994
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 1.4951,
+      "step": 995
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.297142857142857e-06,
+      "loss": 1.1612,
+      "step": 996
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2942857142857146e-06,
+      "loss": 1.4153,
+      "step": 997
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2914285714285718e-06,
+      "loss": 1.617,
+      "step": 998
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.288571428571429e-06,
+      "loss": 1.2513,
+      "step": 999
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 1.4541,
+      "step": 1000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.282857142857143e-06,
+      "loss": 1.6039,
+      "step": 1001
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.28e-06,
+      "loss": 1.4751,
+      "step": 1002
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2771428571428573e-06,
+      "loss": 1.4653,
+      "step": 1003
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2742857142857144e-06,
+      "loss": 1.2647,
+      "step": 1004
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2714285714285715e-06,
+      "loss": 1.3433,
+      "step": 1005
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.268571428571429e-06,
+      "loss": 1.5623,
+      "step": 1006
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2657142857142857e-06,
+      "loss": 1.4052,
+      "step": 1007
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.262857142857143e-06,
+      "loss": 1.3769,
+      "step": 1008
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 1.3326,
+      "step": 1009
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.257142857142857e-06,
+      "loss": 1.5659,
+      "step": 1010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2542857142857146e-06,
+      "loss": 1.3958,
+      "step": 1011
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2514285714285717e-06,
+      "loss": 1.3911,
+      "step": 1012
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.248571428571429e-06,
+      "loss": 1.0775,
+      "step": 1013
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.245714285714286e-06,
+      "loss": 1.246,
+      "step": 1014
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.242857142857143e-06,
+      "loss": 1.4337,
+      "step": 1015
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.24e-06,
+      "loss": 1.5475,
+      "step": 1016
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.2371428571428573e-06,
+      "loss": 1.5011,
+      "step": 1017
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.2342857142857144e-06,
+      "loss": 1.521,
+      "step": 1018
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.2314285714285715e-06,
+      "loss": 1.4863,
+      "step": 1019
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.228571428571429e-06,
+      "loss": 1.5718,
+      "step": 1020
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.2257142857142857e-06,
+      "loss": 1.3349,
+      "step": 1021
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.222857142857143e-06,
+      "loss": 1.8517,
+      "step": 1022
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 1.3088,
+      "step": 1023
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.2171428571428575e-06,
+      "loss": 1.5484,
+      "step": 1024
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.2142857142857146e-06,
+      "loss": 1.354,
+      "step": 1025
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.2114285714285717e-06,
+      "loss": 1.0786,
+      "step": 1026
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.2085714285714288e-06,
+      "loss": 1.2676,
+      "step": 1027
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.205714285714286e-06,
+      "loss": 1.5535,
+      "step": 1028
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.202857142857143e-06,
+      "loss": 1.5599,
+      "step": 1029
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2e-06,
+      "loss": 1.3774,
+      "step": 1030
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1971428571428572e-06,
+      "loss": 1.4077,
+      "step": 1031
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1942857142857143e-06,
+      "loss": 1.2085,
+      "step": 1032
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1914285714285714e-06,
+      "loss": 1.3909,
+      "step": 1033
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.188571428571429e-06,
+      "loss": 1.2159,
+      "step": 1034
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.185714285714286e-06,
+      "loss": 1.4881,
+      "step": 1035
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1828571428571428e-06,
+      "loss": 1.4701,
+      "step": 1036
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 1.2489,
+      "step": 1037
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1771428571428574e-06,
+      "loss": 1.6276,
+      "step": 1038
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1742857142857145e-06,
+      "loss": 1.3159,
+      "step": 1039
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1714285714285716e-06,
+      "loss": 1.4461,
+      "step": 1040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1685714285714287e-06,
+      "loss": 1.5974,
+      "step": 1041
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.165714285714286e-06,
+      "loss": 1.4045,
+      "step": 1042
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.162857142857143e-06,
+      "loss": 1.5935,
+      "step": 1043
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.16e-06,
+      "loss": 1.6288,
+      "step": 1044
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.157142857142857e-06,
+      "loss": 1.0658,
+      "step": 1045
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1542857142857147e-06,
+      "loss": 1.6353,
+      "step": 1046
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1514285714285714e-06,
+      "loss": 1.2384,
+      "step": 1047
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.148571428571429e-06,
+      "loss": 1.1097,
+      "step": 1048
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.145714285714286e-06,
+      "loss": 1.5475,
+      "step": 1049
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 1.3163,
+      "step": 1050
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 1.5872,
+      "step": 1051
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1371428571428574e-06,
+      "loss": 1.268,
+      "step": 1052
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1342857142857145e-06,
+      "loss": 1.3872,
+      "step": 1053
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1314285714285716e-06,
+      "loss": 1.2048,
+      "step": 1054
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1285714285714287e-06,
+      "loss": 1.3657,
+      "step": 1055
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.125714285714286e-06,
+      "loss": 1.5396,
+      "step": 1056
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.122857142857143e-06,
+      "loss": 1.1505,
+      "step": 1057
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.12e-06,
+      "loss": 1.4125,
+      "step": 1058
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.117142857142857e-06,
+      "loss": 1.5346,
+      "step": 1059
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1142857142857147e-06,
+      "loss": 1.3072,
+      "step": 1060
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1114285714285714e-06,
+      "loss": 1.3575,
+      "step": 1061
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.108571428571429e-06,
+      "loss": 1.4158,
+      "step": 1062
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.105714285714286e-06,
+      "loss": 1.5643,
+      "step": 1063
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.102857142857143e-06,
+      "loss": 1.4159,
+      "step": 1064
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.3417,
+      "step": 1065
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0971428571428573e-06,
+      "loss": 1.5956,
+      "step": 1066
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0942857142857144e-06,
+      "loss": 1.0491,
+      "step": 1067
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0914285714285716e-06,
+      "loss": 1.2703,
+      "step": 1068
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0885714285714287e-06,
+      "loss": 1.1041,
+      "step": 1069
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0857142857142858e-06,
+      "loss": 1.5816,
+      "step": 1070
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0828571428571433e-06,
+      "loss": 1.2993,
+      "step": 1071
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.08e-06,
+      "loss": 1.2351,
+      "step": 1072
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.077142857142857e-06,
+      "loss": 1.423,
+      "step": 1073
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0742857142857146e-06,
+      "loss": 1.4748,
+      "step": 1074
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0714285714285717e-06,
+      "loss": 1.5112,
+      "step": 1075
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.068571428571429e-06,
+      "loss": 1.3522,
+      "step": 1076
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.065714285714286e-06,
+      "loss": 1.7162,
+      "step": 1077
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.062857142857143e-06,
+      "loss": 1.1981,
+      "step": 1078
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.06e-06,
+      "loss": 1.5579,
+      "step": 1079
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0571428571428573e-06,
+      "loss": 1.3868,
+      "step": 1080
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8194444444444444,
+      "eval_loss": 1.2311428785324097,
+      "eval_roc_auc": 0.986107248878886,
+      "eval_runtime": 13.9136,
+      "eval_samples_per_second": 25.874,
+      "eval_steps_per_second": 25.874,
+      "step": 1080
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0542857142857144e-06,
+      "loss": 1.3558,
+      "step": 1081
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0514285714285715e-06,
+      "loss": 1.1884,
+      "step": 1082
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0485714285714286e-06,
+      "loss": 1.4385,
+      "step": 1083
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0457142857142857e-06,
+      "loss": 1.6936,
+      "step": 1084
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0428571428571433e-06,
+      "loss": 1.2808,
+      "step": 1085
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.04e-06,
+      "loss": 1.3763,
+      "step": 1086
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.037142857142857e-06,
+      "loss": 1.2007,
+      "step": 1087
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0342857142857146e-06,
+      "loss": 1.3944,
+      "step": 1088
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0314285714285717e-06,
+      "loss": 1.3879,
+      "step": 1089
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.028571428571429e-06,
+      "loss": 1.525,
+      "step": 1090
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.025714285714286e-06,
+      "loss": 1.4839,
+      "step": 1091
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.022857142857143e-06,
+      "loss": 1.6227,
+      "step": 1092
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.02e-06,
+      "loss": 1.2948,
+      "step": 1093
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0171428571428573e-06,
+      "loss": 1.4804,
+      "step": 1094
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0142857142857144e-06,
+      "loss": 1.2514,
+      "step": 1095
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0114285714285715e-06,
+      "loss": 1.2698,
+      "step": 1096
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.0085714285714286e-06,
+      "loss": 1.5843,
+      "step": 1097
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.0057142857142857e-06,
+      "loss": 1.2982,
+      "step": 1098
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.0028571428571432e-06,
+      "loss": 1.6446,
+      "step": 1099
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.3899,
+      "step": 1100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.997142857142857e-06,
+      "loss": 1.7433,
+      "step": 1101
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9942857142857146e-06,
+      "loss": 1.5278,
+      "step": 1102
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9914285714285717e-06,
+      "loss": 1.2616,
+      "step": 1103
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9885714285714288e-06,
+      "loss": 1.1566,
+      "step": 1104
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.985714285714286e-06,
+      "loss": 1.3162,
+      "step": 1105
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.982857142857143e-06,
+      "loss": 1.5371,
+      "step": 1106
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.98e-06,
+      "loss": 1.7397,
+      "step": 1107
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.977142857142857e-06,
+      "loss": 1.6229,
+      "step": 1108
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9742857142857143e-06,
+      "loss": 1.4301,
+      "step": 1109
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9714285714285714e-06,
+      "loss": 1.5361,
+      "step": 1110
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.968571428571429e-06,
+      "loss": 1.6341,
+      "step": 1111
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9657142857142856e-06,
+      "loss": 1.1032,
+      "step": 1112
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.962857142857143e-06,
+      "loss": 1.534,
+      "step": 1113
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 1.4921,
+      "step": 1114
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.957142857142857e-06,
+      "loss": 1.2526,
+      "step": 1115
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9542857142857145e-06,
+      "loss": 1.7662,
+      "step": 1116
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9514285714285716e-06,
+      "loss": 1.7067,
+      "step": 1117
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.9485714285714287e-06,
+      "loss": 1.0632,
+      "step": 1118
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.945714285714286e-06,
+      "loss": 1.6441,
+      "step": 1119
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.942857142857143e-06,
+      "loss": 1.4271,
+      "step": 1120
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.94e-06,
+      "loss": 1.3872,
+      "step": 1121
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.9371428571428576e-06,
+      "loss": 1.6077,
+      "step": 1122
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.9342857142857143e-06,
+      "loss": 1.0191,
+      "step": 1123
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.9314285714285714e-06,
+      "loss": 1.4292,
+      "step": 1124
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.928571428571429e-06,
+      "loss": 1.3432,
+      "step": 1125
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.9257142857142856e-06,
+      "loss": 1.7904,
+      "step": 1126
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.922857142857143e-06,
+      "loss": 1.1395,
+      "step": 1127
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.2404,
+      "step": 1128
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.9171428571428574e-06,
+      "loss": 1.2377,
+      "step": 1129
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.9142857142857145e-06,
+      "loss": 1.1093,
+      "step": 1130
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.9114285714285716e-06,
+      "loss": 1.2642,
+      "step": 1131
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.9085714285714287e-06,
+      "loss": 1.175,
+      "step": 1132
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9057142857142858e-06,
+      "loss": 1.202,
+      "step": 1133
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.902857142857143e-06,
+      "loss": 1.497,
+      "step": 1134
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 1.3667,
+      "step": 1135
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8971428571428573e-06,
+      "loss": 1.7751,
+      "step": 1136
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8942857142857142e-06,
+      "loss": 1.0817,
+      "step": 1137
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8914285714285716e-06,
+      "loss": 1.4432,
+      "step": 1138
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8885714285714289e-06,
+      "loss": 1.5725,
+      "step": 1139
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.885714285714286e-06,
+      "loss": 1.2641,
+      "step": 1140
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8828571428571429e-06,
+      "loss": 1.3357,
+      "step": 1141
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 1.5894,
+      "step": 1142
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8771428571428573e-06,
+      "loss": 1.6488,
+      "step": 1143
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8742857142857142e-06,
+      "loss": 1.6129,
+      "step": 1144
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8714285714285715e-06,
+      "loss": 1.3208,
+      "step": 1145
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8685714285714289e-06,
+      "loss": 1.0734,
+      "step": 1146
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.865714285714286e-06,
+      "loss": 1.1242,
+      "step": 1147
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8628571428571429e-06,
+      "loss": 1.1107,
+      "step": 1148
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 1.5698,
+      "step": 1149
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8571428571428573e-06,
+      "loss": 1.541,
+      "step": 1150
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8542857142857142e-06,
+      "loss": 1.4465,
+      "step": 1151
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8514285714285715e-06,
+      "loss": 1.3876,
+      "step": 1152
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8485714285714288e-06,
+      "loss": 0.9137,
+      "step": 1153
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.845714285714286e-06,
+      "loss": 1.6923,
+      "step": 1154
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8428571428571428e-06,
+      "loss": 1.5347,
+      "step": 1155
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 1.5527,
+      "step": 1156
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8371428571428573e-06,
+      "loss": 1.4542,
+      "step": 1157
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8342857142857146e-06,
+      "loss": 1.5027,
+      "step": 1158
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8314285714285715e-06,
+      "loss": 1.3437,
+      "step": 1159
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8285714285714288e-06,
+      "loss": 1.2157,
+      "step": 1160
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.825714285714286e-06,
+      "loss": 1.3265,
+      "step": 1161
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8228571428571428e-06,
+      "loss": 1.3864,
+      "step": 1162
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 1.5066,
+      "step": 1163
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8171428571428573e-06,
+      "loss": 1.1737,
+      "step": 1164
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8142857142857146e-06,
+      "loss": 1.3972,
+      "step": 1165
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8114285714285715e-06,
+      "loss": 1.4364,
+      "step": 1166
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8085714285714288e-06,
+      "loss": 1.1473,
+      "step": 1167
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.805714285714286e-06,
+      "loss": 1.3474,
+      "step": 1168
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8028571428571432e-06,
+      "loss": 1.1597,
+      "step": 1169
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 1.251,
+      "step": 1170
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7971428571428572e-06,
+      "loss": 1.4534,
+      "step": 1171
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7942857142857146e-06,
+      "loss": 1.3943,
+      "step": 1172
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7914285714285715e-06,
+      "loss": 1.4761,
+      "step": 1173
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7885714285714288e-06,
+      "loss": 1.419,
+      "step": 1174
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 1.1735,
+      "step": 1175
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7828571428571432e-06,
+      "loss": 1.0844,
+      "step": 1176
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 1.5245,
+      "step": 1177
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7771428571428572e-06,
+      "loss": 1.2457,
+      "step": 1178
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7742857142857145e-06,
+      "loss": 1.2294,
+      "step": 1179
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7714285714285714e-06,
+      "loss": 1.2339,
+      "step": 1180
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7685714285714288e-06,
+      "loss": 1.4625,
+      "step": 1181
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7657142857142859e-06,
+      "loss": 1.3742,
+      "step": 1182
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7628571428571432e-06,
+      "loss": 1.1715,
+      "step": 1183
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.76e-06,
+      "loss": 1.3596,
+      "step": 1184
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7571428571428572e-06,
+      "loss": 1.2721,
+      "step": 1185
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7542857142857145e-06,
+      "loss": 1.3902,
+      "step": 1186
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7514285714285716e-06,
+      "loss": 1.0395,
+      "step": 1187
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7485714285714287e-06,
+      "loss": 1.199,
+      "step": 1188
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7457142857142859e-06,
+      "loss": 1.3244,
+      "step": 1189
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7428571428571432e-06,
+      "loss": 1.3725,
+      "step": 1190
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.74e-06,
+      "loss": 1.3808,
+      "step": 1191
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7371428571428572e-06,
+      "loss": 1.4251,
+      "step": 1192
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7342857142857145e-06,
+      "loss": 1.4627,
+      "step": 1193
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7314285714285716e-06,
+      "loss": 1.3622,
+      "step": 1194
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7285714285714287e-06,
+      "loss": 1.5089,
+      "step": 1195
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7257142857142858e-06,
+      "loss": 1.4871,
+      "step": 1196
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7228571428571432e-06,
+      "loss": 1.1344,
+      "step": 1197
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.72e-06,
+      "loss": 1.5562,
+      "step": 1198
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7171428571428572e-06,
+      "loss": 1.4063,
+      "step": 1199
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 1.4793,
+      "step": 1200
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7114285714285716e-06,
+      "loss": 1.4578,
+      "step": 1201
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7085714285714287e-06,
+      "loss": 1.4847,
+      "step": 1202
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7057142857142858e-06,
+      "loss": 1.1693,
+      "step": 1203
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7028571428571431e-06,
+      "loss": 1.3301,
+      "step": 1204
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 1.2535,
+      "step": 1205
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6971428571428571e-06,
+      "loss": 1.4318,
+      "step": 1206
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6942857142857145e-06,
+      "loss": 1.4922,
+      "step": 1207
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6914285714285716e-06,
+      "loss": 1.3059,
+      "step": 1208
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6885714285714287e-06,
+      "loss": 1.3912,
+      "step": 1209
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6857142857142858e-06,
+      "loss": 1.1539,
+      "step": 1210
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6828571428571431e-06,
+      "loss": 1.4254,
+      "step": 1211
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 1.7022,
+      "step": 1212
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6771428571428571e-06,
+      "loss": 1.3449,
+      "step": 1213
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6742857142857144e-06,
+      "loss": 1.5021,
+      "step": 1214
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6714285714285715e-06,
+      "loss": 1.0716,
+      "step": 1215
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6685714285714289e-06,
+      "loss": 1.3544,
+      "step": 1216
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6657142857142858e-06,
+      "loss": 1.4192,
+      "step": 1217
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.662857142857143e-06,
+      "loss": 1.389,
+      "step": 1218
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 1.3609,
+      "step": 1219
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.657142857142857e-06,
+      "loss": 1.2001,
+      "step": 1220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6542857142857144e-06,
+      "loss": 1.4582,
+      "step": 1221
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6514285714285715e-06,
+      "loss": 1.2556,
+      "step": 1222
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6485714285714289e-06,
+      "loss": 1.3055,
+      "step": 1223
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6457142857142857e-06,
+      "loss": 1.5214,
+      "step": 1224
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.642857142857143e-06,
+      "loss": 1.2821,
+      "step": 1225
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 1.2266,
+      "step": 1226
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.637142857142857e-06,
+      "loss": 1.3616,
+      "step": 1227
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6342857142857144e-06,
+      "loss": 1.3445,
+      "step": 1228
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6314285714285715e-06,
+      "loss": 1.3751,
+      "step": 1229
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6285714285714288e-06,
+      "loss": 1.1848,
+      "step": 1230
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6257142857142857e-06,
+      "loss": 1.2689,
+      "step": 1231
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.622857142857143e-06,
+      "loss": 1.3746,
+      "step": 1232
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 1.4136,
+      "step": 1233
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6171428571428575e-06,
+      "loss": 1.3196,
+      "step": 1234
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6142857142857144e-06,
+      "loss": 1.5199,
+      "step": 1235
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6114285714285715e-06,
+      "loss": 1.0738,
+      "step": 1236
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6085714285714288e-06,
+      "loss": 1.215,
+      "step": 1237
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6057142857142857e-06,
+      "loss": 1.8331,
+      "step": 1238
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.602857142857143e-06,
+      "loss": 1.4438,
+      "step": 1239
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.4057,
+      "step": 1240
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5971428571428575e-06,
+      "loss": 1.2631,
+      "step": 1241
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5942857142857144e-06,
+      "loss": 1.6734,
+      "step": 1242
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5914285714285715e-06,
+      "loss": 1.421,
+      "step": 1243
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5885714285714288e-06,
+      "loss": 1.2047,
+      "step": 1244
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5857142857142857e-06,
+      "loss": 1.4622,
+      "step": 1245
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.582857142857143e-06,
+      "loss": 1.3046,
+      "step": 1246
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 1.5943,
+      "step": 1247
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5771428571428574e-06,
+      "loss": 1.2182,
+      "step": 1248
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5742857142857143e-06,
+      "loss": 1.5645,
+      "step": 1249
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5714285714285714e-06,
+      "loss": 1.5568,
+      "step": 1250
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5685714285714288e-06,
+      "loss": 1.4235,
+      "step": 1251
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5657142857142859e-06,
+      "loss": 1.1577,
+      "step": 1252
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.562857142857143e-06,
+      "loss": 1.1961,
+      "step": 1253
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.56e-06,
+      "loss": 0.9715,
+      "step": 1254
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5571428571428574e-06,
+      "loss": 1.4669,
+      "step": 1255
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5542857142857143e-06,
+      "loss": 1.2854,
+      "step": 1256
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5514285714285714e-06,
+      "loss": 1.3285,
+      "step": 1257
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5485714285714287e-06,
+      "loss": 1.2335,
+      "step": 1258
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5457142857142859e-06,
+      "loss": 1.2565,
+      "step": 1259
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.542857142857143e-06,
+      "loss": 1.4971,
+      "step": 1260
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.54e-06,
+      "loss": 1.2405,
+      "step": 1261
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5371428571428574e-06,
+      "loss": 1.0786,
+      "step": 1262
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5342857142857145e-06,
+      "loss": 1.3805,
+      "step": 1263
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5314285714285714e-06,
+      "loss": 1.5344,
+      "step": 1264
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5285714285714287e-06,
+      "loss": 1.2441,
+      "step": 1265
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.5257142857142858e-06,
+      "loss": 0.9799,
+      "step": 1266
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.522857142857143e-06,
+      "loss": 1.4986,
+      "step": 1267
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.52e-06,
+      "loss": 1.6523,
+      "step": 1268
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.5171428571428574e-06,
+      "loss": 1.1332,
+      "step": 1269
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.5142857142857145e-06,
+      "loss": 1.2723,
+      "step": 1270
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.5114285714285714e-06,
+      "loss": 1.4352,
+      "step": 1271
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.5085714285714287e-06,
+      "loss": 1.405,
+      "step": 1272
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.5057142857142858e-06,
+      "loss": 1.4562,
+      "step": 1273
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.502857142857143e-06,
+      "loss": 1.8292,
+      "step": 1274
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.5e-06,
+      "loss": 1.0765,
+      "step": 1275
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4971428571428574e-06,
+      "loss": 1.2282,
+      "step": 1276
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4942857142857145e-06,
+      "loss": 1.5491,
+      "step": 1277
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4914285714285714e-06,
+      "loss": 1.3467,
+      "step": 1278
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4885714285714287e-06,
+      "loss": 1.4163,
+      "step": 1279
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4857142857142858e-06,
+      "loss": 1.3372,
+      "step": 1280
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4828571428571431e-06,
+      "loss": 1.4236,
+      "step": 1281
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.48e-06,
+      "loss": 1.2076,
+      "step": 1282
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4771428571428573e-06,
+      "loss": 1.0454,
+      "step": 1283
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4742857142857144e-06,
+      "loss": 1.3893,
+      "step": 1284
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4714285714285713e-06,
+      "loss": 1.1811,
+      "step": 1285
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4685714285714287e-06,
+      "loss": 1.3924,
+      "step": 1286
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4657142857142858e-06,
+      "loss": 1.2233,
+      "step": 1287
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.462857142857143e-06,
+      "loss": 1.1211,
+      "step": 1288
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.46e-06,
+      "loss": 1.1881,
+      "step": 1289
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4571428571428573e-06,
+      "loss": 1.3666,
+      "step": 1290
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4542857142857144e-06,
+      "loss": 1.2832,
+      "step": 1291
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4514285714285713e-06,
+      "loss": 1.5197,
+      "step": 1292
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4485714285714286e-06,
+      "loss": 1.1325,
+      "step": 1293
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4457142857142858e-06,
+      "loss": 1.2074,
+      "step": 1294
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.442857142857143e-06,
+      "loss": 1.4596,
+      "step": 1295
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.44e-06,
+      "loss": 1.3626,
+      "step": 1296
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4371428571428573e-06,
+      "loss": 1.4585,
+      "step": 1297
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4342857142857144e-06,
+      "loss": 1.3681,
+      "step": 1298
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4314285714285717e-06,
+      "loss": 1.6824,
+      "step": 1299
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.1903,
+      "step": 1300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4257142857142857e-06,
+      "loss": 1.111,
+      "step": 1301
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.422857142857143e-06,
+      "loss": 1.4748,
+      "step": 1302
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.42e-06,
+      "loss": 1.6237,
+      "step": 1303
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4171428571428573e-06,
+      "loss": 1.1288,
+      "step": 1304
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4142857142857144e-06,
+      "loss": 1.2988,
+      "step": 1305
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4114285714285717e-06,
+      "loss": 1.3399,
+      "step": 1306
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4085714285714286e-06,
+      "loss": 1.2095,
+      "step": 1307
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4057142857142857e-06,
+      "loss": 1.5918,
+      "step": 1308
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.402857142857143e-06,
+      "loss": 1.029,
+      "step": 1309
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.5102,
+      "step": 1310
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3971428571428573e-06,
+      "loss": 1.3122,
+      "step": 1311
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3942857142857144e-06,
+      "loss": 1.1902,
+      "step": 1312
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3914285714285717e-06,
+      "loss": 1.3978,
+      "step": 1313
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3885714285714286e-06,
+      "loss": 1.3987,
+      "step": 1314
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3857142857142857e-06,
+      "loss": 1.2071,
+      "step": 1315
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.382857142857143e-06,
+      "loss": 1.3516,
+      "step": 1316
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 1.4847,
+      "step": 1317
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3771428571428572e-06,
+      "loss": 1.4647,
+      "step": 1318
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3742857142857143e-06,
+      "loss": 1.455,
+      "step": 1319
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3714285714285717e-06,
+      "loss": 0.8947,
+      "step": 1320
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3685714285714286e-06,
+      "loss": 1.4683,
+      "step": 1321
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3657142857142857e-06,
+      "loss": 1.0961,
+      "step": 1322
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.362857142857143e-06,
+      "loss": 1.3768,
+      "step": 1323
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 1.5354,
+      "step": 1324
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3571428571428572e-06,
+      "loss": 1.2237,
+      "step": 1325
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3542857142857143e-06,
+      "loss": 1.2298,
+      "step": 1326
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3514285714285716e-06,
+      "loss": 1.3423,
+      "step": 1327
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3485714285714288e-06,
+      "loss": 1.3445,
+      "step": 1328
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3457142857142857e-06,
+      "loss": 1.2062,
+      "step": 1329
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.342857142857143e-06,
+      "loss": 1.4056,
+      "step": 1330
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.34e-06,
+      "loss": 1.3398,
+      "step": 1331
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3371428571428572e-06,
+      "loss": 1.5057,
+      "step": 1332
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3342857142857143e-06,
+      "loss": 1.624,
+      "step": 1333
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.3314285714285716e-06,
+      "loss": 1.3409,
+      "step": 1334
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.3285714285714287e-06,
+      "loss": 1.3849,
+      "step": 1335
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.3257142857142856e-06,
+      "loss": 1.1019,
+      "step": 1336
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.322857142857143e-06,
+      "loss": 1.2818,
+      "step": 1337
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.32e-06,
+      "loss": 1.1495,
+      "step": 1338
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.3171428571428572e-06,
+      "loss": 1.2607,
+      "step": 1339
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.3142857142857143e-06,
+      "loss": 1.3233,
+      "step": 1340
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.3114285714285716e-06,
+      "loss": 1.5148,
+      "step": 1341
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.3085714285714287e-06,
+      "loss": 1.3766,
+      "step": 1342
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.3057142857142856e-06,
+      "loss": 1.3433,
+      "step": 1343
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.302857142857143e-06,
+      "loss": 1.2344,
+      "step": 1344
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3e-06,
+      "loss": 1.4521,
+      "step": 1345
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2971428571428574e-06,
+      "loss": 1.5069,
+      "step": 1346
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2942857142857143e-06,
+      "loss": 1.1729,
+      "step": 1347
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2914285714285716e-06,
+      "loss": 1.3101,
+      "step": 1348
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2885714285714287e-06,
+      "loss": 1.0906,
+      "step": 1349
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2857142857142856e-06,
+      "loss": 1.1831,
+      "step": 1350
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.282857142857143e-06,
+      "loss": 1.4884,
+      "step": 1351
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.28e-06,
+      "loss": 1.4478,
+      "step": 1352
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2771428571428573e-06,
+      "loss": 1.5589,
+      "step": 1353
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2742857142857142e-06,
+      "loss": 1.3046,
+      "step": 1354
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2714285714285716e-06,
+      "loss": 1.4056,
+      "step": 1355
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2685714285714287e-06,
+      "loss": 1.3079,
+      "step": 1356
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.265714285714286e-06,
+      "loss": 1.4111,
+      "step": 1357
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.262857142857143e-06,
+      "loss": 1.4427,
+      "step": 1358
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.26e-06,
+      "loss": 1.2828,
+      "step": 1359
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2571428571428573e-06,
+      "loss": 1.5054,
+      "step": 1360
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2542857142857142e-06,
+      "loss": 1.3919,
+      "step": 1361
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2514285714285715e-06,
+      "loss": 1.4762,
+      "step": 1362
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2485714285714287e-06,
+      "loss": 1.3504,
+      "step": 1363
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2457142857142858e-06,
+      "loss": 1.5425,
+      "step": 1364
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.242857142857143e-06,
+      "loss": 1.3484,
+      "step": 1365
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 1.1972,
+      "step": 1366
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2371428571428573e-06,
+      "loss": 1.1924,
+      "step": 1367
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2342857142857144e-06,
+      "loss": 1.476,
+      "step": 1368
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2314285714285715e-06,
+      "loss": 1.5661,
+      "step": 1369
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2285714285714286e-06,
+      "loss": 1.391,
+      "step": 1370
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2257142857142857e-06,
+      "loss": 1.1377,
+      "step": 1371
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.222857142857143e-06,
+      "loss": 1.3923,
+      "step": 1372
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 1.2332,
+      "step": 1373
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2171428571428573e-06,
+      "loss": 1.4955,
+      "step": 1374
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2142857142857144e-06,
+      "loss": 1.3169,
+      "step": 1375
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2114285714285715e-06,
+      "loss": 1.0662,
+      "step": 1376
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2085714285714286e-06,
+      "loss": 1.3038,
+      "step": 1377
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2057142857142857e-06,
+      "loss": 1.4078,
+      "step": 1378
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.202857142857143e-06,
+      "loss": 1.5029,
+      "step": 1379
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.1636,
+      "step": 1380
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1971428571428573e-06,
+      "loss": 1.3733,
+      "step": 1381
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1942857142857144e-06,
+      "loss": 1.4011,
+      "step": 1382
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1914285714285715e-06,
+      "loss": 1.3733,
+      "step": 1383
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1885714285714286e-06,
+      "loss": 1.3065,
+      "step": 1384
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.185714285714286e-06,
+      "loss": 1.2521,
+      "step": 1385
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.182857142857143e-06,
+      "loss": 1.4415,
+      "step": 1386
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 1.3659,
+      "step": 1387
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1771428571428572e-06,
+      "loss": 1.1511,
+      "step": 1388
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1742857142857144e-06,
+      "loss": 1.3474,
+      "step": 1389
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1714285714285715e-06,
+      "loss": 1.4246,
+      "step": 1390
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1685714285714286e-06,
+      "loss": 1.3736,
+      "step": 1391
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.165714285714286e-06,
+      "loss": 1.4278,
+      "step": 1392
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.162857142857143e-06,
+      "loss": 1.1894,
+      "step": 1393
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 1.3607,
+      "step": 1394
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1571428571428572e-06,
+      "loss": 1.3095,
+      "step": 1395
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1542857142857143e-06,
+      "loss": 1.3531,
+      "step": 1396
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1514285714285714e-06,
+      "loss": 1.1932,
+      "step": 1397
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1485714285714286e-06,
+      "loss": 1.5404,
+      "step": 1398
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1457142857142859e-06,
+      "loss": 1.4742,
+      "step": 1399
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 1.1609,
+      "step": 1400
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.14e-06,
+      "loss": 1.2389,
+      "step": 1401
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1371428571428572e-06,
+      "loss": 1.1877,
+      "step": 1402
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1342857142857145e-06,
+      "loss": 1.2255,
+      "step": 1403
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1314285714285714e-06,
+      "loss": 1.4075,
+      "step": 1404
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1285714285714285e-06,
+      "loss": 1.5356,
+      "step": 1405
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1257142857142859e-06,
+      "loss": 1.15,
+      "step": 1406
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.122857142857143e-06,
+      "loss": 1.2034,
+      "step": 1407
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.12e-06,
+      "loss": 1.2791,
+      "step": 1408
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1171428571428572e-06,
+      "loss": 1.1068,
+      "step": 1409
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1142857142857145e-06,
+      "loss": 1.086,
+      "step": 1410
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1114285714285714e-06,
+      "loss": 1.2135,
+      "step": 1411
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1085714285714287e-06,
+      "loss": 1.2343,
+      "step": 1412
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1057142857142858e-06,
+      "loss": 1.491,
+      "step": 1413
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.102857142857143e-06,
+      "loss": 1.1582,
+      "step": 1414
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1e-06,
+      "loss": 1.3221,
+      "step": 1415
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0971428571428572e-06,
+      "loss": 1.3499,
+      "step": 1416
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0942857142857145e-06,
+      "loss": 1.3823,
+      "step": 1417
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0914285714285714e-06,
+      "loss": 1.2273,
+      "step": 1418
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0885714285714287e-06,
+      "loss": 1.1921,
+      "step": 1419
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0857142857142858e-06,
+      "loss": 1.3878,
+      "step": 1420
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.082857142857143e-06,
+      "loss": 1.4233,
+      "step": 1421
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.08e-06,
+      "loss": 1.2881,
+      "step": 1422
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0771428571428574e-06,
+      "loss": 1.1281,
+      "step": 1423
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0742857142857145e-06,
+      "loss": 1.3113,
+      "step": 1424
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 0.9378,
+      "step": 1425
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0685714285714287e-06,
+      "loss": 1.2381,
+      "step": 1426
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0657142857142858e-06,
+      "loss": 1.4585,
+      "step": 1427
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.062857142857143e-06,
+      "loss": 1.4708,
+      "step": 1428
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.06e-06,
+      "loss": 1.0569,
+      "step": 1429
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0571428571428573e-06,
+      "loss": 1.1907,
+      "step": 1430
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0542857142857144e-06,
+      "loss": 1.2513,
+      "step": 1431
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0514285714285716e-06,
+      "loss": 1.2137,
+      "step": 1432
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0485714285714287e-06,
+      "loss": 1.6102,
+      "step": 1433
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0457142857142858e-06,
+      "loss": 1.3286,
+      "step": 1434
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0428571428571429e-06,
+      "loss": 1.297,
+      "step": 1435
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.04e-06,
+      "loss": 1.2764,
+      "step": 1436
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0371428571428573e-06,
+      "loss": 1.1706,
+      "step": 1437
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0342857142857144e-06,
+      "loss": 0.9072,
+      "step": 1438
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0314285714285715e-06,
+      "loss": 1.3948,
+      "step": 1439
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0285714285714286e-06,
+      "loss": 1.3242,
+      "step": 1440
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.925,
+      "eval_loss": 1.112122654914856,
+      "eval_roc_auc": 0.9827451326408738,
+      "eval_runtime": 14.0141,
+      "eval_samples_per_second": 25.688,
+      "eval_steps_per_second": 25.688,
+      "step": 1440
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0257142857142858e-06,
+      "loss": 1.328,
+      "step": 1441
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0228571428571429e-06,
+      "loss": 1.1811,
+      "step": 1442
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.02e-06,
+      "loss": 1.2103,
+      "step": 1443
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0171428571428573e-06,
+      "loss": 1.2032,
+      "step": 1444
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0142857142857144e-06,
+      "loss": 1.2725,
+      "step": 1445
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0114285714285715e-06,
+      "loss": 1.5754,
+      "step": 1446
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0085714285714286e-06,
+      "loss": 1.3267,
+      "step": 1447
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0057142857142857e-06,
+      "loss": 1.4138,
+      "step": 1448
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.0028571428571428e-06,
+      "loss": 1.1778,
+      "step": 1449
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.3329,
+      "step": 1450
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.971428571428573e-07,
+      "loss": 1.0533,
+      "step": 1451
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.942857142857144e-07,
+      "loss": 1.1611,
+      "step": 1452
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.914285714285715e-07,
+      "loss": 1.3377,
+      "step": 1453
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.885714285714286e-07,
+      "loss": 1.2658,
+      "step": 1454
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.857142857142857e-07,
+      "loss": 1.2995,
+      "step": 1455
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.828571428571428e-07,
+      "loss": 1.3005,
+      "step": 1456
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 1.0043,
+      "step": 1457
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.771428571428573e-07,
+      "loss": 1.1578,
+      "step": 1458
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.742857142857144e-07,
+      "loss": 1.2734,
+      "step": 1459
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.714285714285715e-07,
+      "loss": 1.5071,
+      "step": 1460
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.685714285714288e-07,
+      "loss": 1.1371,
+      "step": 1461
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.657142857142857e-07,
+      "loss": 1.3018,
+      "step": 1462
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.628571428571428e-07,
+      "loss": 1.0808,
+      "step": 1463
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.4283,
+      "step": 1464
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.571428571428572e-07,
+      "loss": 1.4366,
+      "step": 1465
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.542857142857143e-07,
+      "loss": 1.3694,
+      "step": 1466
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.514285714285715e-07,
+      "loss": 1.0832,
+      "step": 1467
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.485714285714287e-07,
+      "loss": 1.2107,
+      "step": 1468
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.457142857142858e-07,
+      "loss": 1.2874,
+      "step": 1469
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.42857142857143e-07,
+      "loss": 1.1907,
+      "step": 1470
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 1.4792,
+      "step": 1471
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.371428571428571e-07,
+      "loss": 1.3166,
+      "step": 1472
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.342857142857144e-07,
+      "loss": 1.2301,
+      "step": 1473
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.314285714285714e-07,
+      "loss": 1.476,
+      "step": 1474
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.285714285714287e-07,
+      "loss": 1.2889,
+      "step": 1475
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.257142857142858e-07,
+      "loss": 1.0904,
+      "step": 1476
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.22857142857143e-07,
+      "loss": 1.3377,
+      "step": 1477
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 1.3485,
+      "step": 1478
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.171428571428573e-07,
+      "loss": 1.1416,
+      "step": 1479
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.142857142857144e-07,
+      "loss": 1.2098,
+      "step": 1480
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.114285714285714e-07,
+      "loss": 1.402,
+      "step": 1481
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.085714285714286e-07,
+      "loss": 1.3369,
+      "step": 1482
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.057142857142857e-07,
+      "loss": 1.3139,
+      "step": 1483
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.02857142857143e-07,
+      "loss": 1.55,
+      "step": 1484
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 1.3555,
+      "step": 1485
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.971428571428573e-07,
+      "loss": 1.3392,
+      "step": 1486
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.942857142857144e-07,
+      "loss": 1.2857,
+      "step": 1487
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.914285714285716e-07,
+      "loss": 1.3844,
+      "step": 1488
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.885714285714286e-07,
+      "loss": 0.8396,
+      "step": 1489
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.857142857142857e-07,
+      "loss": 1.4187,
+      "step": 1490
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.828571428571429e-07,
+      "loss": 1.0237,
+      "step": 1491
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.8e-07,
+      "loss": 1.1561,
+      "step": 1492
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.771428571428573e-07,
+      "loss": 1.2186,
+      "step": 1493
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.742857142857144e-07,
+      "loss": 1.3032,
+      "step": 1494
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.714285714285716e-07,
+      "loss": 1.5154,
+      "step": 1495
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.685714285714286e-07,
+      "loss": 1.1022,
+      "step": 1496
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.657142857142858e-07,
+      "loss": 0.9588,
+      "step": 1497
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.628571428571429e-07,
+      "loss": 1.2375,
+      "step": 1498
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.6e-07,
+      "loss": 1.5377,
+      "step": 1499
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 1.3141,
+      "step": 1500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.542857142857143e-07,
+      "loss": 1.4511,
+      "step": 1501
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.514285714285716e-07,
+      "loss": 1.5137,
+      "step": 1502
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.485714285714286e-07,
+      "loss": 1.3972,
+      "step": 1503
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.457142857142858e-07,
+      "loss": 1.4089,
+      "step": 1504
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.428571428571429e-07,
+      "loss": 1.3114,
+      "step": 1505
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 1.3042,
+      "step": 1506
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.371428571428572e-07,
+      "loss": 1.2078,
+      "step": 1507
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.342857142857144e-07,
+      "loss": 1.1346,
+      "step": 1508
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.314285714285715e-07,
+      "loss": 1.4065,
+      "step": 1509
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.285714285714285e-07,
+      "loss": 1.2529,
+      "step": 1510
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.257142857142858e-07,
+      "loss": 1.2288,
+      "step": 1511
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.228571428571429e-07,
+      "loss": 1.0429,
+      "step": 1512
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 1.4015,
+      "step": 1513
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.171428571428572e-07,
+      "loss": 1.6124,
+      "step": 1514
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.142857142857144e-07,
+      "loss": 1.2477,
+      "step": 1515
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.114285714285715e-07,
+      "loss": 1.4178,
+      "step": 1516
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.085714285714287e-07,
+      "loss": 1.3195,
+      "step": 1517
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.057142857142857e-07,
+      "loss": 0.9961,
+      "step": 1518
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.028571428571429e-07,
+      "loss": 1.374,
+      "step": 1519
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.0939,
+      "step": 1520
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.971428571428572e-07,
+      "loss": 1.1758,
+      "step": 1521
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.942857142857144e-07,
+      "loss": 1.0602,
+      "step": 1522
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.914285714285715e-07,
+      "loss": 1.6085,
+      "step": 1523
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.885714285714287e-07,
+      "loss": 0.9866,
+      "step": 1524
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.857142857142857e-07,
+      "loss": 1.3627,
+      "step": 1525
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.828571428571429e-07,
+      "loss": 1.3584,
+      "step": 1526
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.8e-07,
+      "loss": 1.4903,
+      "step": 1527
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.771428571428572e-07,
+      "loss": 1.1138,
+      "step": 1528
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.742857142857144e-07,
+      "loss": 1.2482,
+      "step": 1529
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.714285714285715e-07,
+      "loss": 1.1615,
+      "step": 1530
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.685714285714287e-07,
+      "loss": 1.3058,
+      "step": 1531
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.657142857142857e-07,
+      "loss": 1.1666,
+      "step": 1532
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.628571428571429e-07,
+      "loss": 1.4284,
+      "step": 1533
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.6e-07,
+      "loss": 1.1864,
+      "step": 1534
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.571428571428572e-07,
+      "loss": 1.1676,
+      "step": 1535
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.542857142857144e-07,
+      "loss": 1.1699,
+      "step": 1536
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.514285714285715e-07,
+      "loss": 1.3104,
+      "step": 1537
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.485714285714287e-07,
+      "loss": 1.2529,
+      "step": 1538
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.457142857142857e-07,
+      "loss": 1.2705,
+      "step": 1539
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.428571428571429e-07,
+      "loss": 1.3788,
+      "step": 1540
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.4e-07,
+      "loss": 1.2961,
+      "step": 1541
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.371428571428572e-07,
+      "loss": 1.5339,
+      "step": 1542
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.342857142857143e-07,
+      "loss": 1.3505,
+      "step": 1543
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.314285714285715e-07,
+      "loss": 0.9157,
+      "step": 1544
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.285714285714287e-07,
+      "loss": 0.9973,
+      "step": 1545
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.257142857142857e-07,
+      "loss": 1.3099,
+      "step": 1546
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.228571428571429e-07,
+      "loss": 1.1476,
+      "step": 1547
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.2e-07,
+      "loss": 1.0576,
+      "step": 1548
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.171428571428572e-07,
+      "loss": 1.6715,
+      "step": 1549
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 1.2887,
+      "step": 1550
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.114285714285715e-07,
+      "loss": 1.158,
+      "step": 1551
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.085714285714286e-07,
+      "loss": 1.4328,
+      "step": 1552
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.057142857142859e-07,
+      "loss": 1.4909,
+      "step": 1553
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.028571428571429e-07,
+      "loss": 1.4315,
+      "step": 1554
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 1.3187,
+      "step": 1555
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.971428571428572e-07,
+      "loss": 1.4031,
+      "step": 1556
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.942857142857143e-07,
+      "loss": 1.2232,
+      "step": 1557
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.914285714285715e-07,
+      "loss": 1.1365,
+      "step": 1558
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.885714285714286e-07,
+      "loss": 1.4568,
+      "step": 1559
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.857142857142858e-07,
+      "loss": 1.185,
+      "step": 1560
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.828571428571428e-07,
+      "loss": 1.471,
+      "step": 1561
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 1.5129,
+      "step": 1562
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.771428571428572e-07,
+      "loss": 1.2321,
+      "step": 1563
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.742857142857144e-07,
+      "loss": 1.4841,
+      "step": 1564
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.714285714285715e-07,
+      "loss": 1.1565,
+      "step": 1565
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.685714285714286e-07,
+      "loss": 1.5507,
+      "step": 1566
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.657142857142858e-07,
+      "loss": 1.2811,
+      "step": 1567
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.628571428571428e-07,
+      "loss": 1.3852,
+      "step": 1568
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.6e-07,
+      "loss": 0.9968,
+      "step": 1569
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.571428571428571e-07,
+      "loss": 1.0757,
+      "step": 1570
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.542857142857144e-07,
+      "loss": 1.1473,
+      "step": 1571
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.514285714285715e-07,
+      "loss": 1.2237,
+      "step": 1572
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.485714285714287e-07,
+      "loss": 1.178,
+      "step": 1573
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.457142857142858e-07,
+      "loss": 1.1438,
+      "step": 1574
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.428571428571428e-07,
+      "loss": 1.0633,
+      "step": 1575
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.4e-07,
+      "loss": 1.3663,
+      "step": 1576
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.371428571428571e-07,
+      "loss": 1.276,
+      "step": 1577
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.342857142857143e-07,
+      "loss": 1.3254,
+      "step": 1578
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.314285714285714e-07,
+      "loss": 1.4993,
+      "step": 1579
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.285714285714287e-07,
+      "loss": 1.1966,
+      "step": 1580
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.257142857142858e-07,
+      "loss": 1.1091,
+      "step": 1581
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.228571428571429e-07,
+      "loss": 1.2837,
+      "step": 1582
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 1.092,
+      "step": 1583
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.171428571428572e-07,
+      "loss": 1.6939,
+      "step": 1584
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.142857142857143e-07,
+      "loss": 1.2989,
+      "step": 1585
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.114285714285715e-07,
+      "loss": 1.3694,
+      "step": 1586
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.085714285714286e-07,
+      "loss": 1.3278,
+      "step": 1587
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.057142857142858e-07,
+      "loss": 1.224,
+      "step": 1588
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.028571428571429e-07,
+      "loss": 1.4432,
+      "step": 1589
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.6383,
+      "step": 1590
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.971428571428572e-07,
+      "loss": 1.2049,
+      "step": 1591
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.942857142857143e-07,
+      "loss": 1.2333,
+      "step": 1592
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.914285714285715e-07,
+      "loss": 1.2903,
+      "step": 1593
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.885714285714286e-07,
+      "loss": 1.1988,
+      "step": 1594
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.857142857142857e-07,
+      "loss": 1.5053,
+      "step": 1595
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.82857142857143e-07,
+      "loss": 1.3295,
+      "step": 1596
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 1.0335,
+      "step": 1597
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.771428571428572e-07,
+      "loss": 1.487,
+      "step": 1598
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.742857142857143e-07,
+      "loss": 1.2056,
+      "step": 1599
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 1.1672,
+      "step": 1600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.685714285714286e-07,
+      "loss": 1.4304,
+      "step": 1601
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.657142857142857e-07,
+      "loss": 1.425,
+      "step": 1602
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.628571428571429e-07,
+      "loss": 1.1867,
+      "step": 1603
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.6e-07,
+      "loss": 1.3692,
+      "step": 1604
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.571428571428573e-07,
+      "loss": 1.3483,
+      "step": 1605
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.542857142857144e-07,
+      "loss": 0.927,
+      "step": 1606
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.514285714285715e-07,
+      "loss": 1.1216,
+      "step": 1607
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.485714285714286e-07,
+      "loss": 1.236,
+      "step": 1608
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.457142857142857e-07,
+      "loss": 1.3209,
+      "step": 1609
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.428571428571429e-07,
+      "loss": 1.1579,
+      "step": 1610
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.4e-07,
+      "loss": 1.3803,
+      "step": 1611
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.371428571428572e-07,
+      "loss": 1.2841,
+      "step": 1612
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.342857142857143e-07,
+      "loss": 1.3376,
+      "step": 1613
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.314285714285715e-07,
+      "loss": 1.2818,
+      "step": 1614
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.285714285714287e-07,
+      "loss": 1.3483,
+      "step": 1615
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.257142857142858e-07,
+      "loss": 1.3367,
+      "step": 1616
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.228571428571429e-07,
+      "loss": 1.2348,
+      "step": 1617
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.2e-07,
+      "loss": 1.3128,
+      "step": 1618
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.171428571428572e-07,
+      "loss": 1.1566,
+      "step": 1619
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.142857142857143e-07,
+      "loss": 1.3588,
+      "step": 1620
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.114285714285714e-07,
+      "loss": 1.2435,
+      "step": 1621
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.085714285714286e-07,
+      "loss": 1.5324,
+      "step": 1622
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.057142857142858e-07,
+      "loss": 1.7156,
+      "step": 1623
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.028571428571429e-07,
+      "loss": 1.3483,
+      "step": 1624
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.3774,
+      "step": 1625
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.971428571428572e-07,
+      "loss": 1.0546,
+      "step": 1626
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.942857142857143e-07,
+      "loss": 1.2291,
+      "step": 1627
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.914285714285714e-07,
+      "loss": 1.2847,
+      "step": 1628
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.885714285714286e-07,
+      "loss": 1.1676,
+      "step": 1629
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.857142857142857e-07,
+      "loss": 1.3691,
+      "step": 1630
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.828571428571428e-07,
+      "loss": 1.5407,
+      "step": 1631
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.3611,
+      "step": 1632
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.771428571428572e-07,
+      "loss": 1.2655,
+      "step": 1633
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.7428571428571434e-07,
+      "loss": 1.2897,
+      "step": 1634
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.714285714285715e-07,
+      "loss": 1.1269,
+      "step": 1635
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.6857142857142855e-07,
+      "loss": 1.4118,
+      "step": 1636
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.657142857142857e-07,
+      "loss": 1.3923,
+      "step": 1637
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.628571428571429e-07,
+      "loss": 1.1708,
+      "step": 1638
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 1.2129,
+      "step": 1639
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.571428571428572e-07,
+      "loss": 1.3341,
+      "step": 1640
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.542857142857143e-07,
+      "loss": 1.1123,
+      "step": 1641
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.514285714285715e-07,
+      "loss": 1.1081,
+      "step": 1642
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.4857142857142864e-07,
+      "loss": 1.1567,
+      "step": 1643
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.457142857142858e-07,
+      "loss": 1.4206,
+      "step": 1644
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.4285714285714286e-07,
+      "loss": 1.6169,
+      "step": 1645
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.4e-07,
+      "loss": 1.2058,
+      "step": 1646
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.371428571428572e-07,
+      "loss": 1.1256,
+      "step": 1647
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.342857142857143e-07,
+      "loss": 1.2355,
+      "step": 1648
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.3142857142857146e-07,
+      "loss": 1.3258,
+      "step": 1649
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.285714285714286e-07,
+      "loss": 1.3027,
+      "step": 1650
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.257142857142858e-07,
+      "loss": 1.5109,
+      "step": 1651
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.228571428571429e-07,
+      "loss": 1.4216,
+      "step": 1652
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 0.9777,
+      "step": 1653
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.171428571428572e-07,
+      "loss": 1.1239,
+      "step": 1654
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.142857142857143e-07,
+      "loss": 1.2414,
+      "step": 1655
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.1142857142857144e-07,
+      "loss": 0.9786,
+      "step": 1656
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.085714285714286e-07,
+      "loss": 1.3094,
+      "step": 1657
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.0571428571428576e-07,
+      "loss": 1.1125,
+      "step": 1658
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.0285714285714287e-07,
+      "loss": 1.0086,
+      "step": 1659
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.1902,
+      "step": 1660
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.971428571428572e-07,
+      "loss": 1.3312,
+      "step": 1661
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.9428571428571436e-07,
+      "loss": 1.5217,
+      "step": 1662
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.9142857142857147e-07,
+      "loss": 1.2494,
+      "step": 1663
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.885714285714286e-07,
+      "loss": 1.2439,
+      "step": 1664
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.8571428571428574e-07,
+      "loss": 1.4043,
+      "step": 1665
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.8285714285714285e-07,
+      "loss": 1.1909,
+      "step": 1666
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.8e-07,
+      "loss": 1.2549,
+      "step": 1667
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.771428571428572e-07,
+      "loss": 1.6456,
+      "step": 1668
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.7428571428571434e-07,
+      "loss": 0.8759,
+      "step": 1669
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.7142857142857145e-07,
+      "loss": 1.1481,
+      "step": 1670
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.685714285714286e-07,
+      "loss": 1.3666,
+      "step": 1671
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.657142857142858e-07,
+      "loss": 1.0474,
+      "step": 1672
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6285714285714283e-07,
+      "loss": 1.2862,
+      "step": 1673
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6e-07,
+      "loss": 1.4036,
+      "step": 1674
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 1.6164,
+      "step": 1675
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.542857142857143e-07,
+      "loss": 1.2115,
+      "step": 1676
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5142857142857143e-07,
+      "loss": 1.1014,
+      "step": 1677
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.485714285714286e-07,
+      "loss": 1.2739,
+      "step": 1678
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4571428571428575e-07,
+      "loss": 1.2749,
+      "step": 1679
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.428571428571429e-07,
+      "loss": 1.211,
+      "step": 1680
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 1.4,
+      "step": 1681
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.371428571428572e-07,
+      "loss": 1.2127,
+      "step": 1682
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.342857142857143e-07,
+      "loss": 1.0338,
+      "step": 1683
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.314285714285714e-07,
+      "loss": 1.3427,
+      "step": 1684
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.2857142857142857e-07,
+      "loss": 1.4227,
+      "step": 1685
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.2571428571428573e-07,
+      "loss": 1.1963,
+      "step": 1686
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.228571428571429e-07,
+      "loss": 1.3191,
+      "step": 1687
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.2e-07,
+      "loss": 1.1268,
+      "step": 1688
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.1714285714285717e-07,
+      "loss": 1.2344,
+      "step": 1689
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.1428571428571433e-07,
+      "loss": 1.2604,
+      "step": 1690
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.1142857142857144e-07,
+      "loss": 1.1162,
+      "step": 1691
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.085714285714286e-07,
+      "loss": 1.5511,
+      "step": 1692
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.0571428571428577e-07,
+      "loss": 1.3329,
+      "step": 1693
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.028571428571429e-07,
+      "loss": 1.2512,
+      "step": 1694
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 1.4232,
+      "step": 1695
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.9714285714285715e-07,
+      "loss": 1.2482,
+      "step": 1696
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.942857142857143e-07,
+      "loss": 1.1815,
+      "step": 1697
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.914285714285715e-07,
+      "loss": 1.0945,
+      "step": 1698
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.885714285714286e-07,
+      "loss": 1.2593,
+      "step": 1699
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 1.216,
+      "step": 1700
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.8285714285714286e-07,
+      "loss": 1.034,
+      "step": 1701
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.8e-07,
+      "loss": 1.2927,
+      "step": 1702
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.771428571428572e-07,
+      "loss": 1.4967,
+      "step": 1703
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.742857142857143e-07,
+      "loss": 1.157,
+      "step": 1704
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.7142857142857145e-07,
+      "loss": 1.3121,
+      "step": 1705
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.685714285714286e-07,
+      "loss": 1.0445,
+      "step": 1706
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.657142857142857e-07,
+      "loss": 1.1268,
+      "step": 1707
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.628571428571429e-07,
+      "loss": 1.3582,
+      "step": 1708
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6e-07,
+      "loss": 1.5967,
+      "step": 1709
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.5714285714285716e-07,
+      "loss": 1.3285,
+      "step": 1710
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.542857142857143e-07,
+      "loss": 1.1139,
+      "step": 1711
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.5142857142857143e-07,
+      "loss": 1.3212,
+      "step": 1712
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.485714285714286e-07,
+      "loss": 1.081,
+      "step": 1713
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.457142857142857e-07,
+      "loss": 1.0219,
+      "step": 1714
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.4285714285714287e-07,
+      "loss": 1.1653,
+      "step": 1715
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 1.3404,
+      "step": 1716
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.3714285714285717e-07,
+      "loss": 1.2039,
+      "step": 1717
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.3428571428571428e-07,
+      "loss": 1.2814,
+      "step": 1718
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.3142857142857144e-07,
+      "loss": 0.9808,
+      "step": 1719
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.285714285714286e-07,
+      "loss": 1.116,
+      "step": 1720
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.2571428571428574e-07,
+      "loss": 1.0969,
+      "step": 1721
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.228571428571429e-07,
+      "loss": 1.3585,
+      "step": 1722
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2e-07,
+      "loss": 1.0163,
+      "step": 1723
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.1714285714285715e-07,
+      "loss": 1.07,
+      "step": 1724
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.142857142857143e-07,
+      "loss": 1.2963,
+      "step": 1725
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.1142857142857145e-07,
+      "loss": 1.1784,
+      "step": 1726
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.085714285714286e-07,
+      "loss": 1.2837,
+      "step": 1727
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0571428571428572e-07,
+      "loss": 1.4468,
+      "step": 1728
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0285714285714288e-07,
+      "loss": 1.4381,
+      "step": 1729
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.9913,
+      "step": 1730
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9714285714285718e-07,
+      "loss": 1.1811,
+      "step": 1731
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.942857142857143e-07,
+      "loss": 1.599,
+      "step": 1732
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9142857142857143e-07,
+      "loss": 1.1201,
+      "step": 1733
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.885714285714286e-07,
+      "loss": 1.1771,
+      "step": 1734
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8571428571428572e-07,
+      "loss": 1.0457,
+      "step": 1735
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.828571428571429e-07,
+      "loss": 1.3198,
+      "step": 1736
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.8e-07,
+      "loss": 1.2868,
+      "step": 1737
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.7714285714285716e-07,
+      "loss": 1.3107,
+      "step": 1738
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.742857142857143e-07,
+      "loss": 1.3594,
+      "step": 1739
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.7142857142857146e-07,
+      "loss": 1.0134,
+      "step": 1740
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.685714285714286e-07,
+      "loss": 1.1972,
+      "step": 1741
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.657142857142857e-07,
+      "loss": 1.5392,
+      "step": 1742
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6285714285714287e-07,
+      "loss": 1.3856,
+      "step": 1743
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6e-07,
+      "loss": 1.4217,
+      "step": 1744
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5714285714285717e-07,
+      "loss": 1.3261,
+      "step": 1745
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.542857142857143e-07,
+      "loss": 1.0982,
+      "step": 1746
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5142857142857144e-07,
+      "loss": 1.3369,
+      "step": 1747
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.4857142857142857e-07,
+      "loss": 1.4063,
+      "step": 1748
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.4571428571428574e-07,
+      "loss": 1.275,
+      "step": 1749
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.4285714285714287e-07,
+      "loss": 1.318,
+      "step": 1750
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.4e-07,
+      "loss": 1.0599,
+      "step": 1751
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.3714285714285715e-07,
+      "loss": 1.1394,
+      "step": 1752
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.342857142857143e-07,
+      "loss": 1.095,
+      "step": 1753
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.3142857142857144e-07,
+      "loss": 0.9051,
+      "step": 1754
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2857142857142858e-07,
+      "loss": 1.3606,
+      "step": 1755
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2571428571428572e-07,
+      "loss": 1.3122,
+      "step": 1756
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2285714285714285e-07,
+      "loss": 1.1528,
+      "step": 1757
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 1.1556,
+      "step": 1758
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.1714285714285714e-07,
+      "loss": 1.0586,
+      "step": 1759
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.142857142857143e-07,
+      "loss": 1.429,
+      "step": 1760
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.1142857142857145e-07,
+      "loss": 1.3722,
+      "step": 1761
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0857142857142857e-07,
+      "loss": 1.0878,
+      "step": 1762
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0571428571428572e-07,
+      "loss": 1.4304,
+      "step": 1763
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0285714285714286e-07,
+      "loss": 1.1535,
+      "step": 1764
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 1.2017,
+      "step": 1765
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.714285714285714e-08,
+      "loss": 1.0915,
+      "step": 1766
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.42857142857143e-08,
+      "loss": 1.0701,
+      "step": 1767
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.142857142857144e-08,
+      "loss": 1.5921,
+      "step": 1768
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.857142857142858e-08,
+      "loss": 1.3195,
+      "step": 1769
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8.571428571428573e-08,
+      "loss": 1.0818,
+      "step": 1770
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8.285714285714285e-08,
+      "loss": 1.4569,
+      "step": 1771
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8e-08,
+      "loss": 1.4378,
+      "step": 1772
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.714285714285715e-08,
+      "loss": 1.3757,
+      "step": 1773
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.428571428571429e-08,
+      "loss": 0.9905,
+      "step": 1774
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.142857142857144e-08,
+      "loss": 1.0939,
+      "step": 1775
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.857142857142857e-08,
+      "loss": 1.1717,
+      "step": 1776
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.571428571428572e-08,
+      "loss": 1.3934,
+      "step": 1777
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.285714285714286e-08,
+      "loss": 1.0477,
+      "step": 1778
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 1.2222,
+      "step": 1779
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.714285714285715e-08,
+      "loss": 1.0024,
+      "step": 1780
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.428571428571429e-08,
+      "loss": 1.065,
+      "step": 1781
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.142857142857143e-08,
+      "loss": 1.1732,
+      "step": 1782
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.857142857142857e-08,
+      "loss": 1.2796,
+      "step": 1783
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.571428571428572e-08,
+      "loss": 1.3353,
+      "step": 1784
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.2857142857142865e-08,
+      "loss": 1.3447,
+      "step": 1785
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4e-08,
+      "loss": 1.2952,
+      "step": 1786
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.7142857142857144e-08,
+      "loss": 1.3102,
+      "step": 1787
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.4285714285714286e-08,
+      "loss": 0.9345,
+      "step": 1788
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.142857142857143e-08,
+      "loss": 1.1161,
+      "step": 1789
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.8571428571428575e-08,
+      "loss": 1.1129,
+      "step": 1790
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5714285714285715e-08,
+      "loss": 1.0069,
+      "step": 1791
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.285714285714286e-08,
+      "loss": 1.3098,
+      "step": 1792
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2e-08,
+      "loss": 1.213,
+      "step": 1793
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7142857142857143e-08,
+      "loss": 1.1657,
+      "step": 1794
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4285714285714288e-08,
+      "loss": 1.3396,
+      "step": 1795
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.142857142857143e-08,
+      "loss": 1.258,
+      "step": 1796
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.571428571428572e-09,
+      "loss": 1.3046,
+      "step": 1797
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.714285714285715e-09,
+      "loss": 1.5405,
+      "step": 1798
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.8571428571428576e-09,
+      "loss": 1.169,
+      "step": 1799
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 1.1869,
+      "step": 1800
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9333333333333333,
+      "eval_loss": 1.063901662826538,
+      "eval_roc_auc": 0.9858799160024966,
+      "eval_runtime": 14.0085,
+      "eval_samples_per_second": 25.699,
+      "eval_steps_per_second": 25.699,
+      "step": 1800
+    }
+  ],
+  "max_steps": 1800,
+  "num_train_epochs": 5,
+  "total_flos": 1.961021954304e+17,
+  "trial_name": null,
+  "trial_params": null
+}