diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16816 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.2493725249595626,
+  "global_step": 2800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4814814814814817e-07,
+      "loss": 1.2395,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.1916,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.2129,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.3046,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.1964,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.346,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.2637,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.1721,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.2269,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.2172,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.1346,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.3154,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.2148,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.248,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.1755,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.215,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.1714,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.21,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 1.1628,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 1.2156,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 1.2705,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 1.2158,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 1.2142,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 1.1957,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.925925925925927e-07,
+      "loss": 1.1827,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.925925925925927e-07,
+      "loss": 1.123,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 1.2762,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 1.1383,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 1.1178,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 1.1631,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 1.1585,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0370370370370371e-06,
+      "loss": 1.2011,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1851851851851854e-06,
+      "loss": 1.1888,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1851851851851854e-06,
+      "loss": 1.1774,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1851851851851854e-06,
+      "loss": 1.1833,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.1248,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.166,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.213,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 1.1453,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 1.1118,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.62962962962963e-06,
+      "loss": 1.1816,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 1.1498,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.925925925925926e-06,
+      "loss": 1.1777,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0740740740740742e-06,
+      "loss": 1.1404,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0740740740740742e-06,
+      "loss": 1.0087,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.9975,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.978,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.0679,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3703703703703707e-06,
+      "loss": 0.9708,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5185185185185186e-06,
+      "loss": 0.976,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.9367,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.8673,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.814814814814815e-06,
+      "loss": 0.9394,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.814814814814815e-06,
+      "loss": 0.8091,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.814814814814815e-06,
+      "loss": 0.7634,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 0.8342,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1111111111111116e-06,
+      "loss": 0.7766,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.25925925925926e-06,
+      "loss": 0.7033,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4074074074074077e-06,
+      "loss": 0.6843,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4074074074074077e-06,
+      "loss": 0.5797,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4074074074074077e-06,
+      "loss": 0.6404,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 0.6248,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 0.607,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.5903,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.851851851851852e-06,
+      "loss": 0.5757,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5247,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.4782,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5269,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.4935,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1481481481481485e-06,
+      "loss": 0.5112,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.296296296296296e-06,
+      "loss": 0.5435,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.4776,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.4394,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.592592592592593e-06,
+      "loss": 0.4765,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7407407407407415e-06,
+      "loss": 0.5083,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 0.4544,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 0.4698,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 0.4547,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.037037037037037e-06,
+      "loss": 0.478,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.4468,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.428,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.4428,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.4717,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.3759,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.4372,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.4482,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.4289,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.481481481481482e-06,
+      "loss": 0.4397,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.481481481481482e-06,
+      "loss": 0.4133,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.62962962962963e-06,
+      "loss": 0.4417,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.62962962962963e-06,
+      "loss": 0.3876,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 0.4369,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 0.4395,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0740740740740745e-06,
+      "loss": 0.4145,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 0.4139,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.370370370370371e-06,
+      "loss": 0.3849,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.370370370370371e-06,
+      "loss": 0.4012,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.370370370370371e-06,
+      "loss": 0.4156,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.370370370370371e-06,
+      "loss": 0.3919,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 0.405,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 0.3972,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 0.409,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 0.424,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 0.364,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 0.3933,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.4046,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.3909,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.3979,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.4278,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.3571,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.4305,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.814814814814815e-06,
+      "loss": 0.387,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.814814814814815e-06,
+      "loss": 0.424,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.962962962962964e-06,
+      "loss": 0.3768,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 0.3918,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2592592592592605e-06,
+      "loss": 0.4111,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2592592592592605e-06,
+      "loss": 0.4059,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.3991,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 0.3667,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.703703703703704e-06,
+      "loss": 0.3799,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.703703703703704e-06,
+      "loss": 0.4143,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.851851851851853e-06,
+      "loss": 0.3663,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.851851851851853e-06,
+      "loss": 0.3528,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.851851851851853e-06,
+      "loss": 0.389,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.851851851851853e-06,
+      "loss": 0.3719,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.4026,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 0.3532,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.296296296296297e-06,
+      "loss": 0.4226,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 0.377,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 0.3737,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.592592592592593e-06,
+      "loss": 0.369,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.740740740740741e-06,
+      "loss": 0.4158,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.4027,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.037037037037037e-06,
+      "loss": 0.3805,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 0.4544,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 0.3755,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 0.4179,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.3476,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.481481481481483e-06,
+      "loss": 0.3592,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.481481481481483e-06,
+      "loss": 0.3402,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.4224,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.3972,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 0.3959,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925925925925927e-06,
+      "loss": 0.3467,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0074074074074074e-05,
+      "loss": 0.4085,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 0.3628,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.3644,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0518518518518519e-05,
+      "loss": 0.3445,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0518518518518519e-05,
+      "loss": 0.3832,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.3733,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0814814814814816e-05,
+      "loss": 0.396,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0962962962962965e-05,
+      "loss": 0.365,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.3586,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.125925925925926e-05,
+      "loss": 0.3736,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1407407407407409e-05,
+      "loss": 0.4107,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1407407407407409e-05,
+      "loss": 0.3754,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1407407407407409e-05,
+      "loss": 0.4017,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 0.3516,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1703703703703703e-05,
+      "loss": 0.3328,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 0.3722,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 0.4009,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2e-05,
+      "loss": 0.3835,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2148148148148149e-05,
+      "loss": 0.4074,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2148148148148149e-05,
+      "loss": 0.3517,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2296296296296298e-05,
+      "loss": 0.3569,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2296296296296298e-05,
+      "loss": 0.379,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 0.3964,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.3787,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.3721,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.4368,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2740740740740742e-05,
+      "loss": 0.4295,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2740740740740742e-05,
+      "loss": 0.396,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 0.3832,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.303703703703704e-05,
+      "loss": 0.3643,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3185185185185185e-05,
+      "loss": 0.3993,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.3652,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3481481481481482e-05,
+      "loss": 0.3708,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.362962962962963e-05,
+      "loss": 0.3507,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.377777777777778e-05,
+      "loss": 0.3754,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3925925925925928e-05,
+      "loss": 0.3634,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.3163,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.3683,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.4118,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.355,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4370370370370372e-05,
+      "loss": 0.3629,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4370370370370372e-05,
+      "loss": 0.3599,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4370370370370372e-05,
+      "loss": 0.3952,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4518518518518521e-05,
+      "loss": 0.3497,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.377,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.3591,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.3952,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4962962962962964e-05,
+      "loss": 0.3493,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4962962962962964e-05,
+      "loss": 0.3686,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4962962962962964e-05,
+      "loss": 0.3916,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4962962962962964e-05,
+      "loss": 0.4107,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.373,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.4226,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.525925925925926e-05,
+      "loss": 0.3905,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5407407407407408e-05,
+      "loss": 0.3486,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5407407407407408e-05,
+      "loss": 0.3517,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5407407407407408e-05,
+      "loss": 0.3406,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5407407407407408e-05,
+      "loss": 0.3536,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.3645,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5703703703703705e-05,
+      "loss": 0.37,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5851851851851852e-05,
+      "loss": 0.3549,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.3584,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.614814814814815e-05,
+      "loss": 0.3503,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.3865,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6444444444444444e-05,
+      "loss": 0.3235,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6592592592592594e-05,
+      "loss": 0.4246,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.6592592592592594e-05,
+      "loss": 0.3731,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.674074074074074e-05,
+      "loss": 0.3962,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 0.3562,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.3098,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.3983,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.3525,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7185185185185185e-05,
+      "loss": 0.3438,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.3362,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.3574,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.3846,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7481481481481483e-05,
+      "loss": 0.3601,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7481481481481483e-05,
+      "loss": 0.3655,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7629629629629633e-05,
+      "loss": 0.3369,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.3573,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7925925925925927e-05,
+      "loss": 0.3209,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8074074074074074e-05,
+      "loss": 0.3719,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 0.3369,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.837037037037037e-05,
+      "loss": 0.373,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.851851851851852e-05,
+      "loss": 0.3539,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.3466,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.3782,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.3485,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8814814814814816e-05,
+      "loss": 0.3573,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8962962962962966e-05,
+      "loss": 0.3791,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 0.3819,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 0.3571,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.3777,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.3462,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.3883,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9407407407407407e-05,
+      "loss": 0.336,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.3662,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.3535,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703703703703704e-05,
+      "loss": 0.3606,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9851851851851855e-05,
+      "loss": 0.3421,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9851851851851855e-05,
+      "loss": 0.3981,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3684,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9999997388497475e-05,
+      "loss": 0.4011,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999998955399126e-05,
+      "loss": 0.3556,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999997649648544e-05,
+      "loss": 0.3611,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9999958215986846e-05,
+      "loss": 0.3595,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999993471250502e-05,
+      "loss": 0.3555,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999990598605224e-05,
+      "loss": 0.4016,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9999872036643514e-05,
+      "loss": 0.3447,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9999832864296565e-05,
+      "loss": 0.3441,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9999788469031855e-05,
+      "loss": 0.3644,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9999738850872576e-05,
+      "loss": 0.3315,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999968400984464e-05,
+      "loss": 0.3635,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999623945976692e-05,
+      "loss": 0.3193,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99995586593001e-05,
+      "loss": 0.4044,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999488149848972e-05,
+      "loss": 0.3316,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999488149848972e-05,
+      "loss": 0.3613,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999412417660125e-05,
+      "loss": 0.387,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999933146277312e-05,
+      "loss": 0.3629,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999245285230237e-05,
+      "loss": 0.384,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999245285230237e-05,
+      "loss": 0.3388,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999153885076488e-05,
+      "loss": 0.3532,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999057262359608e-05,
+      "loss": 0.3581,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9998955417130072e-05,
+      "loss": 0.3602,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999884834944106e-05,
+      "loss": 0.3476,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999873605934851e-05,
+      "loss": 0.3396,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999861854691106e-05,
+      "loss": 0.3641,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999849581219009e-05,
+      "loss": 0.3478,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999849581219009e-05,
+      "loss": 0.3862,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999849581219009e-05,
+      "loss": 0.3251,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99983678552497e-05,
+      "loss": 0.3677,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9998234676156726e-05,
+      "loss": 0.3663,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999809627498073e-05,
+      "loss": 0.3677,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9997952651793994e-05,
+      "loss": 0.3224,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9997803806671534e-05,
+      "loss": 0.3763,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9997649739691095e-05,
+      "loss": 0.3756,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9997490450933143e-05,
+      "loss": 0.3627,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9997325940480875e-05,
+      "loss": 0.3848,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9997156208420217e-05,
+      "loss": 0.3535,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9997156208420217e-05,
+      "loss": 0.3677,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9997156208420217e-05,
+      "loss": 0.3232,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9996981254839816e-05,
+      "loss": 0.3512,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9996801079831057e-05,
+      "loss": 0.3426,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9996801079831057e-05,
+      "loss": 0.3671,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9996801079831057e-05,
+      "loss": 0.3521,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9996615683488038e-05,
+      "loss": 0.3363,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9996425065907595e-05,
+      "loss": 0.3594,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999622922718929e-05,
+      "loss": 0.355,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999622922718929e-05,
+      "loss": 0.3525,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9996028167435404e-05,
+      "loss": 0.321,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9995821886750956e-05,
+      "loss": 0.3567,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9995610385243682e-05,
+      "loss": 0.3623,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9995610385243682e-05,
+      "loss": 0.3656,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9995393663024054e-05,
+      "loss": 0.3708,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9995171720205263e-05,
+      "loss": 0.3636,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999494455690323e-05,
+      "loss": 0.3551,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9994712173236604e-05,
+      "loss": 0.3497,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999447456932676e-05,
+      "loss": 0.3697,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9994231745297792e-05,
+      "loss": 0.3812,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9993983701276536e-05,
+      "loss": 0.359,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999373043739254e-05,
+      "loss": 0.3529,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999373043739254e-05,
+      "loss": 0.4001,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9993471953778084e-05,
+      "loss": 0.4135,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9993471953778084e-05,
+      "loss": 0.3543,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9993471953778084e-05,
+      "loss": 0.3222,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999320825056818e-05,
+      "loss": 0.3958,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999320825056818e-05,
+      "loss": 0.3724,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999320825056818e-05,
+      "loss": 0.3549,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999320825056818e-05,
+      "loss": 0.3715,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999320825056818e-05,
+      "loss": 0.3914,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9992939327900555e-05,
+      "loss": 0.3672,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9992665185915668e-05,
+      "loss": 0.356,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9992665185915668e-05,
+      "loss": 0.3535,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9992665185915668e-05,
+      "loss": 0.3511,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99923858247567e-05,
+      "loss": 0.3436,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99923858247567e-05,
+      "loss": 0.3988,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99923858247567e-05,
+      "loss": 0.3964,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99923858247567e-05,
+      "loss": 0.3574,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999210124456957e-05,
+      "loss": 0.3811,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9991811445502905e-05,
+      "loss": 0.3944,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9991516427708074e-05,
+      "loss": 0.3637,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9991216191339166e-05,
+      "loss": 0.3499,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9991216191339166e-05,
+      "loss": 0.3432,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999091073655299e-05,
+      "loss": 0.3439,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999060006350908e-05,
+      "loss": 0.3636,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9990284172369714e-05,
+      "loss": 0.3822,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9989963063299872e-05,
+      "loss": 0.3446,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9989636736467278e-05,
+      "loss": 0.372,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9989305192042365e-05,
+      "loss": 0.3309,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99889684301983e-05,
+      "loss": 0.3379,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9988626451110976e-05,
+      "loss": 0.3334,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9988626451110976e-05,
+      "loss": 0.3728,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9988279254959007e-05,
+      "loss": 0.3562,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9987926841923736e-05,
+      "loss": 0.364,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9987569212189224e-05,
+      "loss": 0.3625,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998720636594227e-05,
+      "loss": 0.3643,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998720636594227e-05,
+      "loss": 0.3402,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998683830337238e-05,
+      "loss": 0.3674,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998683830337238e-05,
+      "loss": 0.3286,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9986465024671796e-05,
+      "loss": 0.3389,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998608653003548e-05,
+      "loss": 0.3565,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9985702819661128e-05,
+      "loss": 0.389,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9985313893749144e-05,
+      "loss": 0.3933,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9984919752502665e-05,
+      "loss": 0.3745,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9984520396127554e-05,
+      "loss": 0.3883,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.998411582483239e-05,
+      "loss": 0.3411,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.998370603882849e-05,
+      "loss": 0.4061,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.998329103832988e-05,
+      "loss": 0.3413,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.998287082355331e-05,
+      "loss": 0.3178,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9982445394718265e-05,
+      "loss": 0.3663,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9982445394718265e-05,
+      "loss": 0.3978,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9982445394718265e-05,
+      "loss": 0.3578,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9982014752046945e-05,
+      "loss": 0.3369,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9981578895764272e-05,
+      "loss": 0.3398,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9981578895764272e-05,
+      "loss": 0.3575,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99811378260979e-05,
+      "loss": 0.321,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99806915432782e-05,
+      "loss": 0.3546,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.998024004753826e-05,
+      "loss": 0.345,
+      "step": 362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9979783339113897e-05,
+      "loss": 0.3889,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997932141824365e-05,
+      "loss": 0.3225,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9978854285168784e-05,
+      "loss": 0.3858,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9978854285168784e-05,
+      "loss": 0.3968,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9978381940133286e-05,
+      "loss": 0.3892,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997790438338385e-05,
+      "loss": 0.3516,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997790438338385e-05,
+      "loss": 0.3164,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9977421615169916e-05,
+      "loss": 0.3561,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9976933635743627e-05,
+      "loss": 0.3647,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9976440445359858e-05,
+      "loss": 0.3532,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9975942044276206e-05,
+      "loss": 0.3482,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9975942044276206e-05,
+      "loss": 0.3365,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9975438432752977e-05,
+      "loss": 0.3577,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9974929611053214e-05,
+      "loss": 0.342,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9974415579442677e-05,
+      "loss": 0.3528,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9973896338189838e-05,
+      "loss": 0.3169,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99733718875659e-05,
+      "loss": 0.3426,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99733718875659e-05,
+      "loss": 0.324,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9972842227844786e-05,
+      "loss": 0.3408,
+      "step": 381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9972307359303134e-05,
+      "loss": 0.3227,
+      "step": 382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.997176728222031e-05,
+      "loss": 0.3619,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9971221996878395e-05,
+      "loss": 0.3341,
+      "step": 384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9970671503562187e-05,
+      "loss": 0.3307,
+      "step": 385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9970671503562187e-05,
+      "loss": 0.3515,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9970671503562187e-05,
+      "loss": 0.3675,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.997011580255922e-05,
+      "loss": 0.3082,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9969554894159723e-05,
+      "loss": 0.312,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9968988778656672e-05,
+      "loss": 0.3406,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9968988778656672e-05,
+      "loss": 0.3112,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9968988778656672e-05,
+      "loss": 0.3586,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996841745634574e-05,
+      "loss": 0.3426,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996784092752533e-05,
+      "loss": 0.3741,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996725919249657e-05,
+      "loss": 0.3708,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996725919249657e-05,
+      "loss": 0.353,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996725919249657e-05,
+      "loss": 0.3736,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9966672251563293e-05,
+      "loss": 0.3751,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9966672251563293e-05,
+      "loss": 0.3575,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9966080105032062e-05,
+      "loss": 0.3429,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9965482753212154e-05,
+      "loss": 0.3357,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996488019641557e-05,
+      "loss": 0.36,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996427243495702e-05,
+      "loss": 0.4255,
+      "step": 403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996427243495702e-05,
+      "loss": 0.3442,
+      "step": 404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996427243495702e-05,
+      "loss": 0.3368,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996427243495702e-05,
+      "loss": 0.353,
+      "step": 406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996427243495702e-05,
+      "loss": 0.3443,
+      "step": 407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996427243495702e-05,
+      "loss": 0.3801,
+      "step": 408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9963659469153946e-05,
+      "loss": 0.3914,
+      "step": 409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9963659469153946e-05,
+      "loss": 0.3123,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9963041299326492e-05,
+      "loss": 0.3336,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9963041299326492e-05,
+      "loss": 0.368,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9963041299326492e-05,
+      "loss": 0.3669,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9963041299326492e-05,
+      "loss": 0.337,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9963041299326492e-05,
+      "loss": 0.3613,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9963041299326492e-05,
+      "loss": 0.3681,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9962417925797532e-05,
+      "loss": 0.3337,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9961789348892653e-05,
+      "loss": 0.3469,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9961155568940164e-05,
+      "loss": 0.3554,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9961155568940164e-05,
+      "loss": 0.3399,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9960516586271083e-05,
+      "loss": 0.3482,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9959872401219156e-05,
+      "loss": 0.3703,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9959223014120842e-05,
+      "loss": 0.3718,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9958568425315316e-05,
+      "loss": 0.3438,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9957908635144463e-05,
+      "loss": 0.3357,
+      "step": 425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99572436439529e-05,
+      "loss": 0.3422,
+      "step": 426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9956573452087946e-05,
+      "loss": 0.355,
+      "step": 427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.995589805989965e-05,
+      "loss": 0.3466,
+      "step": 428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9955217467740763e-05,
+      "loss": 0.3419,
+      "step": 429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9955217467740763e-05,
+      "loss": 0.3293,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9955217467740763e-05,
+      "loss": 0.3269,
+      "step": 431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.995453167596676e-05,
+      "loss": 0.3216,
+      "step": 432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.995384068493583e-05,
+      "loss": 0.3207,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9953144495008884e-05,
+      "loss": 0.3393,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9953144495008884e-05,
+      "loss": 0.3701,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9952443106549535e-05,
+      "loss": 0.3588,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.995173651992412e-05,
+      "loss": 0.3417,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.995102473550169e-05,
+      "loss": 0.3657,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9950307753654016e-05,
+      "loss": 0.3031,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.994958557475557e-05,
+      "loss": 0.3442,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9948858199183548e-05,
+      "loss": 0.3496,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9948125627317864e-05,
+      "loss": 0.3459,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9948125627317864e-05,
+      "loss": 0.3008,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9948125627317864e-05,
+      "loss": 0.3366,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9947387859541133e-05,
+      "loss": 0.3764,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9947387859541133e-05,
+      "loss": 0.3514,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9946644896238697e-05,
+      "loss": 0.3346,
+      "step": 447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9945896737798603e-05,
+      "loss": 0.3246,
+      "step": 448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.994514338461162e-05,
+      "loss": 0.3404,
+      "step": 449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9944384837071218e-05,
+      "loss": 0.345,
+      "step": 450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9943621095573588e-05,
+      "loss": 0.3366,
+      "step": 451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9943621095573588e-05,
+      "loss": 0.3472,
+      "step": 452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9942852160517633e-05,
+      "loss": 0.3486,
+      "step": 453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9942078032304972e-05,
+      "loss": 0.3822,
+      "step": 454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9941298711339932e-05,
+      "loss": 0.3472,
+      "step": 455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9940514198029545e-05,
+      "loss": 0.389,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9939724492783573e-05,
+      "loss": 0.358,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993892959601447e-05,
+      "loss": 0.3259,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993892959601447e-05,
+      "loss": 0.3794,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993892959601447e-05,
+      "loss": 0.3496,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9938129508137417e-05,
+      "loss": 0.3087,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99373242295703e-05,
+      "loss": 0.3709,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99373242295703e-05,
+      "loss": 0.3634,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9936513760733714e-05,
+      "loss": 0.3787,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993569810205097e-05,
+      "loss": 0.3178,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993569810205097e-05,
+      "loss": 0.3612,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9934877253948086e-05,
+      "loss": 0.3136,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9934877253948086e-05,
+      "loss": 0.3701,
+      "step": 468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993405121685379e-05,
+      "loss": 0.3486,
+      "step": 469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993405121685379e-05,
+      "loss": 0.3683,
+      "step": 470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993405121685379e-05,
+      "loss": 0.3108,
+      "step": 471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993405121685379e-05,
+      "loss": 0.372,
+      "step": 472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993405121685379e-05,
+      "loss": 0.3263,
+      "step": 473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9933219991199523e-05,
+      "loss": 0.3594,
+      "step": 474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9933219991199523e-05,
+      "loss": 0.3809,
+      "step": 475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.3691,
+      "step": 476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993154197595038e-05,
+      "loss": 0.3722,
+      "step": 477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930695187231932e-05,
+      "loss": 0.3817,
+      "step": 478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929843211706372e-05,
+      "loss": 0.3845,
+      "step": 479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929843211706372e-05,
+      "loss": 0.3794,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992898604981868e-05,
+      "loss": 0.3474,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992898604981868e-05,
+      "loss": 0.3168,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992812370201656e-05,
+      "loss": 0.3347,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992812370201656e-05,
+      "loss": 0.3438,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992725616875041e-05,
+      "loss": 0.334,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992725616875041e-05,
+      "loss": 0.3456,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926383450473344e-05,
+      "loss": 0.3705,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925505547641188e-05,
+      "loss": 0.3613,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9924622460712468e-05,
+      "loss": 0.3437,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9923734190148417e-05,
+      "loss": 0.3543,
+      "step": 490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9922840736412987e-05,
+      "loss": 0.3548,
+      "step": 491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992194209997282e-05,
+      "loss": 0.3752,
+      "step": 492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992194209997282e-05,
+      "loss": 0.3732,
+      "step": 493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9921038281297282e-05,
+      "loss": 0.3425,
+      "step": 494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9920129280858434e-05,
+      "loss": 0.3653,
+      "step": 495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9919215099131053e-05,
+      "loss": 0.3353,
+      "step": 496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.991829573659261e-05,
+      "loss": 0.3778,
+      "step": 497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9917371193723287e-05,
+      "loss": 0.3395,
+      "step": 498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.991644147100598e-05,
+      "loss": 0.371,
+      "step": 499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9915506568926283e-05,
+      "loss": 0.3674,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.991456648797249e-05,
+      "loss": 0.3225,
+      "step": 501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.991456648797249e-05,
+      "loss": 0.3437,
+      "step": 502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9913621228635608e-05,
+      "loss": 0.3361,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9912670791409352e-05,
+      "loss": 0.3354,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9912670791409352e-05,
+      "loss": 0.3473,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991171517679013e-05,
+      "loss": 0.3904,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991075438527706e-05,
+      "loss": 0.3911,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9909788417371965e-05,
+      "loss": 0.3351,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9908817273579375e-05,
+      "loss": 0.3502,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9908817273579375e-05,
+      "loss": 0.3384,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990784095440651e-05,
+      "loss": 0.3398,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9906859460363307e-05,
+      "loss": 0.3259,
+      "step": 512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9905872791962408e-05,
+      "loss": 0.3608,
+      "step": 513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9904880949719133e-05,
+      "loss": 0.3812,
+      "step": 514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9903883934151537e-05,
+      "loss": 0.3249,
+      "step": 515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9902881745780357e-05,
+      "loss": 0.3714,
+      "step": 516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990187438512903e-05,
+      "loss": 0.3255,
+      "step": 517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9900861852723716e-05,
+      "loss": 0.3224,
+      "step": 518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9899844149093244e-05,
+      "loss": 0.3407,
+      "step": 519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9899844149093244e-05,
+      "loss": 0.3657,
+      "step": 520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9898821274769174e-05,
+      "loss": 0.3578,
+      "step": 521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989779323028575e-05,
+      "loss": 0.3676,
+      "step": 522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9896760016179918e-05,
+      "loss": 0.3725,
+      "step": 523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9895721632991324e-05,
+      "loss": 0.3477,
+      "step": 524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894678081262323e-05,
+      "loss": 0.3452,
+      "step": 525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894678081262323e-05,
+      "loss": 0.3272,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.989362936153796e-05,
+      "loss": 0.3362,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9892575474365977e-05,
+      "loss": 0.36,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9892575474365977e-05,
+      "loss": 0.3584,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9892575474365977e-05,
+      "loss": 0.3147,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9892575474365977e-05,
+      "loss": 0.362,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9892575474365977e-05,
+      "loss": 0.3293,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9891516420296825e-05,
+      "loss": 0.3412,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9891516420296825e-05,
+      "loss": 0.4041,
+      "step": 534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9891516420296825e-05,
+      "loss": 0.3782,
+      "step": 535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.989045219988365e-05,
+      "loss": 0.3386,
+      "step": 536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.988938281368229e-05,
+      "loss": 0.3654,
+      "step": 537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 0.3274,
+      "step": 538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.988722854615188e-05,
+      "loss": 0.3467,
+      "step": 539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.988614366594801e-05,
+      "loss": 0.3326,
+      "step": 540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.988614366594801e-05,
+      "loss": 0.3611,
+      "step": 541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9885053622206305e-05,
+      "loss": 0.3634,
+      "step": 542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9885053622206305e-05,
+      "loss": 0.3611,
+      "step": 543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9883958415496095e-05,
+      "loss": 0.3619,
+      "step": 544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.988285804638941e-05,
+      "loss": 0.348,
+      "step": 545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9881752515460974e-05,
+      "loss": 0.3561,
+      "step": 546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98806418232882e-05,
+      "loss": 0.3376,
+      "step": 547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9879525970451214e-05,
+      "loss": 0.3809,
+      "step": 548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9878404957532817e-05,
+      "loss": 0.3744,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.987727878511852e-05,
+      "loss": 0.3251,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.987614745379652e-05,
+      "loss": 0.3219,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.987614745379652e-05,
+      "loss": 0.3417,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.987501096415771e-05,
+      "loss": 0.3686,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9873869316795684e-05,
+      "loss": 0.3829,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9872722512306723e-05,
+      "loss": 0.3267,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9871570551289804e-05,
+      "loss": 0.3683,
+      "step": 556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98704134343466e-05,
+      "loss": 0.3971,
+      "step": 557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98704134343466e-05,
+      "loss": 0.3116,
+      "step": 558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98704134343466e-05,
+      "loss": 0.3568,
+      "step": 559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98704134343466e-05,
+      "loss": 0.3568,
+      "step": 560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9869251162081468e-05,
+      "loss": 0.2864,
+      "step": 561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9869251162081468e-05,
+      "loss": 0.3278,
+      "step": 562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9868083735101464e-05,
+      "loss": 0.3364,
+      "step": 563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986691115401634e-05,
+      "loss": 0.3788,
+      "step": 564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9865733419438534e-05,
+      "loss": 0.348,
+      "step": 565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9864550531983172e-05,
+      "loss": 0.3615,
+      "step": 566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9863362492268088e-05,
+      "loss": 0.3091,
+      "step": 567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9862169300913784e-05,
+      "loss": 0.3776,
+      "step": 568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986097095854347e-05,
+      "loss": 0.349,
+      "step": 569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985976746578304e-05,
+      "loss": 0.3536,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985855882326108e-05,
+      "loss": 0.3413,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9857345031608868e-05,
+      "loss": 0.3586,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.985612609146036e-05,
+      "loss": 0.3576,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9854902003452212e-05,
+      "loss": 0.3631,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9853672768223768e-05,
+      "loss": 0.2969,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9852438386417058e-05,
+      "loss": 0.3424,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9852438386417058e-05,
+      "loss": 0.3589,
+      "step": 577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9851198858676796e-05,
+      "loss": 0.3319,
+      "step": 578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9849954185650393e-05,
+      "loss": 0.3697,
+      "step": 579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9849954185650393e-05,
+      "loss": 0.299,
+      "step": 580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9848704367987937e-05,
+      "loss": 0.3113,
+      "step": 581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9847449406342216e-05,
+      "loss": 0.352,
+      "step": 582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984618930136869e-05,
+      "loss": 0.2942,
+      "step": 583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9844924053725517e-05,
+      "loss": 0.3236,
+      "step": 584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9844924053725517e-05,
+      "loss": 0.3114,
+      "step": 585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9844924053725517e-05,
+      "loss": 0.3233,
+      "step": 586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9844924053725517e-05,
+      "loss": 0.3492,
+      "step": 587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9843653664073535e-05,
+      "loss": 0.3427,
+      "step": 588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9843653664073535e-05,
+      "loss": 0.3211,
+      "step": 589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984237813307627e-05,
+      "loss": 0.3482,
+      "step": 590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984109746139993e-05,
+      "loss": 0.3427,
+      "step": 591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839811649713413e-05,
+      "loss": 0.3347,
+      "step": 592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839811649713413e-05,
+      "loss": 0.3202,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9838520698688294e-05,
+      "loss": 0.3597,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9837224608998847e-05,
+      "loss": 0.357,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9835923381322012e-05,
+      "loss": 0.3124,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9834617016337424e-05,
+      "loss": 0.3651,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9833305514727396e-05,
+      "loss": 0.3329,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.983198887717693e-05,
+      "loss": 0.3154,
+      "step": 599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9830667104373703e-05,
+      "loss": 0.3218,
+      "step": 600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9830667104373703e-05,
+      "loss": 0.3236,
+      "step": 601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9829340197008076e-05,
+      "loss": 0.3599,
+      "step": 602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98280081557731e-05,
+      "loss": 0.3372,
+      "step": 603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9826670981364494e-05,
+      "loss": 0.2978,
+      "step": 604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9825328674480672e-05,
+      "loss": 0.3276,
+      "step": 605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9823981235822714e-05,
+      "loss": 0.3288,
+      "step": 606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982262866609439e-05,
+      "loss": 0.3344,
+      "step": 607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821270966002154e-05,
+      "loss": 0.3422,
+      "step": 608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821270966002154e-05,
+      "loss": 0.3339,
+      "step": 609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819908136255124e-05,
+      "loss": 0.3472,
+      "step": 610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818540177565112e-05,
+      "loss": 0.3077,
+      "step": 611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817167090646603e-05,
+      "loss": 0.3176,
+      "step": 612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981578887621676e-05,
+      "loss": 0.3657,
+      "step": 613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981440553499543e-05,
+      "loss": 0.3307,
+      "step": 614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813017067705124e-05,
+      "loss": 0.3469,
+      "step": 615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813017067705124e-05,
+      "loss": 0.308,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9811623475071044e-05,
+      "loss": 0.3391,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9810224757821063e-05,
+      "loss": 0.3155,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9808820916685738e-05,
+      "loss": 0.3568,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9807411952398286e-05,
+      "loss": 0.3556,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9805997865694616e-05,
+      "loss": 0.347,
+      "step": 621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.98045786573133e-05,
+      "loss": 0.317,
+      "step": 622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.98031543279956e-05,
+      "loss": 0.3554,
+      "step": 623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801724878485438e-05,
+      "loss": 0.3347,
+      "step": 624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980029030952942e-05,
+      "loss": 0.3308,
+      "step": 625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798850621876822e-05,
+      "loss": 0.3119,
+      "step": 626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797405816279588e-05,
+      "loss": 0.3166,
+      "step": 627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795955893492344e-05,
+      "loss": 0.3576,
+      "step": 628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794500854272387e-05,
+      "loss": 0.3206,
+      "step": 629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793040699379683e-05,
+      "loss": 0.3547,
+      "step": 630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793040699379683e-05,
+      "loss": 0.3338,
+      "step": 631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979157542957687e-05,
+      "loss": 0.3823,
+      "step": 632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790105045629264e-05,
+      "loss": 0.3581,
+      "step": 633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790105045629264e-05,
+      "loss": 0.3356,
+      "step": 634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788629548304842e-05,
+      "loss": 0.3126,
+      "step": 635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978714893837426e-05,
+      "loss": 0.308,
+      "step": 636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978714893837426e-05,
+      "loss": 0.356,
+      "step": 637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785663216610843e-05,
+      "loss": 0.3171,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9785663216610843e-05,
+      "loss": 0.3174,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9785663216610843e-05,
+      "loss": 0.345,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9784172383790584e-05,
+      "loss": 0.3365,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9782676440692137e-05,
+      "loss": 0.3256,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9781175388096843e-05,
+      "loss": 0.3355,
+      "step": 643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779669226788703e-05,
+      "loss": 0.3289,
+      "step": 644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779669226788703e-05,
+      "loss": 0.3177,
+      "step": 645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977815795755438e-05,
+      "loss": 0.3324,
+      "step": 646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776641581183215e-05,
+      "loss": 0.3356,
+      "step": 647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775120098467212e-05,
+      "loss": 0.3599,
+      "step": 648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773593510201037e-05,
+      "loss": 0.3117,
+      "step": 649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773593510201037e-05,
+      "loss": 0.3345,
+      "step": 650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772061817182036e-05,
+      "loss": 0.3479,
+      "step": 651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772061817182036e-05,
+      "loss": 0.354,
+      "step": 652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770525020210204e-05,
+      "loss": 0.3059,
+      "step": 653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976898312008822e-05,
+      "loss": 0.3638,
+      "step": 654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767436117621416e-05,
+      "loss": 0.3403,
+      "step": 655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765884013617785e-05,
+      "loss": 0.332,
+      "step": 656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764326808888002e-05,
+      "loss": 0.3338,
+      "step": 657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976276450424539e-05,
+      "loss": 0.3335,
+      "step": 658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761197100505943e-05,
+      "loss": 0.3323,
+      "step": 659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975962459848832e-05,
+      "loss": 0.3199,
+      "step": 660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758046999013835e-05,
+      "loss": 0.3309,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756464302906468e-05,
+      "loss": 0.3038,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754876510992863e-05,
+      "loss": 0.326,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754876510992863e-05,
+      "loss": 0.2976,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754876510992863e-05,
+      "loss": 0.3166,
+      "step": 665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975328362410233e-05,
+      "loss": 0.3432,
+      "step": 666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975168564306683e-05,
+      "loss": 0.328,
+      "step": 667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750082568720987e-05,
+      "loss": 0.3535,
+      "step": 668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748474401902095e-05,
+      "loss": 0.3441,
+      "step": 669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974686114345009e-05,
+      "loss": 0.3536,
+      "step": 670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974524279420759e-05,
+      "loss": 0.3399,
+      "step": 671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743619355019844e-05,
+      "loss": 0.3344,
+      "step": 672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743619355019844e-05,
+      "loss": 0.371,
+      "step": 673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743619355019844e-05,
+      "loss": 0.3268,
+      "step": 674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743619355019844e-05,
+      "loss": 0.3342,
+      "step": 675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741990826734793e-05,
+      "loss": 0.3451,
+      "step": 676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740357210203004e-05,
+      "loss": 0.3697,
+      "step": 677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738718506277722e-05,
+      "loss": 0.317,
+      "step": 678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737074715814845e-05,
+      "loss": 0.3373,
+      "step": 679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973542583967292e-05,
+      "loss": 0.3126,
+      "step": 680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733771878713157e-05,
+      "loss": 0.3531,
+      "step": 681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732112833799423e-05,
+      "loss": 0.3034,
+      "step": 682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.3376,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728779495578777e-05,
+      "loss": 0.3378,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972710520401287e-05,
+      "loss": 0.3345,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725425831974993e-05,
+      "loss": 0.3323,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972374138034229e-05,
+      "loss": 0.3444,
+      "step": 687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972374138034229e-05,
+      "loss": 0.3689,
+      "step": 688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722051849994552e-05,
+      "loss": 0.3459,
+      "step": 689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720357241814216e-05,
+      "loss": 0.3271,
+      "step": 690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971865755668638e-05,
+      "loss": 0.3262,
+      "step": 691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716952795498792e-05,
+      "loss": 0.3652,
+      "step": 692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716952795498792e-05,
+      "loss": 0.343,
+      "step": 693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715242959141846e-05,
+      "loss": 0.3092,
+      "step": 694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715242959141846e-05,
+      "loss": 0.3354,
+      "step": 695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971352804850859e-05,
+      "loss": 0.3615,
+      "step": 696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971180806449473e-05,
+      "loss": 0.2784,
+      "step": 697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710083007998606e-05,
+      "loss": 0.3493,
+      "step": 698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970835287992122e-05,
+      "loss": 0.3389,
+      "step": 699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970835287992122e-05,
+      "loss": 0.3488,
+      "step": 700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970835287992122e-05,
+      "loss": 0.3908,
+      "step": 701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970661768116622e-05,
+      "loss": 0.3708,
+      "step": 702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9704877412639895e-05,
+      "loss": 0.3343,
+      "step": 703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9703132075251195e-05,
+      "loss": 0.3438,
+      "step": 704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9701381669911706e-05,
+      "loss": 0.3474,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699626197535667e-05,
+      "loss": 0.3414,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697865659039966e-05,
+      "loss": 0.3711,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697865659039966e-05,
+      "loss": 0.3111,
+      "step": 708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696100055344126e-05,
+      "loss": 0.3003,
+      "step": 709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696100055344126e-05,
+      "loss": 0.3361,
+      "step": 710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694329387370327e-05,
+      "loss": 0.3324,
+      "step": 711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692553656043392e-05,
+      "loss": 0.3366,
+      "step": 712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692553656043392e-05,
+      "loss": 0.3222,
+      "step": 713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692553656043392e-05,
+      "loss": 0.3048,
+      "step": 714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969077286229078e-05,
+      "loss": 0.3243,
+      "step": 715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688987007042606e-05,
+      "loss": 0.3429,
+      "step": 716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968719609123162e-05,
+      "loss": 0.3512,
+      "step": 717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685400115793222e-05,
+      "loss": 0.3798,
+      "step": 718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685400115793222e-05,
+      "loss": 0.3476,
+      "step": 719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683599081665448e-05,
+      "loss": 0.3704,
+      "step": 720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681792989788973e-05,
+      "loss": 0.3233,
+      "step": 721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681792989788973e-05,
+      "loss": 0.3547,
+      "step": 722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.967998184110713e-05,
+      "loss": 0.3296,
+      "step": 723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9678165636565882e-05,
+      "loss": 0.3283,
+      "step": 724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9678165636565882e-05,
+      "loss": 0.3093,
+      "step": 725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9678165636565882e-05,
+      "loss": 0.3605,
+      "step": 726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9676344377113827e-05,
+      "loss": 0.3235,
+      "step": 727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9674518063702216e-05,
+      "loss": 0.3304,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672686697284926e-05,
+      "loss": 0.3181,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967085027881849e-05,
+      "loss": 0.3068,
+      "step": 730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669008809262064e-05,
+      "loss": 0.3342,
+      "step": 731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667162289577447e-05,
+      "loss": 0.3375,
+      "step": 732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665310720729083e-05,
+      "loss": 0.3017,
+      "step": 733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665310720729083e-05,
+      "loss": 0.3451,
+      "step": 734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663454103684043e-05,
+      "loss": 0.3238,
+      "step": 735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966159243941204e-05,
+      "loss": 0.3389,
+      "step": 736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966159243941204e-05,
+      "loss": 0.3468,
+      "step": 737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659725728885423e-05,
+      "loss": 0.3154,
+      "step": 738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657853973079176e-05,
+      "loss": 0.3017,
+      "step": 739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655977172970918e-05,
+      "loss": 0.2992,
+      "step": 740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655977172970918e-05,
+      "loss": 0.3662,
+      "step": 741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96540953295409e-05,
+      "loss": 0.2892,
+      "step": 742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652208443772016e-05,
+      "loss": 0.3302,
+      "step": 743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965031651664978e-05,
+      "loss": 0.3171,
+      "step": 744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.964841954916235e-05,
+      "loss": 0.3482,
+      "step": 745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9646517542300514e-05,
+      "loss": 0.3437,
+      "step": 746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9646517542300514e-05,
+      "loss": 0.3619,
+      "step": 747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9646517542300514e-05,
+      "loss": 0.3447,
+      "step": 748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9646517542300514e-05,
+      "loss": 0.3256,
+      "step": 749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.964461049705769e-05,
+      "loss": 0.3237,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964269841442993e-05,
+      "loss": 0.3476,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964078129541591e-05,
+      "loss": 0.3476,
+      "step": 752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963885914101695e-05,
+      "loss": 0.3257,
+      "step": 753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636931952236987e-05,
+      "loss": 0.3364,
+      "step": 754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634999730082598e-05,
+      "loss": 0.3375,
+      "step": 755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633062475562975e-05,
+      "loss": 0.3553,
+      "step": 756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631120189689956e-05,
+      "loss": 0.3125,
+      "step": 757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629172873477995e-05,
+      "loss": 0.2998,
+      "step": 758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629172873477995e-05,
+      "loss": 0.3207,
+      "step": 759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627220527944173e-05,
+      "loss": 0.3576,
+      "step": 760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627220527944173e-05,
+      "loss": 0.3269,
+      "step": 761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625263154108204e-05,
+      "loss": 0.3227,
+      "step": 762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9623300752992424e-05,
+      "loss": 0.3427,
+      "step": 763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9621333325621798e-05,
+      "loss": 0.3504,
+      "step": 764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9619360873023914e-05,
+      "loss": 0.3234,
+      "step": 765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9619360873023914e-05,
+      "loss": 0.3289,
+      "step": 766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9619360873023914e-05,
+      "loss": 0.2995,
+      "step": 767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9619360873023914e-05,
+      "loss": 0.3471,
+      "step": 768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9619360873023914e-05,
+      "loss": 0.3316,
+      "step": 769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.961738339622898e-05,
+      "loss": 0.3307,
+      "step": 770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.961738339622898e-05,
+      "loss": 0.3088,
+      "step": 771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.961540089626984e-05,
+      "loss": 0.3347,
+      "step": 772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9613413374181954e-05,
+      "loss": 0.3238,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613413374181954e-05,
+      "loss": 0.3196,
+      "step": 774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96114208310034e-05,
+      "loss": 0.2959,
+      "step": 775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96114208310034e-05,
+      "loss": 0.3242,
+      "step": 776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96114208310034e-05,
+      "loss": 0.3302,
+      "step": 777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960942326777489e-05,
+      "loss": 0.3445,
+      "step": 778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960942326777489e-05,
+      "loss": 0.3307,
+      "step": 779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960942326777489e-05,
+      "loss": 0.3318,
+      "step": 780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960742068553975e-05,
+      "loss": 0.3399,
+      "step": 781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960541308534393e-05,
+      "loss": 0.3322,
+      "step": 782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603400468236e-05,
+      "loss": 0.3653,
+      "step": 783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601382835267154e-05,
+      "loss": 0.3281,
+      "step": 784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599360187491195e-05,
+      "loss": 0.3556,
+      "step": 785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597332525964564e-05,
+      "loss": 0.3295,
+      "step": 786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595299851746302e-05,
+      "loss": 0.3388,
+      "step": 787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9593262165898078e-05,
+      "loss": 0.3419,
+      "step": 788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9591219469484174e-05,
+      "loss": 0.3072,
+      "step": 789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9589171763571494e-05,
+      "loss": 0.3127,
+      "step": 790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9589171763571494e-05,
+      "loss": 0.3693,
+      "step": 791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9587119049229558e-05,
+      "loss": 0.3257,
+      "step": 792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9585061327530493e-05,
+      "loss": 0.3398,
+      "step": 793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9582998599549056e-05,
+      "loss": 0.3189,
+      "step": 794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9580930866362605e-05,
+      "loss": 0.3155,
+      "step": 795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957885812905112e-05,
+      "loss": 0.309,
+      "step": 796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576780388697193e-05,
+      "loss": 0.3048,
+      "step": 797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576780388697193e-05,
+      "loss": 0.3322,
+      "step": 798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576780388697193e-05,
+      "loss": 0.3114,
+      "step": 799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576780388697193e-05,
+      "loss": 0.2898,
+      "step": 800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576780388697193e-05,
+      "loss": 0.3199,
+      "step": 801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574697646386027e-05,
+      "loss": 0.322,
+      "step": 802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572609903205442e-05,
+      "loss": 0.3294,
+      "step": 803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957051716024587e-05,
+      "loss": 0.3326,
+      "step": 804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956841941860034e-05,
+      "loss": 0.3512,
+      "step": 805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956631667936452e-05,
+      "loss": 0.3193,
+      "step": 806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956631667936452e-05,
+      "loss": 0.3296,
+      "step": 807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956420894363666e-05,
+      "loss": 0.3669,
+      "step": 808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956209621251763e-05,
+      "loss": 0.3497,
+      "step": 809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9559978487110924e-05,
+      "loss": 0.3271,
+      "step": 810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9557855768522613e-05,
+      "loss": 0.3266,
+      "step": 811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9557855768522613e-05,
+      "loss": 0.3624,
+      "step": 812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.3268,
+      "step": 813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.955359535623861e-05,
+      "loss": 0.336,
+      "step": 814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.955359535623861e-05,
+      "loss": 0.361,
+      "step": 815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9551457664768126e-05,
+      "loss": 0.3644,
+      "step": 816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9549314984566477e-05,
+      "loss": 0.3091,
+      "step": 817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9547167316752786e-05,
+      "loss": 0.3157,
+      "step": 818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954501466244878e-05,
+      "loss": 0.3183,
+      "step": 819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954285702277879e-05,
+      "loss": 0.3466,
+      "step": 820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540694398869754e-05,
+      "loss": 0.3377,
+      "step": 821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538526791851214e-05,
+      "loss": 0.3278,
+      "step": 822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536354202855306e-05,
+      "loss": 0.3464,
+      "step": 823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953417663301678e-05,
+      "loss": 0.3114,
+      "step": 824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531994083472978e-05,
+      "loss": 0.3358,
+      "step": 825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531994083472978e-05,
+      "loss": 0.3292,
+      "step": 826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.952980655536385e-05,
+      "loss": 0.3329,
+      "step": 827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.952761404983194e-05,
+      "loss": 0.3263,
+      "step": 828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95254165680224e-05,
+      "loss": 0.3357,
+      "step": 829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95254165680224e-05,
+      "loss": 0.3298,
+      "step": 830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95254165680224e-05,
+      "loss": 0.317,
+      "step": 831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.952321411108297e-05,
+      "loss": 0.3581,
+      "step": 832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9521006680163995e-05,
+      "loss": 0.3358,
+      "step": 833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.951879427641842e-05,
+      "loss": 0.3099,
+      "step": 834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.951879427641842e-05,
+      "loss": 0.3388,
+      "step": 835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.951879427641842e-05,
+      "loss": 0.2878,
+      "step": 836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.951657690100178e-05,
+      "loss": 0.3319,
+      "step": 837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.951435455507222e-05,
+      "loss": 0.2944,
+      "step": 838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9512127239790463e-05,
+      "loss": 0.3577,
+      "step": 839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9509894956319843e-05,
+      "loss": 0.3101,
+      "step": 840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950765770582628e-05,
+      "loss": 0.3301,
+      "step": 841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505415489478293e-05,
+      "loss": 0.362,
+      "step": 842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950316830844699e-05,
+      "loss": 0.3538,
+      "step": 843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500916163906075e-05,
+      "loss": 0.2982,
+      "step": 844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949865905703185e-05,
+      "loss": 0.2988,
+      "step": 845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9496396989003195e-05,
+      "loss": 0.3197,
+      "step": 846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9494129961001592e-05,
+      "loss": 0.3101,
+      "step": 847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9494129961001592e-05,
+      "loss": 0.3564,
+      "step": 848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9491857974211114e-05,
+      "loss": 0.3399,
+      "step": 849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9491857974211114e-05,
+      "loss": 0.3505,
+      "step": 850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9491857974211114e-05,
+      "loss": 0.3233,
+      "step": 851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9491857974211114e-05,
+      "loss": 0.3246,
+      "step": 852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.948958102981842e-05,
+      "loss": 0.3501,
+      "step": 853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9487299129012757e-05,
+      "loss": 0.3151,
+      "step": 854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9485012272985963e-05,
+      "loss": 0.3298,
+      "step": 855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9485012272985963e-05,
+      "loss": 0.3591,
+      "step": 856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9482720462932464e-05,
+      "loss": 0.2955,
+      "step": 857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9482720462932464e-05,
+      "loss": 0.3116,
+      "step": 858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9482720462932464e-05,
+      "loss": 0.3175,
+      "step": 859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9480423700049275e-05,
+      "loss": 0.3516,
+      "step": 860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9480423700049275e-05,
+      "loss": 0.3405,
+      "step": 861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9478121985535993e-05,
+      "loss": 0.3405,
+      "step": 862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947581532059481e-05,
+      "loss": 0.3458,
+      "step": 863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94735037064305e-05,
+      "loss": 0.3095,
+      "step": 864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471187144250407e-05,
+      "loss": 0.3074,
+      "step": 865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471187144250407e-05,
+      "loss": 0.3373,
+      "step": 866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946886563526449e-05,
+      "loss": 0.3214,
+      "step": 867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946886563526449e-05,
+      "loss": 0.3357,
+      "step": 868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946886563526449e-05,
+      "loss": 0.3594,
+      "step": 869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946653918068526e-05,
+      "loss": 0.3367,
+      "step": 870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946653918068526e-05,
+      "loss": 0.3247,
+      "step": 871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464207781727837e-05,
+      "loss": 0.3054,
+      "step": 872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94618714396099e-05,
+      "loss": 0.3284,
+      "step": 873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945953015555173e-05,
+      "loss": 0.3188,
+      "step": 874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945718393077618e-05,
+      "loss": 0.3266,
+      "step": 875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945718393077618e-05,
+      "loss": 0.3204,
+      "step": 876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945718393077618e-05,
+      "loss": 0.3176,
+      "step": 877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945718393077618e-05,
+      "loss": 0.3544,
+      "step": 878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945483276650868e-05,
+      "loss": 0.3211,
+      "step": 879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945247666397725e-05,
+      "loss": 0.3436,
+      "step": 880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945011562441248e-05,
+      "loss": 0.3218,
+      "step": 881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945011562441248e-05,
+      "loss": 0.3229,
+      "step": 882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.944774964904754e-05,
+      "loss": 0.3284,
+      "step": 883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.944537873911819e-05,
+      "loss": 0.3234,
+      "step": 884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9443002895862745e-05,
+      "loss": 0.3489,
+      "step": 885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440622120522117e-05,
+      "loss": 0.3327,
+      "step": 886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438236414339785e-05,
+      "loss": 0.2745,
+      "step": 887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94358457785618e-05,
+      "loss": 0.3145,
+      "step": 888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94334502144368e-05,
+      "loss": 0.3416,
+      "step": 889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943104972321598e-05,
+      "loss": 0.3528,
+      "step": 890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9428644306153126e-05,
+      "loss": 0.3337,
+      "step": 891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9426233964504586e-05,
+      "loss": 0.3432,
+      "step": 892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942381869952928e-05,
+      "loss": 0.3334,
+      "step": 893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9421398512488704e-05,
+      "loss": 0.3578,
+      "step": 894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9418973404646925e-05,
+      "loss": 0.2876,
+      "step": 895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9418973404646925e-05,
+      "loss": 0.3021,
+      "step": 896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9416543377270572e-05,
+      "loss": 0.3156,
+      "step": 897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9416543377270572e-05,
+      "loss": 0.3319,
+      "step": 898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9414108431628857e-05,
+      "loss": 0.3256,
+      "step": 899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9414108431628857e-05,
+      "loss": 0.3067,
+      "step": 900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.941166856899355e-05,
+      "loss": 0.3105,
+      "step": 901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9409223790638995e-05,
+      "loss": 0.3756,
+      "step": 902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9406774097842095e-05,
+      "loss": 0.3343,
+      "step": 903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9404319491882327e-05,
+      "loss": 0.2831,
+      "step": 904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9401859974041742e-05,
+      "loss": 0.3292,
+      "step": 905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9399395545604935e-05,
+      "loss": 0.3301,
+      "step": 906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9399395545604935e-05,
+      "loss": 0.3538,
+      "step": 907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399395545604935e-05,
+      "loss": 0.3101,
+      "step": 908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.302,
+      "step": 909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394451962093928e-05,
+      "loss": 0.3642,
+      "step": 910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391972809601762e-05,
+      "loss": 0.2962,
+      "step": 911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391972809601762e-05,
+      "loss": 0.3388,
+      "step": 912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.938948875167745e-05,
+      "loss": 0.3611,
+      "step": 913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.938948875167745e-05,
+      "loss": 0.2951,
+      "step": 914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.938699978961842e-05,
+      "loss": 0.3368,
+      "step": 915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9384505924724653e-05,
+      "loss": 0.3305,
+      "step": 916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93820071582987e-05,
+      "loss": 0.3172,
+      "step": 917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9379503491645666e-05,
+      "loss": 0.36,
+      "step": 918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9376994926073215e-05,
+      "loss": 0.3436,
+      "step": 919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9376994926073215e-05,
+      "loss": 0.4003,
+      "step": 920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9374481462891576e-05,
+      "loss": 0.3344,
+      "step": 921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9371963103413533e-05,
+      "loss": 0.315,
+      "step": 922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9369439848954422e-05,
+      "loss": 0.335,
+      "step": 923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9369439848954422e-05,
+      "loss": 0.2878,
+      "step": 924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9369439848954422e-05,
+      "loss": 0.3607,
+      "step": 925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9366911700832146e-05,
+      "loss": 0.3429,
+      "step": 926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.936437866036715e-05,
+      "loss": 0.3092,
+      "step": 927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9361840728882447e-05,
+      "loss": 0.3301,
+      "step": 928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93592979077036e-05,
+      "loss": 0.356,
+      "step": 929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9356750198158724e-05,
+      "loss": 0.3388,
+      "step": 930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354197601578493e-05,
+      "loss": 0.363,
+      "step": 931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354197601578493e-05,
+      "loss": 0.3294,
+      "step": 932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9351640119296126e-05,
+      "loss": 0.4296,
+      "step": 933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9351640119296126e-05,
+      "loss": 0.3719,
+      "step": 934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93490777526474e-05,
+      "loss": 0.3148,
+      "step": 935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93490777526474e-05,
+      "loss": 0.2942,
+      "step": 936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93490777526474e-05,
+      "loss": 0.3173,
+      "step": 937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93490777526474e-05,
+      "loss": 0.311,
+      "step": 938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9346510502970637e-05,
+      "loss": 0.3557,
+      "step": 939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9346510502970637e-05,
+      "loss": 0.3655,
+      "step": 940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9343938371606714e-05,
+      "loss": 0.3523,
+      "step": 941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9341361359899055e-05,
+      "loss": 0.325,
+      "step": 942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9338779469193638e-05,
+      "loss": 0.2885,
+      "step": 943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9338779469193638e-05,
+      "loss": 0.3254,
+      "step": 944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9338779469193638e-05,
+      "loss": 0.3645,
+      "step": 945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9338779469193638e-05,
+      "loss": 0.3336,
+      "step": 946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9336192700838982e-05,
+      "loss": 0.3217,
+      "step": 947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9333601056186165e-05,
+      "loss": 0.3389,
+      "step": 948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9331004536588795e-05,
+      "loss": 0.3413,
+      "step": 949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9328403143403038e-05,
+      "loss": 0.3448,
+      "step": 950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9325796877987607e-05,
+      "loss": 0.3437,
+      "step": 951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9323185741703754e-05,
+      "loss": 0.3403,
+      "step": 952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320569735915273e-05,
+      "loss": 0.3284,
+      "step": 953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317948861988508e-05,
+      "loss": 0.2971,
+      "step": 954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317948861988508e-05,
+      "loss": 0.3467,
+      "step": 955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315323121292344e-05,
+      "loss": 0.3516,
+      "step": 956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9312692515198206e-05,
+      "loss": 0.3162,
+      "step": 957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931005704508006e-05,
+      "loss": 0.2806,
+      "step": 958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9307416712314415e-05,
+      "loss": 0.3391,
+      "step": 959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9304771518280313e-05,
+      "loss": 0.3324,
+      "step": 960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.930212146435935e-05,
+      "loss": 0.2917,
+      "step": 961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.930212146435935e-05,
+      "loss": 0.3263,
+      "step": 962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.929946655193564e-05,
+      "loss": 0.3132,
+      "step": 963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.929680678239585e-05,
+      "loss": 0.3447,
+      "step": 964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.929414215712918e-05,
+      "loss": 0.3226,
+      "step": 965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9291472677527367e-05,
+      "loss": 0.2969,
+      "step": 966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9291472677527367e-05,
+      "loss": 0.3062,
+      "step": 967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9291472677527367e-05,
+      "loss": 0.3283,
+      "step": 968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9291472677527367e-05,
+      "loss": 0.352,
+      "step": 969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9288798344984673e-05,
+      "loss": 0.3252,
+      "step": 970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.928611916089791e-05,
+      "loss": 0.3476,
+      "step": 971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9283435126666416e-05,
+      "loss": 0.338,
+      "step": 972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9280746243692062e-05,
+      "loss": 0.3387,
+      "step": 973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9278052513379256e-05,
+      "loss": 0.3022,
+      "step": 974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278052513379256e-05,
+      "loss": 0.3433,
+      "step": 975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278052513379256e-05,
+      "loss": 0.2962,
+      "step": 976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278052513379256e-05,
+      "loss": 0.321,
+      "step": 977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275353937134932e-05,
+      "loss": 0.2836,
+      "step": 978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275353937134932e-05,
+      "loss": 0.3346,
+      "step": 979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927265051636856e-05,
+      "loss": 0.337,
+      "step": 980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9269942252492134e-05,
+      "loss": 0.3053,
+      "step": 981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9269942252492134e-05,
+      "loss": 0.3108,
+      "step": 982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9267229146920186e-05,
+      "loss": 0.3201,
+      "step": 983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.926451120106977e-05,
+      "loss": 0.3712,
+      "step": 984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9261788416360475e-05,
+      "loss": 0.3689,
+      "step": 985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9259060794214403e-05,
+      "loss": 0.3149,
+      "step": 986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92563283360562e-05,
+      "loss": 0.331,
+      "step": 987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.925359104331303e-05,
+      "loss": 0.3025,
+      "step": 988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9250848917414582e-05,
+      "loss": 0.3143,
+      "step": 989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9250848917414582e-05,
+      "loss": 0.3366,
+      "step": 990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9248101959793066e-05,
+      "loss": 0.339,
+      "step": 991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9245350171883223e-05,
+      "loss": 0.3656,
+      "step": 992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9242593555122316e-05,
+      "loss": 0.3216,
+      "step": 993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9239832110950118e-05,
+      "loss": 0.3243,
+      "step": 994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9237065840808942e-05,
+      "loss": 0.3376,
+      "step": 995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.923429474614361e-05,
+      "loss": 0.3568,
+      "step": 996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9231518828401458e-05,
+      "loss": 0.3665,
+      "step": 997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9228738089032363e-05,
+      "loss": 0.2999,
+      "step": 998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9228738089032363e-05,
+      "loss": 0.3268,
+      "step": 999
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9225952529488695e-05,
+      "loss": 0.3448,
+      "step": 1000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9225952529488695e-05,
+      "loss": 0.3137,
+      "step": 1001
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9223162151225358e-05,
+      "loss": 0.3142,
+      "step": 1002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9220366955699773e-05,
+      "loss": 0.3435,
+      "step": 1003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.921756694437186e-05,
+      "loss": 0.3147,
+      "step": 1004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.921476211870408e-05,
+      "loss": 0.349,
+      "step": 1005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9211952480161382e-05,
+      "loss": 0.3241,
+      "step": 1006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.920913803021125e-05,
+      "loss": 0.3075,
+      "step": 1007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.920631877032367e-05,
+      "loss": 0.3096,
+      "step": 1008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9203494701971142e-05,
+      "loss": 0.3647,
+      "step": 1009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.920066582662868e-05,
+      "loss": 0.3311,
+      "step": 1010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.920066582662868e-05,
+      "loss": 0.3181,
+      "step": 1011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919783214577381e-05,
+      "loss": 0.329,
+      "step": 1012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9194993660886557e-05,
+      "loss": 0.3136,
+      "step": 1013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9192150373449473e-05,
+      "loss": 0.3494,
+      "step": 1014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9189302284947597e-05,
+      "loss": 0.3232,
+      "step": 1015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9186449396868492e-05,
+      "loss": 0.3069,
+      "step": 1016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918359171070223e-05,
+      "loss": 0.3067,
+      "step": 1017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9180729227941374e-05,
+      "loss": 0.3119,
+      "step": 1018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9180729227941374e-05,
+      "loss": 0.3086,
+      "step": 1019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9177861950081002e-05,
+      "loss": 0.3088,
+      "step": 1020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9177861950081002e-05,
+      "loss": 0.3232,
+      "step": 1021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9177861950081002e-05,
+      "loss": 0.3237,
+      "step": 1022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9177861950081002e-05,
+      "loss": 0.3465,
+      "step": 1023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.917498987861869e-05,
+      "loss": 0.3675,
+      "step": 1024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.917211301505453e-05,
+      "loss": 0.3297,
+      "step": 1025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.916923136089111e-05,
+      "loss": 0.3397,
+      "step": 1026
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9166344917633513e-05,
+      "loss": 0.3057,
+      "step": 1027
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9163453686789332e-05,
+      "loss": 0.3623,
+      "step": 1028
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9160557669868657e-05,
+      "loss": 0.3246,
+      "step": 1029
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9157656868384084e-05,
+      "loss": 0.3368,
+      "step": 1030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9154751283850703e-05,
+      "loss": 0.3678,
+      "step": 1031
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9151840917786092e-05,
+      "loss": 0.3248,
+      "step": 1032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9148925771710347e-05,
+      "loss": 0.3342,
+      "step": 1033
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.914600584714605e-05,
+      "loss": 0.3262,
+      "step": 1034
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9143081145618273e-05,
+      "loss": 0.2983,
+      "step": 1035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9140151668654594e-05,
+      "loss": 0.3319,
+      "step": 1036
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9140151668654594e-05,
+      "loss": 0.3027,
+      "step": 1037
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.913721741778508e-05,
+      "loss": 0.3349,
+      "step": 1038
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.913721741778508e-05,
+      "loss": 0.3587,
+      "step": 1039
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.913721741778508e-05,
+      "loss": 0.3522,
+      "step": 1040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.913721741778508e-05,
+      "loss": 0.3055,
+      "step": 1041
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.913427839454229e-05,
+      "loss": 0.2986,
+      "step": 1042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.913427839454229e-05,
+      "loss": 0.2961,
+      "step": 1043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.913427839454229e-05,
+      "loss": 0.3138,
+      "step": 1044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.913427839454229e-05,
+      "loss": 0.2995,
+      "step": 1045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.913427839454229e-05,
+      "loss": 0.3087,
+      "step": 1046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.913427839454229e-05,
+      "loss": 0.2875,
+      "step": 1047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.913427839454229e-05,
+      "loss": 0.3137,
+      "step": 1048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.913427839454229e-05,
+      "loss": 0.3493,
+      "step": 1049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9131334600461274e-05,
+      "loss": 0.3381,
+      "step": 1050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.912838603707959e-05,
+      "loss": 0.2928,
+      "step": 1051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9125432705937254e-05,
+      "loss": 0.3407,
+      "step": 1052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9125432705937254e-05,
+      "loss": 0.316,
+      "step": 1053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9125432705937254e-05,
+      "loss": 0.3343,
+      "step": 1054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.912247460857681e-05,
+      "loss": 0.3479,
+      "step": 1055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9119511746543265e-05,
+      "loss": 0.3193,
+      "step": 1056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9119511746543265e-05,
+      "loss": 0.316,
+      "step": 1057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9119511746543265e-05,
+      "loss": 0.3649,
+      "step": 1058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9119511746543265e-05,
+      "loss": 0.283,
+      "step": 1059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9116544121384124e-05,
+      "loss": 0.3005,
+      "step": 1060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9116544121384124e-05,
+      "loss": 0.3393,
+      "step": 1061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.911357173464938e-05,
+      "loss": 0.3165,
+      "step": 1062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.911059458789152e-05,
+      "loss": 0.3634,
+      "step": 1063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9107612682665495e-05,
+      "loss": 0.3252,
+      "step": 1064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107612682665495e-05,
+      "loss": 0.2992,
+      "step": 1065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9104626020528766e-05,
+      "loss": 0.3062,
+      "step": 1066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9104626020528766e-05,
+      "loss": 0.3301,
+      "step": 1067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9104626020528766e-05,
+      "loss": 0.3583,
+      "step": 1068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9101634603041265e-05,
+      "loss": 0.3217,
+      "step": 1069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9098638431765407e-05,
+      "loss": 0.3553,
+      "step": 1070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90956375082661e-05,
+      "loss": 0.3078,
+      "step": 1071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90956375082661e-05,
+      "loss": 0.3448,
+      "step": 1072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9092631834110723e-05,
+      "loss": 0.2934,
+      "step": 1073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.908962141086915e-05,
+      "loss": 0.339,
+      "step": 1074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9086606240113713e-05,
+      "loss": 0.3385,
+      "step": 1075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9086606240113713e-05,
+      "loss": 0.3063,
+      "step": 1076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9086606240113713e-05,
+      "loss": 0.3084,
+      "step": 1077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9086606240113713e-05,
+      "loss": 0.3078,
+      "step": 1078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9083586323419246e-05,
+      "loss": 0.3375,
+      "step": 1079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.908056166236305e-05,
+      "loss": 0.288,
+      "step": 1080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.907753225852491e-05,
+      "loss": 0.3447,
+      "step": 1081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.907753225852491e-05,
+      "loss": 0.3378,
+      "step": 1082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.907753225852491e-05,
+      "loss": 0.2786,
+      "step": 1083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9074498113487087e-05,
+      "loss": 0.3605,
+      "step": 1084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9071459228834306e-05,
+      "loss": 0.3258,
+      "step": 1085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9071459228834306e-05,
+      "loss": 0.2919,
+      "step": 1086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906841560615379e-05,
+      "loss": 0.3184,
+      "step": 1087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065367247035213e-05,
+      "loss": 0.2941,
+      "step": 1088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065367247035213e-05,
+      "loss": 0.3193,
+      "step": 1089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062314153070746e-05,
+      "loss": 0.3248,
+      "step": 1090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9059256325855017e-05,
+      "loss": 0.3074,
+      "step": 1091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9059256325855017e-05,
+      "loss": 0.3333,
+      "step": 1092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.905619376698513e-05,
+      "loss": 0.3382,
+      "step": 1093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.905312647806066e-05,
+      "loss": 0.324,
+      "step": 1094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9050054460683652e-05,
+      "loss": 0.3149,
+      "step": 1095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9050054460683652e-05,
+      "loss": 0.3384,
+      "step": 1096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9046977716458627e-05,
+      "loss": 0.3096,
+      "step": 1097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.904389624699257e-05,
+      "loss": 0.3152,
+      "step": 1098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.904389624699257e-05,
+      "loss": 0.3045,
+      "step": 1099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.904081005389493e-05,
+      "loss": 0.3347,
+      "step": 1100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.903771913877763e-05,
+      "loss": 0.33,
+      "step": 1101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.903771913877763e-05,
+      "loss": 0.3144,
+      "step": 1102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9034623503255053e-05,
+      "loss": 0.3162,
+      "step": 1103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9034623503255053e-05,
+      "loss": 0.3157,
+      "step": 1104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9034623503255053e-05,
+      "loss": 0.2991,
+      "step": 1105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9031523148944058e-05,
+      "loss": 0.3397,
+      "step": 1106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9028418077463956e-05,
+      "loss": 0.3193,
+      "step": 1107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.902530829043653e-05,
+      "loss": 0.3288,
+      "step": 1108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.902530829043653e-05,
+      "loss": 0.307,
+      "step": 1109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902219378948602e-05,
+      "loss": 0.3414,
+      "step": 1110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019074576239138e-05,
+      "loss": 0.3189,
+      "step": 1111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015950652325045e-05,
+      "loss": 0.3005,
+      "step": 1112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015950652325045e-05,
+      "loss": 0.3305,
+      "step": 1113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901282201937537e-05,
+      "loss": 0.3187,
+      "step": 1114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.3415,
+      "step": 1115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.900655063290807e-05,
+      "loss": 0.3335,
+      "step": 1116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9003407882665998e-05,
+      "loss": 0.2994,
+      "step": 1117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9000260429939435e-05,
+      "loss": 0.3242,
+      "step": 1118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8997108276372305e-05,
+      "loss": 0.3279,
+      "step": 1119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8997108276372305e-05,
+      "loss": 0.3461,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8993951423610976e-05,
+      "loss": 0.2929,
+      "step": 1121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8993951423610976e-05,
+      "loss": 0.3169,
+      "step": 1122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8990789873304268e-05,
+      "loss": 0.3489,
+      "step": 1123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.898762362710347e-05,
+      "loss": 0.3593,
+      "step": 1124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.898762362710347e-05,
+      "loss": 0.3169,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.898762362710347e-05,
+      "loss": 0.3321,
+      "step": 1126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.898762362710347e-05,
+      "loss": 0.329,
+      "step": 1127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.898762362710347e-05,
+      "loss": 0.304,
+      "step": 1128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8984452686662305e-05,
+      "loss": 0.3175,
+      "step": 1129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8981277053636963e-05,
+      "loss": 0.2896,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8981277053636963e-05,
+      "loss": 0.3315,
+      "step": 1131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981277053636963e-05,
+      "loss": 0.3328,
+      "step": 1132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978096729686076e-05,
+      "loss": 0.3188,
+      "step": 1133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978096729686076e-05,
+      "loss": 0.2886,
+      "step": 1134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978096729686076e-05,
+      "loss": 0.318,
+      "step": 1135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978096729686076e-05,
+      "loss": 0.319,
+      "step": 1136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978096729686076e-05,
+      "loss": 0.3169,
+      "step": 1137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974911716470727e-05,
+      "loss": 0.3281,
+      "step": 1138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971722015654457e-05,
+      "loss": 0.3521,
+      "step": 1139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896852762890324e-05,
+      "loss": 0.3001,
+      "step": 1140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896532855788551e-05,
+      "loss": 0.3526,
+      "step": 1141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8962124804272144e-05,
+      "loss": 0.3161,
+      "step": 1142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.895891636973646e-05,
+      "loss": 0.308,
+      "step": 1143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.895891636973646e-05,
+      "loss": 0.3128,
+      "step": 1144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.895891636973646e-05,
+      "loss": 0.3162,
+      "step": 1145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.895891636973646e-05,
+      "loss": 0.3414,
+      "step": 1146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.895570325595423e-05,
+      "loss": 0.2874,
+      "step": 1147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.895570325595423e-05,
+      "loss": 0.3027,
+      "step": 1148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.895570325595423e-05,
+      "loss": 0.3272,
+      "step": 1149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.895570325595423e-05,
+      "loss": 0.3483,
+      "step": 1150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8952485464603664e-05,
+      "loss": 0.3342,
+      "step": 1151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8949262997365413e-05,
+      "loss": 0.3343,
+      "step": 1152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8949262997365413e-05,
+      "loss": 0.3536,
+      "step": 1153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.894603585592258e-05,
+      "loss": 0.2809,
+      "step": 1154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894603585592258e-05,
+      "loss": 0.3364,
+      "step": 1155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894280404196069e-05,
+      "loss": 0.2858,
+      "step": 1156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939567557167736e-05,
+      "loss": 0.3318,
+      "step": 1157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.308,
+      "step": 1158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933080581852717e-05,
+      "loss": 0.3263,
+      "step": 1159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929830094718807e-05,
+      "loss": 0.2914,
+      "step": 1160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926574943530124e-05,
+      "loss": 0.3259,
+      "step": 1161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926574943530124e-05,
+      "loss": 0.323,
+      "step": 1162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8923315129986838e-05,
+      "loss": 0.3372,
+      "step": 1163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920050655791548e-05,
+      "loss": 0.3089,
+      "step": 1164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891678152264929e-05,
+      "loss": 0.366,
+      "step": 1165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891350773226754e-05,
+      "loss": 0.3296,
+      "step": 1166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8910229286356198e-05,
+      "loss": 0.2854,
+      "step": 1167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.890694618662759e-05,
+      "loss": 0.3217,
+      "step": 1168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.890365843479649e-05,
+      "loss": 0.2977,
+      "step": 1169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.890036603258009e-05,
+      "loss": 0.3012,
+      "step": 1170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.890036603258009e-05,
+      "loss": 0.2952,
+      "step": 1171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.890036603258009e-05,
+      "loss": 0.3301,
+      "step": 1172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.889706898169801e-05,
+      "loss": 0.323,
+      "step": 1173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8893767283872306e-05,
+      "loss": 0.3079,
+      "step": 1174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8890460940827452e-05,
+      "loss": 0.3196,
+      "step": 1175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.888714995429036e-05,
+      "loss": 0.3335,
+      "step": 1176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888383432599035e-05,
+      "loss": 0.3595,
+      "step": 1177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880514057659188e-05,
+      "loss": 0.353,
+      "step": 1178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880514057659188e-05,
+      "loss": 0.3471,
+      "step": 1179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887718915103104e-05,
+      "loss": 0.2907,
+      "step": 1180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873859607842512e-05,
+      "loss": 0.3078,
+      "step": 1181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873859607842512e-05,
+      "loss": 0.348,
+      "step": 1182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873859607842512e-05,
+      "loss": 0.3358,
+      "step": 1183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870525429832626e-05,
+      "loss": 0.333,
+      "step": 1184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8867186618742822e-05,
+      "loss": 0.3237,
+      "step": 1185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8867186618742822e-05,
+      "loss": 0.3349,
+      "step": 1186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8867186618742822e-05,
+      "loss": 0.3239,
+      "step": 1187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8867186618742822e-05,
+      "loss": 0.3098,
+      "step": 1188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886384317631697e-05,
+      "loss": 0.3332,
+      "step": 1189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8860495104301346e-05,
+      "loss": 0.325,
+      "step": 1190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.885714240444465e-05,
+      "loss": 0.3232,
+      "step": 1191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8853785078498e-05,
+      "loss": 0.3229,
+      "step": 1192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8850423128214928e-05,
+      "loss": 0.324,
+      "step": 1193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8850423128214928e-05,
+      "loss": 0.2904,
+      "step": 1194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8850423128214928e-05,
+      "loss": 0.3095,
+      "step": 1195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8850423128214928e-05,
+      "loss": 0.3117,
+      "step": 1196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8850423128214928e-05,
+      "loss": 0.3097,
+      "step": 1197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8847056555351383e-05,
+      "loss": 0.3645,
+      "step": 1198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8843685361665724e-05,
+      "loss": 0.3073,
+      "step": 1199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843685361665724e-05,
+      "loss": 0.3157,
+      "step": 1200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840309548918734e-05,
+      "loss": 0.3221,
+      "step": 1201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88369291188736e-05,
+      "loss": 0.3405,
+      "step": 1202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833544073295918e-05,
+      "loss": 0.3316,
+      "step": 1203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833544073295918e-05,
+      "loss": 0.3133,
+      "step": 1204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830154413953704e-05,
+      "loss": 0.3053,
+      "step": 1205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830154413953704e-05,
+      "loss": 0.3339,
+      "step": 1206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826760142617375e-05,
+      "loss": 0.2875,
+      "step": 1207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823361261059762e-05,
+      "loss": 0.3066,
+      "step": 1208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823361261059762e-05,
+      "loss": 0.344,
+      "step": 1209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819957771056106e-05,
+      "loss": 0.3115,
+      "step": 1210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819957771056106e-05,
+      "loss": 0.3356,
+      "step": 1211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8816549674384044e-05,
+      "loss": 0.3396,
+      "step": 1212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881313697282363e-05,
+      "loss": 0.3648,
+      "step": 1213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8809719668157328e-05,
+      "loss": 0.3076,
+      "step": 1214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8809719668157328e-05,
+      "loss": 0.3716,
+      "step": 1215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8809719668157328e-05,
+      "loss": 0.3449,
+      "step": 1216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8809719668157328e-05,
+      "loss": 0.3616,
+      "step": 1217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8809719668157328e-05,
+      "loss": 0.2948,
+      "step": 1218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8806297762169982e-05,
+      "loss": 0.3536,
+      "step": 1219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8802871256648872e-05,
+      "loss": 0.3085,
+      "step": 1220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8802871256648872e-05,
+      "loss": 0.3514,
+      "step": 1221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879944015338365e-05,
+      "loss": 0.3523,
+      "step": 1222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879944015338365e-05,
+      "loss": 0.3385,
+      "step": 1223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8796004454166386e-05,
+      "loss": 0.317,
+      "step": 1224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792564160791555e-05,
+      "loss": 0.2954,
+      "step": 1225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792564160791555e-05,
+      "loss": 0.289,
+      "step": 1226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789119275056018e-05,
+      "loss": 0.2915,
+      "step": 1227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878566979875904e-05,
+      "loss": 0.3491,
+      "step": 1228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782215733702286e-05,
+      "loss": 0.3068,
+      "step": 1229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877875708168982e-05,
+      "loss": 0.3006,
+      "step": 1230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775293844528085e-05,
+      "loss": 0.3706,
+      "step": 1231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771826024025944e-05,
+      "loss": 0.2916,
+      "step": 1232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768353621994636e-05,
+      "loss": 0.3156,
+      "step": 1233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87648766402478e-05,
+      "loss": 0.3302,
+      "step": 1234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8761395080601466e-05,
+      "loss": 0.3022,
+      "step": 1235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8761395080601466e-05,
+      "loss": 0.3691,
+      "step": 1236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.875790894487405e-05,
+      "loss": 0.3175,
+      "step": 1237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.875790894487405e-05,
+      "loss": 0.349,
+      "step": 1238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.875441823488637e-05,
+      "loss": 0.3218,
+      "step": 1239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8750922952461617e-05,
+      "loss": 0.2997,
+      "step": 1240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8750922952461617e-05,
+      "loss": 0.313,
+      "step": 1241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8747423099425387e-05,
+      "loss": 0.2934,
+      "step": 1242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8747423099425387e-05,
+      "loss": 0.2829,
+      "step": 1243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747423099425387e-05,
+      "loss": 0.3232,
+      "step": 1244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747423099425387e-05,
+      "loss": 0.3465,
+      "step": 1245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743918677605647e-05,
+      "loss": 0.3107,
+      "step": 1246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8740409688832762e-05,
+      "loss": 0.3289,
+      "step": 1247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736896134939483e-05,
+      "loss": 0.2808,
+      "step": 1248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733378017760935e-05,
+      "loss": 0.3581,
+      "step": 1249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729855339134633e-05,
+      "loss": 0.2917,
+      "step": 1250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726328100900476e-05,
+      "loss": 0.3114,
+      "step": 1251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872279630490074e-05,
+      "loss": 0.319,
+      "step": 1252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872279630490074e-05,
+      "loss": 0.2838,
+      "step": 1253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872279630490074e-05,
+      "loss": 0.31,
+      "step": 1254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8719259952980086e-05,
+      "loss": 0.3329,
+      "step": 1255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8719259952980086e-05,
+      "loss": 0.3289,
+      "step": 1256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871571904698555e-05,
+      "loss": 0.3133,
+      "step": 1257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8712173588766557e-05,
+      "loss": 0.3288,
+      "step": 1258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8712173588766557e-05,
+      "loss": 0.3028,
+      "step": 1259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.870862358017489e-05,
+      "loss": 0.3328,
+      "step": 1260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.870862358017489e-05,
+      "loss": 0.3192,
+      "step": 1261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.870862358017489e-05,
+      "loss": 0.3192,
+      "step": 1262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8705069023064724e-05,
+      "loss": 0.3264,
+      "step": 1263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8705069023064724e-05,
+      "loss": 0.3012,
+      "step": 1264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8705069023064724e-05,
+      "loss": 0.3478,
+      "step": 1265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.870150991929261e-05,
+      "loss": 0.3247,
+      "step": 1266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697946270717468e-05,
+      "loss": 0.3139,
+      "step": 1267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869437807920059e-05,
+      "loss": 0.3474,
+      "step": 1268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869437807920059e-05,
+      "loss": 0.3223,
+      "step": 1269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869080534660565e-05,
+      "loss": 0.3226,
+      "step": 1270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687228074798684e-05,
+      "loss": 0.3675,
+      "step": 1271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86836462656481e-05,
+      "loss": 0.3241,
+      "step": 1272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680059921024686e-05,
+      "loss": 0.3259,
+      "step": 1273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680059921024686e-05,
+      "loss": 0.3485,
+      "step": 1274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680059921024686e-05,
+      "loss": 0.2923,
+      "step": 1275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867646904280159e-05,
+      "loss": 0.3061,
+      "step": 1276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672873632854322e-05,
+      "loss": 0.3107,
+      "step": 1277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8669273693060776e-05,
+      "loss": 0.3452,
+      "step": 1278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665669225301198e-05,
+      "loss": 0.316,
+      "step": 1279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86620602314582e-05,
+      "loss": 0.3293,
+      "step": 1280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8658446713416773e-05,
+      "loss": 0.3355,
+      "step": 1281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8654828673064245e-05,
+      "loss": 0.3205,
+      "step": 1282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8651206112290326e-05,
+      "loss": 0.2971,
+      "step": 1283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8647579032987085e-05,
+      "loss": 0.3428,
+      "step": 1284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8643947437048944e-05,
+      "loss": 0.3179,
+      "step": 1285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8640311326372686e-05,
+      "loss": 0.2845,
+      "step": 1286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.863667070285746e-05,
+      "loss": 0.3362,
+      "step": 1287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8633025568404757e-05,
+      "loss": 0.3852,
+      "step": 1288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629375924918433e-05,
+      "loss": 0.3099,
+      "step": 1289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625721774304707e-05,
+      "loss": 0.3163,
+      "step": 1290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622063118472135e-05,
+      "loss": 0.2649,
+      "step": 1291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622063118472135e-05,
+      "loss": 0.3237,
+      "step": 1292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618399959331642e-05,
+      "loss": 0.337,
+      "step": 1293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861473229879649e-05,
+      "loss": 0.299,
+      "step": 1294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611060138782306e-05,
+      "loss": 0.2924,
+      "step": 1295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607383481207062e-05,
+      "loss": 0.3431,
+      "step": 1296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8603702327991074e-05,
+      "loss": 0.3205,
+      "step": 1297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860001668105701e-05,
+      "loss": 0.3095,
+      "step": 1298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8596326542329888e-05,
+      "loss": 0.312,
+      "step": 1299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8592631913737067e-05,
+      "loss": 0.3191,
+      "step": 1300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8588932797208255e-05,
+      "loss": 0.3238,
+      "step": 1301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.85852291946755e-05,
+      "loss": 0.3396,
+      "step": 1302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.85815211080732e-05,
+      "loss": 0.3362,
+      "step": 1303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8577808539338087e-05,
+      "loss": 0.271,
+      "step": 1304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8574091490409238e-05,
+      "loss": 0.3015,
+      "step": 1305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.857036996322807e-05,
+      "loss": 0.2988,
+      "step": 1306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.856664395973834e-05,
+      "loss": 0.3203,
+      "step": 1307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8562913481886138e-05,
+      "loss": 0.3399,
+      "step": 1308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.85591785316199e-05,
+      "loss": 0.285,
+      "step": 1309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8555439110890384e-05,
+      "loss": 0.3032,
+      "step": 1310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8555439110890384e-05,
+      "loss": 0.3472,
+      "step": 1311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555439110890384e-05,
+      "loss": 0.3245,
+      "step": 1312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555439110890384e-05,
+      "loss": 0.3263,
+      "step": 1313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555439110890384e-05,
+      "loss": 0.3127,
+      "step": 1314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555439110890384e-05,
+      "loss": 0.3265,
+      "step": 1315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85516952216507e-05,
+      "loss": 0.3333,
+      "step": 1316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8547946865856277e-05,
+      "loss": 0.3261,
+      "step": 1317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8544194045464888e-05,
+      "loss": 0.2841,
+      "step": 1318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8540436762436627e-05,
+      "loss": 0.3285,
+      "step": 1319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8536675018733933e-05,
+      "loss": 0.3517,
+      "step": 1320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8532908816321557e-05,
+      "loss": 0.3261,
+      "step": 1321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8529138157166597e-05,
+      "loss": 0.3351,
+      "step": 1322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8525363043238463e-05,
+      "loss": 0.2823,
+      "step": 1323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8521583476508907e-05,
+      "loss": 0.2594,
+      "step": 1324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8517799458951993e-05,
+      "loss": 0.3713,
+      "step": 1325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8514010992544117e-05,
+      "loss": 0.3085,
+      "step": 1326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8510218079263995e-05,
+      "loss": 0.3535,
+      "step": 1327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8506420721092668e-05,
+      "loss": 0.3256,
+      "step": 1328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85026189200135e-05,
+      "loss": 0.3223,
+      "step": 1329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8498812678012174e-05,
+      "loss": 0.3217,
+      "step": 1330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8498812678012174e-05,
+      "loss": 0.3058,
+      "step": 1331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8495001997076686e-05,
+      "loss": 0.3306,
+      "step": 1332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8495001997076686e-05,
+      "loss": 0.3202,
+      "step": 1333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8495001997076686e-05,
+      "loss": 0.2879,
+      "step": 1334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8495001997076686e-05,
+      "loss": 0.3086,
+      "step": 1335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849118687919737e-05,
+      "loss": 0.3331,
+      "step": 1336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8487367326366845e-05,
+      "loss": 0.3239,
+      "step": 1337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.848354334058008e-05,
+      "loss": 0.3385,
+      "step": 1338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8479714923834343e-05,
+      "loss": 0.3262,
+      "step": 1339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8479714923834343e-05,
+      "loss": 0.3238,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8479714923834343e-05,
+      "loss": 0.275,
+      "step": 1341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8475882078129212e-05,
+      "loss": 0.3127,
+      "step": 1342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8475882078129212e-05,
+      "loss": 0.3376,
+      "step": 1343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8475882078129212e-05,
+      "loss": 0.3027,
+      "step": 1344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8475882078129212e-05,
+      "loss": 0.3109,
+      "step": 1345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8475882078129212e-05,
+      "loss": 0.3379,
+      "step": 1346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.847204480546659e-05,
+      "loss": 0.3267,
+      "step": 1347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.847204480546659e-05,
+      "loss": 0.2833,
+      "step": 1348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.847204480546659e-05,
+      "loss": 0.3476,
+      "step": 1349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.846820310785068e-05,
+      "loss": 0.3306,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8464356987288012e-05,
+      "loss": 0.3009,
+      "step": 1351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.846050644578741e-05,
+      "loss": 0.3071,
+      "step": 1352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.846050644578741e-05,
+      "loss": 0.309,
+      "step": 1353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8456651485360014e-05,
+      "loss": 0.2801,
+      "step": 1354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8456651485360014e-05,
+      "loss": 0.3089,
+      "step": 1355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8452792108019275e-05,
+      "loss": 0.3239,
+      "step": 1356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448928315780942e-05,
+      "loss": 0.314,
+      "step": 1357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844506011066308e-05,
+      "loss": 0.3108,
+      "step": 1358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8441187494686055e-05,
+      "loss": 0.3326,
+      "step": 1359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437310469872535e-05,
+      "loss": 0.3391,
+      "step": 1360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437310469872535e-05,
+      "loss": 0.317,
+      "step": 1361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8433429038247493e-05,
+      "loss": 0.3509,
+      "step": 1362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84295432018382e-05,
+      "loss": 0.3251,
+      "step": 1363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8425652962674233e-05,
+      "loss": 0.3397,
+      "step": 1364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8421758322787467e-05,
+      "loss": 0.3076,
+      "step": 1365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8421758322787467e-05,
+      "loss": 0.3379,
+      "step": 1366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.841785928421207e-05,
+      "loss": 0.3334,
+      "step": 1367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8413955848984515e-05,
+      "loss": 0.2973,
+      "step": 1368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8410048019143568e-05,
+      "loss": 0.3141,
+      "step": 1369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8410048019143568e-05,
+      "loss": 0.3047,
+      "step": 1370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8406135796730285e-05,
+      "loss": 0.3225,
+      "step": 1371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8402219183788033e-05,
+      "loss": 0.3364,
+      "step": 1372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8398298182362454e-05,
+      "loss": 0.3167,
+      "step": 1373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8394372794501486e-05,
+      "loss": 0.3153,
+      "step": 1374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8390443022255367e-05,
+      "loss": 0.3399,
+      "step": 1375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8386508867676617e-05,
+      "loss": 0.3185,
+      "step": 1376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8382570332820045e-05,
+      "loss": 0.3071,
+      "step": 1377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8378627419742753e-05,
+      "loss": 0.3089,
+      "step": 1378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8378627419742753e-05,
+      "loss": 0.3119,
+      "step": 1379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8374680130504122e-05,
+      "loss": 0.3226,
+      "step": 1380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8374680130504122e-05,
+      "loss": 0.3155,
+      "step": 1381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837072846716583e-05,
+      "loss": 0.3597,
+      "step": 1382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8366772431791825e-05,
+      "loss": 0.3154,
+      "step": 1383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8362812026448352e-05,
+      "loss": 0.3165,
+      "step": 1384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8358847253203935e-05,
+      "loss": 0.3622,
+      "step": 1385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.3035,
+      "step": 1386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8350904611297737e-05,
+      "loss": 0.2928,
+      "step": 1387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.834692674678441e-05,
+      "loss": 0.3085,
+      "step": 1388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8342944522667024e-05,
+      "loss": 0.3364,
+      "step": 1389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8338957941025497e-05,
+      "loss": 0.3478,
+      "step": 1390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.833496700394202e-05,
+      "loss": 0.3203,
+      "step": 1391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8330971713501068e-05,
+      "loss": 0.33,
+      "step": 1392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.832697207178938e-05,
+      "loss": 0.3418,
+      "step": 1393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.832697207178938e-05,
+      "loss": 0.3643,
+      "step": 1394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8322968080895966e-05,
+      "loss": 0.3214,
+      "step": 1395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.831895974291212e-05,
+      "loss": 0.2991,
+      "step": 1396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8314947059931394e-05,
+      "loss": 0.3299,
+      "step": 1397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8310930034049616e-05,
+      "loss": 0.2597,
+      "step": 1398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8306908667364883e-05,
+      "loss": 0.2801,
+      "step": 1399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8302882961977552e-05,
+      "loss": 0.2975,
+      "step": 1400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8298852919990254e-05,
+      "loss": 0.3805,
+      "step": 1401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8294818543507878e-05,
+      "loss": 0.3263,
+      "step": 1402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8294818543507878e-05,
+      "loss": 0.3069,
+      "step": 1403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.829077983463759e-05,
+      "loss": 0.3015,
+      "step": 1404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.82867367954888e-05,
+      "loss": 0.332,
+      "step": 1405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.82867367954888e-05,
+      "loss": 0.3013,
+      "step": 1406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.82867367954888e-05,
+      "loss": 0.3274,
+      "step": 1407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8282689428173197e-05,
+      "loss": 0.2822,
+      "step": 1408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8278637734804717e-05,
+      "loss": 0.3315,
+      "step": 1409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8274581717499564e-05,
+      "loss": 0.3534,
+      "step": 1410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8274581717499564e-05,
+      "loss": 0.3487,
+      "step": 1411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8274581717499564e-05,
+      "loss": 0.3248,
+      "step": 1412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8274581717499564e-05,
+      "loss": 0.3212,
+      "step": 1413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.82705213783762e-05,
+      "loss": 0.3248,
+      "step": 1414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8266456719555337e-05,
+      "loss": 0.3312,
+      "step": 1415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8266456719555337e-05,
+      "loss": 0.3103,
+      "step": 1416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.3388,
+      "step": 1417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.825831445131527e-05,
+      "loss": 0.339,
+      "step": 1418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.825423684614878e-05,
+      "loss": 0.311,
+      "step": 1419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8250154929790212e-05,
+      "loss": 0.3342,
+      "step": 1420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8246068704371552e-05,
+      "loss": 0.2813,
+      "step": 1421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8246068704371552e-05,
+      "loss": 0.3079,
+      "step": 1422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8246068704371552e-05,
+      "loss": 0.2791,
+      "step": 1423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8246068704371552e-05,
+      "loss": 0.3403,
+      "step": 1424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8241978172027044e-05,
+      "loss": 0.3427,
+      "step": 1425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8241978172027044e-05,
+      "loss": 0.3146,
+      "step": 1426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.823788333489317e-05,
+      "loss": 0.2703,
+      "step": 1427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.823788333489317e-05,
+      "loss": 0.2982,
+      "step": 1428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.823788333489317e-05,
+      "loss": 0.3389,
+      "step": 1429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8233784195108667e-05,
+      "loss": 0.3197,
+      "step": 1430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8229680754814513e-05,
+      "loss": 0.2972,
+      "step": 1431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8225573016153945e-05,
+      "loss": 0.3115,
+      "step": 1432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8221460981272435e-05,
+      "loss": 0.3078,
+      "step": 1433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8217344652317694e-05,
+      "loss": 0.3048,
+      "step": 1434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.821322403143969e-05,
+      "loss": 0.3203,
+      "step": 1435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8209099120790624e-05,
+      "loss": 0.3174,
+      "step": 1436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8209099120790624e-05,
+      "loss": 0.3063,
+      "step": 1437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8209099120790624e-05,
+      "loss": 0.2869,
+      "step": 1438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.820496992252494e-05,
+      "loss": 0.3323,
+      "step": 1439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8200836438799315e-05,
+      "loss": 0.3092,
+      "step": 1440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.819669867177267e-05,
+      "loss": 0.3141,
+      "step": 1441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8192556623606173e-05,
+      "loss": 0.3108,
+      "step": 1442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.818841029646321e-05,
+      "loss": 0.2927,
+      "step": 1443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8184259692509407e-05,
+      "loss": 0.3209,
+      "step": 1444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.818010481391263e-05,
+      "loss": 0.2995,
+      "step": 1445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8175945662842976e-05,
+      "loss": 0.291,
+      "step": 1446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.817178224147277e-05,
+      "loss": 0.3347,
+      "step": 1447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.817178224147277e-05,
+      "loss": 0.2964,
+      "step": 1448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.816761455197657e-05,
+      "loss": 0.3356,
+      "step": 1449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.816761455197657e-05,
+      "loss": 0.3193,
+      "step": 1450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8163442596531157e-05,
+      "loss": 0.3107,
+      "step": 1451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8163442596531157e-05,
+      "loss": 0.3156,
+      "step": 1452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8159266377315554e-05,
+      "loss": 0.2781,
+      "step": 1453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8155085896510995e-05,
+      "loss": 0.323,
+      "step": 1454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8150901156300956e-05,
+      "loss": 0.3017,
+      "step": 1455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.814671215887112e-05,
+      "loss": 0.3294,
+      "step": 1456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.814671215887112e-05,
+      "loss": 0.3209,
+      "step": 1457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.814671215887112e-05,
+      "loss": 0.3272,
+      "step": 1458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.814671215887112e-05,
+      "loss": 0.309,
+      "step": 1459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81425189064094e-05,
+      "loss": 0.3002,
+      "step": 1460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81425189064094e-05,
+      "loss": 0.2825,
+      "step": 1461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81425189064094e-05,
+      "loss": 0.3105,
+      "step": 1462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81425189064094e-05,
+      "loss": 0.2822,
+      "step": 1463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81425189064094e-05,
+      "loss": 0.326,
+      "step": 1464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81425189064094e-05,
+      "loss": 0.3299,
+      "step": 1465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.813832140110595e-05,
+      "loss": 0.3328,
+      "step": 1466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8134119645153112e-05,
+      "loss": 0.3042,
+      "step": 1467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8129913640745472e-05,
+      "loss": 0.2932,
+      "step": 1468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.812570339007983e-05,
+      "loss": 0.3155,
+      "step": 1469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.812570339007983e-05,
+      "loss": 0.3205,
+      "step": 1470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.812570339007983e-05,
+      "loss": 0.3259,
+      "step": 1471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.812570339007983e-05,
+      "loss": 0.3027,
+      "step": 1472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81214888953552e-05,
+      "loss": 0.3473,
+      "step": 1473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.811727015877282e-05,
+      "loss": 0.3486,
+      "step": 1474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8113047182536128e-05,
+      "loss": 0.3322,
+      "step": 1475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8108819968850794e-05,
+      "loss": 0.3206,
+      "step": 1476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8108819968850794e-05,
+      "loss": 0.2909,
+      "step": 1477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8108819968850794e-05,
+      "loss": 0.3427,
+      "step": 1478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8108819968850794e-05,
+      "loss": 0.2784,
+      "step": 1479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.810458851992469e-05,
+      "loss": 0.381,
+      "step": 1480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8100352837967908e-05,
+      "loss": 0.301,
+      "step": 1481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.809611292519274e-05,
+      "loss": 0.334,
+      "step": 1482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8091868783813703e-05,
+      "loss": 0.3105,
+      "step": 1483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.808762041604751e-05,
+      "loss": 0.3414,
+      "step": 1484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.808762041604751e-05,
+      "loss": 0.3319,
+      "step": 1485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.808762041604751e-05,
+      "loss": 0.3272,
+      "step": 1486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8083367824113086e-05,
+      "loss": 0.3127,
+      "step": 1487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8079111010231563e-05,
+      "loss": 0.3071,
+      "step": 1488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8074849976626273e-05,
+      "loss": 0.3095,
+      "step": 1489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8070584725522763e-05,
+      "loss": 0.3199,
+      "step": 1490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8070584725522763e-05,
+      "loss": 0.2851,
+      "step": 1491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.806631525914877e-05,
+      "loss": 0.2985,
+      "step": 1492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8062041579734238e-05,
+      "loss": 0.3243,
+      "step": 1493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8057763689511315e-05,
+      "loss": 0.3189,
+      "step": 1494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8053481590714347e-05,
+      "loss": 0.3102,
+      "step": 1495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.804919528557987e-05,
+      "loss": 0.3109,
+      "step": 1496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.804490477634663e-05,
+      "loss": 0.334,
+      "step": 1497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8040610065255556e-05,
+      "loss": 0.362,
+      "step": 1498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8036311154549783e-05,
+      "loss": 0.3146,
+      "step": 1499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8032008046474632e-05,
+      "loss": 0.3248,
+      "step": 1500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8032008046474632e-05,
+      "loss": 0.2954,
+      "step": 1501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8032008046474632e-05,
+      "loss": 0.3187,
+      "step": 1502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8027700743277622e-05,
+      "loss": 0.2813,
+      "step": 1503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.802338924720845e-05,
+      "loss": 0.3015,
+      "step": 1504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8019073560519023e-05,
+      "loss": 0.2882,
+      "step": 1505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.801475368546342e-05,
+      "loss": 0.267,
+      "step": 1506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8010429624297918e-05,
+      "loss": 0.3138,
+      "step": 1507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8006101379280974e-05,
+      "loss": 0.3232,
+      "step": 1508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.800176895267323e-05,
+      "loss": 0.3313,
+      "step": 1509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7997432346737524e-05,
+      "loss": 0.3035,
+      "step": 1510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7993091563738856e-05,
+      "loss": 0.3717,
+      "step": 1511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7988746605944428e-05,
+      "loss": 0.2853,
+      "step": 1512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7984397475623608e-05,
+      "loss": 0.3286,
+      "step": 1513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7980044175047952e-05,
+      "loss": 0.3084,
+      "step": 1514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7975686706491187e-05,
+      "loss": 0.2626,
+      "step": 1515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7971325072229227e-05,
+      "loss": 0.3169,
+      "step": 1516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7966959274540153e-05,
+      "loss": 0.2971,
+      "step": 1517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7966959274540153e-05,
+      "loss": 0.3279,
+      "step": 1518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.796258931570422e-05,
+      "loss": 0.3035,
+      "step": 1519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.796258931570422e-05,
+      "loss": 0.3448,
+      "step": 1520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.796258931570422e-05,
+      "loss": 0.3255,
+      "step": 1521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7958215198003866e-05,
+      "loss": 0.3143,
+      "step": 1522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7953836923723692e-05,
+      "loss": 0.2877,
+      "step": 1523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7953836923723692e-05,
+      "loss": 0.3344,
+      "step": 1524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7953836923723692e-05,
+      "loss": 0.3004,
+      "step": 1525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7953836923723692e-05,
+      "loss": 0.3047,
+      "step": 1526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7953836923723692e-05,
+      "loss": 0.3271,
+      "step": 1527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7953836923723692e-05,
+      "loss": 0.3055,
+      "step": 1528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7953836923723692e-05,
+      "loss": 0.348,
+      "step": 1529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.794945449515047e-05,
+      "loss": 0.3152,
+      "step": 1530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7945067914573147e-05,
+      "loss": 0.3488,
+      "step": 1531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.794067718428284e-05,
+      "loss": 0.2815,
+      "step": 1532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7936282306572822e-05,
+      "loss": 0.2769,
+      "step": 1533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7931883283738548e-05,
+      "loss": 0.3133,
+      "step": 1534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7927480118077625e-05,
+      "loss": 0.2985,
+      "step": 1535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.792307281188983e-05,
+      "loss": 0.3297,
+      "step": 1536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79186613674771e-05,
+      "loss": 0.3149,
+      "step": 1537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7914245787143532e-05,
+      "loss": 0.3278,
+      "step": 1538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7914245787143532e-05,
+      "loss": 0.307,
+      "step": 1539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7909826073195394e-05,
+      "loss": 0.3026,
+      "step": 1540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7909826073195394e-05,
+      "loss": 0.2926,
+      "step": 1541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7909826073195394e-05,
+      "loss": 0.3069,
+      "step": 1542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7909826073195394e-05,
+      "loss": 0.2986,
+      "step": 1543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7909826073195394e-05,
+      "loss": 0.3524,
+      "step": 1544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7905402227941098e-05,
+      "loss": 0.3647,
+      "step": 1545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7900974253691222e-05,
+      "loss": 0.3265,
+      "step": 1546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.78965421527585e-05,
+      "loss": 0.3059,
+      "step": 1547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.78965421527585e-05,
+      "loss": 0.3174,
+      "step": 1548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7892105927457822e-05,
+      "loss": 0.3037,
+      "step": 1549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7892105927457822e-05,
+      "loss": 0.3427,
+      "step": 1550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.788766558010623e-05,
+      "loss": 0.3073,
+      "step": 1551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7883221113022916e-05,
+      "loss": 0.2923,
+      "step": 1552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7883221113022916e-05,
+      "loss": 0.3243,
+      "step": 1553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7878772528529232e-05,
+      "loss": 0.3419,
+      "step": 1554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7874319828948677e-05,
+      "loss": 0.3069,
+      "step": 1555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7869863016606893e-05,
+      "loss": 0.2895,
+      "step": 1556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7865402093831673e-05,
+      "loss": 0.3312,
+      "step": 1557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.786093706295297e-05,
+      "loss": 0.3067,
+      "step": 1558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.785646792630286e-05,
+      "loss": 0.3143,
+      "step": 1559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7851994686215592e-05,
+      "loss": 0.3405,
+      "step": 1560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7851994686215592e-05,
+      "loss": 0.3396,
+      "step": 1561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7847517345027522e-05,
+      "loss": 0.2907,
+      "step": 1562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.784303590507718e-05,
+      "loss": 0.2907,
+      "step": 1563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7838550368705218e-05,
+      "loss": 0.3289,
+      "step": 1564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7834060738254435e-05,
+      "loss": 0.2788,
+      "step": 1565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7829567016069773e-05,
+      "loss": 0.3399,
+      "step": 1566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78250692044983e-05,
+      "loss": 0.3218,
+      "step": 1567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7820567305889228e-05,
+      "loss": 0.3294,
+      "step": 1568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7816061322593894e-05,
+      "loss": 0.3203,
+      "step": 1569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7811551256965784e-05,
+      "loss": 0.2978,
+      "step": 1570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7807037111360504e-05,
+      "loss": 0.3392,
+      "step": 1571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3101,
+      "step": 1572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3234,
+      "step": 1573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3121,
+      "step": 1574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3976,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3265,
+      "step": 1576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3367,
+      "step": 1577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3344,
+      "step": 1578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3067,
+      "step": 1579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3036,
+      "step": 1580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3178,
+      "step": 1581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.2964,
+      "step": 1582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7802518888135797e-05,
+      "loss": 0.3009,
+      "step": 1583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.779799658965153e-05,
+      "loss": 0.3235,
+      "step": 1584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.779799658965153e-05,
+      "loss": 0.3078,
+      "step": 1585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.779799658965153e-05,
+      "loss": 0.2579,
+      "step": 1586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.779799658965153e-05,
+      "loss": 0.293,
+      "step": 1587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.779799658965153e-05,
+      "loss": 0.2919,
+      "step": 1588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.779799658965153e-05,
+      "loss": 0.3149,
+      "step": 1589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.77934702182697e-05,
+      "loss": 0.3208,
+      "step": 1590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.778893977635444e-05,
+      "loss": 0.307,
+      "step": 1591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7784405266271996e-05,
+      "loss": 0.3249,
+      "step": 1592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7779866690390747e-05,
+      "loss": 0.3369,
+      "step": 1593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7779866690390747e-05,
+      "loss": 0.3185,
+      "step": 1594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7779866690390747e-05,
+      "loss": 0.3235,
+      "step": 1595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7779866690390747e-05,
+      "loss": 0.2992,
+      "step": 1596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7775324051081194e-05,
+      "loss": 0.3531,
+      "step": 1597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7775324051081194e-05,
+      "loss": 0.3396,
+      "step": 1598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.777077735071596e-05,
+      "loss": 0.2882,
+      "step": 1599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7766226591669787e-05,
+      "loss": 0.2988,
+      "step": 1600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7766226591669787e-05,
+      "loss": 0.3082,
+      "step": 1601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7766226591669787e-05,
+      "loss": 0.3576,
+      "step": 1602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761671776319537e-05,
+      "loss": 0.3182,
+      "step": 1603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.3059,
+      "step": 1604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.3178,
+      "step": 1605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.3271,
+      "step": 1606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.3077,
+      "step": 1607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.3261,
+      "step": 1608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775254998622487e-05,
+      "loss": 0.3326,
+      "step": 1609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7747983016244764e-05,
+      "loss": 0.3037,
+      "step": 1610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774341199948921e-05,
+      "loss": 0.3379,
+      "step": 1611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774341199948921e-05,
+      "loss": 0.2976,
+      "step": 1612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7738836938345656e-05,
+      "loss": 0.2838,
+      "step": 1613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7738836938345656e-05,
+      "loss": 0.2674,
+      "step": 1614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7734257835203657e-05,
+      "loss": 0.3139,
+      "step": 1615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7734257835203657e-05,
+      "loss": 0.3053,
+      "step": 1616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7734257835203657e-05,
+      "loss": 0.3293,
+      "step": 1617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7729674692454882e-05,
+      "loss": 0.2972,
+      "step": 1618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7725087512493112e-05,
+      "loss": 0.3568,
+      "step": 1619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7725087512493112e-05,
+      "loss": 0.308,
+      "step": 1620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7720496297714225e-05,
+      "loss": 0.3019,
+      "step": 1621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.771590105051622e-05,
+      "loss": 0.3237,
+      "step": 1622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7711301773299197e-05,
+      "loss": 0.309,
+      "step": 1623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7706698468465363e-05,
+      "loss": 0.3047,
+      "step": 1624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770209113841902e-05,
+      "loss": 0.3111,
+      "step": 1625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770209113841902e-05,
+      "loss": 0.3178,
+      "step": 1626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770209113841902e-05,
+      "loss": 0.3177,
+      "step": 1627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697479785566586e-05,
+      "loss": 0.32,
+      "step": 1628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697479785566586e-05,
+      "loss": 0.2933,
+      "step": 1629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769286441231657e-05,
+      "loss": 0.3426,
+      "step": 1630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768824502107958e-05,
+      "loss": 0.3249,
+      "step": 1631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768824502107958e-05,
+      "loss": 0.3495,
+      "step": 1632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768362161426833e-05,
+      "loss": 0.3051,
+      "step": 1633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678994194297627e-05,
+      "loss": 0.304,
+      "step": 1634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674362763584377e-05,
+      "loss": 0.2902,
+      "step": 1635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7669727324547574e-05,
+      "loss": 0.319,
+      "step": 1636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7665087879608312e-05,
+      "loss": 0.3224,
+      "step": 1637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7665087879608312e-05,
+      "loss": 0.3341,
+      "step": 1638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.3575,
+      "step": 1639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.765579698171725e-05,
+      "loss": 0.3278,
+      "step": 1640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.765579698171725e-05,
+      "loss": 0.2947,
+      "step": 1641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.765579698171725e-05,
+      "loss": 0.3017,
+      "step": 1642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.765579698171725e-05,
+      "loss": 0.3277,
+      "step": 1643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7651145533618086e-05,
+      "loss": 0.3208,
+      "step": 1644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7646490089321742e-05,
+      "loss": 0.3157,
+      "step": 1645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.764183065125976e-05,
+      "loss": 0.3291,
+      "step": 1646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.764183065125976e-05,
+      "loss": 0.3187,
+      "step": 1647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764183065125976e-05,
+      "loss": 0.3126,
+      "step": 1648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764183065125976e-05,
+      "loss": 0.3474,
+      "step": 1649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7637167221865767e-05,
+      "loss": 0.3396,
+      "step": 1650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632499803575473e-05,
+      "loss": 0.3595,
+      "step": 1651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632499803575473e-05,
+      "loss": 0.2895,
+      "step": 1652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632499803575473e-05,
+      "loss": 0.3517,
+      "step": 1653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627828398826677e-05,
+      "loss": 0.3156,
+      "step": 1654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627828398826677e-05,
+      "loss": 0.3232,
+      "step": 1655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623153010059247e-05,
+      "loss": 0.3027,
+      "step": 1656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7618473639715152e-05,
+      "loss": 0.3311,
+      "step": 1657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761379029023842e-05,
+      "loss": 0.3274,
+      "step": 1658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761379029023842e-05,
+      "loss": 0.3195,
+      "step": 1659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7609102964075173e-05,
+      "loss": 0.3083,
+      "step": 1660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7604411663673605e-05,
+      "loss": 0.3513,
+      "step": 1661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.759971639148398e-05,
+      "loss": 0.2815,
+      "step": 1662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.759501714995864e-05,
+      "loss": 0.344,
+      "step": 1663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7590313941552004e-05,
+      "loss": 0.3524,
+      "step": 1664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7590313941552004e-05,
+      "loss": 0.2795,
+      "step": 1665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7590313941552004e-05,
+      "loss": 0.3058,
+      "step": 1666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7590313941552004e-05,
+      "loss": 0.2918,
+      "step": 1667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7585606768720563e-05,
+      "loss": 0.3375,
+      "step": 1668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7585606768720563e-05,
+      "loss": 0.3167,
+      "step": 1669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7580895633922872e-05,
+      "loss": 0.3262,
+      "step": 1670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757618053961956e-05,
+      "loss": 0.3708,
+      "step": 1671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757618053961956e-05,
+      "loss": 0.3245,
+      "step": 1672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757618053961956e-05,
+      "loss": 0.2781,
+      "step": 1673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757146148827332e-05,
+      "loss": 0.2656,
+      "step": 1674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756673848234892e-05,
+      "loss": 0.2995,
+      "step": 1675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562011524313187e-05,
+      "loss": 0.3002,
+      "step": 1676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562011524313187e-05,
+      "loss": 0.3289,
+      "step": 1677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562011524313187e-05,
+      "loss": 0.3402,
+      "step": 1678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755728061663501e-05,
+      "loss": 0.2959,
+      "step": 1679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755254576178535e-05,
+      "loss": 0.2919,
+      "step": 1680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547806962237222e-05,
+      "loss": 0.3081,
+      "step": 1681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75430642204657e-05,
+      "loss": 0.3116,
+      "step": 1682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7538317538947926e-05,
+      "loss": 0.3392,
+      "step": 1683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7538317538947926e-05,
+      "loss": 0.3214,
+      "step": 1684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7538317538947926e-05,
+      "loss": 0.3341,
+      "step": 1685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7533566920163086e-05,
+      "loss": 0.2982,
+      "step": 1686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7533566920163086e-05,
+      "loss": 0.3212,
+      "step": 1687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.752881236659244e-05,
+      "loss": 0.3238,
+      "step": 1688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.752405388071929e-05,
+      "loss": 0.2968,
+      "step": 1689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7519291465028995e-05,
+      "loss": 0.33,
+      "step": 1690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7514525122008965e-05,
+      "loss": 0.3007,
+      "step": 1691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7509754854148668e-05,
+      "loss": 0.3088,
+      "step": 1692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7504980663939614e-05,
+      "loss": 0.2984,
+      "step": 1693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7504980663939614e-05,
+      "loss": 0.3074,
+      "step": 1694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7500202553875367e-05,
+      "loss": 0.2996,
+      "step": 1695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7495420526451536e-05,
+      "loss": 0.3174,
+      "step": 1696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7495420526451536e-05,
+      "loss": 0.2825,
+      "step": 1697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490634584165774e-05,
+      "loss": 0.2773,
+      "step": 1698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748584472951778e-05,
+      "loss": 0.3625,
+      "step": 1699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481050965009306e-05,
+      "loss": 0.3144,
+      "step": 1700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747625329314413e-05,
+      "loss": 0.2964,
+      "step": 1701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747625329314413e-05,
+      "loss": 0.3423,
+      "step": 1702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747625329314413e-05,
+      "loss": 0.3199,
+      "step": 1703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747625329314413e-05,
+      "loss": 0.2752,
+      "step": 1704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747625329314413e-05,
+      "loss": 0.3029,
+      "step": 1705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747145171642808e-05,
+      "loss": 0.2872,
+      "step": 1706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746664623736902e-05,
+      "loss": 0.3541,
+      "step": 1707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461836858476858e-05,
+      "loss": 0.311,
+      "step": 1708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.745702358226353e-05,
+      "loss": 0.3028,
+      "step": 1709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.745220641124302e-05,
+      "loss": 0.308,
+      "step": 1710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.745220641124302e-05,
+      "loss": 0.3241,
+      "step": 1711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7447385347931337e-05,
+      "loss": 0.351,
+      "step": 1712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7442560394846518e-05,
+      "loss": 0.2746,
+      "step": 1713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7442560394846518e-05,
+      "loss": 0.3274,
+      "step": 1714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7437731554508644e-05,
+      "loss": 0.3018,
+      "step": 1715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743289882943982e-05,
+      "loss": 0.3017,
+      "step": 1716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742806222216418e-05,
+      "loss": 0.348,
+      "step": 1717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742806222216418e-05,
+      "loss": 0.3201,
+      "step": 1718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742806222216418e-05,
+      "loss": 0.2922,
+      "step": 1719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742322173520789e-05,
+      "loss": 0.3384,
+      "step": 1720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418377371099137e-05,
+      "loss": 0.2944,
+      "step": 1721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7413529132368127e-05,
+      "loss": 0.3013,
+      "step": 1722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7413529132368127e-05,
+      "loss": 0.3055,
+      "step": 1723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740867702154711e-05,
+      "loss": 0.3258,
+      "step": 1724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7403821041170336e-05,
+      "loss": 0.3227,
+      "step": 1725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398961193774095e-05,
+      "loss": 0.3244,
+      "step": 1726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394097481896676e-05,
+      "loss": 0.3208,
+      "step": 1727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394097481896676e-05,
+      "loss": 0.2944,
+      "step": 1728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7389229908078407e-05,
+      "loss": 0.3156,
+      "step": 1729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7384358474861626e-05,
+      "loss": 0.3035,
+      "step": 1730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7384358474861626e-05,
+      "loss": 0.3167,
+      "step": 1731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7384358474861626e-05,
+      "loss": 0.2843,
+      "step": 1732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.737948318479068e-05,
+      "loss": 0.3265,
+      "step": 1733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7374604040411934e-05,
+      "loss": 0.3161,
+      "step": 1734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7369721044273777e-05,
+      "loss": 0.3155,
+      "step": 1735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.736483419892659e-05,
+      "loss": 0.2968,
+      "step": 1736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736483419892659e-05,
+      "loss": 0.3061,
+      "step": 1737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736483419892659e-05,
+      "loss": 0.2799,
+      "step": 1738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359943506922775e-05,
+      "loss": 0.354,
+      "step": 1739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359943506922775e-05,
+      "loss": 0.3283,
+      "step": 1740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7355048970816747e-05,
+      "loss": 0.327,
+      "step": 1741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350150593164923e-05,
+      "loss": 0.3004,
+      "step": 1742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345248376525738e-05,
+      "loss": 0.33,
+      "step": 1743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340342323459608e-05,
+      "loss": 0.2949,
+      "step": 1744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733543243652897e-05,
+      "loss": 0.3436,
+      "step": 1745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.3034,
+      "step": 1746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325601171333927e-05,
+      "loss": 0.3095,
+      "step": 1747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325601171333927e-05,
+      "loss": 0.3347,
+      "step": 1748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73206797982044e-05,
+      "loss": 0.2854,
+      "step": 1749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73206797982044e-05,
+      "loss": 0.346,
+      "step": 1750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7315754601480107e-05,
+      "loss": 0.342,
+      "step": 1751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7315754601480107e-05,
+      "loss": 0.2932,
+      "step": 1752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7310825583733493e-05,
+      "loss": 0.3371,
+      "step": 1753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7310825583733493e-05,
+      "loss": 0.3675,
+      "step": 1754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7310825583733493e-05,
+      "loss": 0.3006,
+      "step": 1755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7305892747538978e-05,
+      "loss": 0.3391,
+      "step": 1756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.730095609547299e-05,
+      "loss": 0.3261,
+      "step": 1757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7296015630113945e-05,
+      "loss": 0.3413,
+      "step": 1758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7291071354042247e-05,
+      "loss": 0.3563,
+      "step": 1759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7286123269840292e-05,
+      "loss": 0.3165,
+      "step": 1760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728117138009248e-05,
+      "loss": 0.3295,
+      "step": 1761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728117138009248e-05,
+      "loss": 0.2778,
+      "step": 1762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727621568738517e-05,
+      "loss": 0.299,
+      "step": 1763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271256194306727e-05,
+      "loss": 0.3076,
+      "step": 1764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266292903447502e-05,
+      "loss": 0.2988,
+      "step": 1765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266292903447502e-05,
+      "loss": 0.3239,
+      "step": 1766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266292903447502e-05,
+      "loss": 0.3419,
+      "step": 1767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266292903447502e-05,
+      "loss": 0.3239,
+      "step": 1768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726132581739982e-05,
+      "loss": 0.3664,
+      "step": 1769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725635493875799e-05,
+      "loss": 0.3304,
+      "step": 1770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251380270118313e-05,
+      "loss": 0.2854,
+      "step": 1771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7246401814079053e-05,
+      "loss": 0.2916,
+      "step": 1772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7241419573240463e-05,
+      "loss": 0.3203,
+      "step": 1773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7236433550204773e-05,
+      "loss": 0.3263,
+      "step": 1774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7236433550204773e-05,
+      "loss": 0.3214,
+      "step": 1775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7236433550204773e-05,
+      "loss": 0.3461,
+      "step": 1776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.723144374757618e-05,
+      "loss": 0.2618,
+      "step": 1777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7226450167960862e-05,
+      "loss": 0.3458,
+      "step": 1778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.722145281396697e-05,
+      "loss": 0.3264,
+      "step": 1779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7216451688204623e-05,
+      "loss": 0.3102,
+      "step": 1780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.721144679328591e-05,
+      "loss": 0.3034,
+      "step": 1781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206438131824895e-05,
+      "loss": 0.295,
+      "step": 1782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72014257064376e-05,
+      "loss": 0.2928,
+      "step": 1783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719640951974202e-05,
+      "loss": 0.3336,
+      "step": 1784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719138957435811e-05,
+      "loss": 0.3219,
+      "step": 1785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7186365872907787e-05,
+      "loss": 0.3028,
+      "step": 1786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718133841801494e-05,
+      "loss": 0.3402,
+      "step": 1787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7176307212305406e-05,
+      "loss": 0.3258,
+      "step": 1788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717127225840699e-05,
+      "loss": 0.2825,
+      "step": 1789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7166233558949448e-05,
+      "loss": 0.2817,
+      "step": 1790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7161191116564496e-05,
+      "loss": 0.2936,
+      "step": 1791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7156144933885802e-05,
+      "loss": 0.3442,
+      "step": 1792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7151095013548996e-05,
+      "loss": 0.3329,
+      "step": 1793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7151095013548996e-05,
+      "loss": 0.2899,
+      "step": 1794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7146041358191646e-05,
+      "loss": 0.3007,
+      "step": 1795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7140983970453285e-05,
+      "loss": 0.2884,
+      "step": 1796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7135922852975388e-05,
+      "loss": 0.2881,
+      "step": 1797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7135922852975388e-05,
+      "loss": 0.3447,
+      "step": 1798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.713085800840138e-05,
+      "loss": 0.318,
+      "step": 1799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.713085800840138e-05,
+      "loss": 0.3283,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7125789439376625e-05,
+      "loss": 0.3024,
+      "step": 1801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7120717148548447e-05,
+      "loss": 0.3039,
+      "step": 1802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.71156411385661e-05,
+      "loss": 0.2885,
+      "step": 1803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.71156411385661e-05,
+      "loss": 0.3048,
+      "step": 1804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71156411385661e-05,
+      "loss": 0.3376,
+      "step": 1805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7110561412080795e-05,
+      "loss": 0.2547,
+      "step": 1806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7105477971745668e-05,
+      "loss": 0.3281,
+      "step": 1807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100390820215805e-05,
+      "loss": 0.3113,
+      "step": 1808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7095299960148227e-05,
+      "loss": 0.2951,
+      "step": 1809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7090205394201898e-05,
+      "loss": 0.3511,
+      "step": 1810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7085107125037702e-05,
+      "loss": 0.2934,
+      "step": 1811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7080005155318476e-05,
+      "loss": 0.3168,
+      "step": 1812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.707489948770898e-05,
+      "loss": 0.3377,
+      "step": 1813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7069790124875906e-05,
+      "loss": 0.3387,
+      "step": 1814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7064677069487876e-05,
+      "loss": 0.2884,
+      "step": 1815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.705956032421544e-05,
+      "loss": 0.2967,
+      "step": 1816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.705443989173108e-05,
+      "loss": 0.3184,
+      "step": 1817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.70493157747092e-05,
+      "loss": 0.2904,
+      "step": 1818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7044187975826126e-05,
+      "loss": 0.3134,
+      "step": 1819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.703905649776011e-05,
+      "loss": 0.3436,
+      "step": 1820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.703905649776011e-05,
+      "loss": 0.2789,
+      "step": 1821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7033921343191333e-05,
+      "loss": 0.2846,
+      "step": 1822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7033921343191333e-05,
+      "loss": 0.3084,
+      "step": 1823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.702878251480188e-05,
+      "loss": 0.3274,
+      "step": 1824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7023640015275765e-05,
+      "loss": 0.314,
+      "step": 1825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7018493847298926e-05,
+      "loss": 0.2693,
+      "step": 1826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7013344013559197e-05,
+      "loss": 0.3336,
+      "step": 1827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7013344013559197e-05,
+      "loss": 0.2893,
+      "step": 1828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7013344013559197e-05,
+      "loss": 0.3106,
+      "step": 1829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7008190516746348e-05,
+      "loss": 0.3169,
+      "step": 1830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.700303335955205e-05,
+      "loss": 0.3253,
+      "step": 1831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699787254466989e-05,
+      "loss": 0.3123,
+      "step": 1832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699270807479536e-05,
+      "loss": 0.3148,
+      "step": 1833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.698753995262587e-05,
+      "loss": 0.3323,
+      "step": 1834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.698236818086073e-05,
+      "loss": 0.3184,
+      "step": 1835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697719276220116e-05,
+      "loss": 0.3359,
+      "step": 1836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697719276220116e-05,
+      "loss": 0.3121,
+      "step": 1837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697719276220116e-05,
+      "loss": 0.2929,
+      "step": 1838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6972013699350285e-05,
+      "loss": 0.3131,
+      "step": 1839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6972013699350285e-05,
+      "loss": 0.3487,
+      "step": 1840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6966830995013134e-05,
+      "loss": 0.3303,
+      "step": 1841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6961644651896625e-05,
+      "loss": 0.294,
+      "step": 1842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6956454672709603e-05,
+      "loss": 0.3216,
+      "step": 1843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.3047,
+      "step": 1844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.3335,
+      "step": 1845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.342,
+      "step": 1846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.3062,
+      "step": 1847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.3187,
+      "step": 1848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.3145,
+      "step": 1849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.2752,
+      "step": 1850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.2849,
+      "step": 1851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951261060162783e-05,
+      "loss": 0.3297,
+      "step": 1852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946063816968803e-05,
+      "loss": 0.3083,
+      "step": 1853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694086294584218e-05,
+      "loss": 0.2973,
+      "step": 1854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694086294584218e-05,
+      "loss": 0.2852,
+      "step": 1855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693565844949933e-05,
+      "loss": 0.3086,
+      "step": 1856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693045033065857e-05,
+      "loss": 0.294,
+      "step": 1857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693045033065857e-05,
+      "loss": 0.2923,
+      "step": 1858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6925238592040095e-05,
+      "loss": 0.2926,
+      "step": 1859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6920023236366004e-05,
+      "loss": 0.3188,
+      "step": 1860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.691480426636028e-05,
+      "loss": 0.3089,
+      "step": 1861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6909581684748793e-05,
+      "loss": 0.2808,
+      "step": 1862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69043554942593e-05,
+      "loss": 0.2817,
+      "step": 1863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6899125697621442e-05,
+      "loss": 0.311,
+      "step": 1864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6899125697621442e-05,
+      "loss": 0.2835,
+      "step": 1865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.689389229756675e-05,
+      "loss": 0.3164,
+      "step": 1866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.689389229756675e-05,
+      "loss": 0.3429,
+      "step": 1867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.688865529682862e-05,
+      "loss": 0.3134,
+      "step": 1868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.688865529682862e-05,
+      "loss": 0.3403,
+      "step": 1869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.688341469814235e-05,
+      "loss": 0.3088,
+      "step": 1870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6878170504245105e-05,
+      "loss": 0.3371,
+      "step": 1871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6872922717875923e-05,
+      "loss": 0.3188,
+      "step": 1872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6867671341775736e-05,
+      "loss": 0.2795,
+      "step": 1873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.312,
+      "step": 1874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6857157831355395e-05,
+      "loss": 0.3215,
+      "step": 1875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6857157831355395e-05,
+      "loss": 0.3267,
+      "step": 1876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.685189570252645e-05,
+      "loss": 0.3286,
+      "step": 1877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6846629994948917e-05,
+      "loss": 0.3045,
+      "step": 1878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6841360711373077e-05,
+      "loss": 0.336,
+      "step": 1879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.683608785455108e-05,
+      "loss": 0.289,
+      "step": 1880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.683608785455108e-05,
+      "loss": 0.3156,
+      "step": 1881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6830811427236946e-05,
+      "loss": 0.2943,
+      "step": 1882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6825531432186545e-05,
+      "loss": 0.2984,
+      "step": 1883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6820247872157626e-05,
+      "loss": 0.2964,
+      "step": 1884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68149607499098e-05,
+      "loss": 0.3551,
+      "step": 1885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6809670068204527e-05,
+      "loss": 0.3251,
+      "step": 1886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6804375829805137e-05,
+      "loss": 0.2998,
+      "step": 1887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6799078037476813e-05,
+      "loss": 0.2716,
+      "step": 1888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6793776693986595e-05,
+      "loss": 0.3117,
+      "step": 1889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6788471802103373e-05,
+      "loss": 0.2867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6783163364597896e-05,
+      "loss": 0.3404,
+      "step": 1891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6777851384242768e-05,
+      "loss": 0.3178,
+      "step": 1892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6777851384242768e-05,
+      "loss": 0.2698,
+      "step": 1893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6772535863812434e-05,
+      "loss": 0.3132,
+      "step": 1894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67672168060832e-05,
+      "loss": 0.312,
+      "step": 1895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6761894213833203e-05,
+      "loss": 0.2861,
+      "step": 1896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.675656808984244e-05,
+      "loss": 0.2624,
+      "step": 1897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.675656808984244e-05,
+      "loss": 0.2763,
+      "step": 1898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.675656808984244e-05,
+      "loss": 0.2884,
+      "step": 1899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.675656808984244e-05,
+      "loss": 0.3,
+      "step": 1900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6751238436892754e-05,
+      "loss": 0.2845,
+      "step": 1901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6751238436892754e-05,
+      "loss": 0.3206,
+      "step": 1902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6745905257767815e-05,
+      "loss": 0.2911,
+      "step": 1903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6740568555253153e-05,
+      "loss": 0.2974,
+      "step": 1904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6735228332136132e-05,
+      "loss": 0.2687,
+      "step": 1905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.672988459120594e-05,
+      "loss": 0.3153,
+      "step": 1906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.672453733525363e-05,
+      "loss": 0.294,
+      "step": 1907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.671918656707207e-05,
+      "loss": 0.2832,
+      "step": 1908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.671383228945597e-05,
+      "loss": 0.3361,
+      "step": 1909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.670847450520187e-05,
+      "loss": 0.2632,
+      "step": 1910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.670311321710815e-05,
+      "loss": 0.3354,
+      "step": 1911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6697748427975003e-05,
+      "loss": 0.3278,
+      "step": 1912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.669238014060447e-05,
+      "loss": 0.3059,
+      "step": 1913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6687008357800408e-05,
+      "loss": 0.3108,
+      "step": 1914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6687008357800408e-05,
+      "loss": 0.312,
+      "step": 1915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6687008357800408e-05,
+      "loss": 0.2674,
+      "step": 1916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.66816330823685e-05,
+      "loss": 0.3489,
+      "step": 1917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6676254317116253e-05,
+      "loss": 0.3253,
+      "step": 1918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6670872064853006e-05,
+      "loss": 0.3238,
+      "step": 1919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6670872064853006e-05,
+      "loss": 0.3205,
+      "step": 1920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6670872064853006e-05,
+      "loss": 0.3064,
+      "step": 1921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6670872064853006e-05,
+      "loss": 0.2797,
+      "step": 1922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6665486328389903e-05,
+      "loss": 0.317,
+      "step": 1923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6660097110539926e-05,
+      "loss": 0.2943,
+      "step": 1924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.665470441411786e-05,
+      "loss": 0.3039,
+      "step": 1925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6649308241940313e-05,
+      "loss": 0.2944,
+      "step": 1926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6643908596825717e-05,
+      "loss": 0.286,
+      "step": 1927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6643908596825717e-05,
+      "loss": 0.2916,
+      "step": 1928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6638505481594298e-05,
+      "loss": 0.3041,
+      "step": 1929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6633098899068112e-05,
+      "loss": 0.3277,
+      "step": 1930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.662768885207102e-05,
+      "loss": 0.292,
+      "step": 1931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6622275343428685e-05,
+      "loss": 0.2967,
+      "step": 1932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6616858375968596e-05,
+      "loss": 0.3132,
+      "step": 1933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.661143795252003e-05,
+      "loss": 0.2783,
+      "step": 1934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6606014075914083e-05,
+      "loss": 0.3294,
+      "step": 1935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6606014075914083e-05,
+      "loss": 0.299,
+      "step": 1936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6600586748983642e-05,
+      "loss": 0.3023,
+      "step": 1937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6600586748983642e-05,
+      "loss": 0.2874,
+      "step": 1938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6600586748983642e-05,
+      "loss": 0.3132,
+      "step": 1939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6595155974563407e-05,
+      "loss": 0.3157,
+      "step": 1940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6589721755489877e-05,
+      "loss": 0.3138,
+      "step": 1941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6584284094601338e-05,
+      "loss": 0.3192,
+      "step": 1942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.657884299473789e-05,
+      "loss": 0.3079,
+      "step": 1943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6573398458741424e-05,
+      "loss": 0.2663,
+      "step": 1944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6567950489455618e-05,
+      "loss": 0.2718,
+      "step": 1945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.656249908972595e-05,
+      "loss": 0.2799,
+      "step": 1946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6557044262399692e-05,
+      "loss": 0.2813,
+      "step": 1947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6551586010325902e-05,
+      "loss": 0.2632,
+      "step": 1948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6546124336355428e-05,
+      "loss": 0.2973,
+      "step": 1949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6540659243340904e-05,
+      "loss": 0.3131,
+      "step": 1950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6540659243340904e-05,
+      "loss": 0.3232,
+      "step": 1951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6540659243340904e-05,
+      "loss": 0.2775,
+      "step": 1952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.653519073413675e-05,
+      "loss": 0.2743,
+      "step": 1953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6529718811599175e-05,
+      "loss": 0.3269,
+      "step": 1954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6524243478586163e-05,
+      "loss": 0.2684,
+      "step": 1955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6524243478586163e-05,
+      "loss": 0.337,
+      "step": 1956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6518764737957486e-05,
+      "loss": 0.2508,
+      "step": 1957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6518764737957486e-05,
+      "loss": 0.3177,
+      "step": 1958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.651328259257469e-05,
+      "loss": 0.31,
+      "step": 1959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.651328259257469e-05,
+      "loss": 0.3323,
+      "step": 1960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6507797045301104e-05,
+      "loss": 0.3008,
+      "step": 1961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6502308099001832e-05,
+      "loss": 0.3371,
+      "step": 1962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6496815756543752e-05,
+      "loss": 0.3115,
+      "step": 1963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6496815756543752e-05,
+      "loss": 0.2815,
+      "step": 1964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.649132002079552e-05,
+      "loss": 0.2859,
+      "step": 1965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.648582089462756e-05,
+      "loss": 0.315,
+      "step": 1966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.648031838091207e-05,
+      "loss": 0.289,
+      "step": 1967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.648031838091207e-05,
+      "loss": 0.2942,
+      "step": 1968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6474812482523014e-05,
+      "loss": 0.3244,
+      "step": 1969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6469303202336123e-05,
+      "loss": 0.351,
+      "step": 1970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6463790543228903e-05,
+      "loss": 0.2707,
+      "step": 1971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6458274508080615e-05,
+      "loss": 0.303,
+      "step": 1972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6452755099772285e-05,
+      "loss": 0.3262,
+      "step": 1973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6447232321186706e-05,
+      "loss": 0.293,
+      "step": 1974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6447232321186706e-05,
+      "loss": 0.2627,
+      "step": 1975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6441706175208428e-05,
+      "loss": 0.2885,
+      "step": 1976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.643617666472376e-05,
+      "loss": 0.2931,
+      "step": 1977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.643617666472376e-05,
+      "loss": 0.3272,
+      "step": 1978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.643617666472376e-05,
+      "loss": 0.3046,
+      "step": 1979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.643617666472376e-05,
+      "loss": 0.35,
+      "step": 1980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6430643792620763e-05,
+      "loss": 0.3387,
+      "step": 1981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6430643792620763e-05,
+      "loss": 0.3205,
+      "step": 1982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6425107561789265e-05,
+      "loss": 0.2957,
+      "step": 1983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.641956797512084e-05,
+      "loss": 0.3119,
+      "step": 1984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6414025035508818e-05,
+      "loss": 0.3323,
+      "step": 1985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6414025035508818e-05,
+      "loss": 0.2951,
+      "step": 1986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.640847874584828e-05,
+      "loss": 0.3087,
+      "step": 1987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.640292910903605e-05,
+      "loss": 0.2831,
+      "step": 1988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6397376127970708e-05,
+      "loss": 0.2941,
+      "step": 1989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6391819805552585e-05,
+      "loss": 0.3012,
+      "step": 1990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6386260144683744e-05,
+      "loss": 0.2973,
+      "step": 1991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6386260144683744e-05,
+      "loss": 0.2911,
+      "step": 1992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6386260144683744e-05,
+      "loss": 0.292,
+      "step": 1993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6380697148268007e-05,
+      "loss": 0.3691,
+      "step": 1994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.637513081921092e-05,
+      "loss": 0.2993,
+      "step": 1995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6369561160419783e-05,
+      "loss": 0.3513,
+      "step": 1996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6363988174803638e-05,
+      "loss": 0.3268,
+      "step": 1997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.635841186527325e-05,
+      "loss": 0.2771,
+      "step": 1998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.635283223474113e-05,
+      "loss": 0.3261,
+      "step": 1999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.635283223474113e-05,
+      "loss": 0.3072,
+      "step": 2000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.635283223474113e-05,
+      "loss": 0.3209,
+      "step": 2001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.634724928612152e-05,
+      "loss": 0.339,
+      "step": 2002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6341663022330406e-05,
+      "loss": 0.3272,
+      "step": 2003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6336073446285485e-05,
+      "loss": 0.3263,
+      "step": 2004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6336073446285485e-05,
+      "loss": 0.345,
+      "step": 2005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6330480560906205e-05,
+      "loss": 0.2962,
+      "step": 2006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6324884369113726e-05,
+      "loss": 0.3169,
+      "step": 2007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6319284873830943e-05,
+      "loss": 0.2822,
+      "step": 2008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6319284873830943e-05,
+      "loss": 0.311,
+      "step": 2009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6313682077982478e-05,
+      "loss": 0.3464,
+      "step": 2010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6308075984494674e-05,
+      "loss": 0.3025,
+      "step": 2011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.630246659629559e-05,
+      "loss": 0.3101,
+      "step": 2012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.630246659629559e-05,
+      "loss": 0.3026,
+      "step": 2013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6296853916315022e-05,
+      "loss": 0.3008,
+      "step": 2014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.629123794748447e-05,
+      "loss": 0.3502,
+      "step": 2015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.629123794748447e-05,
+      "loss": 0.3213,
+      "step": 2016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6285618692737156e-05,
+      "loss": 0.317,
+      "step": 2017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6285618692737156e-05,
+      "loss": 0.2968,
+      "step": 2018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6285618692737156e-05,
+      "loss": 0.3018,
+      "step": 2019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6279996155008024e-05,
+      "loss": 0.3527,
+      "step": 2020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.627437033723373e-05,
+      "loss": 0.3058,
+      "step": 2021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.626874124235263e-05,
+      "loss": 0.3617,
+      "step": 2022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.626874124235263e-05,
+      "loss": 0.298,
+      "step": 2023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.626874124235263e-05,
+      "loss": 0.2685,
+      "step": 2024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.626874124235263e-05,
+      "loss": 0.2923,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.626874124235263e-05,
+      "loss": 0.2983,
+      "step": 2026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.626874124235263e-05,
+      "loss": 0.3094,
+      "step": 2027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.626874124235263e-05,
+      "loss": 0.2988,
+      "step": 2028
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6263108873304815e-05,
+      "loss": 0.3009,
+      "step": 2029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.625747323303207e-05,
+      "loss": 0.3278,
+      "step": 2030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.625183432447789e-05,
+      "loss": 0.2933,
+      "step": 2031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.625183432447789e-05,
+      "loss": 0.31,
+      "step": 2032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.625183432447789e-05,
+      "loss": 0.3153,
+      "step": 2033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6246192150587488e-05,
+      "loss": 0.3498,
+      "step": 2034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.624054671430776e-05,
+      "loss": 0.3212,
+      "step": 2035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.2825,
+      "step": 2036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6229246066376524e-05,
+      "loss": 0.3131,
+      "step": 2037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6229246066376524e-05,
+      "loss": 0.292,
+      "step": 2038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6223590860627343e-05,
+      "loss": 0.3087,
+      "step": 2039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6223590860627343e-05,
+      "loss": 0.3057,
+      "step": 2040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6223590860627343e-05,
+      "loss": 0.3002,
+      "step": 2041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6217932404293512e-05,
+      "loss": 0.2703,
+      "step": 2042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6217932404293512e-05,
+      "loss": 0.2789,
+      "step": 2043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6212270700330438e-05,
+      "loss": 0.3139,
+      "step": 2044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6212270700330438e-05,
+      "loss": 0.2984,
+      "step": 2045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.620660575169524e-05,
+      "loss": 0.3007,
+      "step": 2046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6200937561346722e-05,
+      "loss": 0.2838,
+      "step": 2047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.619526613224538e-05,
+      "loss": 0.2976,
+      "step": 2048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.619526613224538e-05,
+      "loss": 0.3043,
+      "step": 2049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6189591467353407e-05,
+      "loss": 0.3136,
+      "step": 2050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6183913569634683e-05,
+      "loss": 0.2754,
+      "step": 2051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6178232442054774e-05,
+      "loss": 0.2974,
+      "step": 2052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.617254808758094e-05,
+      "loss": 0.3193,
+      "step": 2053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6166860509182116e-05,
+      "loss": 0.2613,
+      "step": 2054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6166860509182116e-05,
+      "loss": 0.2546,
+      "step": 2055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6166860509182116e-05,
+      "loss": 0.3043,
+      "step": 2056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6166860509182116e-05,
+      "loss": 0.3301,
+      "step": 2057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6166860509182116e-05,
+      "loss": 0.2877,
+      "step": 2058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6161169709828937e-05,
+      "loss": 0.2984,
+      "step": 2059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.61554756924937e-05,
+      "loss": 0.3232,
+      "step": 2060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.61554756924937e-05,
+      "loss": 0.3426,
+      "step": 2061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.61554756924937e-05,
+      "loss": 0.2706,
+      "step": 2062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.61497784601504e-05,
+      "loss": 0.2508,
+      "step": 2063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.61440780157747e-05,
+      "loss": 0.3036,
+      "step": 2064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6138374362343947e-05,
+      "loss": 0.3003,
+      "step": 2065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6132667502837164e-05,
+      "loss": 0.336,
+      "step": 2066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6132667502837164e-05,
+      "loss": 0.3071,
+      "step": 2067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6126957440235043e-05,
+      "loss": 0.3082,
+      "step": 2068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6121244177519953e-05,
+      "loss": 0.2586,
+      "step": 2069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6121244177519953e-05,
+      "loss": 0.2913,
+      "step": 2070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.611552771767594e-05,
+      "loss": 0.2981,
+      "step": 2071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6109808063688707e-05,
+      "loss": 0.2942,
+      "step": 2072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6104085218545633e-05,
+      "loss": 0.3205,
+      "step": 2073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6098359185235764e-05,
+      "loss": 0.3041,
+      "step": 2074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.609262996674981e-05,
+      "loss": 0.2966,
+      "step": 2075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.609262996674981e-05,
+      "loss": 0.3246,
+      "step": 2076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6086897566080147e-05,
+      "loss": 0.3046,
+      "step": 2077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6081161986220807e-05,
+      "loss": 0.3151,
+      "step": 2078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.607542323016749e-05,
+      "loss": 0.3009,
+      "step": 2079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.606968130091755e-05,
+      "loss": 0.3125,
+      "step": 2080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6063936201469994e-05,
+      "loss": 0.3152,
+      "step": 2081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.60581879348255e-05,
+      "loss": 0.3064,
+      "step": 2082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.605243650398638e-05,
+      "loss": 0.2813,
+      "step": 2083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.605243650398638e-05,
+      "loss": 0.2895,
+      "step": 2084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6046681911956618e-05,
+      "loss": 0.2884,
+      "step": 2085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.604092416174184e-05,
+      "loss": 0.3327,
+      "step": 2086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6035163256349315e-05,
+      "loss": 0.3006,
+      "step": 2087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6035163256349315e-05,
+      "loss": 0.2795,
+      "step": 2088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6029399198787974e-05,
+      "loss": 0.312,
+      "step": 2089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.602363199206838e-05,
+      "loss": 0.3605,
+      "step": 2090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6017861639202756e-05,
+      "loss": 0.2948,
+      "step": 2091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6012088143204953e-05,
+      "loss": 0.2632,
+      "step": 2092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6012088143204953e-05,
+      "loss": 0.2874,
+      "step": 2093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6006311507090477e-05,
+      "loss": 0.2705,
+      "step": 2094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6000531733876462e-05,
+      "loss": 0.3055,
+      "step": 2095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.599474882658169e-05,
+      "loss": 0.3118,
+      "step": 2096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.599474882658169e-05,
+      "loss": 0.2946,
+      "step": 2097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.598896278822658e-05,
+      "loss": 0.3161,
+      "step": 2098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5983173621833174e-05,
+      "loss": 0.29,
+      "step": 2099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5983173621833174e-05,
+      "loss": 0.2975,
+      "step": 2100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5977381330425163e-05,
+      "loss": 0.3119,
+      "step": 2101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.316,
+      "step": 2102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5965787384668222e-05,
+      "loss": 0.3095,
+      "step": 2103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.595998573637481e-05,
+      "loss": 0.2742,
+      "step": 2104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.595998573637481e-05,
+      "loss": 0.3147,
+      "step": 2105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.595998573637481e-05,
+      "loss": 0.2948,
+      "step": 2106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.595998573637481e-05,
+      "loss": 0.2977,
+      "step": 2107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5954180975177837e-05,
+      "loss": 0.294,
+      "step": 2108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5948373104109132e-05,
+      "loss": 0.3072,
+      "step": 2109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.594256212620215e-05,
+      "loss": 0.3255,
+      "step": 2110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.594256212620215e-05,
+      "loss": 0.3146,
+      "step": 2111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.594256212620215e-05,
+      "loss": 0.2949,
+      "step": 2112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.594256212620215e-05,
+      "loss": 0.3153,
+      "step": 2113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.593674804449196e-05,
+      "loss": 0.2937,
+      "step": 2114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5930930862015272e-05,
+      "loss": 0.3016,
+      "step": 2115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5930930862015272e-05,
+      "loss": 0.3117,
+      "step": 2116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5925110581810396e-05,
+      "loss": 0.2918,
+      "step": 2117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5925110581810396e-05,
+      "loss": 0.3035,
+      "step": 2118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5925110581810396e-05,
+      "loss": 0.2823,
+      "step": 2119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5919287206917267e-05,
+      "loss": 0.2997,
+      "step": 2120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.591346074037744e-05,
+      "loss": 0.302,
+      "step": 2121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5907631185234076e-05,
+      "loss": 0.2963,
+      "step": 2122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5907631185234076e-05,
+      "loss": 0.298,
+      "step": 2123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5901798544531958e-05,
+      "loss": 0.3084,
+      "step": 2124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5895962821317482e-05,
+      "loss": 0.295,
+      "step": 2125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5895962821317482e-05,
+      "loss": 0.3221,
+      "step": 2126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5895962821317482e-05,
+      "loss": 0.2908,
+      "step": 2127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.589012401863864e-05,
+      "loss": 0.3059,
+      "step": 2128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.588428213954505e-05,
+      "loss": 0.285,
+      "step": 2129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.587843718708792e-05,
+      "loss": 0.3095,
+      "step": 2130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.587258916432008e-05,
+      "loss": 0.2841,
+      "step": 2131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5866738074295946e-05,
+      "loss": 0.3057,
+      "step": 2132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5860883920071554e-05,
+      "loss": 0.333,
+      "step": 2133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5860883920071554e-05,
+      "loss": 0.2825,
+      "step": 2134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.585502670470453e-05,
+      "loss": 0.2713,
+      "step": 2135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5849166431254097e-05,
+      "loss": 0.3003,
+      "step": 2136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.584330310278108e-05,
+      "loss": 0.2695,
+      "step": 2137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.584330310278108e-05,
+      "loss": 0.2662,
+      "step": 2138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.584330310278108e-05,
+      "loss": 0.3168,
+      "step": 2139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5837436722347902e-05,
+      "loss": 0.3003,
+      "step": 2140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5831567293018572e-05,
+      "loss": 0.3304,
+      "step": 2141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5825694817858693e-05,
+      "loss": 0.2727,
+      "step": 2142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5825694817858693e-05,
+      "loss": 0.2991,
+      "step": 2143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5825694817858693e-05,
+      "loss": 0.2737,
+      "step": 2144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5825694817858693e-05,
+      "loss": 0.3187,
+      "step": 2145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.581981929993547e-05,
+      "loss": 0.2607,
+      "step": 2146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.581981929993547e-05,
+      "loss": 0.2754,
+      "step": 2147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5813940742317687e-05,
+      "loss": 0.2882,
+      "step": 2148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.580805914807571e-05,
+      "loss": 0.2763,
+      "step": 2149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.580805914807571e-05,
+      "loss": 0.3129,
+      "step": 2150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.580805914807571e-05,
+      "loss": 0.2998,
+      "step": 2151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.580805914807571e-05,
+      "loss": 0.2974,
+      "step": 2152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5802174520281512e-05,
+      "loss": 0.265,
+      "step": 2153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5796286862008623e-05,
+      "loss": 0.3108,
+      "step": 2154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5790396176332178e-05,
+      "loss": 0.3175,
+      "step": 2155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5790396176332178e-05,
+      "loss": 0.2933,
+      "step": 2156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5784502466328886e-05,
+      "loss": 0.2608,
+      "step": 2157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5784502466328886e-05,
+      "loss": 0.3032,
+      "step": 2158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.577860573507703e-05,
+      "loss": 0.3069,
+      "step": 2159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5772705985656478e-05,
+      "loss": 0.3098,
+      "step": 2160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5772705985656478e-05,
+      "loss": 0.3128,
+      "step": 2161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5766803221148676e-05,
+      "loss": 0.306,
+      "step": 2162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.576089744463663e-05,
+      "loss": 0.2838,
+      "step": 2163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.575498865920494e-05,
+      "loss": 0.3014,
+      "step": 2164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.575498865920494e-05,
+      "loss": 0.3162,
+      "step": 2165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5749076867939766e-05,
+      "loss": 0.2925,
+      "step": 2166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5743162073928834e-05,
+      "loss": 0.2652,
+      "step": 2167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5737244280261453e-05,
+      "loss": 0.3042,
+      "step": 2168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.573132349002848e-05,
+      "loss": 0.2906,
+      "step": 2169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5725399706322354e-05,
+      "loss": 0.3179,
+      "step": 2170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5725399706322354e-05,
+      "loss": 0.3139,
+      "step": 2171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.571947293223707e-05,
+      "loss": 0.3433,
+      "step": 2172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.571947293223707e-05,
+      "loss": 0.2704,
+      "step": 2173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.571354317086818e-05,
+      "loss": 0.3109,
+      "step": 2174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5707610425312806e-05,
+      "loss": 0.2897,
+      "step": 2175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.570167469866962e-05,
+      "loss": 0.2981,
+      "step": 2176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5695735994038855e-05,
+      "loss": 0.2724,
+      "step": 2177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5689794314522304e-05,
+      "loss": 0.34,
+      "step": 2178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5683849663223306e-05,
+      "loss": 0.3099,
+      "step": 2179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5683849663223306e-05,
+      "loss": 0.2994,
+      "step": 2180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5677902043246758e-05,
+      "loss": 0.3416,
+      "step": 2181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5677902043246758e-05,
+      "loss": 0.3057,
+      "step": 2182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5677902043246758e-05,
+      "loss": 0.291,
+      "step": 2183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5677902043246758e-05,
+      "loss": 0.3033,
+      "step": 2184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5671951457699102e-05,
+      "loss": 0.2968,
+      "step": 2185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5665997909688334e-05,
+      "loss": 0.2678,
+      "step": 2186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5660041402323992e-05,
+      "loss": 0.3271,
+      "step": 2187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5654081938717163e-05,
+      "loss": 0.3211,
+      "step": 2188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.564811952198048e-05,
+      "loss": 0.2943,
+      "step": 2189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5642154155228124e-05,
+      "loss": 0.2874,
+      "step": 2190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5636185841575792e-05,
+      "loss": 0.3463,
+      "step": 2191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5630214584140747e-05,
+      "loss": 0.2893,
+      "step": 2192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5624240386041786e-05,
+      "loss": 0.2837,
+      "step": 2193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.561826325039922e-05,
+      "loss": 0.3085,
+      "step": 2194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.561826325039922e-05,
+      "loss": 0.2688,
+      "step": 2195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.561826325039922e-05,
+      "loss": 0.2997,
+      "step": 2196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.561826325039922e-05,
+      "loss": 0.3078,
+      "step": 2197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5612283180334924e-05,
+      "loss": 0.3084,
+      "step": 2198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5612283180334924e-05,
+      "loss": 0.3041,
+      "step": 2199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.560630017897229e-05,
+      "loss": 0.2973,
+      "step": 2200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.560031424943623e-05,
+      "loss": 0.3055,
+      "step": 2201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.559432539485321e-05,
+      "loss": 0.2994,
+      "step": 2202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.559432539485321e-05,
+      "loss": 0.3048,
+      "step": 2203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.559432539485321e-05,
+      "loss": 0.2921,
+      "step": 2204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.558833361835121e-05,
+      "loss": 0.3266,
+      "step": 2205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5582338923059737e-05,
+      "loss": 0.2716,
+      "step": 2206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5576341312109822e-05,
+      "loss": 0.2981,
+      "step": 2207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.557034078863402e-05,
+      "loss": 0.2981,
+      "step": 2208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5564337355766412e-05,
+      "loss": 0.3044,
+      "step": 2209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5564337355766412e-05,
+      "loss": 0.3291,
+      "step": 2210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5558331016642586e-05,
+      "loss": 0.3066,
+      "step": 2211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5552321774399666e-05,
+      "loss": 0.2905,
+      "step": 2212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5546309632176276e-05,
+      "loss": 0.2576,
+      "step": 2213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5540294593112556e-05,
+      "loss": 0.302,
+      "step": 2214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5534276660350175e-05,
+      "loss": 0.2956,
+      "step": 2215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.55282558370323e-05,
+      "loss": 0.3037,
+      "step": 2216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5522232126303603e-05,
+      "loss": 0.2981,
+      "step": 2217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5516205531310272e-05,
+      "loss": 0.303,
+      "step": 2218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5516205531310272e-05,
+      "loss": 0.3361,
+      "step": 2219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5516205531310272e-05,
+      "loss": 0.2756,
+      "step": 2220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.551017605520001e-05,
+      "loss": 0.3014,
+      "step": 2221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.551017605520001e-05,
+      "loss": 0.3109,
+      "step": 2222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5504143701122005e-05,
+      "loss": 0.3561,
+      "step": 2223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5504143701122005e-05,
+      "loss": 0.3083,
+      "step": 2224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5504143701122005e-05,
+      "loss": 0.2832,
+      "step": 2225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5504143701122005e-05,
+      "loss": 0.3142,
+      "step": 2226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5504143701122005e-05,
+      "loss": 0.3279,
+      "step": 2227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5504143701122005e-05,
+      "loss": 0.303,
+      "step": 2228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5498108472226965e-05,
+      "loss": 0.3254,
+      "step": 2229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5498108472226965e-05,
+      "loss": 0.3217,
+      "step": 2230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.549207037166709e-05,
+      "loss": 0.3064,
+      "step": 2231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5486029402596088e-05,
+      "loss": 0.3018,
+      "step": 2232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5479985568169152e-05,
+      "loss": 0.2717,
+      "step": 2233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5473938871542986e-05,
+      "loss": 0.253,
+      "step": 2234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5473938871542986e-05,
+      "loss": 0.3176,
+      "step": 2235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.546788931587578e-05,
+      "loss": 0.2904,
+      "step": 2236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.546788931587578e-05,
+      "loss": 0.2966,
+      "step": 2237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5461836904327218e-05,
+      "loss": 0.3021,
+      "step": 2238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5455781640058485e-05,
+      "loss": 0.2598,
+      "step": 2239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.544972352623224e-05,
+      "loss": 0.2671,
+      "step": 2240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5443662566012645e-05,
+      "loss": 0.308,
+      "step": 2241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5437598762565342e-05,
+      "loss": 0.2782,
+      "step": 2242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5431532119057454e-05,
+      "loss": 0.3142,
+      "step": 2243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5431532119057454e-05,
+      "loss": 0.2931,
+      "step": 2244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5425462638657597e-05,
+      "loss": 0.2639,
+      "step": 2245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.541939032453586e-05,
+      "loss": 0.2798,
+      "step": 2246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.541939032453586e-05,
+      "loss": 0.3073,
+      "step": 2247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.541939032453586e-05,
+      "loss": 0.2866,
+      "step": 2248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.541939032453586e-05,
+      "loss": 0.2818,
+      "step": 2249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.541939032453586e-05,
+      "loss": 0.2799,
+      "step": 2250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.541939032453586e-05,
+      "loss": 0.3077,
+      "step": 2251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5413315179863816e-05,
+      "loss": 0.3092,
+      "step": 2252
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5407237207814524e-05,
+      "loss": 0.3081,
+      "step": 2253
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5407237207814524e-05,
+      "loss": 0.3375,
+      "step": 2254
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5401156411562498e-05,
+      "loss": 0.302,
+      "step": 2255
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5401156411562498e-05,
+      "loss": 0.2996,
+      "step": 2256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.539507279428375e-05,
+      "loss": 0.2908,
+      "step": 2257
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.538898635915576e-05,
+      "loss": 0.3081,
+      "step": 2258
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5382897109357462e-05,
+      "loss": 0.2673,
+      "step": 2259
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5376805048069288e-05,
+      "loss": 0.3036,
+      "step": 2260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5370710178473117e-05,
+      "loss": 0.3035,
+      "step": 2261
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.536461250375231e-05,
+      "loss": 0.289,
+      "step": 2262
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.536461250375231e-05,
+      "loss": 0.2938,
+      "step": 2263
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.536461250375231e-05,
+      "loss": 0.301,
+      "step": 2264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.536461250375231e-05,
+      "loss": 0.3188,
+      "step": 2265
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.535851202709168e-05,
+      "loss": 0.2824,
+      "step": 2266
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.535851202709168e-05,
+      "loss": 0.2791,
+      "step": 2267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.535851202709168e-05,
+      "loss": 0.279,
+      "step": 2268
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.535240875167751e-05,
+      "loss": 0.3431,
+      "step": 2269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.534630268069754e-05,
+      "loss": 0.2964,
+      "step": 2270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.534019381734098e-05,
+      "loss": 0.2849,
+      "step": 2271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.533408216479849e-05,
+      "loss": 0.2747,
+      "step": 2272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5327967726262193e-05,
+      "loss": 0.3114,
+      "step": 2273
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5327967726262193e-05,
+      "loss": 0.3119,
+      "step": 2274
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5327967726262193e-05,
+      "loss": 0.3185,
+      "step": 2275
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5321850504925654e-05,
+      "loss": 0.3,
+      "step": 2276
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5315730503983908e-05,
+      "loss": 0.3243,
+      "step": 2277
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5315730503983908e-05,
+      "loss": 0.2883,
+      "step": 2278
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5315730503983908e-05,
+      "loss": 0.3133,
+      "step": 2279
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5309607726633433e-05,
+      "loss": 0.303,
+      "step": 2280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5309607726633433e-05,
+      "loss": 0.2859,
+      "step": 2281
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.530348217607216e-05,
+      "loss": 0.3063,
+      "step": 2282
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.530348217607216e-05,
+      "loss": 0.3054,
+      "step": 2283
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5297353855499463e-05,
+      "loss": 0.301,
+      "step": 2284
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5291222768116175e-05,
+      "loss": 0.3119,
+      "step": 2285
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5285088917124555e-05,
+      "loss": 0.3108,
+      "step": 2286
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5285088917124555e-05,
+      "loss": 0.3186,
+      "step": 2287
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5278952305728325e-05,
+      "loss": 0.2907,
+      "step": 2288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5278952305728325e-05,
+      "loss": 0.2956,
+      "step": 2289
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5272812937132638e-05,
+      "loss": 0.2587,
+      "step": 2290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.526667081454409e-05,
+      "loss": 0.3166,
+      "step": 2291
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.526667081454409e-05,
+      "loss": 0.2769,
+      "step": 2292
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.526667081454409e-05,
+      "loss": 0.2745,
+      "step": 2293
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.526052594117071e-05,
+      "loss": 0.3377,
+      "step": 2294
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5254378320221977e-05,
+      "loss": 0.2792,
+      "step": 2295
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.524822795490879e-05,
+      "loss": 0.2997,
+      "step": 2296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5242074848443489e-05,
+      "loss": 0.292,
+      "step": 2297
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5235919004039844e-05,
+      "loss": 0.3102,
+      "step": 2298
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5229760424913056e-05,
+      "loss": 0.2888,
+      "step": 2299
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5223599114279757e-05,
+      "loss": 0.3363,
+      "step": 2300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5223599114279757e-05,
+      "loss": 0.2939,
+      "step": 2301
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5217435075358e-05,
+      "loss": 0.3121,
+      "step": 2302
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5211268311367264e-05,
+      "loss": 0.2995,
+      "step": 2303
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5205098825528462e-05,
+      "loss": 0.297,
+      "step": 2304
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5198926621063905e-05,
+      "loss": 0.3124,
+      "step": 2305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5192751701197347e-05,
+      "loss": 0.3188,
+      "step": 2306
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5186574069153954e-05,
+      "loss": 0.352,
+      "step": 2307
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5180393728160301e-05,
+      "loss": 0.3083,
+      "step": 2308
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5174210681444388e-05,
+      "loss": 0.3212,
+      "step": 2309
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5168024932235619e-05,
+      "loss": 0.2601,
+      "step": 2310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5168024932235619e-05,
+      "loss": 0.3005,
+      "step": 2311
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5161836483764815e-05,
+      "loss": 0.287,
+      "step": 2312
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.515564533926421e-05,
+      "loss": 0.3007,
+      "step": 2313
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5149451501967434e-05,
+      "loss": 0.2812,
+      "step": 2314
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5143254975109538e-05,
+      "loss": 0.2956,
+      "step": 2315
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.513705576192697e-05,
+      "loss": 0.3077,
+      "step": 2316
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.513085386565758e-05,
+      "loss": 0.3184,
+      "step": 2317
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5124649289540623e-05,
+      "loss": 0.2886,
+      "step": 2318
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5124649289540623e-05,
+      "loss": 0.3389,
+      "step": 2319
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5118442036816752e-05,
+      "loss": 0.2928,
+      "step": 2320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5118442036816752e-05,
+      "loss": 0.2888,
+      "step": 2321
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5118442036816752e-05,
+      "loss": 0.3592,
+      "step": 2322
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5112232110728016e-05,
+      "loss": 0.3108,
+      "step": 2323
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5106019514517866e-05,
+      "loss": 0.2559,
+      "step": 2324
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5099804251431142e-05,
+      "loss": 0.3277,
+      "step": 2325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.509358632471408e-05,
+      "loss": 0.311,
+      "step": 2326
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5087365737614308e-05,
+      "loss": 0.3116,
+      "step": 2327
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.508114249338084e-05,
+      "loss": 0.3309,
+      "step": 2328
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.508114249338084e-05,
+      "loss": 0.307,
+      "step": 2329
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5074916595264078e-05,
+      "loss": 0.3039,
+      "step": 2330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5068688046515813e-05,
+      "loss": 0.2604,
+      "step": 2331
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5062456850389222e-05,
+      "loss": 0.3048,
+      "step": 2332
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5056223010138857e-05,
+      "loss": 0.2838,
+      "step": 2333
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5056223010138857e-05,
+      "loss": 0.2867,
+      "step": 2334
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.504998652902066e-05,
+      "loss": 0.3144,
+      "step": 2335
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5043747410291945e-05,
+      "loss": 0.2871,
+      "step": 2336
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5037505657211409e-05,
+      "loss": 0.3168,
+      "step": 2337
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5037505657211409e-05,
+      "loss": 0.2711,
+      "step": 2338
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5031261273039122e-05,
+      "loss": 0.2572,
+      "step": 2339
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5025014261036528e-05,
+      "loss": 0.2863,
+      "step": 2340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5025014261036528e-05,
+      "loss": 0.2536,
+      "step": 2341
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5018764624466449e-05,
+      "loss": 0.2901,
+      "step": 2342
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5012512366593066e-05,
+      "loss": 0.2769,
+      "step": 2343
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5012512366593066e-05,
+      "loss": 0.2996,
+      "step": 2344
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.500625749068194e-05,
+      "loss": 0.2413,
+      "step": 2345
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2889,
+      "step": 2346
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2891,
+      "step": 2347
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2881,
+      "step": 2348
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4993739897815534e-05,
+      "loss": 0.2795,
+      "step": 2349
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4993739897815534e-05,
+      "loss": 0.3095,
+      "step": 2350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4987477187398191e-05,
+      "loss": 0.2701,
+      "step": 2351
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4981211872018998e-05,
+      "loss": 0.2683,
+      "step": 2352
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.497494395495032e-05,
+      "loss": 0.2566,
+      "step": 2353
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4968673439465901e-05,
+      "loss": 0.2506,
+      "step": 2354
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4962400328840834e-05,
+      "loss": 0.2792,
+      "step": 2355
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4956124626351569e-05,
+      "loss": 0.3012,
+      "step": 2356
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4949846335275906e-05,
+      "loss": 0.2683,
+      "step": 2357
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4943565458892999e-05,
+      "loss": 0.3104,
+      "step": 2358
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4937282000483357e-05,
+      "loss": 0.2677,
+      "step": 2359
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4930995963328827e-05,
+      "loss": 0.325,
+      "step": 2360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4924707350712615e-05,
+      "loss": 0.3069,
+      "step": 2361
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4918416165919265e-05,
+      "loss": 0.3142,
+      "step": 2362
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4912122412234665e-05,
+      "loss": 0.3091,
+      "step": 2363
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4905826092946045e-05,
+      "loss": 0.2921,
+      "step": 2364
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4899527211341977e-05,
+      "loss": 0.2854,
+      "step": 2365
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4893225770712372e-05,
+      "loss": 0.2723,
+      "step": 2366
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4893225770712372e-05,
+      "loss": 0.2671,
+      "step": 2367
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4893225770712372e-05,
+      "loss": 0.2588,
+      "step": 2368
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4886921774348473e-05,
+      "loss": 0.2935,
+      "step": 2369
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4880615225542863e-05,
+      "loss": 0.2679,
+      "step": 2370
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4874306127589451e-05,
+      "loss": 0.3136,
+      "step": 2371
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4867994483783485e-05,
+      "loss": 0.2829,
+      "step": 2372
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.486168029742154e-05,
+      "loss": 0.3009,
+      "step": 2373
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4855363571801523e-05,
+      "loss": 0.2907,
+      "step": 2374
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4849044310222654e-05,
+      "loss": 0.2834,
+      "step": 2375
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.484272251598549e-05,
+      "loss": 0.3259,
+      "step": 2376
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4836398192391915e-05,
+      "loss": 0.3112,
+      "step": 2377
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4836398192391915e-05,
+      "loss": 0.2489,
+      "step": 2378
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4830071342745112e-05,
+      "loss": 0.2892,
+      "step": 2379
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4823741970349608e-05,
+      "loss": 0.2681,
+      "step": 2380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4823741970349608e-05,
+      "loss": 0.3123,
+      "step": 2381
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4823741970349608e-05,
+      "loss": 0.2739,
+      "step": 2382
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4817410078511234e-05,
+      "loss": 0.2871,
+      "step": 2383
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4811075670537142e-05,
+      "loss": 0.2803,
+      "step": 2384
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4804738749735797e-05,
+      "loss": 0.3154,
+      "step": 2385
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4798399319416973e-05,
+      "loss": 0.2877,
+      "step": 2386
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.479205738289176e-05,
+      "loss": 0.2435,
+      "step": 2387
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4785712943472552e-05,
+      "loss": 0.2801,
+      "step": 2388
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4779366004473057e-05,
+      "loss": 0.3047,
+      "step": 2389
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4779366004473057e-05,
+      "loss": 0.2856,
+      "step": 2390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4773016569208283e-05,
+      "loss": 0.2708,
+      "step": 2391
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.476666464099454e-05,
+      "loss": 0.3093,
+      "step": 2392
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4760310223149445e-05,
+      "loss": 0.3127,
+      "step": 2393
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4753953318991918e-05,
+      "loss": 0.3111,
+      "step": 2394
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4753953318991918e-05,
+      "loss": 0.3159,
+      "step": 2395
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4747593931842169e-05,
+      "loss": 0.2659,
+      "step": 2396
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4747593931842169e-05,
+      "loss": 0.2663,
+      "step": 2397
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4747593931842169e-05,
+      "loss": 0.274,
+      "step": 2398
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4741232065021706e-05,
+      "loss": 0.3134,
+      "step": 2399
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4734867721853341e-05,
+      "loss": 0.2943,
+      "step": 2400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4728500905661174e-05,
+      "loss": 0.2924,
+      "step": 2401
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4722131619770593e-05,
+      "loss": 0.2696,
+      "step": 2402
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.471575986750828e-05,
+      "loss": 0.296,
+      "step": 2403
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.471575986750828e-05,
+      "loss": 0.2524,
+      "step": 2404
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4709385652202204e-05,
+      "loss": 0.2812,
+      "step": 2405
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4703008977181621e-05,
+      "loss": 0.2682,
+      "step": 2406
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4696629845777072e-05,
+      "loss": 0.2825,
+      "step": 2407
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4690248261320383e-05,
+      "loss": 0.2787,
+      "step": 2408
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4683864227144655e-05,
+      "loss": 0.2592,
+      "step": 2409
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4677477746584274e-05,
+      "loss": 0.2728,
+      "step": 2410
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4677477746584274e-05,
+      "loss": 0.2472,
+      "step": 2411
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4671088822974904e-05,
+      "loss": 0.2952,
+      "step": 2412
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4664697459653476e-05,
+      "loss": 0.3044,
+      "step": 2413
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.465830365995821e-05,
+      "loss": 0.277,
+      "step": 2414
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.465830365995821e-05,
+      "loss": 0.2812,
+      "step": 2415
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.465830365995821e-05,
+      "loss": 0.2693,
+      "step": 2416
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4651907427228589e-05,
+      "loss": 0.2655,
+      "step": 2417
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4645508764805368e-05,
+      "loss": 0.3223,
+      "step": 2418
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4639107676030569e-05,
+      "loss": 0.3198,
+      "step": 2419
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4632704164247487e-05,
+      "loss": 0.3126,
+      "step": 2420
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4626298232800677e-05,
+      "loss": 0.2771,
+      "step": 2421
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4619889885035963e-05,
+      "loss": 0.3222,
+      "step": 2422
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4613479124300427e-05,
+      "loss": 0.3154,
+      "step": 2423
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4613479124300427e-05,
+      "loss": 0.327,
+      "step": 2424
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4607065953942414e-05,
+      "loss": 0.2881,
+      "step": 2425
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4600650377311523e-05,
+      "loss": 0.3168,
+      "step": 2426
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4594232397758614e-05,
+      "loss": 0.2683,
+      "step": 2427
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4594232397758614e-05,
+      "loss": 0.2628,
+      "step": 2428
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4594232397758614e-05,
+      "loss": 0.3119,
+      "step": 2429
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4587812018635803e-05,
+      "loss": 0.2929,
+      "step": 2430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4581389243296455e-05,
+      "loss": 0.2947,
+      "step": 2431
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.457496407509519e-05,
+      "loss": 0.303,
+      "step": 2432
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.457496407509519e-05,
+      "loss": 0.2707,
+      "step": 2433
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.457496407509519e-05,
+      "loss": 0.2687,
+      "step": 2434
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.457496407509519e-05,
+      "loss": 0.292,
+      "step": 2435
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4568536517387876e-05,
+      "loss": 0.2851,
+      "step": 2436
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4562106573531632e-05,
+      "loss": 0.3127,
+      "step": 2437
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4555674246884816e-05,
+      "loss": 0.3102,
+      "step": 2438
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4549239540807038e-05,
+      "loss": 0.2985,
+      "step": 2439
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4542802458659154e-05,
+      "loss": 0.3301,
+      "step": 2440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4536363003803241e-05,
+      "loss": 0.3019,
+      "step": 2441
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4536363003803241e-05,
+      "loss": 0.2718,
+      "step": 2442
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4529921179602644e-05,
+      "loss": 0.3041,
+      "step": 2443
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4523476989421925e-05,
+      "loss": 0.3122,
+      "step": 2444
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4517030436626889e-05,
+      "loss": 0.3364,
+      "step": 2445
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4510581524584571e-05,
+      "loss": 0.2477,
+      "step": 2446
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4510581524584571e-05,
+      "loss": 0.2907,
+      "step": 2447
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4510581524584571e-05,
+      "loss": 0.2304,
+      "step": 2448
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.450413025666324e-05,
+      "loss": 0.2628,
+      "step": 2449
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.450413025666324e-05,
+      "loss": 0.2898,
+      "step": 2450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4497676636232404e-05,
+      "loss": 0.3012,
+      "step": 2451
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4497676636232404e-05,
+      "loss": 0.2833,
+      "step": 2452
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.449122066666279e-05,
+      "loss": 0.2832,
+      "step": 2453
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.449122066666279e-05,
+      "loss": 0.2748,
+      "step": 2454
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4484762351326344e-05,
+      "loss": 0.3388,
+      "step": 2455
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4478301693596259e-05,
+      "loss": 0.2884,
+      "step": 2456
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4471838696846936e-05,
+      "loss": 0.2914,
+      "step": 2457
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4471838696846936e-05,
+      "loss": 0.2711,
+      "step": 2458
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4465373364454001e-05,
+      "loss": 0.3066,
+      "step": 2459
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4458905699794303e-05,
+      "loss": 0.2869,
+      "step": 2460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4452435706245902e-05,
+      "loss": 0.3181,
+      "step": 2461
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4445963387188082e-05,
+      "loss": 0.26,
+      "step": 2462
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4439488746001336e-05,
+      "loss": 0.2861,
+      "step": 2463
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4439488746001336e-05,
+      "loss": 0.3059,
+      "step": 2464
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4433011786067379e-05,
+      "loss": 0.3201,
+      "step": 2465
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.442653251076912e-05,
+      "loss": 0.3119,
+      "step": 2466
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4420050923490696e-05,
+      "loss": 0.2776,
+      "step": 2467
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4413567027617442e-05,
+      "loss": 0.2866,
+      "step": 2468
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4407080826535897e-05,
+      "loss": 0.2428,
+      "step": 2469
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4400592323633809e-05,
+      "loss": 0.2977,
+      "step": 2470
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4394101522300125e-05,
+      "loss": 0.2714,
+      "step": 2471
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4394101522300125e-05,
+      "loss": 0.2588,
+      "step": 2472
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4387608425924998e-05,
+      "loss": 0.2873,
+      "step": 2473
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.438111303789977e-05,
+      "loss": 0.3051,
+      "step": 2474
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.438111303789977e-05,
+      "loss": 0.3378,
+      "step": 2475
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4374615361616985e-05,
+      "loss": 0.266,
+      "step": 2476
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4368115400470393e-05,
+      "loss": 0.3167,
+      "step": 2477
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4361613157854914e-05,
+      "loss": 0.2927,
+      "step": 2478
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4361613157854914e-05,
+      "loss": 0.2455,
+      "step": 2479
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4361613157854914e-05,
+      "loss": 0.2615,
+      "step": 2480
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.435510863716668e-05,
+      "loss": 0.3046,
+      "step": 2481
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.435510863716668e-05,
+      "loss": 0.3143,
+      "step": 2482
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.435510863716668e-05,
+      "loss": 0.322,
+      "step": 2483
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4348601841803005e-05,
+      "loss": 0.3267,
+      "step": 2484
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4342092775162386e-05,
+      "loss": 0.2953,
+      "step": 2485
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.433558144064452e-05,
+      "loss": 0.2481,
+      "step": 2486
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4329067841650274e-05,
+      "loss": 0.275,
+      "step": 2487
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4322551981581707e-05,
+      "loss": 0.3442,
+      "step": 2488
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4322551981581707e-05,
+      "loss": 0.2868,
+      "step": 2489
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4316033863842055e-05,
+      "loss": 0.2504,
+      "step": 2490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4309513491835735e-05,
+      "loss": 0.3021,
+      "step": 2491
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4309513491835735e-05,
+      "loss": 0.2498,
+      "step": 2492
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4309513491835735e-05,
+      "loss": 0.2808,
+      "step": 2493
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.430299086896834e-05,
+      "loss": 0.273,
+      "step": 2494
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4296465998646638e-05,
+      "loss": 0.2682,
+      "step": 2495
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4289938884278574e-05,
+      "loss": 0.3053,
+      "step": 2496
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4283409529273264e-05,
+      "loss": 0.2512,
+      "step": 2497
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4276877937040992e-05,
+      "loss": 0.2797,
+      "step": 2498
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4270344110993208e-05,
+      "loss": 0.3324,
+      "step": 2499
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4263808054542541e-05,
+      "loss": 0.2925,
+      "step": 2500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4263808054542541e-05,
+      "loss": 0.328,
+      "step": 2501
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.425726977110277e-05,
+      "loss": 0.2964,
+      "step": 2502
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4250729264088845e-05,
+      "loss": 0.2656,
+      "step": 2503
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4244186536916877e-05,
+      "loss": 0.2615,
+      "step": 2504
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4237641593004137e-05,
+      "loss": 0.2783,
+      "step": 2505
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4231094435769048e-05,
+      "loss": 0.27,
+      "step": 2506
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4231094435769048e-05,
+      "loss": 0.3061,
+      "step": 2507
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4231094435769048e-05,
+      "loss": 0.2876,
+      "step": 2508
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4231094435769048e-05,
+      "loss": 0.2789,
+      "step": 2509
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.42245450686312e-05,
+      "loss": 0.2665,
+      "step": 2510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4217993495011329e-05,
+      "loss": 0.3028,
+      "step": 2511
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4211439718331321e-05,
+      "loss": 0.2852,
+      "step": 2512
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4204883742014216e-05,
+      "loss": 0.3062,
+      "step": 2513
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4198325569484214e-05,
+      "loss": 0.2525,
+      "step": 2514
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4198325569484214e-05,
+      "loss": 0.3075,
+      "step": 2515
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4191765204166643e-05,
+      "loss": 0.3137,
+      "step": 2516
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4185202649487986e-05,
+      "loss": 0.2649,
+      "step": 2517
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4178637908875873e-05,
+      "loss": 0.2932,
+      "step": 2518
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4178637908875873e-05,
+      "loss": 0.2789,
+      "step": 2519
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4178637908875873e-05,
+      "loss": 0.2919,
+      "step": 2520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4172070985759064e-05,
+      "loss": 0.3192,
+      "step": 2521
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4172070985759064e-05,
+      "loss": 0.2993,
+      "step": 2522
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.416550188356747e-05,
+      "loss": 0.2485,
+      "step": 2523
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4158930605732142e-05,
+      "loss": 0.2673,
+      "step": 2524
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.415235715568525e-05,
+      "loss": 0.318,
+      "step": 2525
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4145781536860121e-05,
+      "loss": 0.295,
+      "step": 2526
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.41392037526912e-05,
+      "loss": 0.2942,
+      "step": 2527
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.41392037526912e-05,
+      "loss": 0.265,
+      "step": 2528
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4132623806614065e-05,
+      "loss": 0.3071,
+      "step": 2529
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4126041702065427e-05,
+      "loss": 0.2775,
+      "step": 2530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4119457442483124e-05,
+      "loss": 0.2875,
+      "step": 2531
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4112871031306118e-05,
+      "loss": 0.277,
+      "step": 2532
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.410628247197449e-05,
+      "loss": 0.2757,
+      "step": 2533
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.410628247197449e-05,
+      "loss": 0.2948,
+      "step": 2534
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4099691767929455e-05,
+      "loss": 0.2715,
+      "step": 2535
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4093098922613338e-05,
+      "loss": 0.3056,
+      "step": 2536
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.408650393946958e-05,
+      "loss": 0.2759,
+      "step": 2537
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.408650393946958e-05,
+      "loss": 0.2813,
+      "step": 2538
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.408650393946958e-05,
+      "loss": 0.2694,
+      "step": 2539
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.408650393946958e-05,
+      "loss": 0.301,
+      "step": 2540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.408650393946958e-05,
+      "loss": 0.3074,
+      "step": 2541
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4079906821942752e-05,
+      "loss": 0.3101,
+      "step": 2542
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4073307573478528e-05,
+      "loss": 0.3313,
+      "step": 2543
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4066706197523701e-05,
+      "loss": 0.3014,
+      "step": 2544
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.406010269752617e-05,
+      "loss": 0.2841,
+      "step": 2545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4053497076934948e-05,
+      "loss": 0.3153,
+      "step": 2546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4053497076934948e-05,
+      "loss": 0.2813,
+      "step": 2547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4046889339200154e-05,
+      "loss": 0.2448,
+      "step": 2548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4040279487773014e-05,
+      "loss": 0.2855,
+      "step": 2549
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4033667526105856e-05,
+      "loss": 0.2585,
+      "step": 2550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.402705345765211e-05,
+      "loss": 0.3004,
+      "step": 2551
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4020437285866306e-05,
+      "loss": 0.3159,
+      "step": 2552
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4013819014204075e-05,
+      "loss": 0.2631,
+      "step": 2553
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4007198646122146e-05,
+      "loss": 0.2829,
+      "step": 2554
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4000576185078338e-05,
+      "loss": 0.2701,
+      "step": 2555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3993951634531566e-05,
+      "loss": 0.2717,
+      "step": 2556
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3987324997941838e-05,
+      "loss": 0.314,
+      "step": 2557
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.398069627877025e-05,
+      "loss": 0.2922,
+      "step": 2558
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.398069627877025e-05,
+      "loss": 0.2781,
+      "step": 2559
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.397406548047898e-05,
+      "loss": 0.2583,
+      "step": 2560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3967432606531302e-05,
+      "loss": 0.3346,
+      "step": 2561
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.278,
+      "step": 2562
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3954160645525217e-05,
+      "loss": 0.2449,
+      "step": 2563
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3947521565398759e-05,
+      "loss": 0.2937,
+      "step": 2564
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3940880423479795e-05,
+      "loss": 0.291,
+      "step": 2565
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3934237223236991e-05,
+      "loss": 0.326,
+      "step": 2566
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3927591968140097e-05,
+      "loss": 0.3141,
+      "step": 2567
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3920944661659934e-05,
+      "loss": 0.3038,
+      "step": 2568
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3914295307268396e-05,
+      "loss": 0.327,
+      "step": 2569
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3907643908438434e-05,
+      "loss": 0.2847,
+      "step": 2570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3900990468644087e-05,
+      "loss": 0.2607,
+      "step": 2571
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3900990468644087e-05,
+      "loss": 0.2805,
+      "step": 2572
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3894334991360448e-05,
+      "loss": 0.3066,
+      "step": 2573
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3887677480063673e-05,
+      "loss": 0.3061,
+      "step": 2574
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3887677480063673e-05,
+      "loss": 0.2513,
+      "step": 2575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3881017938230984e-05,
+      "loss": 0.2885,
+      "step": 2576
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3874356369340667e-05,
+      "loss": 0.2988,
+      "step": 2577
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3867692776872057e-05,
+      "loss": 0.2727,
+      "step": 2578
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3861027164305556e-05,
+      "loss": 0.3019,
+      "step": 2579
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3861027164305556e-05,
+      "loss": 0.2948,
+      "step": 2580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3861027164305556e-05,
+      "loss": 0.2913,
+      "step": 2581
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3861027164305556e-05,
+      "loss": 0.2887,
+      "step": 2582
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3854359535122614e-05,
+      "loss": 0.2857,
+      "step": 2583
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3847689892805743e-05,
+      "loss": 0.2883,
+      "step": 2584
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3841018240838493e-05,
+      "loss": 0.2729,
+      "step": 2585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3834344582705475e-05,
+      "loss": 0.2715,
+      "step": 2586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3827668921892342e-05,
+      "loss": 0.3267,
+      "step": 2587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3820991261885798e-05,
+      "loss": 0.3079,
+      "step": 2588
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.381431160617359e-05,
+      "loss": 0.2722,
+      "step": 2589
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3807629958244498e-05,
+      "loss": 0.2963,
+      "step": 2590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3800946321588353e-05,
+      "loss": 0.2703,
+      "step": 2591
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3794260699696027e-05,
+      "loss": 0.2729,
+      "step": 2592
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3787573096059419e-05,
+      "loss": 0.2964,
+      "step": 2593
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3780883514171468e-05,
+      "loss": 0.3133,
+      "step": 2594
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3774191957526144e-05,
+      "loss": 0.2806,
+      "step": 2595
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3774191957526144e-05,
+      "loss": 0.2791,
+      "step": 2596
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3767498429618456e-05,
+      "loss": 0.3087,
+      "step": 2597
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3767498429618456e-05,
+      "loss": 0.2637,
+      "step": 2598
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3760802933944434e-05,
+      "loss": 0.2976,
+      "step": 2599
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3754105474001138e-05,
+      "loss": 0.2798,
+      "step": 2600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3747406053286656e-05,
+      "loss": 0.2931,
+      "step": 2601
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.37407046753001e-05,
+      "loss": 0.267,
+      "step": 2602
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3734001343541598e-05,
+      "loss": 0.2892,
+      "step": 2603
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3727296061512307e-05,
+      "loss": 0.2845,
+      "step": 2604
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3720588832714403e-05,
+      "loss": 0.2858,
+      "step": 2605
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3720588832714403e-05,
+      "loss": 0.2594,
+      "step": 2606
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3720588832714403e-05,
+      "loss": 0.3226,
+      "step": 2607
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3720588832714403e-05,
+      "loss": 0.2711,
+      "step": 2608
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3713879660651069e-05,
+      "loss": 0.2942,
+      "step": 2609
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3713879660651069e-05,
+      "loss": 0.3044,
+      "step": 2610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3713879660651069e-05,
+      "loss": 0.2616,
+      "step": 2611
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.370716854882651e-05,
+      "loss": 0.2771,
+      "step": 2612
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3700455500745947e-05,
+      "loss": 0.2702,
+      "step": 2613
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3700455500745947e-05,
+      "loss": 0.3036,
+      "step": 2614
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3700455500745947e-05,
+      "loss": 0.3142,
+      "step": 2615
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3700455500745947e-05,
+      "loss": 0.3137,
+      "step": 2616
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3700455500745947e-05,
+      "loss": 0.3248,
+      "step": 2617
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3693740519915604e-05,
+      "loss": 0.304,
+      "step": 2618
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3687023609842718e-05,
+      "loss": 0.3022,
+      "step": 2619
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3687023609842718e-05,
+      "loss": 0.2545,
+      "step": 2620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.368030477403554e-05,
+      "loss": 0.275,
+      "step": 2621
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3673584016003319e-05,
+      "loss": 0.2846,
+      "step": 2622
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3666861339256307e-05,
+      "loss": 0.2727,
+      "step": 2623
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3666861339256307e-05,
+      "loss": 0.2745,
+      "step": 2624
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3660136747305765e-05,
+      "loss": 0.2838,
+      "step": 2625
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.3123,
+      "step": 2626
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3646681831844123e-05,
+      "loss": 0.3018,
+      "step": 2627
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3646681831844123e-05,
+      "loss": 0.2862,
+      "step": 2628
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3639951515360526e-05,
+      "loss": 0.2852,
+      "step": 2629
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3633219297728415e-05,
+      "loss": 0.2788,
+      "step": 2630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3626485182464029e-05,
+      "loss": 0.2826,
+      "step": 2631
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3619749173084598e-05,
+      "loss": 0.2909,
+      "step": 2632
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3613011273108344e-05,
+      "loss": 0.2983,
+      "step": 2633
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3606271486054476e-05,
+      "loss": 0.3266,
+      "step": 2634
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3606271486054476e-05,
+      "loss": 0.2926,
+      "step": 2635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3599529815443186e-05,
+      "loss": 0.2869,
+      "step": 2636
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3592786264795659e-05,
+      "loss": 0.3042,
+      "step": 2637
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3586040837634049e-05,
+      "loss": 0.27,
+      "step": 2638
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3586040837634049e-05,
+      "loss": 0.3127,
+      "step": 2639
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3579293537481495e-05,
+      "loss": 0.3004,
+      "step": 2640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3572544367862118e-05,
+      "loss": 0.2582,
+      "step": 2641
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3572544367862118e-05,
+      "loss": 0.3014,
+      "step": 2642
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3572544367862118e-05,
+      "loss": 0.3166,
+      "step": 2643
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3572544367862118e-05,
+      "loss": 0.2741,
+      "step": 2644
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3572544367862118e-05,
+      "loss": 0.3082,
+      "step": 2645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3572544367862118e-05,
+      "loss": 0.303,
+      "step": 2646
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3572544367862118e-05,
+      "loss": 0.2713,
+      "step": 2647
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3565793332301014e-05,
+      "loss": 0.2785,
+      "step": 2648
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3565793332301014e-05,
+      "loss": 0.2848,
+      "step": 2649
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3559040434324246e-05,
+      "loss": 0.2853,
+      "step": 2650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3559040434324246e-05,
+      "loss": 0.2906,
+      "step": 2651
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3552285677458863e-05,
+      "loss": 0.2886,
+      "step": 2652
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3552285677458863e-05,
+      "loss": 0.2803,
+      "step": 2653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3545529065232874e-05,
+      "loss": 0.2968,
+      "step": 2654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3545529065232874e-05,
+      "loss": 0.2727,
+      "step": 2655
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3538770601175264e-05,
+      "loss": 0.3207,
+      "step": 2656
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3538770601175264e-05,
+      "loss": 0.2736,
+      "step": 2657
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.353201028881598e-05,
+      "loss": 0.322,
+      "step": 2658
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3525248131685936e-05,
+      "loss": 0.2981,
+      "step": 2659
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3525248131685936e-05,
+      "loss": 0.2966,
+      "step": 2660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3518484133317013e-05,
+      "loss": 0.3198,
+      "step": 2661
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3518484133317013e-05,
+      "loss": 0.2991,
+      "step": 2662
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3511718297242048e-05,
+      "loss": 0.2731,
+      "step": 2663
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.350495062699484e-05,
+      "loss": 0.2891,
+      "step": 2664
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.349818112611015e-05,
+      "loss": 0.28,
+      "step": 2665
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3491409798123687e-05,
+      "loss": 0.2806,
+      "step": 2666
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3484636646572122e-05,
+      "loss": 0.2705,
+      "step": 2667
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.347786167499308e-05,
+      "loss": 0.2791,
+      "step": 2668
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3471084886925123e-05,
+      "loss": 0.325,
+      "step": 2669
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3471084886925123e-05,
+      "loss": 0.3025,
+      "step": 2670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3464306285907776e-05,
+      "loss": 0.3057,
+      "step": 2671
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3457525875481504e-05,
+      "loss": 0.2921,
+      "step": 2672
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3450743659187722e-05,
+      "loss": 0.2561,
+      "step": 2673
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.344395964056878e-05,
+      "loss": 0.2431,
+      "step": 2674
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3437173823167979e-05,
+      "loss": 0.2937,
+      "step": 2675
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3430386210529554e-05,
+      "loss": 0.3146,
+      "step": 2676
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3423596806198675e-05,
+      "loss": 0.2839,
+      "step": 2677
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3423596806198675e-05,
+      "loss": 0.285,
+      "step": 2678
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3416805613721456e-05,
+      "loss": 0.2778,
+      "step": 2679
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3410012636644935e-05,
+      "loss": 0.2836,
+      "step": 2680
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3403217878517091e-05,
+      "loss": 0.2665,
+      "step": 2681
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3403217878517091e-05,
+      "loss": 0.297,
+      "step": 2682
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3403217878517091e-05,
+      "loss": 0.2959,
+      "step": 2683
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3396421342886832e-05,
+      "loss": 0.3004,
+      "step": 2684
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3389623033303986e-05,
+      "loss": 0.25,
+      "step": 2685
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3382822953319316e-05,
+      "loss": 0.2608,
+      "step": 2686
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3382822953319316e-05,
+      "loss": 0.2889,
+      "step": 2687
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3382822953319316e-05,
+      "loss": 0.2544,
+      "step": 2688
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3382822953319316e-05,
+      "loss": 0.2649,
+      "step": 2689
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3382822953319316e-05,
+      "loss": 0.2886,
+      "step": 2690
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3382822953319316e-05,
+      "loss": 0.289,
+      "step": 2691
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.337602110648451e-05,
+      "loss": 0.2829,
+      "step": 2692
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3369217496352173e-05,
+      "loss": 0.2858,
+      "step": 2693
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3369217496352173e-05,
+      "loss": 0.2998,
+      "step": 2694
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3369217496352173e-05,
+      "loss": 0.3297,
+      "step": 2695
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3369217496352173e-05,
+      "loss": 0.2951,
+      "step": 2696
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3362412126475835e-05,
+      "loss": 0.2919,
+      "step": 2697
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3355605000409942e-05,
+      "loss": 0.3269,
+      "step": 2698
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3348796121709862e-05,
+      "loss": 0.2743,
+      "step": 2699
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3348796121709862e-05,
+      "loss": 0.2642,
+      "step": 2700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3341985493931877e-05,
+      "loss": 0.2726,
+      "step": 2701
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3335173120633179e-05,
+      "loss": 0.2658,
+      "step": 2702
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3328359005371872e-05,
+      "loss": 0.272,
+      "step": 2703
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3321543151706978e-05,
+      "loss": 0.2709,
+      "step": 2704
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3314725563198412e-05,
+      "loss": 0.2823,
+      "step": 2705
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3314725563198412e-05,
+      "loss": 0.3112,
+      "step": 2706
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3314725563198412e-05,
+      "loss": 0.2644,
+      "step": 2707
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3307906243407011e-05,
+      "loss": 0.2744,
+      "step": 2708
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3301085195894507e-05,
+      "loss": 0.2754,
+      "step": 2709
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3301085195894507e-05,
+      "loss": 0.2762,
+      "step": 2710
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3294262424223541e-05,
+      "loss": 0.3388,
+      "step": 2711
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3287437931957642e-05,
+      "loss": 0.2695,
+      "step": 2712
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.328061172266125e-05,
+      "loss": 0.2988,
+      "step": 2713
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3273783799899697e-05,
+      "loss": 0.2837,
+      "step": 2714
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.326695416723921e-05,
+      "loss": 0.2888,
+      "step": 2715
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.326012282824691e-05,
+      "loss": 0.2955,
+      "step": 2716
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.326012282824691e-05,
+      "loss": 0.2705,
+      "step": 2717
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3253289786490812e-05,
+      "loss": 0.3221,
+      "step": 2718
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3246455045539813e-05,
+      "loss": 0.3165,
+      "step": 2719
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.32396186089637e-05,
+      "loss": 0.2771,
+      "step": 2720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3232780480333153e-05,
+      "loss": 0.2939,
+      "step": 2721
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3225940663219728e-05,
+      "loss": 0.2896,
+      "step": 2722
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3219099161195861e-05,
+      "loss": 0.2675,
+      "step": 2723
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3219099161195861e-05,
+      "loss": 0.336,
+      "step": 2724
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3219099161195861e-05,
+      "loss": 0.2565,
+      "step": 2725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3219099161195861e-05,
+      "loss": 0.3098,
+      "step": 2726
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3219099161195861e-05,
+      "loss": 0.2892,
+      "step": 2727
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3212255977834876e-05,
+      "loss": 0.2735,
+      "step": 2728
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3212255977834876e-05,
+      "loss": 0.3185,
+      "step": 2729
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3205411116710973e-05,
+      "loss": 0.3093,
+      "step": 2730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.319856458139922e-05,
+      "loss": 0.2773,
+      "step": 2731
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3191716375475572e-05,
+      "loss": 0.2762,
+      "step": 2732
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3184866502516846e-05,
+      "loss": 0.2939,
+      "step": 2733
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3184866502516846e-05,
+      "loss": 0.2724,
+      "step": 2734
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3178014966100738e-05,
+      "loss": 0.2494,
+      "step": 2735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3178014966100738e-05,
+      "loss": 0.2662,
+      "step": 2736
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3178014966100738e-05,
+      "loss": 0.2823,
+      "step": 2737
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3171161769805804e-05,
+      "loss": 0.2801,
+      "step": 2738
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3164306917211475e-05,
+      "loss": 0.2977,
+      "step": 2739
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3157450411898047e-05,
+      "loss": 0.2916,
+      "step": 2740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3150592257446671e-05,
+      "loss": 0.291,
+      "step": 2741
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3143732457439369e-05,
+      "loss": 0.2622,
+      "step": 2742
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3136871015459012e-05,
+      "loss": 0.3145,
+      "step": 2743
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.313000793508934e-05,
+      "loss": 0.296,
+      "step": 2744
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.312314321991494e-05,
+      "loss": 0.3181,
+      "step": 2745
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3116276873521261e-05,
+      "loss": 0.3008,
+      "step": 2746
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.310940889949459e-05,
+      "loss": 0.263,
+      "step": 2747
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3102539301422085e-05,
+      "loss": 0.2798,
+      "step": 2748
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3102539301422085e-05,
+      "loss": 0.3061,
+      "step": 2749
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3095668082891733e-05,
+      "loss": 0.281,
+      "step": 2750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3095668082891733e-05,
+      "loss": 0.2818,
+      "step": 2751
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3095668082891733e-05,
+      "loss": 0.2978,
+      "step": 2752
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3088795247492377e-05,
+      "loss": 0.3101,
+      "step": 2753
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3088795247492377e-05,
+      "loss": 0.3053,
+      "step": 2754
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3088795247492377e-05,
+      "loss": 0.2835,
+      "step": 2755
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3088795247492377e-05,
+      "loss": 0.2487,
+      "step": 2756
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3088795247492377e-05,
+      "loss": 0.2897,
+      "step": 2757
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3088795247492377e-05,
+      "loss": 0.2778,
+      "step": 2758
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3081920798813702e-05,
+      "loss": 0.2707,
+      "step": 2759
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3081920798813702e-05,
+      "loss": 0.2539,
+      "step": 2760
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3081920798813702e-05,
+      "loss": 0.3136,
+      "step": 2761
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3075044740446234e-05,
+      "loss": 0.3031,
+      "step": 2762
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3068167075981347e-05,
+      "loss": 0.3073,
+      "step": 2763
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3061287809011243e-05,
+      "loss": 0.292,
+      "step": 2764
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.305440694312897e-05,
+      "loss": 0.2743,
+      "step": 2765
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.305440694312897e-05,
+      "loss": 0.3074,
+      "step": 2766
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3047524481928408e-05,
+      "loss": 0.259,
+      "step": 2767
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3047524481928408e-05,
+      "loss": 0.2975,
+      "step": 2768
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3040640429004267e-05,
+      "loss": 0.2762,
+      "step": 2769
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3040640429004267e-05,
+      "loss": 0.3261,
+      "step": 2770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3040640429004267e-05,
+      "loss": 0.2803,
+      "step": 2771
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3040640429004267e-05,
+      "loss": 0.3215,
+      "step": 2772
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3033754787952094e-05,
+      "loss": 0.2642,
+      "step": 2773
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3026867562368262e-05,
+      "loss": 0.2597,
+      "step": 2774
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3019978755849971e-05,
+      "loss": 0.2765,
+      "step": 2775
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.301308837199525e-05,
+      "loss": 0.3198,
+      "step": 2776
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.300619641440295e-05,
+      "loss": 0.2719,
+      "step": 2777
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2999302886672742e-05,
+      "loss": 0.2793,
+      "step": 2778
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2999302886672742e-05,
+      "loss": 0.2762,
+      "step": 2779
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2999302886672742e-05,
+      "loss": 0.3102,
+      "step": 2780
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2992407792405121e-05,
+      "loss": 0.291,
+      "step": 2781
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2985511135201398e-05,
+      "loss": 0.308,
+      "step": 2782
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2978612918663702e-05,
+      "loss": 0.3136,
+      "step": 2783
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2971713146394971e-05,
+      "loss": 0.2903,
+      "step": 2784
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2971713146394971e-05,
+      "loss": 0.3127,
+      "step": 2785
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2964811821998961e-05,
+      "loss": 0.2666,
+      "step": 2786
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.295790894908024e-05,
+      "loss": 0.2703,
+      "step": 2787
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2951004531244178e-05,
+      "loss": 0.274,
+      "step": 2788
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2944098572096957e-05,
+      "loss": 0.2625,
+      "step": 2789
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2937191075245566e-05,
+      "loss": 0.2764,
+      "step": 2790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2937191075245566e-05,
+      "loss": 0.2901,
+      "step": 2791
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.293028204429779e-05,
+      "loss": 0.28,
+      "step": 2792
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2923371482862222e-05,
+      "loss": 0.2972,
+      "step": 2793
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.291645939454825e-05,
+      "loss": 0.2771,
+      "step": 2794
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.291645939454825e-05,
+      "loss": 0.2621,
+      "step": 2795
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.291645939454825e-05,
+      "loss": 0.2538,
+      "step": 2796
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.291645939454825e-05,
+      "loss": 0.2998,
+      "step": 2797
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.291645939454825e-05,
+      "loss": 0.2642,
+      "step": 2798
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2909545782966064e-05,
+      "loss": 0.3008,
+      "step": 2799
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2909545782966064e-05,
+      "loss": 0.3024,
+      "step": 2800
+    }
+  ],
+  "max_steps": 4482,
+  "num_train_epochs": 2,
+  "total_flos": 5.061149670210798e+18,
+  "trial_name": null,
+  "trial_params": null
+}