diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24030 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.996876951905059,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 3.6509,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 3.8724,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 3.5813,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 3.7054,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 3.5469,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 3.3915,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 2.9396,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 2.9861,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 2.5603,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 2.4441,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 2.076,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 1.6749,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 1.8523,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 1.4658,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-05,
+      "loss": 1.439,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.2239,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8333333333333335e-05,
+      "loss": 1.1319,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-05,
+      "loss": 1.1712,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.8938,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.8565,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5e-05,
+      "loss": 0.9579,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.9346,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8333333333333334e-05,
+      "loss": 0.972,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 0.8496,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.8676,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.9331,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-05,
+      "loss": 1.0395,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.7737,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.6609,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-05,
+      "loss": 0.8102,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.166666666666667e-05,
+      "loss": 1.0367,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 1.1354,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 0.9955,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.666666666666667e-05,
+      "loss": 0.763,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.833333333333334e-05,
+      "loss": 0.8098,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-05,
+      "loss": 0.8816,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.166666666666667e-05,
+      "loss": 0.9731,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.333333333333333e-05,
+      "loss": 0.9083,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.0748,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.7668,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.833333333333333e-05,
+      "loss": 0.9761,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7e-05,
+      "loss": 0.8724,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.166666666666667e-05,
+      "loss": 0.9374,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 0.7787,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.8452,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.666666666666667e-05,
+      "loss": 0.8609,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.833333333333333e-05,
+      "loss": 0.8426,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8e-05,
+      "loss": 0.7829,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.166666666666667e-05,
+      "loss": 0.8437,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.6821,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.5e-05,
+      "loss": 0.8261,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.666666666666667e-05,
+      "loss": 0.9523,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.833333333333333e-05,
+      "loss": 0.5803,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9e-05,
+      "loss": 0.6982,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 0.7902,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 0.7706,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.5e-05,
+      "loss": 0.8136,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.666666666666667e-05,
+      "loss": 0.8685,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.833333333333333e-05,
+      "loss": 0.8771,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001,
+      "loss": 0.7583,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010166666666666667,
+      "loss": 0.8556,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010333333333333334,
+      "loss": 0.8078,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000105,
+      "loss": 0.9972,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 0.8898,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010833333333333333,
+      "loss": 0.7768,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 0.7157,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011166666666666668,
+      "loss": 0.7087,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011333333333333334,
+      "loss": 0.7918,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011499999999999999,
+      "loss": 0.8944,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 0.834,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011833333333333334,
+      "loss": 0.779,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012,
+      "loss": 0.8211,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012166666666666667,
+      "loss": 0.6121,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012333333333333334,
+      "loss": 0.9408,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000125,
+      "loss": 0.9269,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 0.7972,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012833333333333335,
+      "loss": 0.8971,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 0.8007,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013166666666666668,
+      "loss": 0.7287,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.956,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013500000000000003,
+      "loss": 0.7603,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013666666666666666,
+      "loss": 0.8792,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013833333333333333,
+      "loss": 0.8505,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014,
+      "loss": 0.8082,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014166666666666668,
+      "loss": 0.7888,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014333333333333334,
+      "loss": 0.8227,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000145,
+      "loss": 0.9284,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014666666666666666,
+      "loss": 0.7486,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014833333333333335,
+      "loss": 0.9621,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.8752,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015166666666666668,
+      "loss": 0.8513,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015333333333333334,
+      "loss": 0.7876,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000155,
+      "loss": 0.897,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015666666666666666,
+      "loss": 0.7827,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015833333333333332,
+      "loss": 0.7312,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016,
+      "loss": 0.7903,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016166666666666668,
+      "loss": 1.0508,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016333333333333334,
+      "loss": 0.6758,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000165,
+      "loss": 0.9915,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.8401,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00016833333333333335,
+      "loss": 0.9372,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017,
+      "loss": 0.66,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017166666666666667,
+      "loss": 0.7643,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017333333333333334,
+      "loss": 0.8651,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000175,
+      "loss": 0.7401,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017666666666666666,
+      "loss": 0.859,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017833333333333335,
+      "loss": 0.9477,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00018,
+      "loss": 0.9452,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018166666666666667,
+      "loss": 0.757,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.8329,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018500000000000002,
+      "loss": 0.7289,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001866666666666667,
+      "loss": 0.9971,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018833333333333335,
+      "loss": 0.7959,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019,
+      "loss": 0.7427,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019166666666666667,
+      "loss": 0.7877,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 0.9469,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000195,
+      "loss": 0.8896,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019666666666666666,
+      "loss": 0.6943,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019833333333333335,
+      "loss": 0.7861,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8211,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999996722020187,
+      "loss": 0.8396,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999986888082893,
+      "loss": 0.8346,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999970498194573,
+      "loss": 0.9672,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999947552365961,
+      "loss": 0.7467,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999918050612108,
+      "loss": 0.8402,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999988199295235,
+      "loss": 0.8485,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999839379410332,
+      "loss": 0.8158,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999790210013988,
+      "loss": 0.8045,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999734484795555,
+      "loss": 0.7998,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999672203791565,
+      "loss": 0.7494,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999603367042848,
+      "loss": 0.8717,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999952797459453,
+      "loss": 0.7429,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999446026496053,
+      "loss": 0.7994,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999357522801123,
+      "loss": 0.7339,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999262463567773,
+      "loss": 0.8113,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001999916084885832,
+      "loss": 0.7465,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999052678739383,
+      "loss": 0.8394,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001999893795328188,
+      "loss": 0.7644,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998816672561022,
+      "loss": 0.6748,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998688836656323,
+      "loss": 0.9333,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998554445651586,
+      "loss": 0.6719,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998413499634925,
+      "loss": 0.8254,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998265998698737,
+      "loss": 0.8694,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999811194293973,
+      "loss": 0.9538,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999795133245889,
+      "loss": 0.7601,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997784167361527,
+      "loss": 0.8085,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997610447757222,
+      "loss": 0.8888,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997430173759875,
+      "loss": 0.9201,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997243345487665,
+      "loss": 0.7918,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999704996306308,
+      "loss": 0.716,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000199968500266129,
+      "loss": 0.8795,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996643536268204,
+      "loss": 0.7813,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999643049216436,
+      "loss": 0.8511,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996210894441047,
+      "loss": 0.8148,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019995984743242226,
+      "loss": 0.8995,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019995752038716168,
+      "loss": 0.7269,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995512781015423,
+      "loss": 0.7255,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995266970296855,
+      "loss": 0.8454,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995014606721614,
+      "loss": 0.848,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994755690455152,
+      "loss": 1.0063,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994490221667205,
+      "loss": 0.8999,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994218200531822,
+      "loss": 0.8054,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019993939627227335,
+      "loss": 0.8107,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999365450193638,
+      "loss": 0.8921,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993362824845875,
+      "loss": 0.8313,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999306459614705,
+      "loss": 1.2375,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999275981603542,
+      "loss": 0.8284,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019992448484710797,
+      "loss": 0.7206,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999213060237729,
+      "loss": 0.7933,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000199918061692433,
+      "loss": 0.8639,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019991475185521528,
+      "loss": 0.8101,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019991137651428957,
+      "loss": 0.8494,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001999079356718688,
+      "loss": 0.858,
+      "step": 173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990442933020877,
+      "loss": 0.7337,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990085749160822,
+      "loss": 0.8705,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001998972201584088,
+      "loss": 0.8827,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989351733299513,
+      "loss": 0.7868,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019988974901779483,
+      "loss": 0.7252,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019988591521527833,
+      "loss": 0.9468,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001998820159279591,
+      "loss": 0.9483,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019987805115839345,
+      "loss": 0.734,
+      "step": 181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019987402090918067,
+      "loss": 0.8512,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986992518296304,
+      "loss": 0.7809,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986576398242566,
+      "loss": 0.7592,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986153731029656,
+      "loss": 0.7912,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985724516934677,
+      "loss": 0.8442,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998528875623902,
+      "loss": 0.6838,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998484644922837,
+      "loss": 0.7564,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019984397596192697,
+      "loss": 0.7553,
+      "step": 189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001998394219742627,
+      "loss": 0.9651,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983480253227644,
+      "loss": 0.9166,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983011763899673,
+      "loss": 0.8302,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001998253672974949,
+      "loss": 0.7953,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001998205515108853,
+      "loss": 0.7255,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019981567028232514,
+      "loss": 0.8917,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001998107236150145,
+      "loss": 0.8844,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001998057115121964,
+      "loss": 0.7281,
+      "step": 197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019980063397715683,
+      "loss": 0.7675,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001997954910132245,
+      "loss": 0.7229,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019979028262377118,
+      "loss": 0.6383,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019978500881221142,
+      "loss": 0.8268,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019977966958200277,
+      "loss": 0.9039,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019977426493664554,
+      "loss": 0.9156,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001997687948796831,
+      "loss": 0.7017,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019976325941470146,
+      "loss": 0.8159,
+      "step": 205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019975765854532974,
+      "loss": 0.9077,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019975199227523983,
+      "loss": 0.8053,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019974626060814647,
+      "loss": 0.8769,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019974046354780738,
+      "loss": 0.666,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019973460109802305,
+      "loss": 0.9434,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019972867326263692,
+      "loss": 0.889,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997226800455352,
+      "loss": 0.7187,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019971662145064706,
+      "loss": 1.0446,
+      "step": 213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019971049748194447,
+      "loss": 0.7703,
+      "step": 214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001997043081434423,
+      "loss": 0.8324,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019969805343919821,
+      "loss": 0.776,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001996917333733128,
+      "loss": 1.1579,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019968534794992949,
+      "loss": 0.6975,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001996788971732345,
+      "loss": 0.7621,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019967238104745696,
+      "loss": 0.833,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996657995768688,
+      "loss": 0.7199,
+      "step": 221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019965915276578478,
+      "loss": 0.9218,
+      "step": 222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996524406185626,
+      "loss": 0.7541,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019964566313960264,
+      "loss": 0.803,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019963882033334826,
+      "loss": 0.7668,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996319122042855,
+      "loss": 0.9887,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019962493875694335,
+      "loss": 0.7719,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019961789999589356,
+      "loss": 0.8839,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001996107959257507,
+      "loss": 0.8461,
+      "step": 229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019960362655117218,
+      "loss": 0.8146,
+      "step": 230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019959639187685824,
+      "loss": 0.6732,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019958909190755187,
+      "loss": 0.8548,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019958172664803889,
+      "loss": 0.9678,
+      "step": 233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019957429610314797,
+      "loss": 0.9546,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019956680027775051,
+      "loss": 0.8935,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001995592391767608,
+      "loss": 0.8817,
+      "step": 236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995516128051358,
+      "loss": 0.7528,
+      "step": 237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995439211678754,
+      "loss": 0.9104,
+      "step": 238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995361642700221,
+      "loss": 0.8783,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995283421166614,
+      "loss": 0.8012,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995204547129214,
+      "loss": 1.0638,
+      "step": 241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019951250206397313,
+      "loss": 0.8671,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995044841750302,
+      "loss": 0.8329,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019949640105134918,
+      "loss": 0.9014,
+      "step": 244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019948825269822934,
+      "loss": 0.7816,
+      "step": 245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019948003912101273,
+      "loss": 0.8477,
+      "step": 246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000199471760325084,
+      "loss": 0.7748,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019946341631587087,
+      "loss": 0.819,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019945500709884353,
+      "loss": 0.8127,
+      "step": 249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019944653267951504,
+      "loss": 0.7812,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019943799306344125,
+      "loss": 0.899,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019942938825622065,
+      "loss": 0.8734,
+      "step": 252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001994207182634945,
+      "loss": 0.7949,
+      "step": 253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001994119830909469,
+      "loss": 0.8557,
+      "step": 254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019940318274430449,
+      "loss": 0.8166,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001993943172293368,
+      "loss": 0.8648,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019938538655185598,
+      "loss": 0.7316,
+      "step": 257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019937639071771702,
+      "loss": 0.7482,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019936732973281748,
+      "loss": 0.935,
+      "step": 259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019935820360309777,
+      "loss": 0.725,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019934901233454092,
+      "loss": 0.9583,
+      "step": 261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019933975593317262,
+      "loss": 0.8431,
+      "step": 262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019933043440506146,
+      "loss": 0.8585,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019932104775631846,
+      "loss": 0.816,
+      "step": 264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019931159599309757,
+      "loss": 0.8116,
+      "step": 265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019930207912159529,
+      "loss": 0.9234,
+      "step": 266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001992924971480508,
+      "loss": 0.7882,
+      "step": 267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001992828500787461,
+      "loss": 0.7768,
+      "step": 268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019927313792000569,
+      "loss": 0.8658,
+      "step": 269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019926336067819684,
+      "loss": 0.7786,
+      "step": 270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992535183597295,
+      "loss": 1.1715,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019924361097105623,
+      "loss": 0.8415,
+      "step": 272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019923363851867228,
+      "loss": 0.8126,
+      "step": 273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019922360100911552,
+      "loss": 1.0141,
+      "step": 274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019921349844896654,
+      "loss": 0.7947,
+      "step": 275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019920333084484857,
+      "loss": 0.7944,
+      "step": 276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019919309820342737,
+      "loss": 0.7804,
+      "step": 277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019918280053141143,
+      "loss": 0.7211,
+      "step": 278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001991724378355519,
+      "loss": 0.7582,
+      "step": 279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019916201012264254,
+      "loss": 0.812,
+      "step": 280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019915151739951964,
+      "loss": 0.7592,
+      "step": 281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019914095967306223,
+      "loss": 0.7329,
+      "step": 282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019913033695019195,
+      "loss": 0.8257,
+      "step": 283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019911964923787295,
+      "loss": 1.0124,
+      "step": 284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019910889654311208,
+      "loss": 0.7226,
+      "step": 285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001990980788729588,
+      "loss": 1.0185,
+      "step": 286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019908719623450505,
+      "loss": 0.7683,
+      "step": 287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001990762486348855,
+      "loss": 0.753,
+      "step": 288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019906523608127734,
+      "loss": 1.0254,
+      "step": 289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019905415858090036,
+      "loss": 0.9128,
+      "step": 290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001990430161410169,
+      "loss": 0.9133,
+      "step": 291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019903180876893194,
+      "loss": 0.846,
+      "step": 292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019902053647199294,
+      "loss": 0.7402,
+      "step": 293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019900919925759002,
+      "loss": 0.7582,
+      "step": 294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019899779713315575,
+      "loss": 0.6967,
+      "step": 295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019898633010616542,
+      "loss": 0.9939,
+      "step": 296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019897479818413666,
+      "loss": 0.7255,
+      "step": 297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019896320137462983,
+      "loss": 0.9029,
+      "step": 298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019895153968524766,
+      "loss": 1.0026,
+      "step": 299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019893981312363562,
+      "loss": 0.8897,
+      "step": 300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001989280216974815,
+      "loss": 0.8939,
+      "step": 301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001989161654145158,
+      "loss": 0.8435,
+      "step": 302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019890424428251137,
+      "loss": 0.755,
+      "step": 303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019889225830928365,
+      "loss": 0.7969,
+      "step": 304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019888020750269067,
+      "loss": 0.744,
+      "step": 305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019886809187063284,
+      "loss": 0.7558,
+      "step": 306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019885591142105316,
+      "loss": 0.9373,
+      "step": 307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019884366616193706,
+      "loss": 0.831,
+      "step": 308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019883135610131245,
+      "loss": 0.792,
+      "step": 309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019881898124724981,
+      "loss": 0.8501,
+      "step": 310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000198806541607862,
+      "loss": 0.7597,
+      "step": 311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987940371913044,
+      "loss": 0.9629,
+      "step": 312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987814680057749,
+      "loss": 1.0733,
+      "step": 313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019876883405951377,
+      "loss": 0.9062,
+      "step": 314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987561353608038,
+      "loss": 1.0391,
+      "step": 315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987433719179702,
+      "loss": 0.8664,
+      "step": 316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019873054373938058,
+      "loss": 0.8682,
+      "step": 317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019871765083344508,
+      "loss": 0.956,
+      "step": 318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019870469320861628,
+      "loss": 0.8336,
+      "step": 319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019869167087338907,
+      "loss": 0.8174,
+      "step": 320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001986785838363009,
+      "loss": 0.8553,
+      "step": 321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019866543210593154,
+      "loss": 0.7485,
+      "step": 322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019865221569090324,
+      "loss": 0.9945,
+      "step": 323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019863893459988062,
+      "loss": 0.8085,
+      "step": 324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019862558884157068,
+      "loss": 0.6068,
+      "step": 325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001986121784247229,
+      "loss": 0.9127,
+      "step": 326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019859870335812903,
+      "loss": 0.83,
+      "step": 327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019858516365062334,
+      "loss": 0.8814,
+      "step": 328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019857155931108235,
+      "loss": 0.8979,
+      "step": 329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019855789034842504,
+      "loss": 0.7509,
+      "step": 330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019854415677161272,
+      "loss": 0.7644,
+      "step": 331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019853035858964906,
+      "loss": 0.6914,
+      "step": 332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001985164958115801,
+      "loss": 0.7919,
+      "step": 333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019850256844649423,
+      "loss": 0.7903,
+      "step": 334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019848857650352214,
+      "loss": 1.0045,
+      "step": 335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019847451999183694,
+      "loss": 0.8523,
+      "step": 336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000198460398920654,
+      "loss": 0.8218,
+      "step": 337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000198446213299231,
+      "loss": 0.6326,
+      "step": 338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019843196313686803,
+      "loss": 0.7977,
+      "step": 339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019841764844290744,
+      "loss": 0.9348,
+      "step": 340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019840326922673383,
+      "loss": 1.7291,
+      "step": 341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019838882549777425,
+      "loss": 0.8635,
+      "step": 342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019837431726549783,
+      "loss": 0.9324,
+      "step": 343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001983597445394162,
+      "loss": 2.045,
+      "step": 344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019834510732908315,
+      "loss": 1.344,
+      "step": 345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019833040564409476,
+      "loss": 0.6456,
+      "step": 346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019831563949408945,
+      "loss": 1.0397,
+      "step": 347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019830080888874778,
+      "loss": 1.5751,
+      "step": 348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019828591383779266,
+      "loss": 1.0358,
+      "step": 349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019827095435098925,
+      "loss": 0.9787,
+      "step": 350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019825593043814492,
+      "loss": 0.8297,
+      "step": 351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019824084210910925,
+      "loss": 0.8415,
+      "step": 352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019822568937377414,
+      "loss": 0.9048,
+      "step": 353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001982104722420736,
+      "loss": 1.5912,
+      "step": 354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000198195190723984,
+      "loss": 0.9067,
+      "step": 355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019817984482952376,
+      "loss": 0.8979,
+      "step": 356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019816443456875364,
+      "loss": 0.8965,
+      "step": 357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019814895995177653,
+      "loss": 6.3219,
+      "step": 358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019813342098873752,
+      "loss": 1.2196,
+      "step": 359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001981178176898239,
+      "loss": 1.8199,
+      "step": 360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019810215006526517,
+      "loss": 2.4463,
+      "step": 361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019808641812533285,
+      "loss": 1.2921,
+      "step": 362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019807062188034085,
+      "loss": 1.7375,
+      "step": 363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019805476134064507,
+      "loss": 1.1926,
+      "step": 364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001980388365166436,
+      "loss": 1.374,
+      "step": 365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019802284741877673,
+      "loss": 3.0384,
+      "step": 366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019800679405752685,
+      "loss": 1.0249,
+      "step": 367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019799067644341844,
+      "loss": 1.376,
+      "step": 368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001979744945870182,
+      "loss": 2.1662,
+      "step": 369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001979582484989348,
+      "loss": 2.3576,
+      "step": 370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001979419381898192,
+      "loss": 2.4151,
+      "step": 371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019792556367036432,
+      "loss": 1.1511,
+      "step": 372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019790912495130524,
+      "loss": 1.0831,
+      "step": 373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019789262204341916,
+      "loss": 0.9942,
+      "step": 374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019787605495752528,
+      "loss": 1.0966,
+      "step": 375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978594237044849,
+      "loss": 1.1319,
+      "step": 376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978427282952014,
+      "loss": 1.2988,
+      "step": 377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019782596874062027,
+      "loss": 1.4181,
+      "step": 378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019780914505172897,
+      "loss": 2.0592,
+      "step": 379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019779225723955707,
+      "loss": 1.7524,
+      "step": 380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001977753053151761,
+      "loss": 1.8455,
+      "step": 381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019775828928969975,
+      "loss": 1.2825,
+      "step": 382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019774120917428358,
+      "loss": 1.0853,
+      "step": 383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001977240649801253,
+      "loss": 1.0713,
+      "step": 384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019770685671846456,
+      "loss": 1.0005,
+      "step": 385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019768958440058302,
+      "loss": 1.0666,
+      "step": 386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019767224803780433,
+      "loss": 0.9201,
+      "step": 387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019765484764149415,
+      "loss": 0.8887,
+      "step": 388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001976373832230601,
+      "loss": 0.7805,
+      "step": 389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001976198547939518,
+      "loss": 1.3601,
+      "step": 390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001976022623656608,
+      "loss": 0.8677,
+      "step": 391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019758460594972068,
+      "loss": 1.132,
+      "step": 392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019756688555770685,
+      "loss": 1.1691,
+      "step": 393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019754910120123675,
+      "loss": 1.5748,
+      "step": 394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001975312528919697,
+      "loss": 1.8913,
+      "step": 395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019751334064160706,
+      "loss": 0.9006,
+      "step": 396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019749536446189193,
+      "loss": 1.1157,
+      "step": 397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019747732436460952,
+      "loss": 0.8818,
+      "step": 398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019745922036158676,
+      "loss": 0.7211,
+      "step": 399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019744105246469263,
+      "loss": 0.9109,
+      "step": 400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019742282068583786,
+      "loss": 0.9297,
+      "step": 401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019740452503697517,
+      "loss": 1.0601,
+      "step": 402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019738616553009912,
+      "loss": 0.8584,
+      "step": 403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019736774217724614,
+      "loss": 0.7266,
+      "step": 404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019734925499049447,
+      "loss": 0.9935,
+      "step": 405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019733070398196423,
+      "loss": 1.0451,
+      "step": 406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001973120891638174,
+      "loss": 0.8234,
+      "step": 407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019729341054825782,
+      "loss": 1.1063,
+      "step": 408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019727466814753109,
+      "loss": 0.9603,
+      "step": 409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001972558619739246,
+      "loss": 1.0679,
+      "step": 410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019723699203976766,
+      "loss": 0.6256,
+      "step": 411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019721805835743134,
+      "loss": 0.6386,
+      "step": 412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971990609393284,
+      "loss": 0.9978,
+      "step": 413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019717999979791356,
+      "loss": 0.6842,
+      "step": 414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019716087494568317,
+      "loss": 1.0653,
+      "step": 415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019714168639517544,
+      "loss": 0.9135,
+      "step": 416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971224341589703,
+      "loss": 1.0789,
+      "step": 417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971031182496894,
+      "loss": 0.887,
+      "step": 418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019708373867999624,
+      "loss": 0.8216,
+      "step": 419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019706429546259593,
+      "loss": 0.7361,
+      "step": 420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001970447886102354,
+      "loss": 1.0434,
+      "step": 421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019702521813570322,
+      "loss": 0.7624,
+      "step": 422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019700558405182976,
+      "loss": 0.7764,
+      "step": 423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019698588637148703,
+      "loss": 0.8014,
+      "step": 424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019696612510758876,
+      "loss": 0.8821,
+      "step": 425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019694630027309034,
+      "loss": 0.6782,
+      "step": 426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019692641188098886,
+      "loss": 0.8908,
+      "step": 427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019690645994432305,
+      "loss": 0.6177,
+      "step": 428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019688644447617335,
+      "loss": 0.8054,
+      "step": 429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019686636548966178,
+      "loss": 0.7882,
+      "step": 430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019684622299795211,
+      "loss": 0.7332,
+      "step": 431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001968260170142496,
+      "loss": 0.8468,
+      "step": 432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019680574755180126,
+      "loss": 1.1906,
+      "step": 433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019678541462389562,
+      "loss": 0.8267,
+      "step": 434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019676501824386294,
+      "loss": 0.8112,
+      "step": 435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019674455842507492,
+      "loss": 0.8315,
+      "step": 436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000196724035180945,
+      "loss": 0.7515,
+      "step": 437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001967034485249281,
+      "loss": 0.8663,
+      "step": 438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001966827984705208,
+      "loss": 0.8226,
+      "step": 439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019666208503126112,
+      "loss": 0.9114,
+      "step": 440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019664130822072876,
+      "loss": 0.8126,
+      "step": 441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019662046805254488,
+      "loss": 0.8235,
+      "step": 442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019659956454037223,
+      "loss": 0.9065,
+      "step": 443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019657859769791505,
+      "loss": 0.6858,
+      "step": 444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019655756753891916,
+      "loss": 0.6228,
+      "step": 445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019653647407717178,
+      "loss": 0.7511,
+      "step": 446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019651531732650174,
+      "loss": 0.7122,
+      "step": 447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019649409730077935,
+      "loss": 0.8545,
+      "step": 448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001964728140139163,
+      "loss": 0.8005,
+      "step": 449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001964514674798659,
+      "loss": 0.6779,
+      "step": 450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019643005771262278,
+      "loss": 0.6613,
+      "step": 451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019640858472622316,
+      "loss": 1.033,
+      "step": 452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001963870485347446,
+      "loss": 0.8656,
+      "step": 453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001963654491523062,
+      "loss": 0.9862,
+      "step": 454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019634378659306832,
+      "loss": 0.7311,
+      "step": 455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019632206087123296,
+      "loss": 0.814,
+      "step": 456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019630027200104335,
+      "loss": 0.7113,
+      "step": 457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001962784199967842,
+      "loss": 0.7396,
+      "step": 458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019625650487278162,
+      "loss": 0.7611,
+      "step": 459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019623452664340306,
+      "loss": 0.8698,
+      "step": 460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019621248532305732,
+      "loss": 0.8915,
+      "step": 461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019619038092619464,
+      "loss": 0.8448,
+      "step": 462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019616821346730659,
+      "loss": 0.8052,
+      "step": 463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000196145982960926,
+      "loss": 0.7653,
+      "step": 464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961236894216272,
+      "loss": 0.8072,
+      "step": 465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019610133286402563,
+      "loss": 0.8389,
+      "step": 466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019607891330277827,
+      "loss": 0.7022,
+      "step": 467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019605643075258321,
+      "loss": 0.7545,
+      "step": 468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019603388522817996,
+      "loss": 0.7835,
+      "step": 469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019601127674434928,
+      "loss": 0.8487,
+      "step": 470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001959886053159132,
+      "loss": 0.7551,
+      "step": 471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019596587095773495,
+      "loss": 0.9614,
+      "step": 472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019594307368471915,
+      "loss": 0.7785,
+      "step": 473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019592021351181162,
+      "loss": 0.8786,
+      "step": 474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019589729045399934,
+      "loss": 0.8092,
+      "step": 475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001958743045263106,
+      "loss": 0.7172,
+      "step": 476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019585125574381488,
+      "loss": 0.9828,
+      "step": 477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001958281441216229,
+      "loss": 0.8684,
+      "step": 478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019580496967488647,
+      "loss": 0.7016,
+      "step": 479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019578173241879872,
+      "loss": 0.9614,
+      "step": 480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019575843236859387,
+      "loss": 1.0601,
+      "step": 481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001957350695395474,
+      "loss": 0.8885,
+      "step": 482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019571164394697584,
+      "loss": 0.5665,
+      "step": 483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001956881556062369,
+      "loss": 0.7965,
+      "step": 484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019566460453272945,
+      "loss": 0.8825,
+      "step": 485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001956409907418935,
+      "loss": 0.7037,
+      "step": 486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019561731424921016,
+      "loss": 0.8967,
+      "step": 487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019559357507020162,
+      "loss": 0.8995,
+      "step": 488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019556977322043116,
+      "loss": 1.0905,
+      "step": 489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001955459087155033,
+      "loss": 0.7101,
+      "step": 490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019552198157106335,
+      "loss": 0.6981,
+      "step": 491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019549799180279792,
+      "loss": 1.1289,
+      "step": 492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019547393942643466,
+      "loss": 0.6762,
+      "step": 493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019544982445774217,
+      "loss": 0.8395,
+      "step": 494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001954256469125301,
+      "loss": 0.8327,
+      "step": 495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019540140680664913,
+      "loss": 0.8213,
+      "step": 496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019537710415599104,
+      "loss": 0.8692,
+      "step": 497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019535273897648857,
+      "loss": 0.7718,
+      "step": 498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019532831128411533,
+      "loss": 0.748,
+      "step": 499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001953038210948861,
+      "loss": 0.7272,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952792684248565,
+      "loss": 0.6905,
+      "step": 501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019525465329012324,
+      "loss": 0.7664,
+      "step": 502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952299757068238,
+      "loss": 0.8215,
+      "step": 503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019520523569113677,
+      "loss": 0.8606,
+      "step": 504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019518043325928157,
+      "loss": 0.995,
+      "step": 505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019515556842751862,
+      "loss": 0.9492,
+      "step": 506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019513064121214914,
+      "loss": 0.7856,
+      "step": 507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019510565162951537,
+      "loss": 0.8124,
+      "step": 508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019508059969600033,
+      "loss": 0.7319,
+      "step": 509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019505548542802804,
+      "loss": 0.781,
+      "step": 510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001950303088420632,
+      "loss": 0.7732,
+      "step": 511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001950050699546116,
+      "loss": 0.763,
+      "step": 512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019497976878221968,
+      "loss": 0.6412,
+      "step": 513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019495440534147477,
+      "loss": 0.7791,
+      "step": 514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019492897964900512,
+      "loss": 0.843,
+      "step": 515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019490349172147963,
+      "loss": 0.8018,
+      "step": 516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019487794157560813,
+      "loss": 0.8308,
+      "step": 517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019485232922814117,
+      "loss": 0.7731,
+      "step": 518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001948266546958701,
+      "loss": 0.8159,
+      "step": 519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019480091799562704,
+      "loss": 1.0027,
+      "step": 520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001947751191442849,
+      "loss": 0.7653,
+      "step": 521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019474925815875729,
+      "loss": 0.8923,
+      "step": 522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019472333505599854,
+      "loss": 0.6521,
+      "step": 523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019469734985300371,
+      "loss": 0.6805,
+      "step": 524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019467130256680868,
+      "loss": 0.7221,
+      "step": 525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001946451932144899,
+      "loss": 0.7155,
+      "step": 526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019461902181316454,
+      "loss": 0.7742,
+      "step": 527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019459278837999046,
+      "loss": 0.6602,
+      "step": 528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019456649293216622,
+      "loss": 0.7018,
+      "step": 529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019454013548693102,
+      "loss": 0.8856,
+      "step": 530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019451371606156464,
+      "loss": 0.6427,
+      "step": 531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019448723467338763,
+      "loss": 0.8967,
+      "step": 532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019446069133976102,
+      "loss": 0.6785,
+      "step": 533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001944340860780865,
+      "loss": 0.8484,
+      "step": 534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019440741890580643,
+      "loss": 0.6799,
+      "step": 535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019438068984040365,
+      "loss": 0.8792,
+      "step": 536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019435389889940166,
+      "loss": 0.6023,
+      "step": 537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019432704610036446,
+      "loss": 0.9402,
+      "step": 538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001943001314608967,
+      "loss": 1.1932,
+      "step": 539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019427315499864344,
+      "loss": 0.7145,
+      "step": 540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019424611673129035,
+      "loss": 0.7456,
+      "step": 541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019421901667656365,
+      "loss": 0.7625,
+      "step": 542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019419185485223,
+      "loss": 0.641,
+      "step": 543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019416463127609656,
+      "loss": 0.7488,
+      "step": 544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019413734596601104,
+      "loss": 0.8391,
+      "step": 545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019410999893986156,
+      "loss": 0.8041,
+      "step": 546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019408259021557668,
+      "loss": 0.847,
+      "step": 547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001940551198111255,
+      "loss": 0.6587,
+      "step": 548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019402758774451754,
+      "loss": 1.034,
+      "step": 549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019399999403380266,
+      "loss": 0.774,
+      "step": 550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019397233869707114,
+      "loss": 0.7039,
+      "step": 551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019394462175245381,
+      "loss": 0.8096,
+      "step": 552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019391684321812171,
+      "loss": 0.787,
+      "step": 553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019388900311228638,
+      "loss": 1.0114,
+      "step": 554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019386110145319963,
+      "loss": 0.74,
+      "step": 555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001938331382591537,
+      "loss": 0.8035,
+      "step": 556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001938051135484812,
+      "loss": 0.8018,
+      "step": 557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019377702733955495,
+      "loss": 0.7815,
+      "step": 558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019374887965078815,
+      "loss": 0.8881,
+      "step": 559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019372067050063438,
+      "loss": 0.7119,
+      "step": 560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019369239990758738,
+      "loss": 0.8882,
+      "step": 561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019366406789018126,
+      "loss": 0.824,
+      "step": 562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019363567446699037,
+      "loss": 1.1157,
+      "step": 563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019360721965662933,
+      "loss": 0.7087,
+      "step": 564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000193578703477753,
+      "loss": 0.8689,
+      "step": 565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019355012594905646,
+      "loss": 0.6984,
+      "step": 566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019352148708927508,
+      "loss": 0.7447,
+      "step": 567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019349278691718427,
+      "loss": 0.7454,
+      "step": 568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019346402545159983,
+      "loss": 0.8421,
+      "step": 569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019343520271137763,
+      "loss": 0.8498,
+      "step": 570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019340631871541376,
+      "loss": 0.755,
+      "step": 571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019337737348264447,
+      "loss": 0.7672,
+      "step": 572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001933483670320461,
+      "loss": 0.8534,
+      "step": 573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019331929938263515,
+      "loss": 0.7932,
+      "step": 574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001932901705534683,
+      "loss": 0.8736,
+      "step": 575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019326098056364222,
+      "loss": 0.6475,
+      "step": 576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019323172943229387,
+      "loss": 0.8494,
+      "step": 577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019320241717860005,
+      "loss": 1.0929,
+      "step": 578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019317304382177782,
+      "loss": 0.7348,
+      "step": 579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019314360938108425,
+      "loss": 0.8822,
+      "step": 580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001931141138758164,
+      "loss": 0.7978,
+      "step": 581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019308455732531139,
+      "loss": 0.7134,
+      "step": 582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019305493974894642,
+      "loss": 0.8921,
+      "step": 583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019302526116613864,
+      "loss": 0.6289,
+      "step": 584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019299552159634517,
+      "loss": 0.8326,
+      "step": 585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019296572105906323,
+      "loss": 0.7289,
+      "step": 586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019293585957382987,
+      "loss": 0.9605,
+      "step": 587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019290593716022217,
+      "loss": 0.7592,
+      "step": 588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019287595383785715,
+      "loss": 0.8413,
+      "step": 589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019284590962639176,
+      "loss": 0.857,
+      "step": 590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019281580454552288,
+      "loss": 0.7063,
+      "step": 591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019278563861498723,
+      "loss": 0.7862,
+      "step": 592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019275541185456154,
+      "loss": 0.7532,
+      "step": 593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001927251242840623,
+      "loss": 0.7902,
+      "step": 594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001926947759233459,
+      "loss": 0.7466,
+      "step": 595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019266436679230865,
+      "loss": 0.8959,
+      "step": 596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019263389691088665,
+      "loss": 0.901,
+      "step": 597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001926033662990558,
+      "loss": 0.8807,
+      "step": 598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019257277497683188,
+      "loss": 0.8533,
+      "step": 599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019254212296427044,
+      "loss": 0.7948,
+      "step": 600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001925114102814668,
+      "loss": 0.8897,
+      "step": 601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019248063694855602,
+      "loss": 0.7787,
+      "step": 602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019244980298571305,
+      "loss": 0.7488,
+      "step": 603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019241890841315248,
+      "loss": 0.6055,
+      "step": 604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 0.8675,
+      "step": 605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001923569375199357,
+      "loss": 0.7248,
+      "step": 606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019232586123990738,
+      "loss": 0.8154,
+      "step": 607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001922947244314172,
+      "loss": 0.8757,
+      "step": 608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019226352711487822,
+      "loss": 0.915,
+      "step": 609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001922322693107434,
+      "loss": 0.8014,
+      "step": 610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019220095103950516,
+      "loss": 0.7581,
+      "step": 611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001921695723216957,
+      "loss": 0.7049,
+      "step": 612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001921381331778867,
+      "loss": 0.6386,
+      "step": 613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019210663362868955,
+      "loss": 0.7832,
+      "step": 614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001920750736947553,
+      "loss": 0.7197,
+      "step": 615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019204345339677442,
+      "loss": 0.8425,
+      "step": 616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001920117727554771,
+      "loss": 0.8899,
+      "step": 617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019198003179163306,
+      "loss": 0.7559,
+      "step": 618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001919482305260515,
+      "loss": 0.9371,
+      "step": 619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019191636897958122,
+      "loss": 0.8289,
+      "step": 620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019188444717311055,
+      "loss": 0.8349,
+      "step": 621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019185246512756727,
+      "loss": 0.8907,
+      "step": 622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019182042286391865,
+      "loss": 0.8293,
+      "step": 623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019178832040317155,
+      "loss": 0.778,
+      "step": 624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001917561577663721,
+      "loss": 0.818,
+      "step": 625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001917239349746061,
+      "loss": 0.8333,
+      "step": 626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019169165204899864,
+      "loss": 0.8751,
+      "step": 627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001916593090107143,
+      "loss": 0.7381,
+      "step": 628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000191626905880957,
+      "loss": 1.0315,
+      "step": 629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019159444268097012,
+      "loss": 0.744,
+      "step": 630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001915619194320364,
+      "loss": 0.8938,
+      "step": 631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019152933615547798,
+      "loss": 1.0249,
+      "step": 632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019149669287265626,
+      "loss": 0.9576,
+      "step": 633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001914639896049721,
+      "loss": 0.8832,
+      "step": 634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019143122637386566,
+      "loss": 0.8184,
+      "step": 635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001913984032008163,
+      "loss": 0.808,
+      "step": 636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001913655201073428,
+      "loss": 0.8897,
+      "step": 637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001913325771150032,
+      "loss": 0.7782,
+      "step": 638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019129957424539472,
+      "loss": 0.6742,
+      "step": 639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019126651152015403,
+      "loss": 0.7387,
+      "step": 640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001912333889609568,
+      "loss": 0.6528,
+      "step": 641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019120020658951813,
+      "loss": 0.8312,
+      "step": 642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019116696442759222,
+      "loss": 0.7113,
+      "step": 643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001911336624969725,
+      "loss": 0.6102,
+      "step": 644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019110030081949156,
+      "loss": 0.8695,
+      "step": 645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019106687941702118,
+      "loss": 0.6868,
+      "step": 646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001910333983114723,
+      "loss": 0.8666,
+      "step": 647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019099985752479506,
+      "loss": 0.6739,
+      "step": 648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019096625707897856,
+      "loss": 0.6963,
+      "step": 649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019093259699605125,
+      "loss": 0.7062,
+      "step": 650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019089887729808037,
+      "loss": 0.8559,
+      "step": 651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019086509800717258,
+      "loss": 0.8813,
+      "step": 652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019083125914547337,
+      "loss": 0.6612,
+      "step": 653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019079736073516736,
+      "loss": 0.7493,
+      "step": 654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001907634027984782,
+      "loss": 0.7552,
+      "step": 655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019072938535766865,
+      "loss": 0.8365,
+      "step": 656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019069530843504031,
+      "loss": 0.8745,
+      "step": 657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019066117205293392,
+      "loss": 0.882,
+      "step": 658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001906269762337292,
+      "loss": 0.791,
+      "step": 659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001905927209998447,
+      "loss": 0.8376,
+      "step": 660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019055840637373806,
+      "loss": 0.7178,
+      "step": 661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019052403237790582,
+      "loss": 0.7847,
+      "step": 662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001904895990348834,
+      "loss": 0.7411,
+      "step": 663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001904551063672452,
+      "loss": 0.7904,
+      "step": 664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019042055439760444,
+      "loss": 0.8403,
+      "step": 665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001903859431486133,
+      "loss": 0.6887,
+      "step": 666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019035127264296268,
+      "loss": 1.0925,
+      "step": 667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019031654290338254,
+      "loss": 0.7751,
+      "step": 668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019028175395264147,
+      "loss": 1.0082,
+      "step": 669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000190246905813547,
+      "loss": 0.65,
+      "step": 670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019021199850894545,
+      "loss": 0.7327,
+      "step": 671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019017703206172185,
+      "loss": 0.8718,
+      "step": 672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001901420064948001,
+      "loss": 0.7973,
+      "step": 673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019010692183114286,
+      "loss": 0.7636,
+      "step": 674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001900717780937514,
+      "loss": 0.8395,
+      "step": 675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001900365753056659,
+      "loss": 0.817,
+      "step": 676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001900013134899651,
+      "loss": 0.8808,
+      "step": 677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018996599266976656,
+      "loss": 0.6495,
+      "step": 678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018993061286822643,
+      "loss": 0.9393,
+      "step": 679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018989517410853955,
+      "loss": 0.8133,
+      "step": 680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018985967641393944,
+      "loss": 0.7114,
+      "step": 681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001898241198076983,
+      "loss": 0.6773,
+      "step": 682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018978850431312686,
+      "loss": 0.7994,
+      "step": 683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018975282995357446,
+      "loss": 0.7968,
+      "step": 684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001897170967524291,
+      "loss": 0.7133,
+      "step": 685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018968130473311732,
+      "loss": 0.9301,
+      "step": 686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018964545391910424,
+      "loss": 0.7954,
+      "step": 687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018960954433389345,
+      "loss": 0.7357,
+      "step": 688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001895735760010272,
+      "loss": 0.8854,
+      "step": 689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018953754894408616,
+      "loss": 0.8041,
+      "step": 690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001895014631866895,
+      "loss": 0.8201,
+      "step": 691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018946531875249493,
+      "loss": 0.7836,
+      "step": 692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018942911566519857,
+      "loss": 0.7724,
+      "step": 693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018939285394853502,
+      "loss": 0.8421,
+      "step": 694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001893565336262773,
+      "loss": 1.0076,
+      "step": 695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018932015472223693,
+      "loss": 0.8518,
+      "step": 696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001892837172602637,
+      "loss": 0.7997,
+      "step": 697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001892472212642459,
+      "loss": 0.7941,
+      "step": 698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018921066675811016,
+      "loss": 0.8069,
+      "step": 699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018917405376582145,
+      "loss": 0.7401,
+      "step": 700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018913738231138313,
+      "loss": 0.6356,
+      "step": 701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001891006524188368,
+      "loss": 0.7705,
+      "step": 702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001890638641122625,
+      "loss": 0.6332,
+      "step": 703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001890270174157784,
+      "loss": 0.7877,
+      "step": 704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018899011235354115,
+      "loss": 0.7323,
+      "step": 705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018895314894974553,
+      "loss": 0.819,
+      "step": 706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018891612722862455,
+      "loss": 0.7393,
+      "step": 707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018887904721444953,
+      "loss": 0.5993,
+      "step": 708
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018884190893153,
+      "loss": 0.7651,
+      "step": 709
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018880471240421365,
+      "loss": 0.6474,
+      "step": 710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001887674576568864,
+      "loss": 0.8727,
+      "step": 711
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018873014471397224,
+      "loss": 0.7029,
+      "step": 712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018869277359993345,
+      "loss": 0.687,
+      "step": 713
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018865534433927034,
+      "loss": 0.856,
+      "step": 714
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018861785695652142,
+      "loss": 0.8123,
+      "step": 715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018858031147626325,
+      "loss": 0.8782,
+      "step": 716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018854270792311047,
+      "loss": 0.8027,
+      "step": 717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001885050463217159,
+      "loss": 0.8702,
+      "step": 718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018846732669677025,
+      "loss": 0.9548,
+      "step": 719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018842954907300236,
+      "loss": 0.8639,
+      "step": 720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018839171347517913,
+      "loss": 0.77,
+      "step": 721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001883538199281054,
+      "loss": 0.828,
+      "step": 722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018831586845662404,
+      "loss": 0.778,
+      "step": 723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018827785908561584,
+      "loss": 0.7944,
+      "step": 724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018823979183999964,
+      "loss": 0.9976,
+      "step": 725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018820166674473216,
+      "loss": 0.7956,
+      "step": 726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018816348382480805,
+      "loss": 0.7295,
+      "step": 727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001881252431052599,
+      "loss": 0.8993,
+      "step": 728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018808694461115813,
+      "loss": 0.7765,
+      "step": 729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018804858836761107,
+      "loss": 0.7637,
+      "step": 730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000188010174399765,
+      "loss": 0.9371,
+      "step": 731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018797170273280388,
+      "loss": 0.7909,
+      "step": 732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018793317339194963,
+      "loss": 0.8053,
+      "step": 733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001878945864024619,
+      "loss": 0.6514,
+      "step": 734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001878559417896382,
+      "loss": 0.8744,
+      "step": 735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018781723957881372,
+      "loss": 0.7529,
+      "step": 736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018777847979536155,
+      "loss": 0.6732,
+      "step": 737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018773966246469237,
+      "loss": 0.8312,
+      "step": 738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018770078761225475,
+      "loss": 0.7741,
+      "step": 739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001876618552635348,
+      "loss": 0.8016,
+      "step": 740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018762286544405643,
+      "loss": 0.7435,
+      "step": 741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018758381817938127,
+      "loss": 0.7518,
+      "step": 742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001875447134951085,
+      "loss": 0.8366,
+      "step": 743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000187505551416875,
+      "loss": 0.8691,
+      "step": 744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018746633197035527,
+      "loss": 0.7876,
+      "step": 745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001874270551812614,
+      "loss": 0.7981,
+      "step": 746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018738772107534314,
+      "loss": 0.7551,
+      "step": 747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018734832967838775,
+      "loss": 1.1028,
+      "step": 748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018730888101622006,
+      "loss": 0.9075,
+      "step": 749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018726937511470246,
+      "loss": 0.8651,
+      "step": 750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018722981199973489,
+      "loss": 0.885,
+      "step": 751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018719019169725472,
+      "loss": 0.8346,
+      "step": 752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018715051423323685,
+      "loss": 0.8147,
+      "step": 753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018711077963369375,
+      "loss": 0.8039,
+      "step": 754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018707098792467515,
+      "loss": 0.8944,
+      "step": 755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018703113913226847,
+      "loss": 0.8617,
+      "step": 756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018699123328259828,
+      "loss": 0.7269,
+      "step": 757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018695127040182675,
+      "loss": 0.907,
+      "step": 758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018691125051615342,
+      "loss": 0.7895,
+      "step": 759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018687117365181512,
+      "loss": 0.9355,
+      "step": 760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001868310398350861,
+      "loss": 0.8549,
+      "step": 761
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001867908490922779,
+      "loss": 0.7331,
+      "step": 762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018675060144973944,
+      "loss": 0.7482,
+      "step": 763
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001867102969338569,
+      "loss": 0.6274,
+      "step": 764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018666993557105377,
+      "loss": 0.7205,
+      "step": 765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018662951738779076,
+      "loss": 0.6207,
+      "step": 766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018658904241056594,
+      "loss": 0.8382,
+      "step": 767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018654851066591448,
+      "loss": 0.902,
+      "step": 768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001865079221804088,
+      "loss": 0.8208,
+      "step": 769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018646727698065865,
+      "loss": 0.868,
+      "step": 770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018642657509331076,
+      "loss": 0.8533,
+      "step": 771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001863858165450492,
+      "loss": 0.7038,
+      "step": 772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018634500136259502,
+      "loss": 0.8563,
+      "step": 773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001863041295727066,
+      "loss": 0.8869,
+      "step": 774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018626320120217923,
+      "loss": 0.7725,
+      "step": 775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001862222162778454,
+      "loss": 0.7933,
+      "step": 776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001861811748265747,
+      "loss": 0.6886,
+      "step": 777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018614007687527373,
+      "loss": 0.6821,
+      "step": 778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001860989224508861,
+      "loss": 0.7628,
+      "step": 779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018605771158039253,
+      "loss": 0.7537,
+      "step": 780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001860164442908107,
+      "loss": 0.6721,
+      "step": 781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018597512060919522,
+      "loss": 0.7666,
+      "step": 782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001859337405626378,
+      "loss": 0.7936,
+      "step": 783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018589230417826697,
+      "loss": 0.7869,
+      "step": 784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018585081148324832,
+      "loss": 0.9065,
+      "step": 785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018580926250478426,
+      "loss": 0.9016,
+      "step": 786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001857676572701141,
+      "loss": 0.8175,
+      "step": 787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018572599580651415,
+      "loss": 0.8704,
+      "step": 788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001856842781412974,
+      "loss": 0.8327,
+      "step": 789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018564250430181387,
+      "loss": 0.7853,
+      "step": 790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018560067431545022,
+      "loss": 0.6945,
+      "step": 791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018555878820963013,
+      "loss": 0.7118,
+      "step": 792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001855168460118139,
+      "loss": 0.6509,
+      "step": 793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018547484774949867,
+      "loss": 0.9692,
+      "step": 794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018543279345021834,
+      "loss": 0.7238,
+      "step": 795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018539068314154354,
+      "loss": 0.945,
+      "step": 796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001853485168510816,
+      "loss": 0.767,
+      "step": 797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018530629460647657,
+      "loss": 0.7866,
+      "step": 798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018526401643540922,
+      "loss": 0.94,
+      "step": 799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018522168236559695,
+      "loss": 0.7814,
+      "step": 800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018517929242479374,
+      "loss": 0.8348,
+      "step": 801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018513684664079035,
+      "loss": 0.827,
+      "step": 802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018509434504141397,
+      "loss": 0.7679,
+      "step": 803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018505178765452853,
+      "loss": 0.7376,
+      "step": 804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001850091745080345,
+      "loss": 0.836,
+      "step": 805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018496650562986887,
+      "loss": 0.7615,
+      "step": 806
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018492378104800517,
+      "loss": 0.6513,
+      "step": 807
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018488100079045344,
+      "loss": 0.7145,
+      "step": 808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001848381648852603,
+      "loss": 0.8169,
+      "step": 809
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018479527336050878,
+      "loss": 0.6889,
+      "step": 810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001847523262443184,
+      "loss": 0.6449,
+      "step": 811
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018470932356484508,
+      "loss": 0.7576,
+      "step": 812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018466626535028121,
+      "loss": 0.7149,
+      "step": 813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018462315162885563,
+      "loss": 0.7683,
+      "step": 814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018457998242883344,
+      "loss": 0.8299,
+      "step": 815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018453675777851627,
+      "loss": 0.8349,
+      "step": 816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018449347770624198,
+      "loss": 0.7851,
+      "step": 817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018445014224038485,
+      "loss": 0.5875,
+      "step": 818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001844067514093554,
+      "loss": 0.7516,
+      "step": 819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018436330524160047,
+      "loss": 0.6599,
+      "step": 820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018431980376560324,
+      "loss": 0.7259,
+      "step": 821
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018427624700988307,
+      "loss": 0.8296,
+      "step": 822
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001842326350029956,
+      "loss": 0.7817,
+      "step": 823
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001841889677735327,
+      "loss": 0.7002,
+      "step": 824
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018414524535012244,
+      "loss": 0.8322,
+      "step": 825
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000184101467761429,
+      "loss": 0.6261,
+      "step": 826
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018405763503615284,
+      "loss": 0.7882,
+      "step": 827
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018401374720303056,
+      "loss": 0.7369,
+      "step": 828
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018396980429083475,
+      "loss": 0.582,
+      "step": 829
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018392580632837423,
+      "loss": 0.7864,
+      "step": 830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018388175334449396,
+      "loss": 0.8018,
+      "step": 831
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018383764536807485,
+      "loss": 0.8805,
+      "step": 832
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018379348242803385,
+      "loss": 0.7097,
+      "step": 833
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001837492645533241,
+      "loss": 0.8331,
+      "step": 834
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018370499177293464,
+      "loss": 0.9276,
+      "step": 835
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001836606641158905,
+      "loss": 0.7617,
+      "step": 836
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001836162816112527,
+      "loss": 0.7274,
+      "step": 837
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018357184428811828,
+      "loss": 0.6609,
+      "step": 838
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018352735217562013,
+      "loss": 0.7582,
+      "step": 839
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018348280530292713,
+      "loss": 0.719,
+      "step": 840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000183438203699244,
+      "loss": 0.8751,
+      "step": 841
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001833935473938114,
+      "loss": 0.739,
+      "step": 842
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018334883641590575,
+      "loss": 0.76,
+      "step": 843
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018330407079483952,
+      "loss": 0.7853,
+      "step": 844
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018325925055996076,
+      "loss": 0.8082,
+      "step": 845
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018321437574065347,
+      "loss": 0.7367,
+      "step": 846
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018316944636633743,
+      "loss": 0.7813,
+      "step": 847
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001831244624664681,
+      "loss": 0.8694,
+      "step": 848
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018307942407053674,
+      "loss": 0.8614,
+      "step": 849
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001830343312080704,
+      "loss": 0.7012,
+      "step": 850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018298918390863175,
+      "loss": 0.9371,
+      "step": 851
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018294398220181917,
+      "loss": 0.9336,
+      "step": 852
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018289872611726673,
+      "loss": 0.7229,
+      "step": 853
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018285341568464414,
+      "loss": 0.7194,
+      "step": 854
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018280805093365672,
+      "loss": 0.5526,
+      "step": 855
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001827626318940454,
+      "loss": 0.7177,
+      "step": 856
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018271715859558677,
+      "loss": 0.7777,
+      "step": 857
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018267163106809288,
+      "loss": 0.6419,
+      "step": 858
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018262604934141146,
+      "loss": 0.8032,
+      "step": 859
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018258041344542566,
+      "loss": 0.7169,
+      "step": 860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001825347234100542,
+      "loss": 1.02,
+      "step": 861
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001824889792652513,
+      "loss": 0.8615,
+      "step": 862
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018244318104100658,
+      "loss": 0.8184,
+      "step": 863
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018239732876734527,
+      "loss": 0.834,
+      "step": 864
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018235142247432782,
+      "loss": 0.7535,
+      "step": 865
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018230546219205032,
+      "loss": 0.8242,
+      "step": 866
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018225944795064404,
+      "loss": 0.7736,
+      "step": 867
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018221337978027583,
+      "loss": 0.811,
+      "step": 868
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018216725771114773,
+      "loss": 0.81,
+      "step": 869
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001821210817734972,
+      "loss": 0.7197,
+      "step": 870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000182074851997597,
+      "loss": 0.9024,
+      "step": 871
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018202856841375518,
+      "loss": 0.6468,
+      "step": 872
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001819822310523151,
+      "loss": 0.7431,
+      "step": 873
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001819358399436553,
+      "loss": 0.6912,
+      "step": 874
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018188939511818965,
+      "loss": 0.7594,
+      "step": 875
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018184289660636715,
+      "loss": 0.6277,
+      "step": 876
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018179634443867207,
+      "loss": 0.7571,
+      "step": 877
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001817497386456238,
+      "loss": 0.8256,
+      "step": 878
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001817030792577769,
+      "loss": 0.7068,
+      "step": 879
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001816563663057211,
+      "loss": 0.6752,
+      "step": 880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018160959982008122,
+      "loss": 0.8982,
+      "step": 881
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001815627798315172,
+      "loss": 0.8652,
+      "step": 882
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018151590637072396,
+      "loss": 0.6783,
+      "step": 883
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018146897946843163,
+      "loss": 0.6434,
+      "step": 884
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018142199915540527,
+      "loss": 0.7475,
+      "step": 885
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018137496546244498,
+      "loss": 0.7745,
+      "step": 886
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001813278784203859,
+      "loss": 0.8309,
+      "step": 887
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000181280738060098,
+      "loss": 0.6819,
+      "step": 888
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001812335444124864,
+      "loss": 0.6933,
+      "step": 889
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018118629750849105,
+      "loss": 0.5936,
+      "step": 890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001811389973790868,
+      "loss": 0.7178,
+      "step": 891
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001810916440552835,
+      "loss": 0.8156,
+      "step": 892
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001810442375681257,
+      "loss": 0.9001,
+      "step": 893
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018099677794869296,
+      "loss": 0.815,
+      "step": 894
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001809492652280996,
+      "loss": 0.7136,
+      "step": 895
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 0.7121,
+      "step": 896
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001808540806080624,
+      "loss": 0.9789,
+      "step": 897
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001808064087710212,
+      "loss": 0.6437,
+      "step": 898
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018075868395762466,
+      "loss": 0.8143,
+      "step": 899
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018071090619916093,
+      "loss": 0.7523,
+      "step": 900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018066307552695298,
+      "loss": 0.8618,
+      "step": 901
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018061519197235836,
+      "loss": 0.7881,
+      "step": 902
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018056725556676932,
+      "loss": 0.5721,
+      "step": 903
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018051926634161282,
+      "loss": 0.8736,
+      "step": 904
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018047122432835038,
+      "loss": 0.7147,
+      "step": 905
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018042312955847818,
+      "loss": 0.723,
+      "step": 906
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001803749820635269,
+      "loss": 0.5867,
+      "step": 907
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018032678187506187,
+      "loss": 0.8015,
+      "step": 908
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018027852902468294,
+      "loss": 0.7489,
+      "step": 909
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001802302235440245,
+      "loss": 0.6513,
+      "step": 910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001801818654647554,
+      "loss": 0.7774,
+      "step": 911
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018013345481857903,
+      "loss": 0.7811,
+      "step": 912
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001800849916372332,
+      "loss": 0.7134,
+      "step": 913
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018003647595249013,
+      "loss": 0.7153,
+      "step": 914
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001799879077961566,
+      "loss": 0.7692,
+      "step": 915
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001799392872000736,
+      "loss": 0.7273,
+      "step": 916
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017989061419611667,
+      "loss": 0.6219,
+      "step": 917
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017984188881619564,
+      "loss": 0.7013,
+      "step": 918
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001797931110922546,
+      "loss": 0.7175,
+      "step": 919
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017974428105627208,
+      "loss": 0.8797,
+      "step": 920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017969539874026087,
+      "loss": 0.7152,
+      "step": 921
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017964646417626797,
+      "loss": 0.9644,
+      "step": 922
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017959747739637472,
+      "loss": 0.8494,
+      "step": 923
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017954843843269664,
+      "loss": 0.7647,
+      "step": 924
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017949934731738347,
+      "loss": 0.6678,
+      "step": 925
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017945020408261916,
+      "loss": 0.6749,
+      "step": 926
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017940100876062182,
+      "loss": 0.655,
+      "step": 927
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001793517613836437,
+      "loss": 0.7376,
+      "step": 928
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017930246198397115,
+      "loss": 0.7938,
+      "step": 929
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001792531105939247,
+      "loss": 0.7795,
+      "step": 930
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017920370724585895,
+      "loss": 0.7104,
+      "step": 931
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017915425197216245,
+      "loss": 0.6705,
+      "step": 932
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017910474480525794,
+      "loss": 0.7389,
+      "step": 933
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017905518577760208,
+      "loss": 0.8265,
+      "step": 934
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001790055749216856,
+      "loss": 0.8465,
+      "step": 935
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017895591227003315,
+      "loss": 0.6147,
+      "step": 936
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017890619785520338,
+      "loss": 0.8657,
+      "step": 937
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001788564317097889,
+      "loss": 0.7079,
+      "step": 938
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017880661386641614,
+      "loss": 0.9004,
+      "step": 939
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017875674435774547,
+      "loss": 0.8611,
+      "step": 940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017870682321647114,
+      "loss": 0.8298,
+      "step": 941
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001786568504753213,
+      "loss": 0.6698,
+      "step": 942
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017860682616705785,
+      "loss": 0.85,
+      "step": 943
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017855675032447648,
+      "loss": 0.7686,
+      "step": 944
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017850662298040678,
+      "loss": 0.5775,
+      "step": 945
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017845644416771198,
+      "loss": 0.7421,
+      "step": 946
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017840621391928917,
+      "loss": 1.0828,
+      "step": 947
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017835593226806903,
+      "loss": 0.6465,
+      "step": 948
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017830559924701604,
+      "loss": 0.8314,
+      "step": 949
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001782552148891283,
+      "loss": 0.7396,
+      "step": 950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017820477922743762,
+      "loss": 0.9174,
+      "step": 951
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017815429229500946,
+      "loss": 0.7693,
+      "step": 952
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017810375412494276,
+      "loss": 0.7952,
+      "step": 953
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017805316475037018,
+      "loss": 0.7138,
+      "step": 954
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017800252420445788,
+      "loss": 0.6924,
+      "step": 955
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017795183252040567,
+      "loss": 0.6545,
+      "step": 956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017790108973144678,
+      "loss": 1.0184,
+      "step": 957
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017785029587084794,
+      "loss": 0.7687,
+      "step": 958
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017779945097190942,
+      "loss": 0.6105,
+      "step": 959
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017774855506796496,
+      "loss": 0.83,
+      "step": 960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017769760819238165,
+      "loss": 0.7243,
+      "step": 961
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001776466103785601,
+      "loss": 0.8047,
+      "step": 962
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017759556165993428,
+      "loss": 0.8478,
+      "step": 963
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001775444620699715,
+      "loss": 0.757,
+      "step": 964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001774933116421725,
+      "loss": 0.6599,
+      "step": 965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017744211041007118,
+      "loss": 0.842,
+      "step": 966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017739085840723498,
+      "loss": 0.7881,
+      "step": 967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001773395556672644,
+      "loss": 0.7875,
+      "step": 968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017728820222379338,
+      "loss": 0.6974,
+      "step": 969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017723679811048904,
+      "loss": 0.7585,
+      "step": 970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017718534336105165,
+      "loss": 0.8348,
+      "step": 971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017713383800921478,
+      "loss": 0.6044,
+      "step": 972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001770822820887451,
+      "loss": 0.7534,
+      "step": 973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001770306756334425,
+      "loss": 0.7952,
+      "step": 974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017697901867713995,
+      "loss": 0.8525,
+      "step": 975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017692731125370354,
+      "loss": 0.7961,
+      "step": 976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017687555339703244,
+      "loss": 1.0254,
+      "step": 977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017682374514105888,
+      "loss": 0.7888,
+      "step": 978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017677188651974817,
+      "loss": 0.8766,
+      "step": 979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017671997756709863,
+      "loss": 0.7805,
+      "step": 980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001766680183171415,
+      "loss": 0.5599,
+      "step": 981
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001766160088039411,
+      "loss": 0.7275,
+      "step": 982
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017656394906159464,
+      "loss": 0.6505,
+      "step": 983
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017651183912423228,
+      "loss": 0.8124,
+      "step": 984
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001764596790260171,
+      "loss": 0.8291,
+      "step": 985
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017640746880114505,
+      "loss": 0.7428,
+      "step": 986
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001763552084838449,
+      "loss": 0.6476,
+      "step": 987
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017630289810837834,
+      "loss": 0.7687,
+      "step": 988
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001762505377090398,
+      "loss": 0.8618,
+      "step": 989
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017619812732015664,
+      "loss": 0.7814,
+      "step": 990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001761456669760888,
+      "loss": 0.7989,
+      "step": 991
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001760931567112291,
+      "loss": 0.8835,
+      "step": 992
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001760405965600031,
+      "loss": 0.8057,
+      "step": 993
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000175987986556869,
+      "loss": 0.7426,
+      "step": 994
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017593532673631766,
+      "loss": 0.6891,
+      "step": 995
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017588261713287267,
+      "loss": 0.7632,
+      "step": 996
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017582985778109026,
+      "loss": 0.7792,
+      "step": 997
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001757770487155592,
+      "loss": 0.674,
+      "step": 998
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000175724189970901,
+      "loss": 0.6418,
+      "step": 999
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017567128158176953,
+      "loss": 0.7169,
+      "step": 1000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017561832358285138,
+      "loss": 0.7671,
+      "step": 1001
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017556531600886554,
+      "loss": 0.7621,
+      "step": 1002
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017551225889456365,
+      "loss": 0.7583,
+      "step": 1003
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017545915227472965,
+      "loss": 0.8833,
+      "step": 1004
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017540599618418007,
+      "loss": 0.725,
+      "step": 1005
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001753527906577638,
+      "loss": 0.7384,
+      "step": 1006
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017529953573036224,
+      "loss": 1.037,
+      "step": 1007
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017524623143688902,
+      "loss": 0.6814,
+      "step": 1008
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001751928778122903,
+      "loss": 0.7488,
+      "step": 1009
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017513947489154443,
+      "loss": 0.8158,
+      "step": 1010
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017508602270966216,
+      "loss": 0.6863,
+      "step": 1011
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017503252130168657,
+      "loss": 0.6875,
+      "step": 1012
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001749789707026929,
+      "loss": 0.8221,
+      "step": 1013
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001749253709477888,
+      "loss": 0.9427,
+      "step": 1014
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017487172207211396,
+      "loss": 0.7597,
+      "step": 1015
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017481802411084042,
+      "loss": 0.7965,
+      "step": 1016
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001747642770991723,
+      "loss": 0.6991,
+      "step": 1017
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017471048107234598,
+      "loss": 0.7611,
+      "step": 1018
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017465663606562988,
+      "loss": 0.7399,
+      "step": 1019
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001746027421143246,
+      "loss": 0.6661,
+      "step": 1020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001745487992537628,
+      "loss": 0.8229,
+      "step": 1021
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017449480751930912,
+      "loss": 0.7213,
+      "step": 1022
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017444076694636041,
+      "loss": 0.73,
+      "step": 1023
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017438667757034546,
+      "loss": 0.9427,
+      "step": 1024
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017433253942672496,
+      "loss": 0.6311,
+      "step": 1025
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017427835255099172,
+      "loss": 0.8465,
+      "step": 1026
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017422411697867046,
+      "loss": 0.9116,
+      "step": 1027
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017416983274531775,
+      "loss": 0.8761,
+      "step": 1028
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017411549988652212,
+      "loss": 0.7848,
+      "step": 1029
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000174061118437904,
+      "loss": 0.7138,
+      "step": 1030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001740066884351156,
+      "loss": 0.6889,
+      "step": 1031
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001739522099138411,
+      "loss": 0.7285,
+      "step": 1032
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001738976829097963,
+      "loss": 0.7797,
+      "step": 1033
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017384310745872895,
+      "loss": 0.6561,
+      "step": 1034
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017378848359641847,
+      "loss": 0.9094,
+      "step": 1035
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017373381135867604,
+      "loss": 0.7241,
+      "step": 1036
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017367909078134453,
+      "loss": 0.7219,
+      "step": 1037
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017362432190029862,
+      "loss": 0.8374,
+      "step": 1038
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001735695047514445,
+      "loss": 0.8104,
+      "step": 1039
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017351463937072004,
+      "loss": 0.9037,
+      "step": 1040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017345972579409488,
+      "loss": 0.7776,
+      "step": 1041
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017340476405756998,
+      "loss": 0.7263,
+      "step": 1042
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017334975419717815,
+      "loss": 0.6951,
+      "step": 1043
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001732946962489836,
+      "loss": 0.8177,
+      "step": 1044
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017323959024908209,
+      "loss": 0.7882,
+      "step": 1045
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001731844362336009,
+      "loss": 0.7204,
+      "step": 1046
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001731292342386988,
+      "loss": 0.68,
+      "step": 1047
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017307398430056593,
+      "loss": 0.8076,
+      "step": 1048
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017301868645542401,
+      "loss": 0.8671,
+      "step": 1049
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017296334073952605,
+      "loss": 0.8185,
+      "step": 1050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017290794718915643,
+      "loss": 0.6169,
+      "step": 1051
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000172852505840631,
+      "loss": 0.8338,
+      "step": 1052
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017279701673029687,
+      "loss": 0.7491,
+      "step": 1053
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017274147989453247,
+      "loss": 0.859,
+      "step": 1054
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001726858953697475,
+      "loss": 0.8591,
+      "step": 1055
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017263026319238301,
+      "loss": 0.7975,
+      "step": 1056
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017257458339891118,
+      "loss": 0.8403,
+      "step": 1057
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017251885602583545,
+      "loss": 0.6687,
+      "step": 1058
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017246308110969053,
+      "loss": 0.7616,
+      "step": 1059
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017240725868704218,
+      "loss": 0.6852,
+      "step": 1060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017235138879448733,
+      "loss": 0.8182,
+      "step": 1061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001722954714686541,
+      "loss": 1.0798,
+      "step": 1062
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017223950674620164,
+      "loss": 0.7648,
+      "step": 1063
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017218349466382023,
+      "loss": 0.6719,
+      "step": 1064
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017212743525823112,
+      "loss": 0.6476,
+      "step": 1065
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017207132856618667,
+      "loss": 0.7717,
+      "step": 1066
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017201517462447013,
+      "loss": 0.8554,
+      "step": 1067
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001719589734698959,
+      "loss": 0.8765,
+      "step": 1068
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017190272513930915,
+      "loss": 0.7007,
+      "step": 1069
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001718464296695861,
+      "loss": 0.8068,
+      "step": 1070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001717900870976338,
+      "loss": 0.67,
+      "step": 1071
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017173369746039025,
+      "loss": 0.7765,
+      "step": 1072
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017167726079482426,
+      "loss": 0.8964,
+      "step": 1073
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017162077713793545,
+      "loss": 0.7219,
+      "step": 1074
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001715642465267543,
+      "loss": 0.7028,
+      "step": 1075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017150766899834204,
+      "loss": 0.7724,
+      "step": 1076
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017145104458979074,
+      "loss": 0.7078,
+      "step": 1077
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000171394373338223,
+      "loss": 0.8496,
+      "step": 1078
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017133765528079239,
+      "loss": 0.9062,
+      "step": 1079
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017128089045468294,
+      "loss": 0.7347,
+      "step": 1080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001712240788971095,
+      "loss": 0.8078,
+      "step": 1081
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017116722064531748,
+      "loss": 0.7295,
+      "step": 1082
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017111031573658294,
+      "loss": 0.8403,
+      "step": 1083
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017105336420821247,
+      "loss": 0.6432,
+      "step": 1084
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017099636609754329,
+      "loss": 0.644,
+      "step": 1085
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001709393214419431,
+      "loss": 0.847,
+      "step": 1086
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017088223027881023,
+      "loss": 0.7044,
+      "step": 1087
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001708250926455733,
+      "loss": 0.8025,
+      "step": 1088
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017076790857969163,
+      "loss": 0.7007,
+      "step": 1089
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.6679,
+      "step": 1090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001706534012999828,
+      "loss": 0.7731,
+      "step": 1091
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017059607816122618,
+      "loss": 0.753,
+      "step": 1092
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017053870873996572,
+      "loss": 0.7222,
+      "step": 1093
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017048129307381266,
+      "loss": 0.7462,
+      "step": 1094
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017042383120040834,
+      "loss": 0.6969,
+      "step": 1095
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017036632315742462,
+      "loss": 0.9344,
+      "step": 1096
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017030876898256354,
+      "loss": 0.7562,
+      "step": 1097
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017025116871355735,
+      "loss": 0.8233,
+      "step": 1098
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001701935223881686,
+      "loss": 0.6335,
+      "step": 1099
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017013583004418993,
+      "loss": 0.736,
+      "step": 1100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017007809171944423,
+      "loss": 0.7096,
+      "step": 1101
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017002030745178455,
+      "loss": 0.7125,
+      "step": 1102
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016996247727909397,
+      "loss": 0.6483,
+      "step": 1103
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016990460123928575,
+      "loss": 0.8764,
+      "step": 1104
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016984667937030318,
+      "loss": 0.8556,
+      "step": 1105
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001697887117101196,
+      "loss": 0.6606,
+      "step": 1106
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016973069829673837,
+      "loss": 0.8142,
+      "step": 1107
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016967263916819287,
+      "loss": 0.7089,
+      "step": 1108
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001696145343625464,
+      "loss": 0.7784,
+      "step": 1109
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016955638391789228,
+      "loss": 0.8951,
+      "step": 1110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016949818787235366,
+      "loss": 0.792,
+      "step": 1111
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016943994626408363,
+      "loss": 0.8853,
+      "step": 1112
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001693816591312652,
+      "loss": 0.9822,
+      "step": 1113
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016932332651211116,
+      "loss": 0.746,
+      "step": 1114
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016926494844486412,
+      "loss": 0.7391,
+      "step": 1115
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001692065249677965,
+      "loss": 0.9108,
+      "step": 1116
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016914805611921056,
+      "loss": 0.7977,
+      "step": 1117
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016908954193743816,
+      "loss": 1.0081,
+      "step": 1118
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016903098246084098,
+      "loss": 0.6629,
+      "step": 1119
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016897237772781044,
+      "loss": 0.6262,
+      "step": 1120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016891372777676748,
+      "loss": 0.7589,
+      "step": 1121
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016885503264616283,
+      "loss": 0.7699,
+      "step": 1122
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016879629237447676,
+      "loss": 0.7152,
+      "step": 1123
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016873750700021915,
+      "loss": 0.8177,
+      "step": 1124
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016867867656192946,
+      "loss": 0.6904,
+      "step": 1125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001686198010981767,
+      "loss": 0.6232,
+      "step": 1126
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016856088064755938,
+      "loss": 0.6911,
+      "step": 1127
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016850191524870546,
+      "loss": 0.6231,
+      "step": 1128
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001684429049402725,
+      "loss": 0.7527,
+      "step": 1129
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016838384976094738,
+      "loss": 0.8166,
+      "step": 1130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016832474974944642,
+      "loss": 0.8718,
+      "step": 1131
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016826560494451537,
+      "loss": 0.8681,
+      "step": 1132
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016820641538492934,
+      "loss": 0.6555,
+      "step": 1133
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016814718110949275,
+      "loss": 0.8215,
+      "step": 1134
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016808790215703935,
+      "loss": 0.6504,
+      "step": 1135
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016802857856643215,
+      "loss": 0.7562,
+      "step": 1136
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001679692103765635,
+      "loss": 0.6297,
+      "step": 1137
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016790979762635496,
+      "loss": 0.7602,
+      "step": 1138
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016785034035475725,
+      "loss": 0.7967,
+      "step": 1139
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016779083860075033,
+      "loss": 0.8625,
+      "step": 1140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001677312924033433,
+      "loss": 0.7811,
+      "step": 1141
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016767170180157444,
+      "loss": 0.749,
+      "step": 1142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016761206683451108,
+      "loss": 0.806,
+      "step": 1143
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016755238754124965,
+      "loss": 0.8395,
+      "step": 1144
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001674926639609157,
+      "loss": 0.6652,
+      "step": 1145
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001674328961326637,
+      "loss": 0.8078,
+      "step": 1146
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016737308409567727,
+      "loss": 0.7759,
+      "step": 1147
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016731322788916892,
+      "loss": 0.8967,
+      "step": 1148
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001672533275523801,
+      "loss": 0.7249,
+      "step": 1149
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016719338312458124,
+      "loss": 0.6671,
+      "step": 1150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016713339464507172,
+      "loss": 0.7989,
+      "step": 1151
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016707336215317968,
+      "loss": 0.702,
+      "step": 1152
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001670132856882622,
+      "loss": 0.6556,
+      "step": 1153
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016695316528970517,
+      "loss": 0.7699,
+      "step": 1154
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016689300099692332,
+      "loss": 0.8325,
+      "step": 1155
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016683279284936004,
+      "loss": 0.7143,
+      "step": 1156
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016677254088648757,
+      "loss": 0.7666,
+      "step": 1157
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016671224514780693,
+      "loss": 0.7173,
+      "step": 1158
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016665190567284764,
+      "loss": 0.6302,
+      "step": 1159
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016659152250116812,
+      "loss": 0.7839,
+      "step": 1160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016653109567235527,
+      "loss": 0.6993,
+      "step": 1161
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016647062522602473,
+      "loss": 0.6919,
+      "step": 1162
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016641011120182065,
+      "loss": 0.7547,
+      "step": 1163
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016634955363941574,
+      "loss": 0.7296,
+      "step": 1164
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016628895257851135,
+      "loss": 0.8771,
+      "step": 1165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001662283080588373,
+      "loss": 0.6203,
+      "step": 1166
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001661676201201518,
+      "loss": 0.8908,
+      "step": 1167
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016610688880224178,
+      "loss": 0.716,
+      "step": 1168
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001660461141449223,
+      "loss": 0.7355,
+      "step": 1169
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000165985296188037,
+      "loss": 0.747,
+      "step": 1170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016592443497145793,
+      "loss": 0.7416,
+      "step": 1171
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001658635305350855,
+      "loss": 0.729,
+      "step": 1172
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001658025829188483,
+      "loss": 0.7984,
+      "step": 1173
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001657415921627034,
+      "loss": 0.7819,
+      "step": 1174
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001656805583066361,
+      "loss": 0.8155,
+      "step": 1175
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016561948139065996,
+      "loss": 0.8535,
+      "step": 1176
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016555836145481674,
+      "loss": 0.7974,
+      "step": 1177
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001654971985391764,
+      "loss": 0.7551,
+      "step": 1178
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016543599268383714,
+      "loss": 0.7544,
+      "step": 1179
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016537474392892528,
+      "loss": 0.6385,
+      "step": 1180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001653134523145952,
+      "loss": 0.826,
+      "step": 1181
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016525211788102946,
+      "loss": 0.9611,
+      "step": 1182
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001651907406684387,
+      "loss": 0.718,
+      "step": 1183
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016512932071706152,
+      "loss": 0.8885,
+      "step": 1184
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016506785806716465,
+      "loss": 0.7014,
+      "step": 1185
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016500635275904272,
+      "loss": 0.7309,
+      "step": 1186
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016494480483301836,
+      "loss": 0.7353,
+      "step": 1187
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001648832143294422,
+      "loss": 0.9406,
+      "step": 1188
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016482158128869258,
+      "loss": 1.0239,
+      "step": 1189
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016475990575117605,
+      "loss": 0.7183,
+      "step": 1190
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016469818775732668,
+      "loss": 0.7745,
+      "step": 1191
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001646364273476067,
+      "loss": 0.6641,
+      "step": 1192
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016457462456250584,
+      "loss": 0.7575,
+      "step": 1193
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016451277944254185,
+      "loss": 0.8114,
+      "step": 1194
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001644508920282601,
+      "loss": 0.7699,
+      "step": 1195
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016438896236023375,
+      "loss": 0.7707,
+      "step": 1196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016432699047906363,
+      "loss": 0.737,
+      "step": 1197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016426497642537825,
+      "loss": 0.8471,
+      "step": 1198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016420292023983378,
+      "loss": 0.9261,
+      "step": 1199
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000164140821963114,
+      "loss": 0.7724,
+      "step": 1200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016407868163593027,
+      "loss": 0.7028,
+      "step": 1201
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001640164992990216,
+      "loss": 0.7371,
+      "step": 1202
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016395427499315442,
+      "loss": 0.6352,
+      "step": 1203
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016389200875912278,
+      "loss": 0.7697,
+      "step": 1204
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001638297006377481,
+      "loss": 0.873,
+      "step": 1205
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001637673506698794,
+      "loss": 0.7766,
+      "step": 1206
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016370495889639304,
+      "loss": 0.7073,
+      "step": 1207
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016364252535819282,
+      "loss": 0.7323,
+      "step": 1208
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016358005009620992,
+      "loss": 0.786,
+      "step": 1209
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016351753315140287,
+      "loss": 0.7615,
+      "step": 1210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001634549745647575,
+      "loss": 0.8211,
+      "step": 1211
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000163392374377287,
+      "loss": 0.745,
+      "step": 1212
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016332973263003176,
+      "loss": 0.6718,
+      "step": 1213
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016326704936405953,
+      "loss": 0.9161,
+      "step": 1214
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016320432462046516,
+      "loss": 0.8297,
+      "step": 1215
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016314155844037074,
+      "loss": 0.7377,
+      "step": 1216
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016307875086492548,
+      "loss": 0.5759,
+      "step": 1217
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016301590193530584,
+      "loss": 0.795,
+      "step": 1218
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016295301169271531,
+      "loss": 0.7401,
+      "step": 1219
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016289008017838445,
+      "loss": 0.8508,
+      "step": 1220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016282710743357096,
+      "loss": 0.6911,
+      "step": 1221
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016276409349955944,
+      "loss": 0.8146,
+      "step": 1222
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016270103841766167,
+      "loss": 0.6548,
+      "step": 1223
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001626379422292162,
+      "loss": 0.807,
+      "step": 1224
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016257480497558873,
+      "loss": 0.7765,
+      "step": 1225
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001625116266981717,
+      "loss": 0.79,
+      "step": 1226
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001624484074383846,
+      "loss": 0.7595,
+      "step": 1227
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016238514723767374,
+      "loss": 0.6465,
+      "step": 1228
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016232184613751218,
+      "loss": 0.7716,
+      "step": 1229
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001622585041793999,
+      "loss": 0.9568,
+      "step": 1230
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016219512140486363,
+      "loss": 0.7801,
+      "step": 1231
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001621316978554569,
+      "loss": 0.7859,
+      "step": 1232
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016206823357275984,
+      "loss": 0.7454,
+      "step": 1233
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016200472859837945,
+      "loss": 1.1258,
+      "step": 1234
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016194118297394936,
+      "loss": 0.89,
+      "step": 1235
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016187759674112973,
+      "loss": 0.8703,
+      "step": 1236
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016181396994160752,
+      "loss": 0.8565,
+      "step": 1237
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016175030261709615,
+      "loss": 0.8997,
+      "step": 1238
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001616865948093357,
+      "loss": 0.7286,
+      "step": 1239
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016162284656009274,
+      "loss": 0.8531,
+      "step": 1240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016155905791116038,
+      "loss": 0.9182,
+      "step": 1241
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016149522890435814,
+      "loss": 0.7161,
+      "step": 1242
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001614313595815321,
+      "loss": 0.7212,
+      "step": 1243
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016136744998455476,
+      "loss": 0.8635,
+      "step": 1244
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016130350015532496,
+      "loss": 0.8798,
+      "step": 1245
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016123951013576794,
+      "loss": 0.7718,
+      "step": 1246
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016117547996783533,
+      "loss": 0.7027,
+      "step": 1247
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016111140969350503,
+      "loss": 0.7919,
+      "step": 1248
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016104729935478124,
+      "loss": 0.6695,
+      "step": 1249
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016098314899369446,
+      "loss": 0.7681,
+      "step": 1250
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016091895865230141,
+      "loss": 0.7496,
+      "step": 1251
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016085472837268502,
+      "loss": 0.7029,
+      "step": 1252
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016079045819695437,
+      "loss": 0.8447,
+      "step": 1253
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016072614816724478,
+      "loss": 0.832,
+      "step": 1254
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001606617983257176,
+      "loss": 0.6667,
+      "step": 1255
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016059740871456036,
+      "loss": 0.7167,
+      "step": 1256
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001605329793759866,
+      "loss": 0.7535,
+      "step": 1257
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016046851035223593,
+      "loss": 0.8269,
+      "step": 1258
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016040400168557405,
+      "loss": 0.8114,
+      "step": 1259
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016033945341829248,
+      "loss": 0.7501,
+      "step": 1260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016027486559270886,
+      "loss": 0.7137,
+      "step": 1261
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016021023825116672,
+      "loss": 0.7913,
+      "step": 1262
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016014557143603545,
+      "loss": 0.7648,
+      "step": 1263
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016008086518971037,
+      "loss": 0.7205,
+      "step": 1264
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016001611955461265,
+      "loss": 0.9152,
+      "step": 1265
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001599513345731892,
+      "loss": 0.6588,
+      "step": 1266
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015988651028791287,
+      "loss": 0.7564,
+      "step": 1267
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001598216467412822,
+      "loss": 0.7539,
+      "step": 1268
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001597567439758214,
+      "loss": 0.8606,
+      "step": 1269
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001596918020340805,
+      "loss": 0.6577,
+      "step": 1270
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015962682095863523,
+      "loss": 0.8534,
+      "step": 1271
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015956180079208682,
+      "loss": 0.8389,
+      "step": 1272
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015949674157706228,
+      "loss": 0.6407,
+      "step": 1273
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001594316433562142,
+      "loss": 0.7298,
+      "step": 1274
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015936650617222063,
+      "loss": 0.764,
+      "step": 1275
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001593013300677853,
+      "loss": 0.8274,
+      "step": 1276
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001592361150856374,
+      "loss": 0.6794,
+      "step": 1277
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001591708612685316,
+      "loss": 0.756,
+      "step": 1278
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015910556865924802,
+      "loss": 0.9038,
+      "step": 1279
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015904023730059228,
+      "loss": 0.7789,
+      "step": 1280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015897486723539533,
+      "loss": 0.7292,
+      "step": 1281
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015890945850651346,
+      "loss": 0.9091,
+      "step": 1282
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015884401115682847,
+      "loss": 0.818,
+      "step": 1283
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015877852522924732,
+      "loss": 0.7368,
+      "step": 1284
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015871300076670234,
+      "loss": 0.759,
+      "step": 1285
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001586474378121511,
+      "loss": 0.76,
+      "step": 1286
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001585818364085764,
+      "loss": 0.9963,
+      "step": 1287
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015851619659898623,
+      "loss": 0.8492,
+      "step": 1288
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015845051842641383,
+      "loss": 0.7923,
+      "step": 1289
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015838480193391754,
+      "loss": 0.8058,
+      "step": 1290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001583190471645808,
+      "loss": 0.7064,
+      "step": 1291
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015825325416151222,
+      "loss": 0.6573,
+      "step": 1292
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015818742296784535,
+      "loss": 0.7897,
+      "step": 1293
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015812155362673896,
+      "loss": 0.7828,
+      "step": 1294
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001580556461813766,
+      "loss": 0.668,
+      "step": 1295
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000157989700674967,
+      "loss": 0.7288,
+      "step": 1296
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015792371715074376,
+      "loss": 0.7361,
+      "step": 1297
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001578576956519654,
+      "loss": 0.7025,
+      "step": 1298
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015779163622191538,
+      "loss": 0.7684,
+      "step": 1299
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015772553890390197,
+      "loss": 0.6602,
+      "step": 1300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001576594037412583,
+      "loss": 0.9337,
+      "step": 1301
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001575932307773423,
+      "loss": 0.779,
+      "step": 1302
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001575270200555367,
+      "loss": 0.8531,
+      "step": 1303
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015746077161924905,
+      "loss": 0.7833,
+      "step": 1304
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001573944855119115,
+      "loss": 0.6894,
+      "step": 1305
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015732816177698098,
+      "loss": 0.8026,
+      "step": 1306
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000157261800457939,
+      "loss": 0.8205,
+      "step": 1307
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015719540159829184,
+      "loss": 0.6721,
+      "step": 1308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001571289652415703,
+      "loss": 0.753,
+      "step": 1309
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015706249143132982,
+      "loss": 0.7912,
+      "step": 1310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015699598021115028,
+      "loss": 0.5783,
+      "step": 1311
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015692943162463628,
+      "loss": 0.7996,
+      "step": 1312
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015686284571541671,
+      "loss": 0.6549,
+      "step": 1313
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015679622252714507,
+      "loss": 0.751,
+      "step": 1314
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015672956210349923,
+      "loss": 0.7803,
+      "step": 1315
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001566628644881815,
+      "loss": 0.8471,
+      "step": 1316
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015659612972491858,
+      "loss": 0.772,
+      "step": 1317
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001565293578574615,
+      "loss": 0.8832,
+      "step": 1318
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015646254892958566,
+      "loss": 0.8665,
+      "step": 1319
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015639570298509064,
+      "loss": 0.8066,
+      "step": 1320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015632882006780046,
+      "loss": 0.9439,
+      "step": 1321
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015626190022156327,
+      "loss": 0.9167,
+      "step": 1322
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001561949434902514,
+      "loss": 0.8207,
+      "step": 1323
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015612794991776147,
+      "loss": 0.9199,
+      "step": 1324
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001560609195480142,
+      "loss": 0.747,
+      "step": 1325
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015599385242495438,
+      "loss": 0.7274,
+      "step": 1326
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015592674859255096,
+      "loss": 0.8337,
+      "step": 1327
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015585960809479696,
+      "loss": 0.7462,
+      "step": 1328
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001557924309757094,
+      "loss": 0.7635,
+      "step": 1329
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015572521727932935,
+      "loss": 0.9531,
+      "step": 1330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015565796704972184,
+      "loss": 0.7138,
+      "step": 1331
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015559068033097582,
+      "loss": 0.7285,
+      "step": 1332
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001555233571672042,
+      "loss": 0.8221,
+      "step": 1333
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015545599760254382,
+      "loss": 0.6298,
+      "step": 1334
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015538860168115527,
+      "loss": 0.807,
+      "step": 1335
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015532116944722308,
+      "loss": 0.6649,
+      "step": 1336
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015525370094495556,
+      "loss": 0.7202,
+      "step": 1337
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015518619621858476,
+      "loss": 0.6853,
+      "step": 1338
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001551186553123665,
+      "loss": 0.7626,
+      "step": 1339
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015505107827058036,
+      "loss": 0.6392,
+      "step": 1340
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015498346513752957,
+      "loss": 0.7727,
+      "step": 1341
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000154915815957541,
+      "loss": 0.802,
+      "step": 1342
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001548481307749652,
+      "loss": 0.7913,
+      "step": 1343
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001547804096341763,
+      "loss": 0.7507,
+      "step": 1344
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015471265257957202,
+      "loss": 0.749,
+      "step": 1345
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001546448596555736,
+      "loss": 0.8569,
+      "step": 1346
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015457703090662576,
+      "loss": 0.8356,
+      "step": 1347
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015450916637719684,
+      "loss": 0.7783,
+      "step": 1348
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015444126611177848,
+      "loss": 0.9263,
+      "step": 1349
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015437333015488587,
+      "loss": 0.7359,
+      "step": 1350
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015430535855105753,
+      "loss": 1.0603,
+      "step": 1351
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015423735134485536,
+      "loss": 0.8113,
+      "step": 1352
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001541693085808646,
+      "loss": 0.742,
+      "step": 1353
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015410123030369386,
+      "loss": 0.9023,
+      "step": 1354
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015403311655797492,
+      "loss": 0.685,
+      "step": 1355
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015396496738836292,
+      "loss": 0.8507,
+      "step": 1356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015389678283953616,
+      "loss": 0.7255,
+      "step": 1357
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001538285629561962,
+      "loss": 0.7862,
+      "step": 1358
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015376030778306766,
+      "loss": 0.8411,
+      "step": 1359
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001536920173648984,
+      "loss": 0.8774,
+      "step": 1360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001536236917464593,
+      "loss": 0.7173,
+      "step": 1361
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015355533097254436,
+      "loss": 0.7336,
+      "step": 1362
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001534869350879707,
+      "loss": 0.891,
+      "step": 1363
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001534185041375783,
+      "loss": 0.6952,
+      "step": 1364
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015335003816623028,
+      "loss": 0.7578,
+      "step": 1365
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001532815372188126,
+      "loss": 0.8295,
+      "step": 1366
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015321300134023424,
+      "loss": 0.7273,
+      "step": 1367
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015314443057542703,
+      "loss": 0.7864,
+      "step": 1368
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001530758249693457,
+      "loss": 0.7327,
+      "step": 1369
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015300718456696778,
+      "loss": 0.7818,
+      "step": 1370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001529385094132937,
+      "loss": 0.713,
+      "step": 1371
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015286979955334652,
+      "loss": 0.7514,
+      "step": 1372
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015280105503217225,
+      "loss": 0.6877,
+      "step": 1373
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015273227589483946,
+      "loss": 0.7906,
+      "step": 1374
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015266346218643947,
+      "loss": 0.8441,
+      "step": 1375
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015259461395208628,
+      "loss": 0.8263,
+      "step": 1376
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015252573123691654,
+      "loss": 0.7415,
+      "step": 1377
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015245681408608945,
+      "loss": 0.7761,
+      "step": 1378
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015238786254478682,
+      "loss": 0.9678,
+      "step": 1379
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000152318876658213,
+      "loss": 0.8058,
+      "step": 1380
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001522498564715949,
+      "loss": 0.7609,
+      "step": 1381
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015218080203018182,
+      "loss": 0.7379,
+      "step": 1382
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001521117133792456,
+      "loss": 0.858,
+      "step": 1383
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015204259056408046,
+      "loss": 0.8597,
+      "step": 1384
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015197343363000307,
+      "loss": 0.698,
+      "step": 1385
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001519042426223524,
+      "loss": 0.7371,
+      "step": 1386
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015183501758648987,
+      "loss": 0.8358,
+      "step": 1387
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015176575856779904,
+      "loss": 0.7387,
+      "step": 1388
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001516964656116859,
+      "loss": 0.7181,
+      "step": 1389
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015162713876357858,
+      "loss": 0.7505,
+      "step": 1390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015155777806892754,
+      "loss": 1.019,
+      "step": 1391
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015148838357320537,
+      "loss": 0.7116,
+      "step": 1392
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015141895532190677,
+      "loss": 0.7876,
+      "step": 1393
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015134949336054865,
+      "loss": 0.6581,
+      "step": 1394
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015127999773467002,
+      "loss": 0.7123,
+      "step": 1395
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001512104684898319,
+      "loss": 0.6713,
+      "step": 1396
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015114090567161734,
+      "loss": 0.8461,
+      "step": 1397
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001510713093256315,
+      "loss": 0.9173,
+      "step": 1398
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001510016794975015,
+      "loss": 0.6969,
+      "step": 1399
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015093201623287631,
+      "loss": 0.7124,
+      "step": 1400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001508623195774269,
+      "loss": 0.6165,
+      "step": 1401
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001507925895768461,
+      "loss": 0.8286,
+      "step": 1402
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015072282627684867,
+      "loss": 0.7192,
+      "step": 1403
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015065302972317108,
+      "loss": 0.9173,
+      "step": 1404
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015058319996157172,
+      "loss": 0.6801,
+      "step": 1405
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015051333703783068,
+      "loss": 0.6879,
+      "step": 1406
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001504434409977498,
+      "loss": 0.598,
+      "step": 1407
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015037351188715265,
+      "loss": 0.7261,
+      "step": 1408
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001503035497518845,
+      "loss": 0.8939,
+      "step": 1409
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001502335546378122,
+      "loss": 0.7161,
+      "step": 1410
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015016352659082428,
+      "loss": 0.7742,
+      "step": 1411
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015009346565683087,
+      "loss": 0.7509,
+      "step": 1412
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001500233718817636,
+      "loss": 0.7152,
+      "step": 1413
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001499532453115757,
+      "loss": 0.8825,
+      "step": 1414
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014988308599224183,
+      "loss": 0.6963,
+      "step": 1415
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014981289396975817,
+      "loss": 0.8952,
+      "step": 1416
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014974266929014234,
+      "loss": 0.7149,
+      "step": 1417
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014967241199943332,
+      "loss": 0.6696,
+      "step": 1418
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014960212214369155,
+      "loss": 0.5835,
+      "step": 1419
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014953179976899878,
+      "loss": 0.7761,
+      "step": 1420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000149461444921458,
+      "loss": 0.8417,
+      "step": 1421
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014939105764719368,
+      "loss": 0.778,
+      "step": 1422
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014932063799235135,
+      "loss": 0.6825,
+      "step": 1423
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014925018600309785,
+      "loss": 0.7067,
+      "step": 1424
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001491797017256212,
+      "loss": 0.7655,
+      "step": 1425
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014910918520613073,
+      "loss": 0.8152,
+      "step": 1426
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014903863649085668,
+      "loss": 0.7448,
+      "step": 1427
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001489680556260505,
+      "loss": 0.7509,
+      "step": 1428
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014889744265798478,
+      "loss": 0.7651,
+      "step": 1429
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014882679763295306,
+      "loss": 0.5907,
+      "step": 1430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014875612059726993,
+      "loss": 0.7993,
+      "step": 1431
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014868541159727096,
+      "loss": 0.8609,
+      "step": 1432
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014861467067931272,
+      "loss": 0.8653,
+      "step": 1433
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014854389788977266,
+      "loss": 0.7315,
+      "step": 1434
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001484730932750491,
+      "loss": 0.714,
+      "step": 1435
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001484022568815613,
+      "loss": 0.7004,
+      "step": 1436
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001483313887557493,
+      "loss": 0.7298,
+      "step": 1437
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014826048894407397,
+      "loss": 0.9142,
+      "step": 1438
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001481895574930169,
+      "loss": 0.6334,
+      "step": 1439
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014811859444908052,
+      "loss": 0.8139,
+      "step": 1440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014804759985878786,
+      "loss": 0.6127,
+      "step": 1441
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014797657376868273,
+      "loss": 0.9046,
+      "step": 1442
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001479055162253295,
+      "loss": 0.7271,
+      "step": 1443
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014783442727531328,
+      "loss": 0.8972,
+      "step": 1444
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014776330696523963,
+      "loss": 0.846,
+      "step": 1445
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014769215534173475,
+      "loss": 0.9177,
+      "step": 1446
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001476209724514454,
+      "loss": 0.8201,
+      "step": 1447
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014754975834103877,
+      "loss": 0.6874,
+      "step": 1448
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014747851305720256,
+      "loss": 0.7929,
+      "step": 1449
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014740723664664483,
+      "loss": 0.8737,
+      "step": 1450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014733592915609414,
+      "loss": 0.9325,
+      "step": 1451
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014726459063229945,
+      "loss": 0.7913,
+      "step": 1452
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014719322112202992,
+      "loss": 0.6657,
+      "step": 1453
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014712182067207517,
+      "loss": 0.7492,
+      "step": 1454
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014705038932924503,
+      "loss": 0.8277,
+      "step": 1455
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014697892714036958,
+      "loss": 0.776,
+      "step": 1456
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014690743415229918,
+      "loss": 0.683,
+      "step": 1457
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001468359104119043,
+      "loss": 0.7246,
+      "step": 1458
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001467643559660757,
+      "loss": 0.7263,
+      "step": 1459
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014669277086172406,
+      "loss": 0.7353,
+      "step": 1460
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014662115514578038,
+      "loss": 0.7482,
+      "step": 1461
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014654950886519564,
+      "loss": 0.6343,
+      "step": 1462
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001464778320669408,
+      "loss": 0.7966,
+      "step": 1463
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014640612479800686,
+      "loss": 0.8045,
+      "step": 1464
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014633438710540489,
+      "loss": 0.8755,
+      "step": 1465
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014626261903616578,
+      "loss": 0.8442,
+      "step": 1466
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001461908206373404,
+      "loss": 0.8017,
+      "step": 1467
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014611899195599953,
+      "loss": 0.6298,
+      "step": 1468
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001460471330392337,
+      "loss": 0.6288,
+      "step": 1469
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014597524393415335,
+      "loss": 0.922,
+      "step": 1470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014590332468788868,
+      "loss": 0.9224,
+      "step": 1471
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014583137534758967,
+      "loss": 0.7782,
+      "step": 1472
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014575939596042604,
+      "loss": 0.6592,
+      "step": 1473
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014568738657358714,
+      "loss": 0.7492,
+      "step": 1474
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014561534723428205,
+      "loss": 0.8096,
+      "step": 1475
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001455432779897395,
+      "loss": 0.677,
+      "step": 1476
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014547117888720777,
+      "loss": 0.8121,
+      "step": 1477
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014539904997395468,
+      "loss": 0.9615,
+      "step": 1478
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014532689129726777,
+      "loss": 0.7023,
+      "step": 1479
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014525470290445392,
+      "loss": 0.7383,
+      "step": 1480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014518248484283948,
+      "loss": 0.5984,
+      "step": 1481
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014511023715977047,
+      "loss": 0.7832,
+      "step": 1482
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014503795990261205,
+      "loss": 0.6561,
+      "step": 1483
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014496565311874902,
+      "loss": 0.7158,
+      "step": 1484
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014489331685558525,
+      "loss": 0.723,
+      "step": 1485
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014482095116054422,
+      "loss": 0.7918,
+      "step": 1486
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014474855608106858,
+      "loss": 0.7664,
+      "step": 1487
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014467613166462023,
+      "loss": 0.9447,
+      "step": 1488
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014460367795868034,
+      "loss": 0.9244,
+      "step": 1489
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014453119501074924,
+      "loss": 0.8869,
+      "step": 1490
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014445868286834648,
+      "loss": 0.9234,
+      "step": 1491
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001443861415790107,
+      "loss": 0.7654,
+      "step": 1492
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001443135711902997,
+      "loss": 0.7178,
+      "step": 1493
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014424097174979038,
+      "loss": 0.6083,
+      "step": 1494
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014416834330507856,
+      "loss": 0.8265,
+      "step": 1495
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014409568590377918,
+      "loss": 0.7071,
+      "step": 1496
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014402299959352614,
+      "loss": 0.873,
+      "step": 1497
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001439502844219723,
+      "loss": 0.7955,
+      "step": 1498
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014387754043678947,
+      "loss": 0.8353,
+      "step": 1499
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014380476768566824,
+      "loss": 1.0436,
+      "step": 1500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014373196621631816,
+      "loss": 0.974,
+      "step": 1501
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014365913607646761,
+      "loss": 0.8572,
+      "step": 1502
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001435862773138637,
+      "loss": 0.6742,
+      "step": 1503
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014351338997627234,
+      "loss": 0.7399,
+      "step": 1504
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014344047411147818,
+      "loss": 0.7241,
+      "step": 1505
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001433675297672846,
+      "loss": 0.6266,
+      "step": 1506
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014329455699151354,
+      "loss": 0.769,
+      "step": 1507
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014322155583200576,
+      "loss": 0.7495,
+      "step": 1508
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014314852633662045,
+      "loss": 0.8028,
+      "step": 1509
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014307546855323549,
+      "loss": 0.7671,
+      "step": 1510
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014300238252974723,
+      "loss": 0.7795,
+      "step": 1511
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014292926831407061,
+      "loss": 0.6647,
+      "step": 1512
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000142856125954139,
+      "loss": 0.6447,
+      "step": 1513
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001427829554979042,
+      "loss": 0.797,
+      "step": 1514
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014270975699333654,
+      "loss": 0.7916,
+      "step": 1515
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001426365304884246,
+      "loss": 0.8226,
+      "step": 1516
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001425632760311754,
+      "loss": 0.823,
+      "step": 1517
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001424899936696143,
+      "loss": 0.7964,
+      "step": 1518
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014241668345178486,
+      "loss": 0.7626,
+      "step": 1519
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014234334542574906,
+      "loss": 0.6746,
+      "step": 1520
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014226997963958686,
+      "loss": 0.8372,
+      "step": 1521
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014219658614139674,
+      "loss": 0.7628,
+      "step": 1522
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014212316497929506,
+      "loss": 0.7627,
+      "step": 1523
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014204971620141647,
+      "loss": 1.0511,
+      "step": 1524
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014197623985591373,
+      "loss": 0.6545,
+      "step": 1525
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014190273599095762,
+      "loss": 0.6587,
+      "step": 1526
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014182920465473692,
+      "loss": 0.7417,
+      "step": 1527
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014175564589545854,
+      "loss": 0.7774,
+      "step": 1528
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001416820597613473,
+      "loss": 0.7665,
+      "step": 1529
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014160844630064595,
+      "loss": 0.9658,
+      "step": 1530
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001415348055616152,
+      "loss": 0.7743,
+      "step": 1531
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014146113759253362,
+      "loss": 0.6876,
+      "step": 1532
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014138744244169762,
+      "loss": 0.7549,
+      "step": 1533
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014131372015742142,
+      "loss": 0.6643,
+      "step": 1534
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014123997078803707,
+      "loss": 0.7086,
+      "step": 1535
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001411661943818944,
+      "loss": 0.7412,
+      "step": 1536
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001410923909873609,
+      "loss": 0.9003,
+      "step": 1537
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014101856065282172,
+      "loss": 0.7405,
+      "step": 1538
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001409447034266798,
+      "loss": 0.7827,
+      "step": 1539
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014087081935735564,
+      "loss": 0.8676,
+      "step": 1540
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001407969084932873,
+      "loss": 0.7087,
+      "step": 1541
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014072297088293042,
+      "loss": 0.869,
+      "step": 1542
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001406490065747583,
+      "loss": 0.7207,
+      "step": 1543
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014057501561726157,
+      "loss": 0.7011,
+      "step": 1544
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014050099805894837,
+      "loss": 0.9816,
+      "step": 1545
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014042695394834436,
+      "loss": 0.9845,
+      "step": 1546
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014035288333399257,
+      "loss": 0.735,
+      "step": 1547
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001402787862644534,
+      "loss": 0.8069,
+      "step": 1548
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014020466278830452,
+      "loss": 0.7518,
+      "step": 1549
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014013051295414108,
+      "loss": 0.7873,
+      "step": 1550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014005633681057536,
+      "loss": 0.616,
+      "step": 1551
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001399821344062369,
+      "loss": 0.7763,
+      "step": 1552
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013990790578977257,
+      "loss": 0.8089,
+      "step": 1553
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013983365100984633,
+      "loss": 0.7346,
+      "step": 1554
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013975937011513932,
+      "loss": 0.7517,
+      "step": 1555
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013968506315434974,
+      "loss": 0.7635,
+      "step": 1556
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000139610730176193,
+      "loss": 0.827,
+      "step": 1557
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013953637122940147,
+      "loss": 0.7624,
+      "step": 1558
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013946198636272458,
+      "loss": 0.7902,
+      "step": 1559
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013938757562492873,
+      "loss": 0.7368,
+      "step": 1560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013931313906479732,
+      "loss": 0.7421,
+      "step": 1561
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013923867673113066,
+      "loss": 0.8186,
+      "step": 1562
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013916418867274596,
+      "loss": 0.7311,
+      "step": 1563
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001390896749384773,
+      "loss": 0.7166,
+      "step": 1564
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013901513557717553,
+      "loss": 0.7409,
+      "step": 1565
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001389405706377084,
+      "loss": 0.7116,
+      "step": 1566
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013886598016896038,
+      "loss": 0.7832,
+      "step": 1567
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013879136421983266,
+      "loss": 1.0161,
+      "step": 1568
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013871672283924318,
+      "loss": 0.8046,
+      "step": 1569
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013864205607612648,
+      "loss": 0.731,
+      "step": 1570
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013856736397943386,
+      "loss": 0.7836,
+      "step": 1571
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013849264659813312,
+      "loss": 0.8645,
+      "step": 1572
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013841790398120868,
+      "loss": 1.0,
+      "step": 1573
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013834313617766146,
+      "loss": 0.6616,
+      "step": 1574
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000138268343236509,
+      "loss": 0.6907,
+      "step": 1575
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001381935252067852,
+      "loss": 0.866,
+      "step": 1576
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013811868213754042,
+      "loss": 0.8047,
+      "step": 1577
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001380438140778416,
+      "loss": 0.7615,
+      "step": 1578
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013796892107677182,
+      "loss": 0.826,
+      "step": 1579
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013789400318343068,
+      "loss": 0.6562,
+      "step": 1580
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013781906044693404,
+      "loss": 0.8871,
+      "step": 1581
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013774409291641407,
+      "loss": 0.7366,
+      "step": 1582
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013766910064101913,
+      "loss": 0.8775,
+      "step": 1583
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001375940836699139,
+      "loss": 0.6714,
+      "step": 1584
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001375190420522792,
+      "loss": 0.7455,
+      "step": 1585
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013744397583731203,
+      "loss": 0.8168,
+      "step": 1586
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013736888507422546,
+      "loss": 0.8646,
+      "step": 1587
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001372937698122487,
+      "loss": 0.7896,
+      "step": 1588
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000137218630100627,
+      "loss": 0.7025,
+      "step": 1589
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013714346598862166,
+      "loss": 0.86,
+      "step": 1590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013706827752550996,
+      "loss": 0.8737,
+      "step": 1591
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001369930647605852,
+      "loss": 0.8184,
+      "step": 1592
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013691782774315653,
+      "loss": 0.8627,
+      "step": 1593
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013684256652254906,
+      "loss": 0.7043,
+      "step": 1594
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013676728114810367,
+      "loss": 0.8508,
+      "step": 1595
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013669197166917723,
+      "loss": 0.6009,
+      "step": 1596
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001366166381351423,
+      "loss": 0.7551,
+      "step": 1597
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001365412805953872,
+      "loss": 0.8174,
+      "step": 1598
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013646589909931608,
+      "loss": 0.8675,
+      "step": 1599
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013639049369634876,
+      "loss": 0.6946,
+      "step": 1600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001363150644359207,
+      "loss": 0.6405,
+      "step": 1601
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013623961136748295,
+      "loss": 0.7259,
+      "step": 1602
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013616413454050233,
+      "loss": 0.7548,
+      "step": 1603
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013608863400446113,
+      "loss": 0.7419,
+      "step": 1604
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013601310980885714,
+      "loss": 0.7651,
+      "step": 1605
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013593756200320372,
+      "loss": 0.8546,
+      "step": 1606
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001358619906370298,
+      "loss": 0.7113,
+      "step": 1607
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013578639575987958,
+      "loss": 0.7299,
+      "step": 1608
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001357107774213128,
+      "loss": 0.7393,
+      "step": 1609
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001356351356709045,
+      "loss": 0.756,
+      "step": 1610
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013555947055824511,
+      "loss": 0.73,
+      "step": 1611
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001354837821329404,
+      "loss": 0.8241,
+      "step": 1612
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013540807044461135,
+      "loss": 0.701,
+      "step": 1613
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013533233554289433,
+      "loss": 0.7865,
+      "step": 1614
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013525657747744072,
+      "loss": 0.6917,
+      "step": 1615
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013518079629791724,
+      "loss": 0.709,
+      "step": 1616
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001351049920540058,
+      "loss": 0.7352,
+      "step": 1617
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013502916479540326,
+      "loss": 0.8024,
+      "step": 1618
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001349533145718217,
+      "loss": 0.7791,
+      "step": 1619
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013487744143298822,
+      "loss": 0.8598,
+      "step": 1620
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013480154542864497,
+      "loss": 0.711,
+      "step": 1621
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013472562660854902,
+      "loss": 0.7014,
+      "step": 1622
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013464968502247245,
+      "loss": 0.8542,
+      "step": 1623
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001345737207202023,
+      "loss": 0.8301,
+      "step": 1624
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001344977337515404,
+      "loss": 0.8141,
+      "step": 1625
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013442172416630355,
+      "loss": 0.7161,
+      "step": 1626
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013434569201432325,
+      "loss": 0.6533,
+      "step": 1627
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000134269637345446,
+      "loss": 0.732,
+      "step": 1628
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013419356020953282,
+      "loss": 0.7168,
+      "step": 1629
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001341174606564596,
+      "loss": 0.6825,
+      "step": 1630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013404133873611696,
+      "loss": 0.7046,
+      "step": 1631
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013396519449841005,
+      "loss": 0.8678,
+      "step": 1632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013388902799325874,
+      "loss": 0.9132,
+      "step": 1633
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013381283927059752,
+      "loss": 0.7131,
+      "step": 1634
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013373662838037537,
+      "loss": 0.7573,
+      "step": 1635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001336603953725559,
+      "loss": 0.8548,
+      "step": 1636
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013358414029711706,
+      "loss": 0.792,
+      "step": 1637
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013350786320405144,
+      "loss": 0.7557,
+      "step": 1638
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000133431564143366,
+      "loss": 0.7326,
+      "step": 1639
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013335524316508208,
+      "loss": 0.776,
+      "step": 1640
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001332789003192354,
+      "loss": 0.833,
+      "step": 1641
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000133202535655876,
+      "loss": 0.7576,
+      "step": 1642
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001331261492250683,
+      "loss": 0.764,
+      "step": 1643
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013304974107689087,
+      "loss": 0.6934,
+      "step": 1644
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013297331126143667,
+      "loss": 0.6613,
+      "step": 1645
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001328968598288127,
+      "loss": 0.8234,
+      "step": 1646
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013282038682914025,
+      "loss": 0.6504,
+      "step": 1647
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013274389231255466,
+      "loss": 0.7804,
+      "step": 1648
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001326673763292055,
+      "loss": 0.7235,
+      "step": 1649
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013259083892925633,
+      "loss": 0.6231,
+      "step": 1650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013251428016288467,
+      "loss": 0.7425,
+      "step": 1651
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013243770008028224,
+      "loss": 0.6285,
+      "step": 1652
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001323610987316546,
+      "loss": 0.6435,
+      "step": 1653
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013228447616722128,
+      "loss": 0.656,
+      "step": 1654
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013220783243721572,
+      "loss": 0.7094,
+      "step": 1655
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013213116759188523,
+      "loss": 0.5815,
+      "step": 1656
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013205448168149102,
+      "loss": 0.5334,
+      "step": 1657
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000131977774756308,
+      "loss": 0.7491,
+      "step": 1658
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013190104686662497,
+      "loss": 0.7694,
+      "step": 1659
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001318242980627444,
+      "loss": 0.8098,
+      "step": 1660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001317475283949825,
+      "loss": 0.7439,
+      "step": 1661
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013167073791366915,
+      "loss": 0.697,
+      "step": 1662
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013159392666914788,
+      "loss": 0.7043,
+      "step": 1663
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013151709471177588,
+      "loss": 0.604,
+      "step": 1664
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001314402420919238,
+      "loss": 0.9331,
+      "step": 1665
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001313633688599759,
+      "loss": 0.8426,
+      "step": 1666
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013128647506633004,
+      "loss": 0.8403,
+      "step": 1667
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013120956076139746,
+      "loss": 0.6865,
+      "step": 1668
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001311326259956028,
+      "loss": 0.807,
+      "step": 1669
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013105567081938424,
+      "loss": 0.7211,
+      "step": 1670
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001309786952831933,
+      "loss": 0.6761,
+      "step": 1671
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 0.7429,
+      "step": 1672
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013082468333276682,
+      "loss": 0.6979,
+      "step": 1673
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013074764701950095,
+      "loss": 0.551,
+      "step": 1674
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013067059054820183,
+      "loss": 0.5818,
+      "step": 1675
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001305935139693874,
+      "loss": 0.6953,
+      "step": 1676
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013051641733358868,
+      "loss": 0.8221,
+      "step": 1677
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013043930069134998,
+      "loss": 0.6278,
+      "step": 1678
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013036216409322864,
+      "loss": 0.941,
+      "step": 1679
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013028500758979506,
+      "loss": 0.7997,
+      "step": 1680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013020783123163284,
+      "loss": 0.885,
+      "step": 1681
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013013063506933837,
+      "loss": 0.8119,
+      "step": 1682
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013005341915352123,
+      "loss": 0.8996,
+      "step": 1683
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00012997618353480377,
+      "loss": 0.8067,
+      "step": 1684
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00012989892826382145,
+      "loss": 0.6805,
+      "step": 1685
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012982165339122246,
+      "loss": 0.7776,
+      "step": 1686
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001297443589676679,
+      "loss": 0.5873,
+      "step": 1687
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012966704504383168,
+      "loss": 0.8729,
+      "step": 1688
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012958971167040053,
+      "loss": 0.8561,
+      "step": 1689
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012951235889807386,
+      "loss": 0.9101,
+      "step": 1690
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012943498677756382,
+      "loss": 0.7356,
+      "step": 1691
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012935759535959528,
+      "loss": 0.78,
+      "step": 1692
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012928018469490578,
+      "loss": 0.8559,
+      "step": 1693
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012920275483424538,
+      "loss": 0.6114,
+      "step": 1694
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012912530582837682,
+      "loss": 0.6431,
+      "step": 1695
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012904783772807533,
+      "loss": 0.7935,
+      "step": 1696
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012897035058412874,
+      "loss": 0.8003,
+      "step": 1697
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012889284444733722,
+      "loss": 0.7724,
+      "step": 1698
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012881531936851355,
+      "loss": 0.8919,
+      "step": 1699
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012873777539848283,
+      "loss": 0.7477,
+      "step": 1700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012866021258808256,
+      "loss": 0.7821,
+      "step": 1701
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012858263098816265,
+      "loss": 0.7828,
+      "step": 1702
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012850503064958528,
+      "loss": 0.8093,
+      "step": 1703
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012842741162322487,
+      "loss": 0.9681,
+      "step": 1704
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012834977395996818,
+      "loss": 0.768,
+      "step": 1705
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001282721177107141,
+      "loss": 1.0626,
+      "step": 1706
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012819444292637384,
+      "loss": 0.8518,
+      "step": 1707
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012811674965787056,
+      "loss": 0.7713,
+      "step": 1708
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001280390379561398,
+      "loss": 0.6365,
+      "step": 1709
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001279613078721289,
+      "loss": 0.7617,
+      "step": 1710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012788355945679748,
+      "loss": 0.8057,
+      "step": 1711
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012780579276111702,
+      "loss": 0.9079,
+      "step": 1712
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012772800783607111,
+      "loss": 0.7535,
+      "step": 1713
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012765020473265519,
+      "loss": 0.7586,
+      "step": 1714
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001275723835018767,
+      "loss": 1.0038,
+      "step": 1715
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012749454419475487,
+      "loss": 0.7273,
+      "step": 1716
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001274166868623209,
+      "loss": 0.7104,
+      "step": 1717
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001273388115556177,
+      "loss": 0.822,
+      "step": 1718
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001272609183257,
+      "loss": 0.7358,
+      "step": 1719
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001271830072236343,
+      "loss": 0.8069,
+      "step": 1720
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001271050783004988,
+      "loss": 0.6804,
+      "step": 1721
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012702713160738345,
+      "loss": 0.8252,
+      "step": 1722
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001269491671953897,
+      "loss": 0.7545,
+      "step": 1723
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012687118511563075,
+      "loss": 1.0889,
+      "step": 1724
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001267931854192313,
+      "loss": 0.7931,
+      "step": 1725
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012671516815732767,
+      "loss": 0.8303,
+      "step": 1726
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012663713338106764,
+      "loss": 0.882,
+      "step": 1727
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001265590811416105,
+      "loss": 0.7592,
+      "step": 1728
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012648101149012703,
+      "loss": 0.6661,
+      "step": 1729
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001264029244777993,
+      "loss": 0.7085,
+      "step": 1730
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001263248201558209,
+      "loss": 0.7846,
+      "step": 1731
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001262466985753967,
+      "loss": 0.7736,
+      "step": 1732
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012616855978774281,
+      "loss": 0.8949,
+      "step": 1733
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012609040384408684,
+      "loss": 0.8278,
+      "step": 1734
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012601223079566743,
+      "loss": 0.7773,
+      "step": 1735
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001259340406937345,
+      "loss": 0.9449,
+      "step": 1736
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012585583358954921,
+      "loss": 0.8258,
+      "step": 1737
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012577760953438383,
+      "loss": 0.6397,
+      "step": 1738
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012569936857952167,
+      "loss": 0.7894,
+      "step": 1739
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012562111077625722,
+      "loss": 0.7743,
+      "step": 1740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012554283617589602,
+      "loss": 0.7292,
+      "step": 1741
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012546454482975454,
+      "loss": 0.7304,
+      "step": 1742
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012538623678916026,
+      "loss": 0.6272,
+      "step": 1743
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012530791210545162,
+      "loss": 0.5835,
+      "step": 1744
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000125229570829978,
+      "loss": 0.6849,
+      "step": 1745
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001251512130140996,
+      "loss": 0.7808,
+      "step": 1746
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012507283870918747,
+      "loss": 0.6427,
+      "step": 1747
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012499444796662353,
+      "loss": 0.7767,
+      "step": 1748
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001249160408378004,
+      "loss": 0.7942,
+      "step": 1749
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001248376173741215,
+      "loss": 0.7944,
+      "step": 1750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012475917762700092,
+      "loss": 0.7133,
+      "step": 1751
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001246807216478634,
+      "loss": 0.6308,
+      "step": 1752
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001246022494881445,
+      "loss": 0.7438,
+      "step": 1753
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012452376119929007,
+      "loss": 0.7291,
+      "step": 1754
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012444525683275688,
+      "loss": 0.7187,
+      "step": 1755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012436673644001197,
+      "loss": 0.7083,
+      "step": 1756
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012428820007253302,
+      "loss": 0.7486,
+      "step": 1757
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012420964778180814,
+      "loss": 0.6836,
+      "step": 1758
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012413107961933595,
+      "loss": 0.8365,
+      "step": 1759
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012405249563662537,
+      "loss": 0.7618,
+      "step": 1760
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012397389588519574,
+      "loss": 0.8009,
+      "step": 1761
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001238952804165768,
+      "loss": 0.6492,
+      "step": 1762
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012381664928230846,
+      "loss": 0.8065,
+      "step": 1763
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012373800253394102,
+      "loss": 0.7293,
+      "step": 1764
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012365934022303491,
+      "loss": 0.6439,
+      "step": 1765
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012358066240116092,
+      "loss": 0.7535,
+      "step": 1766
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012350196911989984,
+      "loss": 0.8839,
+      "step": 1767
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012342326043084266,
+      "loss": 0.6529,
+      "step": 1768
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012334453638559057,
+      "loss": 0.8266,
+      "step": 1769
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012326579703575462,
+      "loss": 0.7534,
+      "step": 1770
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001231870424329561,
+      "loss": 0.7628,
+      "step": 1771
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012310827262882615,
+      "loss": 0.8637,
+      "step": 1772
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012302948767500596,
+      "loss": 0.7958,
+      "step": 1773
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001229506876231466,
+      "loss": 0.5572,
+      "step": 1774
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012287187252490913,
+      "loss": 0.8441,
+      "step": 1775
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012279304243196436,
+      "loss": 0.805,
+      "step": 1776
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000122714197395993,
+      "loss": 0.6483,
+      "step": 1777
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012263533746868552,
+      "loss": 0.7742,
+      "step": 1778
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012255646270174217,
+      "loss": 0.6571,
+      "step": 1779
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012247757314687297,
+      "loss": 0.7705,
+      "step": 1780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012239866885579753,
+      "loss": 0.8438,
+      "step": 1781
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001223197498802452,
+      "loss": 0.5883,
+      "step": 1782
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000122240816271955,
+      "loss": 0.7757,
+      "step": 1783
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012216186808267546,
+      "loss": 0.9212,
+      "step": 1784
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012208290536416463,
+      "loss": 0.7701,
+      "step": 1785
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012200392816819022,
+      "loss": 0.5605,
+      "step": 1786
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012192493654652933,
+      "loss": 0.7841,
+      "step": 1787
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012184593055096854,
+      "loss": 0.7709,
+      "step": 1788
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001217669102333039,
+      "loss": 0.7195,
+      "step": 1789
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012168787564534078,
+      "loss": 0.584,
+      "step": 1790
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012160882683889396,
+      "loss": 0.6112,
+      "step": 1791
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001215297638657875,
+      "loss": 0.8635,
+      "step": 1792
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012145068677785478,
+      "loss": 0.9767,
+      "step": 1793
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012137159562693838,
+      "loss": 0.6971,
+      "step": 1794
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001212924904648902,
+      "loss": 0.8747,
+      "step": 1795
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001212133713435712,
+      "loss": 0.7379,
+      "step": 1796
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001211342383148516,
+      "loss": 0.7468,
+      "step": 1797
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012105509143061071,
+      "loss": 0.7411,
+      "step": 1798
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012097593074273687,
+      "loss": 0.702,
+      "step": 1799
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012089675630312754,
+      "loss": 0.7952,
+      "step": 1800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001208175681636891,
+      "loss": 0.6818,
+      "step": 1801
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012073836637633705,
+      "loss": 0.8502,
+      "step": 1802
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012065915099299569,
+      "loss": 0.7757,
+      "step": 1803
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012057992206559837,
+      "loss": 0.6375,
+      "step": 1804
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012050067964608724,
+      "loss": 0.8021,
+      "step": 1805
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001204214237864133,
+      "loss": 0.7282,
+      "step": 1806
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012034215453853637,
+      "loss": 0.6956,
+      "step": 1807
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012026287195442503,
+      "loss": 0.9006,
+      "step": 1808
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012018357608605666,
+      "loss": 0.7684,
+      "step": 1809
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012010426698541728,
+      "loss": 0.659,
+      "step": 1810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012002494470450163,
+      "loss": 0.8728,
+      "step": 1811
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00011994560929531309,
+      "loss": 0.6111,
+      "step": 1812
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00011986626080986362,
+      "loss": 0.8388,
+      "step": 1813
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011978689930017379,
+      "loss": 0.7989,
+      "step": 1814
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001197075248182726,
+      "loss": 0.7575,
+      "step": 1815
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011962813741619777,
+      "loss": 0.7172,
+      "step": 1816
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011954873714599526,
+      "loss": 0.6372,
+      "step": 1817
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001194693240597196,
+      "loss": 0.837,
+      "step": 1818
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011938989820943369,
+      "loss": 0.6056,
+      "step": 1819
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011931045964720881,
+      "loss": 0.7064,
+      "step": 1820
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011923100842512451,
+      "loss": 0.7395,
+      "step": 1821
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011915154459526875,
+      "loss": 0.6654,
+      "step": 1822
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011907206820973769,
+      "loss": 0.7689,
+      "step": 1823
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001189925793206357,
+      "loss": 0.6055,
+      "step": 1824
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011891307798007536,
+      "loss": 0.8137,
+      "step": 1825
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011883356424017748,
+      "loss": 0.7405,
+      "step": 1826
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011875403815307091,
+      "loss": 0.779,
+      "step": 1827
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011867449977089265,
+      "loss": 0.9176,
+      "step": 1828
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011859494914578773,
+      "loss": 0.6756,
+      "step": 1829
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011851538632990921,
+      "loss": 0.7424,
+      "step": 1830
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011843581137541819,
+      "loss": 0.7219,
+      "step": 1831
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011835622433448361,
+      "loss": 0.8267,
+      "step": 1832
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011827662525928252,
+      "loss": 0.8077,
+      "step": 1833
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011819701420199969,
+      "loss": 0.6856,
+      "step": 1834
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011811739121482777,
+      "loss": 0.8236,
+      "step": 1835
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011803775634996734,
+      "loss": 0.7006,
+      "step": 1836
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011795810965962669,
+      "loss": 0.8708,
+      "step": 1837
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011787845119602183,
+      "loss": 0.8589,
+      "step": 1838
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011779878101137653,
+      "loss": 0.7354,
+      "step": 1839
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001177190991579223,
+      "loss": 0.6936,
+      "step": 1840
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011763940568789817,
+      "loss": 0.7386,
+      "step": 1841
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011755970065355086,
+      "loss": 0.8982,
+      "step": 1842
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011747998410713475,
+      "loss": 0.8543,
+      "step": 1843
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011740025610091159,
+      "loss": 0.7502,
+      "step": 1844
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011732051668715081,
+      "loss": 0.6577,
+      "step": 1845
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011724076591812918,
+      "loss": 0.6973,
+      "step": 1846
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011716100384613104,
+      "loss": 0.8598,
+      "step": 1847
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011708123052344804,
+      "loss": 0.7492,
+      "step": 1848
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001170014460023793,
+      "loss": 0.7403,
+      "step": 1849
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011692165033523117,
+      "loss": 0.8144,
+      "step": 1850
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001168418435743174,
+      "loss": 0.6341,
+      "step": 1851
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011676202577195901,
+      "loss": 0.8301,
+      "step": 1852
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011668219698048419,
+      "loss": 0.821,
+      "step": 1853
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011660235725222835,
+      "loss": 0.8346,
+      "step": 1854
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011652250663953415,
+      "loss": 0.6766,
+      "step": 1855
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001164426451947513,
+      "loss": 0.6047,
+      "step": 1856
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011636277297023661,
+      "loss": 0.6111,
+      "step": 1857
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011628289001835404,
+      "loss": 0.6734,
+      "step": 1858
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011620299639147453,
+      "loss": 0.7648,
+      "step": 1859
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011612309214197599,
+      "loss": 0.853,
+      "step": 1860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011604317732224333,
+      "loss": 0.7228,
+      "step": 1861
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001159632519846684,
+      "loss": 0.806,
+      "step": 1862
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011588331618164991,
+      "loss": 0.786,
+      "step": 1863
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011580336996559343,
+      "loss": 0.6549,
+      "step": 1864
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011572341338891144,
+      "loss": 0.661,
+      "step": 1865
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001156434465040231,
+      "loss": 0.7965,
+      "step": 1866
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011556346936335438,
+      "loss": 0.6803,
+      "step": 1867
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011548348201933798,
+      "loss": 0.7869,
+      "step": 1868
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011540348452441327,
+      "loss": 0.8671,
+      "step": 1869
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011532347693102632,
+      "loss": 0.7158,
+      "step": 1870
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011524345929162975,
+      "loss": 0.5902,
+      "step": 1871
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011516343165868279,
+      "loss": 0.7299,
+      "step": 1872
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011508339408465127,
+      "loss": 0.7172,
+      "step": 1873
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011500334662200749,
+      "loss": 0.7513,
+      "step": 1874
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011492328932323022,
+      "loss": 0.641,
+      "step": 1875
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011484322224080472,
+      "loss": 0.8385,
+      "step": 1876
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011476314542722266,
+      "loss": 0.8149,
+      "step": 1877
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011468305893498203,
+      "loss": 0.7774,
+      "step": 1878
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011460296281658727,
+      "loss": 0.7731,
+      "step": 1879
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011452285712454904,
+      "loss": 0.8845,
+      "step": 1880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011444274191138431,
+      "loss": 0.635,
+      "step": 1881
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011436261722961628,
+      "loss": 0.9447,
+      "step": 1882
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011428248313177439,
+      "loss": 0.616,
+      "step": 1883
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011420233967039422,
+      "loss": 0.6779,
+      "step": 1884
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011412218689801748,
+      "loss": 0.7719,
+      "step": 1885
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011404202486719205,
+      "loss": 0.7335,
+      "step": 1886
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001139618536304718,
+      "loss": 0.75,
+      "step": 1887
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011388167324041669,
+      "loss": 0.7051,
+      "step": 1888
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001138014837495926,
+      "loss": 0.8995,
+      "step": 1889
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011372128521057155,
+      "loss": 0.7561,
+      "step": 1890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011364107767593126,
+      "loss": 0.8329,
+      "step": 1891
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011356086119825553,
+      "loss": 0.8739,
+      "step": 1892
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011348063583013398,
+      "loss": 0.716,
+      "step": 1893
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011340040162416197,
+      "loss": 0.8757,
+      "step": 1894
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011332015863294076,
+      "loss": 0.8925,
+      "step": 1895
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011323990690907733,
+      "loss": 0.7387,
+      "step": 1896
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011315964650518438,
+      "loss": 0.7511,
+      "step": 1897
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011307937747388034,
+      "loss": 0.7827,
+      "step": 1898
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011299909986778918,
+      "loss": 0.6888,
+      "step": 1899
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011291881373954065,
+      "loss": 0.9315,
+      "step": 1900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011283851914177,
+      "loss": 0.7057,
+      "step": 1901
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011275821612711803,
+      "loss": 0.899,
+      "step": 1902
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011267790474823105,
+      "loss": 0.7977,
+      "step": 1903
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011259758505776092,
+      "loss": 0.8183,
+      "step": 1904
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011251725710836489,
+      "loss": 0.7827,
+      "step": 1905
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011243692095270564,
+      "loss": 0.6608,
+      "step": 1906
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001123565766434512,
+      "loss": 0.6641,
+      "step": 1907
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011227622423327502,
+      "loss": 0.648,
+      "step": 1908
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011219586377485578,
+      "loss": 0.8332,
+      "step": 1909
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011211549532087749,
+      "loss": 0.9994,
+      "step": 1910
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011203511892402938,
+      "loss": 0.6389,
+      "step": 1911
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001119547346370059,
+      "loss": 0.7792,
+      "step": 1912
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011187434251250666,
+      "loss": 0.6804,
+      "step": 1913
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001117939426032364,
+      "loss": 0.7547,
+      "step": 1914
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011171353496190498,
+      "loss": 0.7884,
+      "step": 1915
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011163311964122734,
+      "loss": 0.747,
+      "step": 1916
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011155269669392341,
+      "loss": 0.8856,
+      "step": 1917
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001114722661727182,
+      "loss": 0.8093,
+      "step": 1918
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011139182813034157,
+      "loss": 0.7216,
+      "step": 1919
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011131138261952845,
+      "loss": 0.7484,
+      "step": 1920
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001112309296930185,
+      "loss": 0.777,
+      "step": 1921
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011115046940355642,
+      "loss": 0.6185,
+      "step": 1922
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011107000180389162,
+      "loss": 0.7116,
+      "step": 1923
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011098952694677829,
+      "loss": 0.8237,
+      "step": 1924
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011090904488497549,
+      "loss": 0.6079,
+      "step": 1925
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011082855567124692,
+      "loss": 0.9299,
+      "step": 1926
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011074805935836096,
+      "loss": 0.8304,
+      "step": 1927
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011066755599909064,
+      "loss": 0.6821,
+      "step": 1928
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011058704564621371,
+      "loss": 0.6618,
+      "step": 1929
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001105065283525124,
+      "loss": 0.6658,
+      "step": 1930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011042600417077351,
+      "loss": 0.712,
+      "step": 1931
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011034547315378838,
+      "loss": 0.6362,
+      "step": 1932
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011026493535435282,
+      "loss": 0.6682,
+      "step": 1933
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011018439082526707,
+      "loss": 0.6751,
+      "step": 1934
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011010383961933581,
+      "loss": 0.7213,
+      "step": 1935
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011002328178936811,
+      "loss": 0.8638,
+      "step": 1936
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010994271738817734,
+      "loss": 0.7483,
+      "step": 1937
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010986214646858115,
+      "loss": 0.9325,
+      "step": 1938
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010978156908340158,
+      "loss": 0.7541,
+      "step": 1939
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010970098528546481,
+      "loss": 0.8097,
+      "step": 1940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010962039512760128,
+      "loss": 0.7014,
+      "step": 1941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010953979866264548,
+      "loss": 0.7915,
+      "step": 1942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010945919594343624,
+      "loss": 0.8108,
+      "step": 1943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010937858702281631,
+      "loss": 0.7255,
+      "step": 1944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010929797195363259,
+      "loss": 0.7637,
+      "step": 1945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010921735078873598,
+      "loss": 0.6124,
+      "step": 1946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010913672358098143,
+      "loss": 0.6483,
+      "step": 1947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010905609038322779,
+      "loss": 0.6482,
+      "step": 1948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010897545124833783,
+      "loss": 0.7312,
+      "step": 1949
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001088948062291783,
+      "loss": 0.7522,
+      "step": 1950
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001088141553786197,
+      "loss": 0.8263,
+      "step": 1951
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001087334987495364,
+      "loss": 0.6402,
+      "step": 1952
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001086528363948066,
+      "loss": 0.7303,
+      "step": 1953
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010857216836731222,
+      "loss": 0.8417,
+      "step": 1954
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010849149471993882,
+      "loss": 0.7552,
+      "step": 1955
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010841081550557578,
+      "loss": 0.7288,
+      "step": 1956
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010833013077711607,
+      "loss": 0.7234,
+      "step": 1957
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010824944058745623,
+      "loss": 0.8175,
+      "step": 1958
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010816874498949643,
+      "loss": 0.7855,
+      "step": 1959
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010808804403614043,
+      "loss": 0.8037,
+      "step": 1960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010800733778029542,
+      "loss": 0.7427,
+      "step": 1961
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010792662627487207,
+      "loss": 0.7333,
+      "step": 1962
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001078459095727845,
+      "loss": 0.6383,
+      "step": 1963
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010776518772695034,
+      "loss": 0.7577,
+      "step": 1964
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010768446079029044,
+      "loss": 0.8111,
+      "step": 1965
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010760372881572905,
+      "loss": 0.7905,
+      "step": 1966
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010752299185619377,
+      "loss": 0.7351,
+      "step": 1967
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001074422499646154,
+      "loss": 0.6686,
+      "step": 1968
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010736150319392802,
+      "loss": 0.7804,
+      "step": 1969
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001072807515970688,
+      "loss": 0.9291,
+      "step": 1970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010719999522697828,
+      "loss": 0.6902,
+      "step": 1971
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010711923413659995,
+      "loss": 0.6258,
+      "step": 1972
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010703846837888045,
+      "loss": 0.7622,
+      "step": 1973
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010695769800676949,
+      "loss": 0.7381,
+      "step": 1974
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010687692307321984,
+      "loss": 0.7515,
+      "step": 1975
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010679614363118717,
+      "loss": 0.8314,
+      "step": 1976
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010671535973363018,
+      "loss": 0.9125,
+      "step": 1977
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010663457143351044,
+      "loss": 0.6826,
+      "step": 1978
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010655377878379247,
+      "loss": 0.8385,
+      "step": 1979
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010647298183744359,
+      "loss": 0.703,
+      "step": 1980
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010639218064743392,
+      "loss": 0.7842,
+      "step": 1981
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010631137526673647,
+      "loss": 0.7024,
+      "step": 1982
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010623056574832687,
+      "loss": 0.6624,
+      "step": 1983
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001061497521451835,
+      "loss": 0.7331,
+      "step": 1984
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010606893451028743,
+      "loss": 0.7104,
+      "step": 1985
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010598811289662243,
+      "loss": 0.6324,
+      "step": 1986
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010590728735717479,
+      "loss": 0.7158,
+      "step": 1987
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010582645794493337,
+      "loss": 0.8078,
+      "step": 1988
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010574562471288969,
+      "loss": 0.759,
+      "step": 1989
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010566478771403763,
+      "loss": 0.6603,
+      "step": 1990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010558394700137362,
+      "loss": 0.6431,
+      "step": 1991
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010550310262789649,
+      "loss": 0.6772,
+      "step": 1992
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010542225464660749,
+      "loss": 0.8119,
+      "step": 1993
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010534140311051026,
+      "loss": 0.8,
+      "step": 1994
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010526054807261067,
+      "loss": 0.8161,
+      "step": 1995
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010517968958591705,
+      "loss": 0.7785,
+      "step": 1996
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010509882770343983,
+      "loss": 0.7712,
+      "step": 1997
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010501796247819175,
+      "loss": 0.7116,
+      "step": 1998
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010493709396318771,
+      "loss": 0.7456,
+      "step": 1999
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010485622221144484,
+      "loss": 0.68,
+      "step": 2000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010477534727598229,
+      "loss": 0.6633,
+      "step": 2001
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010469446920982129,
+      "loss": 0.6131,
+      "step": 2002
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010461358806598527,
+      "loss": 0.7044,
+      "step": 2003
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010453270389749957,
+      "loss": 0.6119,
+      "step": 2004
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010445181675739144,
+      "loss": 0.7501,
+      "step": 2005
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010437092669869024,
+      "loss": 0.7921,
+      "step": 2006
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010429003377442715,
+      "loss": 0.9771,
+      "step": 2007
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010420913803763521,
+      "loss": 0.8328,
+      "step": 2008
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010412823954134938,
+      "loss": 0.8843,
+      "step": 2009
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010404733833860639,
+      "loss": 0.765,
+      "step": 2010
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010396643448244472,
+      "loss": 0.7074,
+      "step": 2011
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010388552802590462,
+      "loss": 0.8087,
+      "step": 2012
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010380461902202801,
+      "loss": 0.7042,
+      "step": 2013
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010372370752385853,
+      "loss": 0.5895,
+      "step": 2014
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010364279358444144,
+      "loss": 0.7156,
+      "step": 2015
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010356187725682359,
+      "loss": 0.7299,
+      "step": 2016
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010348095859405338,
+      "loss": 0.7607,
+      "step": 2017
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010340003764918078,
+      "loss": 0.7795,
+      "step": 2018
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010331911447525724,
+      "loss": 0.6723,
+      "step": 2019
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010323818912533561,
+      "loss": 0.9001,
+      "step": 2020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010315726165247028,
+      "loss": 0.5957,
+      "step": 2021
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010307633210971697,
+      "loss": 0.7042,
+      "step": 2022
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010299540055013272,
+      "loss": 0.631,
+      "step": 2023
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010291446702677599,
+      "loss": 0.5871,
+      "step": 2024
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010283353159270643,
+      "loss": 0.8197,
+      "step": 2025
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000102752594300985,
+      "loss": 0.8031,
+      "step": 2026
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010267165520467386,
+      "loss": 0.7761,
+      "step": 2027
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010259071435683636,
+      "loss": 0.7135,
+      "step": 2028
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010250977181053698,
+      "loss": 0.7715,
+      "step": 2029
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010242882761884131,
+      "loss": 0.8672,
+      "step": 2030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010234788183481607,
+      "loss": 0.9106,
+      "step": 2031
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000102266934511529,
+      "loss": 0.8094,
+      "step": 2032
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010218598570204883,
+      "loss": 0.8561,
+      "step": 2033
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010210503545944521,
+      "loss": 0.7685,
+      "step": 2034
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010202408383678888,
+      "loss": 0.7284,
+      "step": 2035
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010194313088715135,
+      "loss": 0.7851,
+      "step": 2036
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010186217666360507,
+      "loss": 0.5829,
+      "step": 2037
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010178122121922324,
+      "loss": 0.5843,
+      "step": 2038
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010170026460708,
+      "loss": 0.6555,
+      "step": 2039
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010161930688025017,
+      "loss": 0.6592,
+      "step": 2040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010153834809180925,
+      "loss": 0.7806,
+      "step": 2041
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010145738829483353,
+      "loss": 0.6187,
+      "step": 2042
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010137642754239995,
+      "loss": 0.7172,
+      "step": 2043
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010129546588758605,
+      "loss": 0.6033,
+      "step": 2044
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001012145033834699,
+      "loss": 0.8653,
+      "step": 2045
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010113354008313025,
+      "loss": 0.6914,
+      "step": 2046
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001010525760396463,
+      "loss": 0.8284,
+      "step": 2047
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010097161130609773,
+      "loss": 0.7278,
+      "step": 2048
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010089064593556474,
+      "loss": 0.8752,
+      "step": 2049
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010080967998112787,
+      "loss": 0.6988,
+      "step": 2050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010072871349586808,
+      "loss": 0.671,
+      "step": 2051
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010064774653286661,
+      "loss": 0.7689,
+      "step": 2052
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010056677914520518,
+      "loss": 0.9217,
+      "step": 2053
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010048581138596563,
+      "loss": 0.8251,
+      "step": 2054
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010040484330823006,
+      "loss": 0.5996,
+      "step": 2055
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010032387496508089,
+      "loss": 0.8445,
+      "step": 2056
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001002429064096006,
+      "loss": 0.5986,
+      "step": 2057
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010016193769487181,
+      "loss": 0.8224,
+      "step": 2058
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010008096887397733,
+      "loss": 0.7757,
+      "step": 2059
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001,
+      "loss": 0.7436,
+      "step": 2060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.991903112602269e-05,
+      "loss": 0.6999,
+      "step": 2061
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.98380623051282e-05,
+      "loss": 0.5878,
+      "step": 2062
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.975709359039944e-05,
+      "loss": 0.7301,
+      "step": 2063
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.967612503491914e-05,
+      "loss": 0.7549,
+      "step": 2064
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.959515669176996e-05,
+      "loss": 0.7683,
+      "step": 2065
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95141886140344e-05,
+      "loss": 0.7435,
+      "step": 2066
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943322085479484e-05,
+      "loss": 0.7437,
+      "step": 2067
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.935225346713341e-05,
+      "loss": 0.7077,
+      "step": 2068
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.927128650413196e-05,
+      "loss": 0.7834,
+      "step": 2069
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.919032001887215e-05,
+      "loss": 0.6674,
+      "step": 2070
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.910935406443528e-05,
+      "loss": 0.6724,
+      "step": 2071
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.902838869390229e-05,
+      "loss": 0.6665,
+      "step": 2072
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.894742396035371e-05,
+      "loss": 0.8993,
+      "step": 2073
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.886645991686976e-05,
+      "loss": 0.8569,
+      "step": 2074
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.878549661653012e-05,
+      "loss": 0.8687,
+      "step": 2075
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.870453411241399e-05,
+      "loss": 0.7592,
+      "step": 2076
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.862357245760006e-05,
+      "loss": 0.6714,
+      "step": 2077
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.854261170516648e-05,
+      "loss": 0.754,
+      "step": 2078
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.846165190819078e-05,
+      "loss": 0.7679,
+      "step": 2079
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.838069311974986e-05,
+      "loss": 0.6928,
+      "step": 2080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.829973539292001e-05,
+      "loss": 0.6997,
+      "step": 2081
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.821877878077678e-05,
+      "loss": 0.8398,
+      "step": 2082
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.813782333639496e-05,
+      "loss": 0.6465,
+      "step": 2083
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.805686911284868e-05,
+      "loss": 0.4927,
+      "step": 2084
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.797591616321114e-05,
+      "loss": 0.8897,
+      "step": 2085
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.789496454055482e-05,
+      "loss": 0.6658,
+      "step": 2086
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.781401429795119e-05,
+      "loss": 0.6872,
+      "step": 2087
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.7733065488471e-05,
+      "loss": 0.783,
+      "step": 2088
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.765211816518394e-05,
+      "loss": 0.7792,
+      "step": 2089
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.757117238115871e-05,
+      "loss": 0.8266,
+      "step": 2090
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.749022818946306e-05,
+      "loss": 0.9758,
+      "step": 2091
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.740928564316368e-05,
+      "loss": 0.6612,
+      "step": 2092
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.732834479532619e-05,
+      "loss": 0.8427,
+      "step": 2093
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.724740569901503e-05,
+      "loss": 0.6651,
+      "step": 2094
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.716646840729361e-05,
+      "loss": 0.6584,
+      "step": 2095
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.708553297322406e-05,
+      "loss": 0.7825,
+      "step": 2096
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.70045994498673e-05,
+      "loss": 0.6608,
+      "step": 2097
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.692366789028307e-05,
+      "loss": 0.6443,
+      "step": 2098
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.684273834752976e-05,
+      "loss": 0.7893,
+      "step": 2099
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.676181087466444e-05,
+      "loss": 0.6082,
+      "step": 2100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.668088552474281e-05,
+      "loss": 0.7754,
+      "step": 2101
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.659996235081926e-05,
+      "loss": 0.6626,
+      "step": 2102
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.651904140594667e-05,
+      "loss": 0.7934,
+      "step": 2103
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.643812274317644e-05,
+      "loss": 0.7163,
+      "step": 2104
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.635720641555858e-05,
+      "loss": 0.7113,
+      "step": 2105
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.627629247614152e-05,
+      "loss": 0.6598,
+      "step": 2106
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.619538097797205e-05,
+      "loss": 0.6361,
+      "step": 2107
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.611447197409543e-05,
+      "loss": 0.6359,
+      "step": 2108
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.603356551755533e-05,
+      "loss": 0.654,
+      "step": 2109
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.595266166139366e-05,
+      "loss": 0.8619,
+      "step": 2110
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.587176045865064e-05,
+      "loss": 0.8212,
+      "step": 2111
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.579086196236482e-05,
+      "loss": 0.8254,
+      "step": 2112
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.570996622557291e-05,
+      "loss": 0.6412,
+      "step": 2113
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.562907330130981e-05,
+      "loss": 0.636,
+      "step": 2114
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.554818324260859e-05,
+      "loss": 0.6904,
+      "step": 2115
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.54672961025005e-05,
+      "loss": 0.7011,
+      "step": 2116
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.538641193401477e-05,
+      "loss": 0.6656,
+      "step": 2117
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.530553079017872e-05,
+      "loss": 0.6085,
+      "step": 2118
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.522465272401777e-05,
+      "loss": 0.6016,
+      "step": 2119
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.514377778855521e-05,
+      "loss": 0.7361,
+      "step": 2120
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.506290603681229e-05,
+      "loss": 0.7609,
+      "step": 2121
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.498203752180826e-05,
+      "loss": 0.6107,
+      "step": 2122
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.490117229656018e-05,
+      "loss": 0.8273,
+      "step": 2123
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.482031041408296e-05,
+      "loss": 0.9795,
+      "step": 2124
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.473945192738933e-05,
+      "loss": 0.6879,
+      "step": 2125
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.465859688948977e-05,
+      "loss": 0.7663,
+      "step": 2126
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.45777453533925e-05,
+      "loss": 0.844,
+      "step": 2127
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.449689737210352e-05,
+      "loss": 0.8779,
+      "step": 2128
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.441605299862639e-05,
+      "loss": 0.7234,
+      "step": 2129
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.433521228596237e-05,
+      "loss": 0.713,
+      "step": 2130
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.42543752871103e-05,
+      "loss": 0.7439,
+      "step": 2131
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.417354205506663e-05,
+      "loss": 0.8104,
+      "step": 2132
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.409271264282524e-05,
+      "loss": 0.7732,
+      "step": 2133
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.401188710337758e-05,
+      "loss": 0.7885,
+      "step": 2134
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.393106548971256e-05,
+      "loss": 0.8051,
+      "step": 2135
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.385024785481654e-05,
+      "loss": 0.6853,
+      "step": 2136
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.376943425167314e-05,
+      "loss": 0.7504,
+      "step": 2137
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.368862473326354e-05,
+      "loss": 0.7567,
+      "step": 2138
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.360781935256607e-05,
+      "loss": 0.7205,
+      "step": 2139
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.352701816255643e-05,
+      "loss": 0.8081,
+      "step": 2140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.344622121620753e-05,
+      "loss": 0.8056,
+      "step": 2141
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.336542856648956e-05,
+      "loss": 0.847,
+      "step": 2142
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.328464026636984e-05,
+      "loss": 0.8909,
+      "step": 2143
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.320385636881283e-05,
+      "loss": 0.7299,
+      "step": 2144
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.312307692678017e-05,
+      "loss": 0.7915,
+      "step": 2145
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.304230199323049e-05,
+      "loss": 0.8418,
+      "step": 2146
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.296153162111956e-05,
+      "loss": 0.7666,
+      "step": 2147
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.288076586340006e-05,
+      "loss": 0.6936,
+      "step": 2148
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.280000477302173e-05,
+      "loss": 0.7963,
+      "step": 2149
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.27192484029312e-05,
+      "loss": 0.7948,
+      "step": 2150
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.2638496806072e-05,
+      "loss": 0.7703,
+      "step": 2151
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255775003538462e-05,
+      "loss": 0.8168,
+      "step": 2152
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.247700814380624e-05,
+      "loss": 0.6834,
+      "step": 2153
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.239627118427096e-05,
+      "loss": 0.7005,
+      "step": 2154
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.231553920970958e-05,
+      "loss": 0.6284,
+      "step": 2155
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.223481227304968e-05,
+      "loss": 0.7647,
+      "step": 2156
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.215409042721552e-05,
+      "loss": 0.718,
+      "step": 2157
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.207337372512796e-05,
+      "loss": 0.7383,
+      "step": 2158
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.199266221970462e-05,
+      "loss": 0.903,
+      "step": 2159
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19119559638596e-05,
+      "loss": 0.7483,
+      "step": 2160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.183125501050358e-05,
+      "loss": 0.7185,
+      "step": 2161
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.175055941254379e-05,
+      "loss": 0.5695,
+      "step": 2162
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.166986922288397e-05,
+      "loss": 0.7706,
+      "step": 2163
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.158918449442423e-05,
+      "loss": 0.8572,
+      "step": 2164
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.150850528006119e-05,
+      "loss": 0.5849,
+      "step": 2165
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.142783163268781e-05,
+      "loss": 0.783,
+      "step": 2166
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.13471636051934e-05,
+      "loss": 0.7248,
+      "step": 2167
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.126650125046361e-05,
+      "loss": 0.5595,
+      "step": 2168
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.118584462138031e-05,
+      "loss": 0.8597,
+      "step": 2169
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.110519377082172e-05,
+      "loss": 0.7005,
+      "step": 2170
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.102454875166218e-05,
+      "loss": 0.7486,
+      "step": 2171
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.094390961677223e-05,
+      "loss": 0.7958,
+      "step": 2172
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.086327641901859e-05,
+      "loss": 0.6535,
+      "step": 2173
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.078264921126404e-05,
+      "loss": 0.7787,
+      "step": 2174
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.070202804636745e-05,
+      "loss": 0.6842,
+      "step": 2175
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.062141297718371e-05,
+      "loss": 0.8536,
+      "step": 2176
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.054080405656379e-05,
+      "loss": 0.9343,
+      "step": 2177
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.046020133735454e-05,
+      "loss": 0.7542,
+      "step": 2178
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.037960487239876e-05,
+      "loss": 0.7554,
+      "step": 2179
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.02990147145352e-05,
+      "loss": 0.7821,
+      "step": 2180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.021843091659843e-05,
+      "loss": 0.6783,
+      "step": 2181
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.013785353141886e-05,
+      "loss": 0.6486,
+      "step": 2182
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.005728261182268e-05,
+      "loss": 0.7589,
+      "step": 2183
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.997671821063191e-05,
+      "loss": 0.794,
+      "step": 2184
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.98961603806642e-05,
+      "loss": 0.7155,
+      "step": 2185
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.981560917473293e-05,
+      "loss": 1.0281,
+      "step": 2186
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.97350646456472e-05,
+      "loss": 0.8645,
+      "step": 2187
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.965452684621164e-05,
+      "loss": 0.888,
+      "step": 2188
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.957399582922653e-05,
+      "loss": 0.7091,
+      "step": 2189
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.949347164748762e-05,
+      "loss": 0.7683,
+      "step": 2190
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.94129543537863e-05,
+      "loss": 0.7165,
+      "step": 2191
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.933244400090937e-05,
+      "loss": 0.6682,
+      "step": 2192
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.925194064163906e-05,
+      "loss": 0.6587,
+      "step": 2193
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.91714443287531e-05,
+      "loss": 0.6632,
+      "step": 2194
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.909095511502452e-05,
+      "loss": 0.7441,
+      "step": 2195
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.901047305322172e-05,
+      "loss": 0.9829,
+      "step": 2196
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.892999819610841e-05,
+      "loss": 0.8008,
+      "step": 2197
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.88495305964436e-05,
+      "loss": 0.6156,
+      "step": 2198
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.876907030698151e-05,
+      "loss": 0.5607,
+      "step": 2199
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.868861738047158e-05,
+      "loss": 0.7847,
+      "step": 2200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.860817186965844e-05,
+      "loss": 0.7385,
+      "step": 2201
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.852773382728183e-05,
+      "loss": 0.8593,
+      "step": 2202
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.844730330607661e-05,
+      "loss": 0.732,
+      "step": 2203
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.836688035877267e-05,
+      "loss": 0.7723,
+      "step": 2204
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.828646503809504e-05,
+      "loss": 0.8353,
+      "step": 2205
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.820605739676363e-05,
+      "loss": 0.6978,
+      "step": 2206
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.812565748749336e-05,
+      "loss": 0.6289,
+      "step": 2207
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.804526536299413e-05,
+      "loss": 0.7798,
+      "step": 2208
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.796488107597065e-05,
+      "loss": 0.6638,
+      "step": 2209
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.788450467912255e-05,
+      "loss": 0.6794,
+      "step": 2210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.780413622514424e-05,
+      "loss": 0.74,
+      "step": 2211
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.772377576672502e-05,
+      "loss": 0.6647,
+      "step": 2212
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.764342335654883e-05,
+      "loss": 0.8224,
+      "step": 2213
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.756307904729439e-05,
+      "loss": 0.6551,
+      "step": 2214
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.748274289163514e-05,
+      "loss": 0.7631,
+      "step": 2215
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.740241494223911e-05,
+      "loss": 0.7263,
+      "step": 2216
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.732209525176898e-05,
+      "loss": 0.7453,
+      "step": 2217
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.724178387288201e-05,
+      "loss": 0.8067,
+      "step": 2218
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.716148085823004e-05,
+      "loss": 0.7468,
+      "step": 2219
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.70811862604594e-05,
+      "loss": 0.6288,
+      "step": 2220
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.700090013221085e-05,
+      "loss": 0.7844,
+      "step": 2221
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.692062252611973e-05,
+      "loss": 0.8089,
+      "step": 2222
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.684035349481567e-05,
+      "loss": 0.7792,
+      "step": 2223
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.676009309092272e-05,
+      "loss": 0.8764,
+      "step": 2224
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.667984136705928e-05,
+      "loss": 0.7154,
+      "step": 2225
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.659959837583807e-05,
+      "loss": 0.7216,
+      "step": 2226
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.651936416986607e-05,
+      "loss": 0.6194,
+      "step": 2227
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.643913880174448e-05,
+      "loss": 0.6728,
+      "step": 2228
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.635892232406877e-05,
+      "loss": 0.6986,
+      "step": 2229
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.627871478942851e-05,
+      "loss": 0.7309,
+      "step": 2230
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.619851625040743e-05,
+      "loss": 0.7372,
+      "step": 2231
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.611832675958336e-05,
+      "loss": 0.728,
+      "step": 2232
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.603814636952824e-05,
+      "loss": 0.75,
+      "step": 2233
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.595797513280799e-05,
+      "loss": 0.716,
+      "step": 2234
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.587781310198255e-05,
+      "loss": 0.7283,
+      "step": 2235
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.579766032960582e-05,
+      "loss": 0.7685,
+      "step": 2236
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.571751686822566e-05,
+      "loss": 0.6708,
+      "step": 2237
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.563738277038377e-05,
+      "loss": 0.7838,
+      "step": 2238
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.555725808861573e-05,
+      "loss": 0.8678,
+      "step": 2239
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.5477142875451e-05,
+      "loss": 0.8324,
+      "step": 2240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.539703718341277e-05,
+      "loss": 0.8329,
+      "step": 2241
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.531694106501797e-05,
+      "loss": 0.6748,
+      "step": 2242
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.523685457277736e-05,
+      "loss": 0.7081,
+      "step": 2243
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.515677775919527e-05,
+      "loss": 0.6422,
+      "step": 2244
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.507671067676979e-05,
+      "loss": 0.5794,
+      "step": 2245
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.499665337799254e-05,
+      "loss": 0.7236,
+      "step": 2246
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.491660591534872e-05,
+      "loss": 0.7952,
+      "step": 2247
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.48365683413172e-05,
+      "loss": 0.716,
+      "step": 2248
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.475654070837026e-05,
+      "loss": 0.7204,
+      "step": 2249
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.467652306897369e-05,
+      "loss": 0.7459,
+      "step": 2250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.459651547558671e-05,
+      "loss": 0.7864,
+      "step": 2251
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.451651798066203e-05,
+      "loss": 0.6773,
+      "step": 2252
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.443653063664563e-05,
+      "loss": 0.7429,
+      "step": 2253
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.435655349597689e-05,
+      "loss": 0.6312,
+      "step": 2254
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.427658661108857e-05,
+      "loss": 0.7869,
+      "step": 2255
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.419663003440657e-05,
+      "loss": 0.7114,
+      "step": 2256
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.411668381835012e-05,
+      "loss": 0.7666,
+      "step": 2257
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.403674801533161e-05,
+      "loss": 0.8254,
+      "step": 2258
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.395682267775668e-05,
+      "loss": 0.7892,
+      "step": 2259
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.387690785802402e-05,
+      "loss": 0.7308,
+      "step": 2260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.379700360852548e-05,
+      "loss": 0.6002,
+      "step": 2261
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.371710998164594e-05,
+      "loss": 0.9391,
+      "step": 2262
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36372270297634e-05,
+      "loss": 0.6838,
+      "step": 2263
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.355735480524874e-05,
+      "loss": 0.8056,
+      "step": 2264
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.347749336046586e-05,
+      "loss": 0.7218,
+      "step": 2265
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.339764274777165e-05,
+      "loss": 0.6961,
+      "step": 2266
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.331780301951583e-05,
+      "loss": 0.6629,
+      "step": 2267
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.323797422804099e-05,
+      "loss": 0.6305,
+      "step": 2268
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.31581564256826e-05,
+      "loss": 0.6236,
+      "step": 2269
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.307834966476884e-05,
+      "loss": 0.8001,
+      "step": 2270
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.299855399762073e-05,
+      "loss": 0.6721,
+      "step": 2271
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.291876947655196e-05,
+      "loss": 0.7006,
+      "step": 2272
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.283899615386898e-05,
+      "loss": 0.8546,
+      "step": 2273
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.275923408187086e-05,
+      "loss": 0.7405,
+      "step": 2274
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.267948331284923e-05,
+      "loss": 0.8515,
+      "step": 2275
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.259974389908842e-05,
+      "loss": 0.8379,
+      "step": 2276
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.252001589286529e-05,
+      "loss": 0.8405,
+      "step": 2277
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.244029934644916e-05,
+      "loss": 0.8263,
+      "step": 2278
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.236059431210186e-05,
+      "loss": 0.7541,
+      "step": 2279
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.228090084207774e-05,
+      "loss": 0.796,
+      "step": 2280
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.220121898862348e-05,
+      "loss": 0.7715,
+      "step": 2281
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.212154880397818e-05,
+      "loss": 0.714,
+      "step": 2282
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.204189034037332e-05,
+      "loss": 0.5867,
+      "step": 2283
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.196224365003267e-05,
+      "loss": 0.8508,
+      "step": 2284
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.188260878517224e-05,
+      "loss": 0.8511,
+      "step": 2285
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.180298579800035e-05,
+      "loss": 0.7545,
+      "step": 2286
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.17233747407175e-05,
+      "loss": 0.7715,
+      "step": 2287
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16437756655164e-05,
+      "loss": 0.6578,
+      "step": 2288
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.156418862458184e-05,
+      "loss": 0.6742,
+      "step": 2289
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.14846136700908e-05,
+      "loss": 0.7058,
+      "step": 2290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.140505085421229e-05,
+      "loss": 0.8087,
+      "step": 2291
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.132550022910737e-05,
+      "loss": 0.5936,
+      "step": 2292
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.12459618469291e-05,
+      "loss": 0.684,
+      "step": 2293
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.116643575982253e-05,
+      "loss": 0.8786,
+      "step": 2294
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.108692201992465e-05,
+      "loss": 1.0615,
+      "step": 2295
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.100742067936431e-05,
+      "loss": 0.7618,
+      "step": 2296
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.092793179026232e-05,
+      "loss": 0.8744,
+      "step": 2297
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.084845540473126e-05,
+      "loss": 0.877,
+      "step": 2298
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.076899157487551e-05,
+      "loss": 0.6661,
+      "step": 2299
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.068954035279121e-05,
+      "loss": 0.781,
+      "step": 2300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.061010179056633e-05,
+      "loss": 0.6117,
+      "step": 2301
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053067594028044e-05,
+      "loss": 0.6215,
+      "step": 2302
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.045126285400476e-05,
+      "loss": 0.7328,
+      "step": 2303
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.037186258380226e-05,
+      "loss": 0.8742,
+      "step": 2304
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02924751817274e-05,
+      "loss": 0.8059,
+      "step": 2305
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.021310069982624e-05,
+      "loss": 0.7726,
+      "step": 2306
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.013373919013639e-05,
+      "loss": 0.8052,
+      "step": 2307
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.005439070468692e-05,
+      "loss": 0.7897,
+      "step": 2308
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.99750552954984e-05,
+      "loss": 0.7599,
+      "step": 2309
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.989573301458273e-05,
+      "loss": 0.9599,
+      "step": 2310
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.981642391394337e-05,
+      "loss": 0.695,
+      "step": 2311
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.973712804557501e-05,
+      "loss": 0.7415,
+      "step": 2312
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.965784546146366e-05,
+      "loss": 0.8911,
+      "step": 2313
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.957857621358674e-05,
+      "loss": 0.7343,
+      "step": 2314
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.949932035391278e-05,
+      "loss": 0.7423,
+      "step": 2315
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.942007793440164e-05,
+      "loss": 0.8168,
+      "step": 2316
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.934084900700432e-05,
+      "loss": 0.6335,
+      "step": 2317
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.926163362366299e-05,
+      "loss": 0.8912,
+      "step": 2318
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.918243183631093e-05,
+      "loss": 0.6991,
+      "step": 2319
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91032436968725e-05,
+      "loss": 0.7096,
+      "step": 2320
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.902406925726315e-05,
+      "loss": 0.7495,
+      "step": 2321
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.894490856938932e-05,
+      "loss": 0.8511,
+      "step": 2322
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.886576168514841e-05,
+      "loss": 0.7776,
+      "step": 2323
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.878662865642881e-05,
+      "loss": 0.6397,
+      "step": 2324
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.870750953510984e-05,
+      "loss": 0.7288,
+      "step": 2325
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.862840437306165e-05,
+      "loss": 0.7668,
+      "step": 2326
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.854931322214525e-05,
+      "loss": 0.7353,
+      "step": 2327
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.847023613421251e-05,
+      "loss": 0.7889,
+      "step": 2328
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.839117316110608e-05,
+      "loss": 0.7718,
+      "step": 2329
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.831212435465924e-05,
+      "loss": 0.6946,
+      "step": 2330
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82330897666961e-05,
+      "loss": 0.7522,
+      "step": 2331
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.815406944903147e-05,
+      "loss": 0.6726,
+      "step": 2332
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.807506345347071e-05,
+      "loss": 0.7787,
+      "step": 2333
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.799607183180982e-05,
+      "loss": 0.6878,
+      "step": 2334
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.79170946358354e-05,
+      "loss": 0.6479,
+      "step": 2335
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.78381319173246e-05,
+      "loss": 0.731,
+      "step": 2336
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.775918372804504e-05,
+      "loss": 0.7796,
+      "step": 2337
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.768025011975481e-05,
+      "loss": 0.7546,
+      "step": 2338
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.760133114420252e-05,
+      "loss": 0.6705,
+      "step": 2339
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.75224268531271e-05,
+      "loss": 0.8177,
+      "step": 2340
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.744353729825786e-05,
+      "loss": 0.7305,
+      "step": 2341
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.736466253131452e-05,
+      "loss": 0.7266,
+      "step": 2342
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.728580260400705e-05,
+      "loss": 0.736,
+      "step": 2343
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72069575680357e-05,
+      "loss": 0.6585,
+      "step": 2344
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.71281274750909e-05,
+      "loss": 0.7782,
+      "step": 2345
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.704931237685342e-05,
+      "loss": 0.7217,
+      "step": 2346
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.697051232499409e-05,
+      "loss": 0.8326,
+      "step": 2347
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.689172737117389e-05,
+      "loss": 0.6534,
+      "step": 2348
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.681295756704395e-05,
+      "loss": 0.704,
+      "step": 2349
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.673420296424541e-05,
+      "loss": 0.8751,
+      "step": 2350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66554636144095e-05,
+      "loss": 0.7862,
+      "step": 2351
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.657673956915735e-05,
+      "loss": 0.5698,
+      "step": 2352
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64980308801002e-05,
+      "loss": 0.6797,
+      "step": 2353
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.641933759883913e-05,
+      "loss": 0.7571,
+      "step": 2354
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.634065977696511e-05,
+      "loss": 0.7094,
+      "step": 2355
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.626199746605903e-05,
+      "loss": 0.8548,
+      "step": 2356
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.618335071769158e-05,
+      "loss": 0.7919,
+      "step": 2357
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.610471958342326e-05,
+      "loss": 0.7596,
+      "step": 2358
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.602610411480427e-05,
+      "loss": 0.7547,
+      "step": 2359
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.594750436337467e-05,
+      "loss": 0.794,
+      "step": 2360
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.586892038066408e-05,
+      "loss": 0.7217,
+      "step": 2361
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.579035221819187e-05,
+      "loss": 0.7139,
+      "step": 2362
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.571179992746702e-05,
+      "loss": 0.7993,
+      "step": 2363
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.563326355998803e-05,
+      "loss": 0.8273,
+      "step": 2364
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.555474316724313e-05,
+      "loss": 0.8638,
+      "step": 2365
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.547623880070993e-05,
+      "loss": 0.5596,
+      "step": 2366
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.539775051185551e-05,
+      "loss": 0.7809,
+      "step": 2367
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.531927835213656e-05,
+      "loss": 0.6814,
+      "step": 2368
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.52408223729991e-05,
+      "loss": 0.6972,
+      "step": 2369
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.516238262587851e-05,
+      "loss": 0.7944,
+      "step": 2370
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.508395916219959e-05,
+      "loss": 0.7282,
+      "step": 2371
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.500555203337647e-05,
+      "loss": 0.7314,
+      "step": 2372
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.492716129081253e-05,
+      "loss": 0.7154,
+      "step": 2373
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.48487869859004e-05,
+      "loss": 0.8379,
+      "step": 2374
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4770429170022e-05,
+      "loss": 0.7518,
+      "step": 2375
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.469208789454838e-05,
+      "loss": 0.7781,
+      "step": 2376
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.461376321083975e-05,
+      "loss": 0.7672,
+      "step": 2377
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.453545517024547e-05,
+      "loss": 0.6594,
+      "step": 2378
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.445716382410398e-05,
+      "loss": 0.8446,
+      "step": 2379
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.437888922374276e-05,
+      "loss": 0.9074,
+      "step": 2380
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.430063142047831e-05,
+      "loss": 0.6465,
+      "step": 2381
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.422239046561619e-05,
+      "loss": 0.6926,
+      "step": 2382
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.41441664104508e-05,
+      "loss": 0.6466,
+      "step": 2383
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.40659593062655e-05,
+      "loss": 0.7887,
+      "step": 2384
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.398776920433258e-05,
+      "loss": 0.777,
+      "step": 2385
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.390959615591316e-05,
+      "loss": 0.6326,
+      "step": 2386
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.383144021225717e-05,
+      "loss": 0.8081,
+      "step": 2387
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.37533014246033e-05,
+      "loss": 0.9853,
+      "step": 2388
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.36751798441791e-05,
+      "loss": 0.7513,
+      "step": 2389
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.35970755222007e-05,
+      "loss": 0.761,
+      "step": 2390
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.351898850987299e-05,
+      "loss": 0.6888,
+      "step": 2391
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.344091885838948e-05,
+      "loss": 0.8088,
+      "step": 2392
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.336286661893237e-05,
+      "loss": 0.7203,
+      "step": 2393
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.328483184267235e-05,
+      "loss": 0.6644,
+      "step": 2394
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.320681458076871e-05,
+      "loss": 0.9015,
+      "step": 2395
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.312881488436927e-05,
+      "loss": 0.7852,
+      "step": 2396
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.305083280461032e-05,
+      "loss": 0.5865,
+      "step": 2397
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.297286839261658e-05,
+      "loss": 0.7166,
+      "step": 2398
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.289492169950118e-05,
+      "loss": 0.7402,
+      "step": 2399
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.281699277636572e-05,
+      "loss": 0.7876,
+      "step": 2400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.273908167430004e-05,
+      "loss": 0.7534,
+      "step": 2401
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.266118844438235e-05,
+      "loss": 0.9019,
+      "step": 2402
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.258331313767914e-05,
+      "loss": 0.7311,
+      "step": 2403
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.250545580524515e-05,
+      "loss": 0.9113,
+      "step": 2404
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.242761649812335e-05,
+      "loss": 0.8763,
+      "step": 2405
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.234979526734482e-05,
+      "loss": 0.7224,
+      "step": 2406
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.227199216392892e-05,
+      "loss": 0.7691,
+      "step": 2407
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2194207238883e-05,
+      "loss": 0.8828,
+      "step": 2408
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.211644054320254e-05,
+      "loss": 0.599,
+      "step": 2409
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.20386921278711e-05,
+      "loss": 0.7165,
+      "step": 2410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.196096204386022e-05,
+      "loss": 0.7151,
+      "step": 2411
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.188325034212943e-05,
+      "loss": 0.6387,
+      "step": 2412
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.180555707362618e-05,
+      "loss": 0.7354,
+      "step": 2413
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.172788228928591e-05,
+      "loss": 0.7929,
+      "step": 2414
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.165022604003186e-05,
+      "loss": 0.7197,
+      "step": 2415
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.157258837677514e-05,
+      "loss": 0.8234,
+      "step": 2416
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.149496935041476e-05,
+      "loss": 0.7235,
+      "step": 2417
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.141736901183736e-05,
+      "loss": 0.6741,
+      "step": 2418
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.133978741191745e-05,
+      "loss": 0.7936,
+      "step": 2419
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.126222460151719e-05,
+      "loss": 0.7044,
+      "step": 2420
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.118468063148646e-05,
+      "loss": 0.6982,
+      "step": 2421
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.110715555266281e-05,
+      "loss": 0.6448,
+      "step": 2422
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.102964941587129e-05,
+      "loss": 0.6316,
+      "step": 2423
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.095216227192467e-05,
+      "loss": 0.7873,
+      "step": 2424
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.08746941716232e-05,
+      "loss": 0.787,
+      "step": 2425
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.079724516575466e-05,
+      "loss": 0.7776,
+      "step": 2426
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.071981530509424e-05,
+      "loss": 0.7189,
+      "step": 2427
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.064240464040473e-05,
+      "loss": 0.8225,
+      "step": 2428
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.056501322243622e-05,
+      "loss": 0.6398,
+      "step": 2429
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.048764110192618e-05,
+      "loss": 0.6382,
+      "step": 2430
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.04102883295995e-05,
+      "loss": 0.7683,
+      "step": 2431
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.033295495616834e-05,
+      "loss": 0.6836,
+      "step": 2432
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.025564103233213e-05,
+      "loss": 0.7196,
+      "step": 2433
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.017834660877756e-05,
+      "loss": 0.7546,
+      "step": 2434
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.010107173617857e-05,
+      "loss": 0.7087,
+      "step": 2435
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.002381646519625e-05,
+      "loss": 0.6198,
+      "step": 2436
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.994658084647881e-05,
+      "loss": 0.7183,
+      "step": 2437
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.986936493066165e-05,
+      "loss": 0.8834,
+      "step": 2438
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.97921687683672e-05,
+      "loss": 0.7635,
+      "step": 2439
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.971499241020495e-05,
+      "loss": 0.7624,
+      "step": 2440
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.963783590677138e-05,
+      "loss": 0.6364,
+      "step": 2441
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.956069930865004e-05,
+      "loss": 0.7203,
+      "step": 2442
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.948358266641135e-05,
+      "loss": 0.7349,
+      "step": 2443
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.940648603061263e-05,
+      "loss": 0.6215,
+      "step": 2444
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.932940945179818e-05,
+      "loss": 0.6638,
+      "step": 2445
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.925235298049906e-05,
+      "loss": 0.5951,
+      "step": 2446
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.917531666723321e-05,
+      "loss": 0.7193,
+      "step": 2447
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.6822,
+      "step": 2448
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.902130471680675e-05,
+      "loss": 0.7825,
+      "step": 2449
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.894432918061579e-05,
+      "loss": 0.6985,
+      "step": 2450
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.886737400439721e-05,
+      "loss": 0.5583,
+      "step": 2451
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.879043923860257e-05,
+      "loss": 0.7553,
+      "step": 2452
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.871352493366997e-05,
+      "loss": 0.77,
+      "step": 2453
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.863663114002411e-05,
+      "loss": 0.672,
+      "step": 2454
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.855975790807623e-05,
+      "loss": 0.7333,
+      "step": 2455
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.848290528822416e-05,
+      "loss": 0.7179,
+      "step": 2456
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.840607333085214e-05,
+      "loss": 0.7309,
+      "step": 2457
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.832926208633086e-05,
+      "loss": 0.7289,
+      "step": 2458
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.825247160501753e-05,
+      "loss": 0.865,
+      "step": 2459
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.817570193725564e-05,
+      "loss": 0.7329,
+      "step": 2460
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.809895313337508e-05,
+      "loss": 0.725,
+      "step": 2461
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.802222524369202e-05,
+      "loss": 0.6374,
+      "step": 2462
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.794551831850902e-05,
+      "loss": 0.7045,
+      "step": 2463
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.786883240811479e-05,
+      "loss": 0.5344,
+      "step": 2464
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.77921675627843e-05,
+      "loss": 0.9198,
+      "step": 2465
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.771552383277875e-05,
+      "loss": 0.6753,
+      "step": 2466
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.763890126834543e-05,
+      "loss": 0.737,
+      "step": 2467
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.756229991971779e-05,
+      "loss": 0.7448,
+      "step": 2468
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.748571983711534e-05,
+      "loss": 0.6173,
+      "step": 2469
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.740916107074372e-05,
+      "loss": 0.7897,
+      "step": 2470
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733262367079453e-05,
+      "loss": 0.6522,
+      "step": 2471
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.725610768744534e-05,
+      "loss": 0.5612,
+      "step": 2472
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.717961317085979e-05,
+      "loss": 0.5554,
+      "step": 2473
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.710314017118734e-05,
+      "loss": 0.8011,
+      "step": 2474
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.702668873856338e-05,
+      "loss": 0.6443,
+      "step": 2475
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.695025892310914e-05,
+      "loss": 0.765,
+      "step": 2476
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.687385077493174e-05,
+      "loss": 0.6826,
+      "step": 2477
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.679746434412404e-05,
+      "loss": 0.7919,
+      "step": 2478
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.672109968076465e-05,
+      "loss": 0.516,
+      "step": 2479
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.664475683491796e-05,
+      "loss": 0.8268,
+      "step": 2480
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.656843585663403e-05,
+      "loss": 0.5922,
+      "step": 2481
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.649213679594859e-05,
+      "loss": 0.8128,
+      "step": 2482
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.641585970288297e-05,
+      "loss": 0.8269,
+      "step": 2483
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.633960462744416e-05,
+      "loss": 0.7101,
+      "step": 2484
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.626337161962461e-05,
+      "loss": 0.7049,
+      "step": 2485
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.618716072940248e-05,
+      "loss": 0.6792,
+      "step": 2486
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.611097200674125e-05,
+      "loss": 0.7382,
+      "step": 2487
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.603480550158995e-05,
+      "loss": 0.893,
+      "step": 2488
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.595866126388305e-05,
+      "loss": 0.6765,
+      "step": 2489
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.588253934354039e-05,
+      "loss": 0.7335,
+      "step": 2490
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.580643979046717e-05,
+      "loss": 0.8172,
+      "step": 2491
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5730362654554e-05,
+      "loss": 0.7475,
+      "step": 2492
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.565430798567673e-05,
+      "loss": 0.8856,
+      "step": 2493
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.557827583369647e-05,
+      "loss": 0.7962,
+      "step": 2494
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.550226624845961e-05,
+      "loss": 0.6308,
+      "step": 2495
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.542627927979771e-05,
+      "loss": 0.632,
+      "step": 2496
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.535031497752754e-05,
+      "loss": 0.7097,
+      "step": 2497
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.527437339145098e-05,
+      "loss": 0.7664,
+      "step": 2498
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.519845457135504e-05,
+      "loss": 0.7715,
+      "step": 2499
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.512255856701177e-05,
+      "loss": 0.6746,
+      "step": 2500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.504668542817831e-05,
+      "loss": 0.7355,
+      "step": 2501
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.497083520459674e-05,
+      "loss": 0.7095,
+      "step": 2502
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.489500794599421e-05,
+      "loss": 0.6612,
+      "step": 2503
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.481920370208274e-05,
+      "loss": 0.671,
+      "step": 2504
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.474342252255927e-05,
+      "loss": 0.7809,
+      "step": 2505
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.466766445710568e-05,
+      "loss": 0.6668,
+      "step": 2506
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.459192955538863e-05,
+      "loss": 0.6836,
+      "step": 2507
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.451621786705962e-05,
+      "loss": 0.6211,
+      "step": 2508
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.444052944175488e-05,
+      "loss": 0.6149,
+      "step": 2509
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.43648643290955e-05,
+      "loss": 0.7946,
+      "step": 2510
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.428922257868722e-05,
+      "loss": 0.938,
+      "step": 2511
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.42136042401204e-05,
+      "loss": 0.5667,
+      "step": 2512
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.413800936297019e-05,
+      "loss": 0.7271,
+      "step": 2513
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.406243799679626e-05,
+      "loss": 0.8119,
+      "step": 2514
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.398689019114289e-05,
+      "loss": 0.7309,
+      "step": 2515
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.39113659955389e-05,
+      "loss": 0.816,
+      "step": 2516
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.383586545949768e-05,
+      "loss": 0.8262,
+      "step": 2517
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.376038863251705e-05,
+      "loss": 0.7314,
+      "step": 2518
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.368493556407933e-05,
+      "loss": 0.7167,
+      "step": 2519
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.360950630365126e-05,
+      "loss": 0.8018,
+      "step": 2520
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.353410090068392e-05,
+      "loss": 0.8891,
+      "step": 2521
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.345871940461283e-05,
+      "loss": 0.843,
+      "step": 2522
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.338336186485774e-05,
+      "loss": 0.7473,
+      "step": 2523
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.330802833082279e-05,
+      "loss": 0.7173,
+      "step": 2524
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.323271885189635e-05,
+      "loss": 0.8303,
+      "step": 2525
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.315743347745097e-05,
+      "loss": 0.6188,
+      "step": 2526
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.308217225684348e-05,
+      "loss": 0.7432,
+      "step": 2527
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.300693523941482e-05,
+      "loss": 0.6928,
+      "step": 2528
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.293172247449005e-05,
+      "loss": 0.7576,
+      "step": 2529
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.285653401137837e-05,
+      "loss": 0.7265,
+      "step": 2530
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.278136989937305e-05,
+      "loss": 0.7186,
+      "step": 2531
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.270623018775135e-05,
+      "loss": 0.6506,
+      "step": 2532
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.263111492577456e-05,
+      "loss": 0.7082,
+      "step": 2533
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.255602416268799e-05,
+      "loss": 0.7348,
+      "step": 2534
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.248095794772079e-05,
+      "loss": 0.7679,
+      "step": 2535
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.24059163300861e-05,
+      "loss": 0.5934,
+      "step": 2536
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.233089935898088e-05,
+      "loss": 0.7287,
+      "step": 2537
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.225590708358596e-05,
+      "loss": 0.6793,
+      "step": 2538
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.218093955306598e-05,
+      "loss": 0.6902,
+      "step": 2539
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.210599681656933e-05,
+      "loss": 0.746,
+      "step": 2540
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.203107892322819e-05,
+      "loss": 0.7777,
+      "step": 2541
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.195618592215843e-05,
+      "loss": 0.691,
+      "step": 2542
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.188131786245959e-05,
+      "loss": 0.6089,
+      "step": 2543
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.180647479321485e-05,
+      "loss": 0.6872,
+      "step": 2544
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.173165676349103e-05,
+      "loss": 0.7273,
+      "step": 2545
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.165686382233855e-05,
+      "loss": 0.6593,
+      "step": 2546
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.158209601879134e-05,
+      "loss": 0.7259,
+      "step": 2547
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.15073534018669e-05,
+      "loss": 0.76,
+      "step": 2548
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.143263602056615e-05,
+      "loss": 0.7319,
+      "step": 2549
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.135794392387353e-05,
+      "loss": 0.6582,
+      "step": 2550
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.128327716075683e-05,
+      "loss": 0.7743,
+      "step": 2551
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.120863578016735e-05,
+      "loss": 0.6363,
+      "step": 2552
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.113401983103966e-05,
+      "loss": 0.8139,
+      "step": 2553
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.105942936229161e-05,
+      "loss": 0.6334,
+      "step": 2554
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0984864422824496e-05,
+      "loss": 0.6589,
+      "step": 2555
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.091032506152274e-05,
+      "loss": 0.6473,
+      "step": 2556
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.083581132725407e-05,
+      "loss": 0.7228,
+      "step": 2557
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.076132326886934e-05,
+      "loss": 0.8345,
+      "step": 2558
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0686860935202694e-05,
+      "loss": 0.7443,
+      "step": 2559
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.061242437507131e-05,
+      "loss": 0.772,
+      "step": 2560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.053801363727545e-05,
+      "loss": 0.7736,
+      "step": 2561
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.046362877059857e-05,
+      "loss": 0.8095,
+      "step": 2562
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0389269823807036e-05,
+      "loss": 0.7914,
+      "step": 2563
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.031493684565029e-05,
+      "loss": 0.7269,
+      "step": 2564
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.024062988486072e-05,
+      "loss": 0.7921,
+      "step": 2565
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.016634899015369e-05,
+      "loss": 0.7535,
+      "step": 2566
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.009209421022744e-05,
+      "loss": 0.8988,
+      "step": 2567
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.00178655937631e-05,
+      "loss": 0.7246,
+      "step": 2568
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.994366318942468e-05,
+      "loss": 0.732,
+      "step": 2569
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.986948704585895e-05,
+      "loss": 0.7149,
+      "step": 2570
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.97953372116955e-05,
+      "loss": 0.8137,
+      "step": 2571
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.972121373554664e-05,
+      "loss": 0.7464,
+      "step": 2572
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9647116666007455e-05,
+      "loss": 0.805,
+      "step": 2573
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.957304605165567e-05,
+      "loss": 0.6894,
+      "step": 2574
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.949900194105167e-05,
+      "loss": 0.789,
+      "step": 2575
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.942498438273849e-05,
+      "loss": 0.5962,
+      "step": 2576
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.935099342524175e-05,
+      "loss": 0.7421,
+      "step": 2577
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9277029117069604e-05,
+      "loss": 0.827,
+      "step": 2578
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.920309150671275e-05,
+      "loss": 0.6649,
+      "step": 2579
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9129180642644414e-05,
+      "loss": 0.5851,
+      "step": 2580
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9055296573320226e-05,
+      "loss": 0.7707,
+      "step": 2581
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8981439347178304e-05,
+      "loss": 0.7082,
+      "step": 2582
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.890760901263915e-05,
+      "loss": 0.6365,
+      "step": 2583
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.883380561810563e-05,
+      "loss": 0.6476,
+      "step": 2584
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8760029211962954e-05,
+      "loss": 0.6256,
+      "step": 2585
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8686279842578615e-05,
+      "loss": 0.7587,
+      "step": 2586
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.861255755830244e-05,
+      "loss": 0.6907,
+      "step": 2587
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8538862407466425e-05,
+      "loss": 0.7745,
+      "step": 2588
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8465194438384816e-05,
+      "loss": 0.8004,
+      "step": 2589
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.839155369935407e-05,
+      "loss": 0.7186,
+      "step": 2590
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8317940238652714e-05,
+      "loss": 0.6458,
+      "step": 2591
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.82443541045415e-05,
+      "loss": 0.8524,
+      "step": 2592
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8170795345263106e-05,
+      "loss": 0.7729,
+      "step": 2593
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.809726400904242e-05,
+      "loss": 0.8253,
+      "step": 2594
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.802376014408632e-05,
+      "loss": 0.8208,
+      "step": 2595
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.795028379858355e-05,
+      "loss": 0.6729,
+      "step": 2596
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.7876835020705e-05,
+      "loss": 0.7047,
+      "step": 2597
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.780341385860333e-05,
+      "loss": 0.6933,
+      "step": 2598
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.773002036041315e-05,
+      "loss": 0.6589,
+      "step": 2599
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.765665457425102e-05,
+      "loss": 0.7539,
+      "step": 2600
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.758331654821516e-05,
+      "loss": 0.8596,
+      "step": 2601
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.751000633038572e-05,
+      "loss": 0.5354,
+      "step": 2602
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.743672396882464e-05,
+      "loss": 0.7826,
+      "step": 2603
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.736346951157544e-05,
+      "loss": 0.6813,
+      "step": 2604
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.729024300666349e-05,
+      "loss": 0.7611,
+      "step": 2605
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7217044502095806e-05,
+      "loss": 0.6351,
+      "step": 2606
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.714387404586101e-05,
+      "loss": 0.7804,
+      "step": 2607
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.707073168592942e-05,
+      "loss": 0.6416,
+      "step": 2608
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.699761747025278e-05,
+      "loss": 0.7768,
+      "step": 2609
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6924531446764504e-05,
+      "loss": 0.6593,
+      "step": 2610
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6851473663379564e-05,
+      "loss": 0.7075,
+      "step": 2611
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.677844416799424e-05,
+      "loss": 0.8571,
+      "step": 2612
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.670544300848643e-05,
+      "loss": 0.6766,
+      "step": 2613
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6632470232715426e-05,
+      "loss": 0.7602,
+      "step": 2614
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6559525888521815e-05,
+      "loss": 0.9211,
+      "step": 2615
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.648661002372768e-05,
+      "loss": 0.7262,
+      "step": 2616
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.641372268613631e-05,
+      "loss": 0.6234,
+      "step": 2617
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6340863923532396e-05,
+      "loss": 0.7539,
+      "step": 2618
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.626803378368185e-05,
+      "loss": 0.7176,
+      "step": 2619
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6195232314331766e-05,
+      "loss": 0.7247,
+      "step": 2620
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.612245956321053e-05,
+      "loss": 0.712,
+      "step": 2621
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.60497155780277e-05,
+      "loss": 0.6421,
+      "step": 2622
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.597700040647385e-05,
+      "loss": 0.7797,
+      "step": 2623
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.590431409622081e-05,
+      "loss": 0.8409,
+      "step": 2624
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5831656694921465e-05,
+      "loss": 0.7742,
+      "step": 2625
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.575902825020962e-05,
+      "loss": 0.7277,
+      "step": 2626
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.568642880970026e-05,
+      "loss": 0.7137,
+      "step": 2627
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.56138584209893e-05,
+      "loss": 0.7527,
+      "step": 2628
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.554131713165353e-05,
+      "loss": 0.6276,
+      "step": 2629
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5468804989250786e-05,
+      "loss": 0.7864,
+      "step": 2630
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.539632204131968e-05,
+      "loss": 0.7586,
+      "step": 2631
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.532386833537977e-05,
+      "loss": 0.6876,
+      "step": 2632
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.525144391893144e-05,
+      "loss": 0.7628,
+      "step": 2633
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.517904883945577e-05,
+      "loss": 0.7147,
+      "step": 2634
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.510668314441474e-05,
+      "loss": 0.8237,
+      "step": 2635
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.503434688125104e-05,
+      "loss": 0.7778,
+      "step": 2636
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.496204009738795e-05,
+      "loss": 0.7622,
+      "step": 2637
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.488976284022953e-05,
+      "loss": 0.6646,
+      "step": 2638
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4817515157160515e-05,
+      "loss": 0.6707,
+      "step": 2639
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.474529709554612e-05,
+      "loss": 0.6026,
+      "step": 2640
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.467310870273224e-05,
+      "loss": 0.6737,
+      "step": 2641
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4600950026045326e-05,
+      "loss": 0.7882,
+      "step": 2642
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.452882111279227e-05,
+      "loss": 0.7879,
+      "step": 2643
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.445672201026054e-05,
+      "loss": 0.6573,
+      "step": 2644
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.438465276571796e-05,
+      "loss": 0.812,
+      "step": 2645
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.431261342641286e-05,
+      "loss": 0.6547,
+      "step": 2646
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4240604039573984e-05,
+      "loss": 0.6894,
+      "step": 2647
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.416862465241033e-05,
+      "loss": 0.6796,
+      "step": 2648
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.409667531211131e-05,
+      "loss": 0.6921,
+      "step": 2649
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.402475606584669e-05,
+      "loss": 0.7982,
+      "step": 2650
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.395286696076632e-05,
+      "loss": 0.7225,
+      "step": 2651
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.388100804400049e-05,
+      "loss": 0.7148,
+      "step": 2652
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.380917936265961e-05,
+      "loss": 0.8328,
+      "step": 2653
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.373738096383423e-05,
+      "loss": 0.7578,
+      "step": 2654
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.366561289459512e-05,
+      "loss": 0.7143,
+      "step": 2655
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3593875201993174e-05,
+      "loss": 0.8177,
+      "step": 2656
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3522167933059243e-05,
+      "loss": 0.6649,
+      "step": 2657
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3450491134804414e-05,
+      "loss": 0.893,
+      "step": 2658
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.337884485421962e-05,
+      "loss": 0.6458,
+      "step": 2659
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3307229138275936e-05,
+      "loss": 0.6816,
+      "step": 2660
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.323564403392436e-05,
+      "loss": 0.7116,
+      "step": 2661
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.31640895880957e-05,
+      "loss": 0.7358,
+      "step": 2662
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3092565847700834e-05,
+      "loss": 0.7647,
+      "step": 2663
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.302107285963045e-05,
+      "loss": 0.7122,
+      "step": 2664
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2949610670755e-05,
+      "loss": 0.7058,
+      "step": 2665
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.287817932792485e-05,
+      "loss": 0.6237,
+      "step": 2666
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.280677887797012e-05,
+      "loss": 0.741,
+      "step": 2667
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.273540936770058e-05,
+      "loss": 0.879,
+      "step": 2668
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.266407084390586e-05,
+      "loss": 0.7692,
+      "step": 2669
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.259276335335521e-05,
+      "loss": 0.6391,
+      "step": 2670
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.252148694279748e-05,
+      "loss": 0.7541,
+      "step": 2671
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.245024165896126e-05,
+      "loss": 0.54,
+      "step": 2672
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.237902754855462e-05,
+      "loss": 0.7436,
+      "step": 2673
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.230784465826524e-05,
+      "loss": 0.6506,
+      "step": 2674
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.223669303476041e-05,
+      "loss": 0.7735,
+      "step": 2675
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2165572724686754e-05,
+      "loss": 0.8176,
+      "step": 2676
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2094483774670497e-05,
+      "loss": 0.747,
+      "step": 2677
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.202342623131731e-05,
+      "loss": 0.7075,
+      "step": 2678
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1952400141212164e-05,
+      "loss": 0.7082,
+      "step": 2679
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1881405550919493e-05,
+      "loss": 0.7959,
+      "step": 2680
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.181044250698312e-05,
+      "loss": 0.6419,
+      "step": 2681
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1739511055926047e-05,
+      "loss": 0.7703,
+      "step": 2682
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1668611244250684e-05,
+      "loss": 0.7224,
+      "step": 2683
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1597743118438726e-05,
+      "loss": 0.6709,
+      "step": 2684
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.152690672495091e-05,
+      "loss": 0.5835,
+      "step": 2685
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1456102110227375e-05,
+      "loss": 0.6547,
+      "step": 2686
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1385329320687294e-05,
+      "loss": 0.6885,
+      "step": 2687
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1314588402729044e-05,
+      "loss": 0.7383,
+      "step": 2688
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.124387940273011e-05,
+      "loss": 0.6867,
+      "step": 2689
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.117320236704697e-05,
+      "loss": 0.7023,
+      "step": 2690
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1102557342015224e-05,
+      "loss": 0.7567,
+      "step": 2691
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.103194437394952e-05,
+      "loss": 0.6996,
+      "step": 2692
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.096136350914335e-05,
+      "loss": 0.8404,
+      "step": 2693
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.089081479386928e-05,
+      "loss": 0.7904,
+      "step": 2694
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.08202982743788e-05,
+      "loss": 0.7218,
+      "step": 2695
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.074981399690218e-05,
+      "loss": 0.6913,
+      "step": 2696
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.067936200764869e-05,
+      "loss": 0.7725,
+      "step": 2697
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0608942352806364e-05,
+      "loss": 0.7411,
+      "step": 2698
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0538555078542005e-05,
+      "loss": 0.7427,
+      "step": 2699
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0468200231001286e-05,
+      "loss": 0.7478,
+      "step": 2700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.039787785630847e-05,
+      "loss": 0.8113,
+      "step": 2701
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0327588000566695e-05,
+      "loss": 0.6636,
+      "step": 2702
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.025733070985772e-05,
+      "loss": 0.8221,
+      "step": 2703
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.018710603024187e-05,
+      "loss": 0.6825,
+      "step": 2704
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.01169140077582e-05,
+      "loss": 0.619,
+      "step": 2705
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.004675468842436e-05,
+      "loss": 0.7122,
+      "step": 2706
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.997662811823642e-05,
+      "loss": 0.745,
+      "step": 2707
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9906534343169144e-05,
+      "loss": 0.8198,
+      "step": 2708
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9836473409175754e-05,
+      "loss": 0.8252,
+      "step": 2709
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.976644536218783e-05,
+      "loss": 0.664,
+      "step": 2710
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9696450248115524e-05,
+      "loss": 0.7003,
+      "step": 2711
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.962648811284738e-05,
+      "loss": 0.7166,
+      "step": 2712
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9556559002250235e-05,
+      "loss": 0.8048,
+      "step": 2713
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.948666296216938e-05,
+      "loss": 0.643,
+      "step": 2714
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9416800038428324e-05,
+      "loss": 0.6391,
+      "step": 2715
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.934697027682894e-05,
+      "loss": 0.821,
+      "step": 2716
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.927717372315139e-05,
+      "loss": 0.8238,
+      "step": 2717
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.920741042315392e-05,
+      "loss": 0.5706,
+      "step": 2718
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.913768042257313e-05,
+      "loss": 0.6465,
+      "step": 2719
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9067983767123736e-05,
+      "loss": 0.66,
+      "step": 2720
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.899832050249853e-05,
+      "loss": 0.7169,
+      "step": 2721
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.89286906743685e-05,
+      "loss": 0.6601,
+      "step": 2722
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.885909432838269e-05,
+      "loss": 0.7123,
+      "step": 2723
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8789531510168163e-05,
+      "loss": 0.6422,
+      "step": 2724
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8720002265330015e-05,
+      "loss": 0.6611,
+      "step": 2725
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.865050663945139e-05,
+      "loss": 0.7804,
+      "step": 2726
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.858104467809324e-05,
+      "loss": 0.8916,
+      "step": 2727
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.851161642679466e-05,
+      "loss": 0.7832,
+      "step": 2728
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8442221931072466e-05,
+      "loss": 0.6926,
+      "step": 2729
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.837286123642141e-05,
+      "loss": 0.8913,
+      "step": 2730
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.830353438831413e-05,
+      "loss": 0.6641,
+      "step": 2731
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8234241432200965e-05,
+      "loss": 0.6908,
+      "step": 2732
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8164982413510177e-05,
+      "loss": 0.7409,
+      "step": 2733
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.809575737764759e-05,
+      "loss": 0.7816,
+      "step": 2734
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8026566369996926e-05,
+      "loss": 0.8089,
+      "step": 2735
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.795740943591955e-05,
+      "loss": 0.6895,
+      "step": 2736
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7888286620754406e-05,
+      "loss": 0.7576,
+      "step": 2737
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7819197969818175e-05,
+      "loss": 0.7121,
+      "step": 2738
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7750143528405126e-05,
+      "loss": 0.7826,
+      "step": 2739
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.768112334178699e-05,
+      "loss": 0.8603,
+      "step": 2740
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.761213745521317e-05,
+      "loss": 0.7321,
+      "step": 2741
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.754318591391057e-05,
+      "loss": 0.8115,
+      "step": 2742
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7474268763083464e-05,
+      "loss": 0.7361,
+      "step": 2743
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.74053860479137e-05,
+      "loss": 0.7155,
+      "step": 2744
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.733653781356055e-05,
+      "loss": 0.6928,
+      "step": 2745
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.726772410516055e-05,
+      "loss": 0.749,
+      "step": 2746
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7198944967827774e-05,
+      "loss": 0.8049,
+      "step": 2747
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7130200446653475e-05,
+      "loss": 0.6278,
+      "step": 2748
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7061490586706305e-05,
+      "loss": 0.821,
+      "step": 2749
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.699281543303222e-05,
+      "loss": 0.7185,
+      "step": 2750
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6924175030654304e-05,
+      "loss": 0.707,
+      "step": 2751
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6855569424572955e-05,
+      "loss": 0.6567,
+      "step": 2752
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6786998659765766e-05,
+      "loss": 0.8059,
+      "step": 2753
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.67184627811874e-05,
+      "loss": 0.6674,
+      "step": 2754
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6649961833769715e-05,
+      "loss": 0.7591,
+      "step": 2755
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.65814958624217e-05,
+      "loss": 0.66,
+      "step": 2756
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6513064912029304e-05,
+      "loss": 0.7361,
+      "step": 2757
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.644466902745561e-05,
+      "loss": 0.8713,
+      "step": 2758
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6376308253540726e-05,
+      "loss": 0.684,
+      "step": 2759
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.630798263510162e-05,
+      "loss": 0.7005,
+      "step": 2760
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.623969221693236e-05,
+      "loss": 0.6121,
+      "step": 2761
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.617143704380381e-05,
+      "loss": 0.6308,
+      "step": 2762
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.610321716046382e-05,
+      "loss": 0.7153,
+      "step": 2763
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6035032611637094e-05,
+      "loss": 0.7845,
+      "step": 2764
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.596688344202509e-05,
+      "loss": 0.8432,
+      "step": 2765
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5898769696306155e-05,
+      "loss": 0.6058,
+      "step": 2766
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.583069141913543e-05,
+      "loss": 0.7352,
+      "step": 2767
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5762648655144666e-05,
+      "loss": 0.6885,
+      "step": 2768
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5694641448942485e-05,
+      "loss": 0.7257,
+      "step": 2769
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.562666984511416e-05,
+      "loss": 0.836,
+      "step": 2770
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.555873388822153e-05,
+      "loss": 0.7716,
+      "step": 2771
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.549083362280317e-05,
+      "loss": 0.7716,
+      "step": 2772
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.542296909337426e-05,
+      "loss": 0.7863,
+      "step": 2773
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.535514034442644e-05,
+      "loss": 0.8777,
+      "step": 2774
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.528734742042803e-05,
+      "loss": 0.7374,
+      "step": 2775
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5219590365823714e-05,
+      "loss": 0.6106,
+      "step": 2776
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5151869225034806e-05,
+      "loss": 0.6947,
+      "step": 2777
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.508418404245903e-05,
+      "loss": 0.6859,
+      "step": 2778
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.501653486247046e-05,
+      "loss": 0.8493,
+      "step": 2779
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.494892172941965e-05,
+      "loss": 0.6054,
+      "step": 2780
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.488134468763353e-05,
+      "loss": 0.8118,
+      "step": 2781
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.481380378141527e-05,
+      "loss": 0.7298,
+      "step": 2782
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.474629905504446e-05,
+      "loss": 0.7301,
+      "step": 2783
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.467883055277695e-05,
+      "loss": 0.6953,
+      "step": 2784
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.461139831884474e-05,
+      "loss": 0.7508,
+      "step": 2785
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.454400239745619e-05,
+      "loss": 0.6568,
+      "step": 2786
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.44766428327958e-05,
+      "loss": 0.7498,
+      "step": 2787
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.440931966902418e-05,
+      "loss": 0.7896,
+      "step": 2788
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.43420329502782e-05,
+      "loss": 0.7187,
+      "step": 2789
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.427478272067066e-05,
+      "loss": 0.6244,
+      "step": 2790
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4207569024290596e-05,
+      "loss": 0.7448,
+      "step": 2791
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.414039190520308e-05,
+      "loss": 0.8247,
+      "step": 2792
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.407325140744907e-05,
+      "loss": 0.7668,
+      "step": 2793
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.400614757504564e-05,
+      "loss": 0.6496,
+      "step": 2794
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.393908045198585e-05,
+      "loss": 0.8899,
+      "step": 2795
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.387205008223854e-05,
+      "loss": 0.7862,
+      "step": 2796
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3805056509748586e-05,
+      "loss": 0.517,
+      "step": 2797
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.373809977843676e-05,
+      "loss": 0.6701,
+      "step": 2798
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.367117993219955e-05,
+      "loss": 0.7365,
+      "step": 2799
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.360429701490934e-05,
+      "loss": 0.7009,
+      "step": 2800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.353745107041438e-05,
+      "loss": 0.6843,
+      "step": 2801
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.34706421425385e-05,
+      "loss": 0.6483,
+      "step": 2802
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.340387027508145e-05,
+      "loss": 0.9249,
+      "step": 2803
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.333713551181852e-05,
+      "loss": 0.7288,
+      "step": 2804
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.327043789650078e-05,
+      "loss": 0.6811,
+      "step": 2805
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.320377747285497e-05,
+      "loss": 0.7997,
+      "step": 2806
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.31371542845833e-05,
+      "loss": 0.6522,
+      "step": 2807
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.307056837536373e-05,
+      "loss": 0.7622,
+      "step": 2808
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.300401978884973e-05,
+      "loss": 0.7135,
+      "step": 2809
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2937508568670194e-05,
+      "loss": 0.6894,
+      "step": 2810
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2871034758429686e-05,
+      "loss": 0.6637,
+      "step": 2811
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2804598401708175e-05,
+      "loss": 0.8291,
+      "step": 2812
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2738199542061005e-05,
+      "loss": 0.7039,
+      "step": 2813
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2671838223019036e-05,
+      "loss": 0.6073,
+      "step": 2814
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2605514488088515e-05,
+      "loss": 0.7236,
+      "step": 2815
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.253922838075095e-05,
+      "loss": 0.7725,
+      "step": 2816
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.247297994446331e-05,
+      "loss": 0.7011,
+      "step": 2817
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.240676922265774e-05,
+      "loss": 0.7098,
+      "step": 2818
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.234059625874174e-05,
+      "loss": 0.7017,
+      "step": 2819
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.227446109609809e-05,
+      "loss": 0.6856,
+      "step": 2820
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.220836377808465e-05,
+      "loss": 0.8549,
+      "step": 2821
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.21423043480346e-05,
+      "loss": 0.7382,
+      "step": 2822
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.207628284925628e-05,
+      "loss": 0.7248,
+      "step": 2823
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2010299325033034e-05,
+      "loss": 0.8919,
+      "step": 2824
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1944353818623424e-05,
+      "loss": 0.7865,
+      "step": 2825
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.18784463732611e-05,
+      "loss": 0.7372,
+      "step": 2826
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.181257703215466e-05,
+      "loss": 0.761,
+      "step": 2827
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.17467458384878e-05,
+      "loss": 0.6281,
+      "step": 2828
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1680952835419216e-05,
+      "loss": 0.759,
+      "step": 2829
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.161519806608247e-05,
+      "loss": 0.6285,
+      "step": 2830
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.15494815735862e-05,
+      "loss": 0.848,
+      "step": 2831
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1483803401013796e-05,
+      "loss": 0.6014,
+      "step": 2832
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.141816359142364e-05,
+      "loss": 0.8041,
+      "step": 2833
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1352562187848954e-05,
+      "loss": 0.821,
+      "step": 2834
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.12869992332977e-05,
+      "loss": 0.795,
+      "step": 2835
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.12214747707527e-05,
+      "loss": 0.7797,
+      "step": 2836
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1155988843171565e-05,
+      "loss": 0.689,
+      "step": 2837
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.109054149348655e-05,
+      "loss": 0.6424,
+      "step": 2838
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.102513276460471e-05,
+      "loss": 0.6614,
+      "step": 2839
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0959762699407766e-05,
+      "loss": 0.6848,
+      "step": 2840
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0894431340751994e-05,
+      "loss": 0.6655,
+      "step": 2841
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0829138731468416e-05,
+      "loss": 0.7573,
+      "step": 2842
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0763884914362635e-05,
+      "loss": 0.7774,
+      "step": 2843
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0698669932214727e-05,
+      "loss": 0.7641,
+      "step": 2844
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0633493827779425e-05,
+      "loss": 0.776,
+      "step": 2845
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0568356643785856e-05,
+      "loss": 0.835,
+      "step": 2846
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.050325842293773e-05,
+      "loss": 0.6695,
+      "step": 2847
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.043819920791322e-05,
+      "loss": 0.6048,
+      "step": 2848
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.037317904136477e-05,
+      "loss": 0.7548,
+      "step": 2849
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.030819796591949e-05,
+      "loss": 0.8898,
+      "step": 2850
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.02432560241786e-05,
+      "loss": 0.6424,
+      "step": 2851
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0178353258717804e-05,
+      "loss": 0.6338,
+      "step": 2852
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0113489712087125e-05,
+      "loss": 0.7473,
+      "step": 2853
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0048665426810794e-05,
+      "loss": 0.6693,
+      "step": 2854
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9983880445387366e-05,
+      "loss": 0.6598,
+      "step": 2855
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.991913481028965e-05,
+      "loss": 0.7745,
+      "step": 2856
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9854428563964564e-05,
+      "loss": 0.7472,
+      "step": 2857
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.978976174883329e-05,
+      "loss": 0.8453,
+      "step": 2858
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9725134407291154e-05,
+      "loss": 0.6937,
+      "step": 2859
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.966054658170754e-05,
+      "loss": 0.6364,
+      "step": 2860
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.959599831442596e-05,
+      "loss": 0.7933,
+      "step": 2861
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.953148964776408e-05,
+      "loss": 0.822,
+      "step": 2862
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.946702062401341e-05,
+      "loss": 0.7253,
+      "step": 2863
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.940259128543967e-05,
+      "loss": 0.6483,
+      "step": 2864
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9338201674282406e-05,
+      "loss": 0.6983,
+      "step": 2865
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9273851832755214e-05,
+      "loss": 0.7912,
+      "step": 2866
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9209541803045635e-05,
+      "loss": 0.7821,
+      "step": 2867
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9145271627314986e-05,
+      "loss": 0.8756,
+      "step": 2868
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9081041347698574e-05,
+      "loss": 0.675,
+      "step": 2869
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9016851006305545e-05,
+      "loss": 0.756,
+      "step": 2870
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8952700645218766e-05,
+      "loss": 0.7474,
+      "step": 2871
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8888590306494974e-05,
+      "loss": 0.6701,
+      "step": 2872
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.882452003216469e-05,
+      "loss": 0.7634,
+      "step": 2873
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8760489864232066e-05,
+      "loss": 0.7713,
+      "step": 2874
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.869649984467504e-05,
+      "loss": 0.7232,
+      "step": 2875
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8632550015445256e-05,
+      "loss": 0.7406,
+      "step": 2876
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.856864041846789e-05,
+      "loss": 0.6416,
+      "step": 2877
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8504771095641904e-05,
+      "loss": 0.6791,
+      "step": 2878
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.844094208883966e-05,
+      "loss": 0.683,
+      "step": 2879
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8377153439907266e-05,
+      "loss": 0.7268,
+      "step": 2880
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.831340519066433e-05,
+      "loss": 0.7651,
+      "step": 2881
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.824969738290386e-05,
+      "loss": 0.7041,
+      "step": 2882
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.81860300583925e-05,
+      "loss": 0.8141,
+      "step": 2883
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.81224032588703e-05,
+      "loss": 0.6316,
+      "step": 2884
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8058817026050677e-05,
+      "loss": 0.7407,
+      "step": 2885
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.799527140162055e-05,
+      "loss": 0.6551,
+      "step": 2886
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.793176642724019e-05,
+      "loss": 0.6148,
+      "step": 2887
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.786830214454315e-05,
+      "loss": 0.7614,
+      "step": 2888
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.780487859513636e-05,
+      "loss": 0.6965,
+      "step": 2889
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.774149582060012e-05,
+      "loss": 0.6854,
+      "step": 2890
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.767815386248784e-05,
+      "loss": 0.6633,
+      "step": 2891
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7614852762326305e-05,
+      "loss": 0.8191,
+      "step": 2892
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7551592561615403e-05,
+      "loss": 0.7205,
+      "step": 2893
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7488373301828296e-05,
+      "loss": 0.684,
+      "step": 2894
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.742519502441132e-05,
+      "loss": 0.7654,
+      "step": 2895
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.736205777078381e-05,
+      "loss": 0.7517,
+      "step": 2896
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.729896158233835e-05,
+      "loss": 0.7461,
+      "step": 2897
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7235906500440574e-05,
+      "loss": 0.8077,
+      "step": 2898
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.717289256642906e-05,
+      "loss": 0.7304,
+      "step": 2899
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.710991982161555e-05,
+      "loss": 0.7112,
+      "step": 2900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.704698830728472e-05,
+      "loss": 0.6429,
+      "step": 2901
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.698409806469417e-05,
+      "loss": 0.7241,
+      "step": 2902
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6921249135074524e-05,
+      "loss": 0.6498,
+      "step": 2903
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6858441559629306e-05,
+      "loss": 0.9153,
+      "step": 2904
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.679567537953485e-05,
+      "loss": 0.8049,
+      "step": 2905
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.673295063594049e-05,
+      "loss": 0.6003,
+      "step": 2906
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.667026736996823e-05,
+      "loss": 0.7328,
+      "step": 2907
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6607625622713e-05,
+      "loss": 0.7377,
+      "step": 2908
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.654502543524253e-05,
+      "loss": 0.6705,
+      "step": 2909
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.648246684859716e-05,
+      "loss": 0.6651,
+      "step": 2910
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6419949903790085e-05,
+      "loss": 0.6744,
+      "step": 2911
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.63574746418072e-05,
+      "loss": 0.7428,
+      "step": 2912
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.629504110360697e-05,
+      "loss": 0.7946,
+      "step": 2913
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6232649330120605e-05,
+      "loss": 0.7558,
+      "step": 2914
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.617029936225193e-05,
+      "loss": 0.7047,
+      "step": 2915
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.610799124087725e-05,
+      "loss": 0.7149,
+      "step": 2916
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.604572500684558e-05,
+      "loss": 0.678,
+      "step": 2917
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5983500700978425e-05,
+      "loss": 0.584,
+      "step": 2918
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.592131836406972e-05,
+      "loss": 0.7768,
+      "step": 2919
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.585917803688603e-05,
+      "loss": 0.6653,
+      "step": 2920
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.579707976016624e-05,
+      "loss": 0.8123,
+      "step": 2921
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.573502357462176e-05,
+      "loss": 0.7705,
+      "step": 2922
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.567300952093641e-05,
+      "loss": 0.6264,
+      "step": 2923
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5611037639766265e-05,
+      "loss": 0.6712,
+      "step": 2924
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.55491079717399e-05,
+      "loss": 0.6838,
+      "step": 2925
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5487220557458176e-05,
+      "loss": 0.6565,
+      "step": 2926
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.542537543749417e-05,
+      "loss": 0.8221,
+      "step": 2927
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5363572652393326e-05,
+      "loss": 0.6631,
+      "step": 2928
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.530181224267333e-05,
+      "loss": 0.6951,
+      "step": 2929
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5240094248824e-05,
+      "loss": 0.5913,
+      "step": 2930
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5178418711307414e-05,
+      "loss": 0.5944,
+      "step": 2931
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.511678567055786e-05,
+      "loss": 0.7174,
+      "step": 2932
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5055195166981645e-05,
+      "loss": 0.7476,
+      "step": 2933
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4993647240957304e-05,
+      "loss": 0.7939,
+      "step": 2934
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.493214193283536e-05,
+      "loss": 0.816,
+      "step": 2935
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.487067928293848e-05,
+      "loss": 0.5908,
+      "step": 2936
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.480925933156134e-05,
+      "loss": 0.707,
+      "step": 2937
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4747882118970565e-05,
+      "loss": 0.8311,
+      "step": 2938
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.468654768540483e-05,
+      "loss": 0.8066,
+      "step": 2939
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4625256071074773e-05,
+      "loss": 0.6111,
+      "step": 2940
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.456400731616288e-05,
+      "loss": 0.7178,
+      "step": 2941
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4502801460823607e-05,
+      "loss": 0.5499,
+      "step": 2942
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.444163854518331e-05,
+      "loss": 0.869,
+      "step": 2943
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4380518609340076e-05,
+      "loss": 0.6707,
+      "step": 2944
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4319441693363906e-05,
+      "loss": 0.8913,
+      "step": 2945
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.425840783729664e-05,
+      "loss": 0.6768,
+      "step": 2946
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.419741708115174e-05,
+      "loss": 0.6123,
+      "step": 2947
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4136469464914575e-05,
+      "loss": 0.6966,
+      "step": 2948
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.407556502854209e-05,
+      "loss": 0.6916,
+      "step": 2949
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4014703811963025e-05,
+      "loss": 0.8758,
+      "step": 2950
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3953885855077774e-05,
+      "loss": 0.7017,
+      "step": 2951
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.389311119775828e-05,
+      "loss": 0.7668,
+      "step": 2952
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.38323798798482e-05,
+      "loss": 0.6312,
+      "step": 2953
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.377169194116275e-05,
+      "loss": 0.6305,
+      "step": 2954
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3711047421488675e-05,
+      "loss": 0.6499,
+      "step": 2955
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3650446360584275e-05,
+      "loss": 0.6119,
+      "step": 2956
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3589888798179416e-05,
+      "loss": 0.713,
+      "step": 2957
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.35293747739753e-05,
+      "loss": 0.7515,
+      "step": 2958
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3468904327644736e-05,
+      "loss": 0.6239,
+      "step": 2959
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.340847749883191e-05,
+      "loss": 0.8624,
+      "step": 2960
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.334809432715238e-05,
+      "loss": 0.6415,
+      "step": 2961
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3287754852193144e-05,
+      "loss": 0.8464,
+      "step": 2962
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.322745911351245e-05,
+      "loss": 0.633,
+      "step": 2963
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.316720715064e-05,
+      "loss": 0.7774,
+      "step": 2964
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3106999003076746e-05,
+      "loss": 0.678,
+      "step": 2965
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.304683471029485e-05,
+      "loss": 0.6667,
+      "step": 2966
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2986714311737824e-05,
+      "loss": 0.6331,
+      "step": 2967
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.292663784682036e-05,
+      "loss": 0.7183,
+      "step": 2968
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2866605354928315e-05,
+      "loss": 0.6625,
+      "step": 2969
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.280661687541876e-05,
+      "loss": 0.7132,
+      "step": 2970
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.274667244761991e-05,
+      "loss": 0.6766,
+      "step": 2971
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.268677211083109e-05,
+      "loss": 0.6175,
+      "step": 2972
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2626915904322744e-05,
+      "loss": 0.74,
+      "step": 2973
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.256710386733629e-05,
+      "loss": 0.614,
+      "step": 2974
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2507336039084314e-05,
+      "loss": 0.721,
+      "step": 2975
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2447612458750365e-05,
+      "loss": 0.6411,
+      "step": 2976
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.238793316548894e-05,
+      "loss": 0.6661,
+      "step": 2977
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.232829819842555e-05,
+      "loss": 0.7069,
+      "step": 2978
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2268707596656714e-05,
+      "loss": 0.7285,
+      "step": 2979
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2209161399249674e-05,
+      "loss": 0.7317,
+      "step": 2980
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.214965964524278e-05,
+      "loss": 0.7298,
+      "step": 2981
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.209020237364505e-05,
+      "loss": 0.8565,
+      "step": 2982
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2030789623436485e-05,
+      "loss": 0.7006,
+      "step": 2983
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.197142143356787e-05,
+      "loss": 0.785,
+      "step": 2984
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.191209784296068e-05,
+      "loss": 0.6926,
+      "step": 2985
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.185281889050725e-05,
+      "loss": 0.7075,
+      "step": 2986
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.179358461507068e-05,
+      "loss": 0.6426,
+      "step": 2987
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.173439505548462e-05,
+      "loss": 0.7944,
+      "step": 2988
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.167525025055357e-05,
+      "loss": 0.6844,
+      "step": 2989
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.161615023905265e-05,
+      "loss": 0.7558,
+      "step": 2990
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.15570950597275e-05,
+      "loss": 0.6388,
+      "step": 2991
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.149808475129452e-05,
+      "loss": 0.659,
+      "step": 2992
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.143911935244065e-05,
+      "loss": 0.686,
+      "step": 2993
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.138019890182331e-05,
+      "loss": 0.6085,
+      "step": 2994
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132132343807056e-05,
+      "loss": 0.674,
+      "step": 2995
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.126249299978086e-05,
+      "loss": 0.6892,
+      "step": 2996
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.120370762552325e-05,
+      "loss": 0.7351,
+      "step": 2997
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1144967353837196e-05,
+      "loss": 0.7213,
+      "step": 2998
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.108627222323253e-05,
+      "loss": 0.6966,
+      "step": 2999
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.102762227218957e-05,
+      "loss": 0.6944,
+      "step": 3000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.096901753915903e-05,
+      "loss": 0.7204,
+      "step": 3001
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.091045806256187e-05,
+      "loss": 0.6358,
+      "step": 3002
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.085194388078946e-05,
+      "loss": 0.6018,
+      "step": 3003
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.079347503220351e-05,
+      "loss": 0.621,
+      "step": 3004
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.073505155513591e-05,
+      "loss": 0.5979,
+      "step": 3005
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.067667348788885e-05,
+      "loss": 0.7458,
+      "step": 3006
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.061834086873482e-05,
+      "loss": 0.7776,
+      "step": 3007
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.056005373591637e-05,
+      "loss": 0.7395,
+      "step": 3008
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.050181212764639e-05,
+      "loss": 0.6737,
+      "step": 3009
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.044361608210775e-05,
+      "loss": 0.6894,
+      "step": 3010
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0385465637453604e-05,
+      "loss": 1.0309,
+      "step": 3011
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.032736083180716e-05,
+      "loss": 0.6945,
+      "step": 3012
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.026930170326164e-05,
+      "loss": 0.7493,
+      "step": 3013
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.02112882898804e-05,
+      "loss": 0.8388,
+      "step": 3014
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0153320629696846e-05,
+      "loss": 0.7042,
+      "step": 3015
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0095398760714267e-05,
+      "loss": 0.7494,
+      "step": 3016
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.003752272090603e-05,
+      "loss": 0.6994,
+      "step": 3017
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9979692548215477e-05,
+      "loss": 0.6553,
+      "step": 3018
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9921908280555778e-05,
+      "loss": 0.7313,
+      "step": 3019
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9864169955810084e-05,
+      "loss": 0.7419,
+      "step": 3020
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.980647761183144e-05,
+      "loss": 0.7731,
+      "step": 3021
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9748831286442657e-05,
+      "loss": 0.7237,
+      "step": 3022
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9691231017436494e-05,
+      "loss": 0.6521,
+      "step": 3023
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9633676842575387e-05,
+      "loss": 0.7137,
+      "step": 3024
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9576168799591664e-05,
+      "loss": 0.6998,
+      "step": 3025
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.951870692618739e-05,
+      "loss": 0.8611,
+      "step": 3026
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9461291260034264e-05,
+      "loss": 0.7364,
+      "step": 3027
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.940392183877382e-05,
+      "loss": 0.652,
+      "step": 3028
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9346598700017226e-05,
+      "loss": 0.7365,
+      "step": 3029
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.6576,
+      "step": 3030
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9232091420308383e-05,
+      "loss": 0.5657,
+      "step": 3031
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9174907354426696e-05,
+      "loss": 0.7131,
+      "step": 3032
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.911776972118978e-05,
+      "loss": 0.7075,
+      "step": 3033
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9060678558056874e-05,
+      "loss": 0.6957,
+      "step": 3034
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.900363390245674e-05,
+      "loss": 0.7524,
+      "step": 3035
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8946635791787545e-05,
+      "loss": 0.7197,
+      "step": 3036
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.888968426341707e-05,
+      "loss": 0.7454,
+      "step": 3037
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8832779354682536e-05,
+      "loss": 0.7464,
+      "step": 3038
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8775921102890513e-05,
+      "loss": 0.8166,
+      "step": 3039
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8719109545317103e-05,
+      "loss": 0.6796,
+      "step": 3040
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8662344719207658e-05,
+      "loss": 0.7903,
+      "step": 3041
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8605626661776996e-05,
+      "loss": 0.6623,
+      "step": 3042
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.854895541020931e-05,
+      "loss": 0.6707,
+      "step": 3043
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8492331001657945e-05,
+      "loss": 0.7224,
+      "step": 3044
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8435753473245698e-05,
+      "loss": 0.639,
+      "step": 3045
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8379222862064568e-05,
+      "loss": 0.7693,
+      "step": 3046
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8322739205175763e-05,
+      "loss": 0.7514,
+      "step": 3047
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8266302539609745e-05,
+      "loss": 0.7614,
+      "step": 3048
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.820991290236622e-05,
+      "loss": 0.6995,
+      "step": 3049
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8153570330413925e-05,
+      "loss": 0.6454,
+      "step": 3050
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8097274860690857e-05,
+      "loss": 0.7605,
+      "step": 3051
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.804102653010414e-05,
+      "loss": 0.7318,
+      "step": 3052
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.7984825375529877e-05,
+      "loss": 0.6778,
+      "step": 3053
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7928671433813392e-05,
+      "loss": 0.5937,
+      "step": 3054
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7872564741768913e-05,
+      "loss": 0.7342,
+      "step": 3055
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7816505336179798e-05,
+      "loss": 0.6962,
+      "step": 3056
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7760493253798393e-05,
+      "loss": 0.5959,
+      "step": 3057
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.770452853134593e-05,
+      "loss": 0.8126,
+      "step": 3058
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7648611205512685e-05,
+      "loss": 0.7277,
+      "step": 3059
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.759274131295787e-05,
+      "loss": 0.6525,
+      "step": 3060
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7536918890309493e-05,
+      "loss": 0.6761,
+      "step": 3061
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7481143974164547e-05,
+      "loss": 0.7845,
+      "step": 3062
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.742541660108886e-05,
+      "loss": 0.7773,
+      "step": 3063
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.736973680761702e-05,
+      "loss": 0.6165,
+      "step": 3064
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.73141046302525e-05,
+      "loss": 0.6223,
+      "step": 3065
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7258520105467567e-05,
+      "loss": 0.7605,
+      "step": 3066
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.720298326970315e-05,
+      "loss": 0.5629,
+      "step": 3067
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7147494159369036e-05,
+      "loss": 0.7498,
+      "step": 3068
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7092052810843593e-05,
+      "loss": 0.6361,
+      "step": 3069
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7036659260473974e-05,
+      "loss": 0.7289,
+      "step": 3070
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6981313544576016e-05,
+      "loss": 0.8828,
+      "step": 3071
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6926015699434072e-05,
+      "loss": 0.6026,
+      "step": 3072
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.687076576130122e-05,
+      "loss": 0.7732,
+      "step": 3073
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.681556376639912e-05,
+      "loss": 0.6778,
+      "step": 3074
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6760409750917927e-05,
+      "loss": 0.6975,
+      "step": 3075
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6705303751016408e-05,
+      "loss": 0.6264,
+      "step": 3076
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6650245802821884e-05,
+      "loss": 0.6789,
+      "step": 3077
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.659523594243004e-05,
+      "loss": 0.702,
+      "step": 3078
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6540274205905168e-05,
+      "loss": 0.9042,
+      "step": 3079
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6485360629279987e-05,
+      "loss": 0.6892,
+      "step": 3080
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6430495248555544e-05,
+      "loss": 0.6651,
+      "step": 3081
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6375678099701428e-05,
+      "loss": 0.7098,
+      "step": 3082
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6320909218655486e-05,
+      "loss": 0.7337,
+      "step": 3083
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6266188641323996e-05,
+      "loss": 0.5448,
+      "step": 3084
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6211516403581582e-05,
+      "loss": 0.7427,
+      "step": 3085
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6156892541271084e-05,
+      "loss": 0.7691,
+      "step": 3086
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.610231709020372e-05,
+      "loss": 0.8486,
+      "step": 3087
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6047790086158952e-05,
+      "loss": 0.686,
+      "step": 3088
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5993311564884417e-05,
+      "loss": 0.7776,
+      "step": 3089
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.593888156209603e-05,
+      "loss": 0.6824,
+      "step": 3090
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5884500113477894e-05,
+      "loss": 0.637,
+      "step": 3091
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5830167254682257e-05,
+      "loss": 0.6734,
+      "step": 3092
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5775883021329562e-05,
+      "loss": 0.687,
+      "step": 3093
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.572164744900827e-05,
+      "loss": 0.6944,
+      "step": 3094
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5667460573275028e-05,
+      "loss": 0.782,
+      "step": 3095
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5613322429654574e-05,
+      "loss": 0.6375,
+      "step": 3096
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5559233053639575e-05,
+      "loss": 0.7817,
+      "step": 3097
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5505192480690865e-05,
+      "loss": 0.7291,
+      "step": 3098
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5451200746237237e-05,
+      "loss": 0.5891,
+      "step": 3099
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5397257885675397e-05,
+      "loss": 0.7397,
+      "step": 3100
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5343363934370122e-05,
+      "loss": 0.6579,
+      "step": 3101
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5289518927654023e-05,
+      "loss": 0.8571,
+      "step": 3102
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5235722900827684e-05,
+      "loss": 0.5451,
+      "step": 3103
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5181975889159615e-05,
+      "loss": 0.9301,
+      "step": 3104
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5128277927886055e-05,
+      "loss": 0.6174,
+      "step": 3105
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5074629052211217e-05,
+      "loss": 0.6359,
+      "step": 3106
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5021029297307107e-05,
+      "loss": 0.7649,
+      "step": 3107
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.496747869831345e-05,
+      "loss": 0.7556,
+      "step": 3108
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4913977290337842e-05,
+      "loss": 0.8617,
+      "step": 3109
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.48605251084556e-05,
+      "loss": 0.7836,
+      "step": 3110
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.480712218770972e-05,
+      "loss": 0.6031,
+      "step": 3111
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.475376856311097e-05,
+      "loss": 0.7999,
+      "step": 3112
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.470046426963778e-05,
+      "loss": 0.8265,
+      "step": 3113
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.464720934223619e-05,
+      "loss": 0.7921,
+      "step": 3114
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4594003815819966e-05,
+      "loss": 0.7095,
+      "step": 3115
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4540847725270378e-05,
+      "loss": 0.7534,
+      "step": 3116
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.448774110543638e-05,
+      "loss": 0.7081,
+      "step": 3117
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4434683991134477e-05,
+      "loss": 0.7413,
+      "step": 3118
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4381676417148648e-05,
+      "loss": 0.6881,
+      "step": 3119
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.432871841823047e-05,
+      "loss": 0.655,
+      "step": 3120
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4275810029099024e-05,
+      "loss": 0.7231,
+      "step": 3121
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4222951284440776e-05,
+      "loss": 0.7015,
+      "step": 3122
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4170142218909742e-05,
+      "loss": 0.6438,
+      "step": 3123
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.411738286712735e-05,
+      "loss": 0.6888,
+      "step": 3124
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.406467326368237e-05,
+      "loss": 0.7553,
+      "step": 3125
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.401201344313102e-05,
+      "loss": 0.5768,
+      "step": 3126
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3959403439996907e-05,
+      "loss": 0.7068,
+      "step": 3127
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3906843288770886e-05,
+      "loss": 0.6629,
+      "step": 3128
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3854333023911225e-05,
+      "loss": 0.7573,
+      "step": 3129
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3801872679843385e-05,
+      "loss": 0.7663,
+      "step": 3130
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.374946229096019e-05,
+      "loss": 0.6664,
+      "step": 3131
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3697101891621697e-05,
+      "loss": 0.729,
+      "step": 3132
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.364479151615513e-05,
+      "loss": 0.6841,
+      "step": 3133
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3592531198854974e-05,
+      "loss": 0.7572,
+      "step": 3134
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3540320973982922e-05,
+      "loss": 0.721,
+      "step": 3135
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3488160875767717e-05,
+      "loss": 0.7077,
+      "step": 3136
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3436050938405363e-05,
+      "loss": 0.6842,
+      "step": 3137
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.338399119605892e-05,
+      "loss": 0.7126,
+      "step": 3138
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3331981682858505e-05,
+      "loss": 0.7343,
+      "step": 3139
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3280022432901383e-05,
+      "loss": 0.702,
+      "step": 3140
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3228113480251846e-05,
+      "loss": 0.776,
+      "step": 3141
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3176254858941127e-05,
+      "loss": 0.7751,
+      "step": 3142
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3124446602967597e-05,
+      "loss": 0.6545,
+      "step": 3143
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.307268874629649e-05,
+      "loss": 0.7138,
+      "step": 3144
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.302098132286006e-05,
+      "loss": 0.706,
+      "step": 3145
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2969324366557522e-05,
+      "loss": 0.6488,
+      "step": 3146
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.291771791125491e-05,
+      "loss": 0.7156,
+      "step": 3147
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2866161990785228e-05,
+      "loss": 0.724,
+      "step": 3148
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2814656638948384e-05,
+      "loss": 0.6404,
+      "step": 3149
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2763201889510987e-05,
+      "loss": 0.7384,
+      "step": 3150
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2711797776206624e-05,
+      "loss": 0.7625,
+      "step": 3151
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.266044433273562e-05,
+      "loss": 0.7898,
+      "step": 3152
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.260914159276506e-05,
+      "loss": 0.7434,
+      "step": 3153
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2557889589928815e-05,
+      "loss": 0.6486,
+      "step": 3154
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2506688357827544e-05,
+      "loss": 0.6276,
+      "step": 3155
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.245553793002849e-05,
+      "loss": 0.7543,
+      "step": 3156
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2404438340065736e-05,
+      "loss": 0.7806,
+      "step": 3157
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2353389621439903e-05,
+      "loss": 0.6806,
+      "step": 3158
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.230239180761836e-05,
+      "loss": 0.6416,
+      "step": 3159
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2251444932035094e-05,
+      "loss": 0.6174,
+      "step": 3160
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.22005490280906e-05,
+      "loss": 0.6585,
+      "step": 3161
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2149704129152084e-05,
+      "loss": 0.7731,
+      "step": 3162
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2098910268553263e-05,
+      "loss": 0.7195,
+      "step": 3163
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.204816747959434e-05,
+      "loss": 0.691,
+      "step": 3164
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.199747579554211e-05,
+      "loss": 0.6855,
+      "step": 3165
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.194683524962986e-05,
+      "loss": 0.6972,
+      "step": 3166
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1896245875057265e-05,
+      "loss": 0.6637,
+      "step": 3167
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.184570770499056e-05,
+      "loss": 0.7567,
+      "step": 3168
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1795220772562386e-05,
+      "loss": 0.7841,
+      "step": 3169
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.174478511087171e-05,
+      "loss": 0.6571,
+      "step": 3170
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1694400752984013e-05,
+      "loss": 0.7274,
+      "step": 3171
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1644067731931007e-05,
+      "loss": 0.6374,
+      "step": 3172
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1593786080710864e-05,
+      "loss": 0.6825,
+      "step": 3173
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1543555832288054e-05,
+      "loss": 0.6977,
+      "step": 3174
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.149337701959325e-05,
+      "loss": 0.6492,
+      "step": 3175
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1443249675523536e-05,
+      "loss": 0.6597,
+      "step": 3176
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1393173832942205e-05,
+      "loss": 0.6028,
+      "step": 3177
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.134314952467873e-05,
+      "loss": 0.7685,
+      "step": 3178
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1293176783528867e-05,
+      "loss": 0.7196,
+      "step": 3179
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1243255642254578e-05,
+      "loss": 0.6479,
+      "step": 3180
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1193386133583904e-05,
+      "loss": 0.7614,
+      "step": 3181
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1143568290211114e-05,
+      "loss": 0.6816,
+      "step": 3182
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1093802144796627e-05,
+      "loss": 0.7594,
+      "step": 3183
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1044087729966856e-05,
+      "loss": 0.6665,
+      "step": 3184
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.099442507831444e-05,
+      "loss": 0.6588,
+      "step": 3185
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0944814222397944e-05,
+      "loss": 0.7978,
+      "step": 3186
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.089525519474209e-05,
+      "loss": 0.7243,
+      "step": 3187
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0845748027837586e-05,
+      "loss": 0.7428,
+      "step": 3188
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0796292754141088e-05,
+      "loss": 0.7041,
+      "step": 3189
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.074688940607529e-05,
+      "loss": 0.6377,
+      "step": 3190
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0697538016028862e-05,
+      "loss": 0.6814,
+      "step": 3191
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0648238616356332e-05,
+      "loss": 0.7488,
+      "step": 3192
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0598991239378185e-05,
+      "loss": 0.7016,
+      "step": 3193
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0549795917380864e-05,
+      "loss": 0.7111,
+      "step": 3194
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.050065268261655e-05,
+      "loss": 0.8881,
+      "step": 3195
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.045156156730338e-05,
+      "loss": 0.7649,
+      "step": 3196
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0402522603625308e-05,
+      "loss": 0.7176,
+      "step": 3197
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.035353582373205e-05,
+      "loss": 0.7165,
+      "step": 3198
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0304601259739175e-05,
+      "loss": 0.6536,
+      "step": 3199
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.025571894372794e-05,
+      "loss": 0.7246,
+      "step": 3200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0206888907745415e-05,
+      "loss": 0.7693,
+      "step": 3201
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0158111183804407e-05,
+      "loss": 0.7736,
+      "step": 3202
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.010938580388334e-05,
+      "loss": 0.7311,
+      "step": 3203
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0060712799926408e-05,
+      "loss": 0.662,
+      "step": 3204
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.001209220384346e-05,
+      "loss": 0.6213,
+      "step": 3205
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.9963524047509897e-05,
+      "loss": 0.7469,
+      "step": 3206
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.991500836276685e-05,
+      "loss": 0.5494,
+      "step": 3207
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9866545181421013e-05,
+      "loss": 0.6462,
+      "step": 3208
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9818134535244624e-05,
+      "loss": 0.704,
+      "step": 3209
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.976977645597552e-05,
+      "loss": 0.729,
+      "step": 3210
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9721470975317092e-05,
+      "loss": 0.6594,
+      "step": 3211
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.967321812493813e-05,
+      "loss": 0.701,
+      "step": 3212
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9625017936473124e-05,
+      "loss": 0.6732,
+      "step": 3213
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9576870441521833e-05,
+      "loss": 0.6238,
+      "step": 3214
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9528775671649592e-05,
+      "loss": 0.7944,
+      "step": 3215
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9480733658387175e-05,
+      "loss": 0.6794,
+      "step": 3216
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9432744433230666e-05,
+      "loss": 0.7424,
+      "step": 3217
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9384808027641666e-05,
+      "loss": 0.6073,
+      "step": 3218
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9336924473047037e-05,
+      "loss": 0.6668,
+      "step": 3219
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9289093800839066e-05,
+      "loss": 0.6922,
+      "step": 3220
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9241316042375366e-05,
+      "loss": 0.7422,
+      "step": 3221
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9193591228978814e-05,
+      "loss": 0.68,
+      "step": 3222
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9145919391937616e-05,
+      "loss": 0.6408,
+      "step": 3223
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.7263,
+      "step": 3224
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9050734771900413e-05,
+      "loss": 0.6976,
+      "step": 3225
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9003222051307047e-05,
+      "loss": 0.576,
+      "step": 3226
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8955762431874312e-05,
+      "loss": 0.758,
+      "step": 3227
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8908355944716517e-05,
+      "loss": 0.6367,
+      "step": 3228
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8861002620913172e-05,
+      "loss": 0.6995,
+      "step": 3229
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8813702491508955e-05,
+      "loss": 0.7699,
+      "step": 3230
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.87664555875136e-05,
+      "loss": 0.7118,
+      "step": 3231
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.871926193990202e-05,
+      "loss": 0.6419,
+      "step": 3232
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8672121579614144e-05,
+      "loss": 0.6712,
+      "step": 3233
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8625034537555018e-05,
+      "loss": 0.6847,
+      "step": 3234
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8578000844594747e-05,
+      "loss": 0.5491,
+      "step": 3235
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8531020531568378e-05,
+      "loss": 0.6546,
+      "step": 3236
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8484093629276032e-05,
+      "loss": 0.5877,
+      "step": 3237
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8437220168482837e-05,
+      "loss": 0.7664,
+      "step": 3238
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8390400179918786e-05,
+      "loss": 0.6371,
+      "step": 3239
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8343633694278895e-05,
+      "loss": 0.6142,
+      "step": 3240
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8296920742223122e-05,
+      "loss": 0.6295,
+      "step": 3241
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.825026135437622e-05,
+      "loss": 0.6407,
+      "step": 3242
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8203655561327938e-05,
+      "loss": 0.6669,
+      "step": 3243
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8157103393632868e-05,
+      "loss": 0.7123,
+      "step": 3244
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8110604881810355e-05,
+      "loss": 0.5909,
+      "step": 3245
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8064160056344716e-05,
+      "loss": 0.6804,
+      "step": 3246
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.801776894768492e-05,
+      "loss": 0.7808,
+      "step": 3247
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7971431586244815e-05,
+      "loss": 0.6845,
+      "step": 3248
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7925148002403026e-05,
+      "loss": 0.6425,
+      "step": 3249
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7878918226502816e-05,
+      "loss": 0.7069,
+      "step": 3250
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.783274228885229e-05,
+      "loss": 0.6969,
+      "step": 3251
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7786620219724204e-05,
+      "loss": 0.6531,
+      "step": 3252
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7740552049355962e-05,
+      "loss": 0.6585,
+      "step": 3253
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7694537807949708e-05,
+      "loss": 0.7001,
+      "step": 3254
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7648577525672194e-05,
+      "loss": 0.7631,
+      "step": 3255
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7602671232654754e-05,
+      "loss": 0.6829,
+      "step": 3256
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7556818958993403e-05,
+      "loss": 0.6099,
+      "step": 3257
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.751102073474873e-05,
+      "loss": 0.709,
+      "step": 3258
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.746527658994581e-05,
+      "loss": 0.8643,
+      "step": 3259
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.741958655457436e-05,
+      "loss": 0.7701,
+      "step": 3260
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7373950658588557e-05,
+      "loss": 0.6955,
+      "step": 3261
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7328368931907113e-05,
+      "loss": 0.6581,
+      "step": 3262
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7282841404413264e-05,
+      "loss": 0.7879,
+      "step": 3263
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.723736810595461e-05,
+      "loss": 0.6529,
+      "step": 3264
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7191949066343305e-05,
+      "loss": 0.7637,
+      "step": 3265
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7146584315355885e-05,
+      "loss": 0.6176,
+      "step": 3266
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.710127388273327e-05,
+      "loss": 0.6326,
+      "step": 3267
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7056017798180824e-05,
+      "loss": 0.6485,
+      "step": 3268
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7010816091368265e-05,
+      "loss": 0.8045,
+      "step": 3269
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.69656687919296e-05,
+      "loss": 0.6509,
+      "step": 3270
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6920575929463256e-05,
+      "loss": 0.7548,
+      "step": 3271
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6875537533531948e-05,
+      "loss": 0.7417,
+      "step": 3272
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6830553633662606e-05,
+      "loss": 0.6007,
+      "step": 3273
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6785624259346557e-05,
+      "loss": 0.6032,
+      "step": 3274
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6740749440039262e-05,
+      "loss": 0.7323,
+      "step": 3275
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6695929205160487e-05,
+      "loss": 0.6936,
+      "step": 3276
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.665116358409424e-05,
+      "loss": 0.7821,
+      "step": 3277
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.660645260618864e-05,
+      "loss": 0.6713,
+      "step": 3278
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6561796300756006e-05,
+      "loss": 0.7688,
+      "step": 3279
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.65171946970729e-05,
+      "loss": 0.7459,
+      "step": 3280
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6472647824379882e-05,
+      "loss": 0.6855,
+      "step": 3281
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6428155711881722e-05,
+      "loss": 0.6401,
+      "step": 3282
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6383718388747316e-05,
+      "loss": 0.6522,
+      "step": 3283
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6339335884109518e-05,
+      "loss": 0.5665,
+      "step": 3284
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6295008227065367e-05,
+      "loss": 0.7228,
+      "step": 3285
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6250735446675912e-05,
+      "loss": 0.7127,
+      "step": 3286
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.620651757196615e-05,
+      "loss": 0.6596,
+      "step": 3287
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6162354631925204e-05,
+      "loss": 0.7576,
+      "step": 3288
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6118246655506065e-05,
+      "loss": 0.7588,
+      "step": 3289
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.607419367162577e-05,
+      "loss": 0.7071,
+      "step": 3290
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6030195709165295e-05,
+      "loss": 0.7232,
+      "step": 3291
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.598625279696948e-05,
+      "loss": 0.6292,
+      "step": 3292
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5942364963847156e-05,
+      "loss": 0.7072,
+      "step": 3293
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.589853223857103e-05,
+      "loss": 0.5998,
+      "step": 3294
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5854754649877603e-05,
+      "loss": 0.6431,
+      "step": 3295
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5811032226467305e-05,
+      "loss": 0.6734,
+      "step": 3296
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.576736499700442e-05,
+      "loss": 0.646,
+      "step": 3297
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5723752990116947e-05,
+      "loss": 0.7127,
+      "step": 3298
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5680196234396783e-05,
+      "loss": 0.7218,
+      "step": 3299
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.563669475839956e-05,
+      "loss": 0.6639,
+      "step": 3300
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.559324859064464e-05,
+      "loss": 0.7533,
+      "step": 3301
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5549857759615194e-05,
+      "loss": 0.7024,
+      "step": 3302
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.550652229375804e-05,
+      "loss": 0.7989,
+      "step": 3303
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5463242221483743e-05,
+      "loss": 0.7309,
+      "step": 3304
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.542001757116658e-05,
+      "loss": 0.7007,
+      "step": 3305
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5376848371144402e-05,
+      "loss": 0.738,
+      "step": 3306
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.533373464971879e-05,
+      "loss": 0.7312,
+      "step": 3307
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.529067643515495e-05,
+      "loss": 0.6555,
+      "step": 3308
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5247673755681624e-05,
+      "loss": 0.8253,
+      "step": 3309
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5204726639491218e-05,
+      "loss": 0.6797,
+      "step": 3310
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5161835114739708e-05,
+      "loss": 0.694,
+      "step": 3311
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5118999209546559e-05,
+      "loss": 0.6936,
+      "step": 3312
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5076218951994858e-05,
+      "loss": 0.6164,
+      "step": 3313
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5033494370131163e-05,
+      "loss": 0.6596,
+      "step": 3314
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4990825491965522e-05,
+      "loss": 0.7225,
+      "step": 3315
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4948212345471491e-05,
+      "loss": 0.7005,
+      "step": 3316
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4905654958586068e-05,
+      "loss": 0.714,
+      "step": 3317
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4863153359209692e-05,
+      "loss": 0.7454,
+      "step": 3318
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4820707575206294e-05,
+      "loss": 0.6974,
+      "step": 3319
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4778317634403083e-05,
+      "loss": 0.8554,
+      "step": 3320
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4735983564590783e-05,
+      "loss": 0.7199,
+      "step": 3321
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4693705393523449e-05,
+      "loss": 0.6265,
+      "step": 3322
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4651483148918432e-05,
+      "loss": 0.7087,
+      "step": 3323
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.460931685845649e-05,
+      "loss": 0.699,
+      "step": 3324
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4567206549781698e-05,
+      "loss": 0.705,
+      "step": 3325
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4525152250501361e-05,
+      "loss": 0.731,
+      "step": 3326
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4483153988186116e-05,
+      "loss": 0.7884,
+      "step": 3327
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.444121179036989e-05,
+      "loss": 0.6553,
+      "step": 3328
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4399325684549781e-05,
+      "loss": 0.7723,
+      "step": 3329
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4357495698186186e-05,
+      "loss": 0.6328,
+      "step": 3330
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4315721858702613e-05,
+      "loss": 0.5567,
+      "step": 3331
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.427400419348588e-05,
+      "loss": 0.6091,
+      "step": 3332
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4232342729885916e-05,
+      "loss": 0.6114,
+      "step": 3333
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4190737495215745e-05,
+      "loss": 0.7248,
+      "step": 3334
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.41491885167517e-05,
+      "loss": 0.5502,
+      "step": 3335
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4107695821733025e-05,
+      "loss": 0.6711,
+      "step": 3336
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4066259437362205e-05,
+      "loss": 0.681,
+      "step": 3337
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.402487939080479e-05,
+      "loss": 0.8116,
+      "step": 3338
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3983555709189333e-05,
+      "loss": 0.7097,
+      "step": 3339
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3942288419607475e-05,
+      "loss": 0.7368,
+      "step": 3340
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3901077549113906e-05,
+      "loss": 0.5746,
+      "step": 3341
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3859923124726281e-05,
+      "loss": 0.7134,
+      "step": 3342
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3818825173425287e-05,
+      "loss": 0.6099,
+      "step": 3343
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3777783722154603e-05,
+      "loss": 0.7361,
+      "step": 3344
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3736798797820782e-05,
+      "loss": 0.7597,
+      "step": 3345
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.369587042729341e-05,
+      "loss": 0.6406,
+      "step": 3346
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3654998637404981e-05,
+      "loss": 0.7219,
+      "step": 3347
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3614183454950824e-05,
+      "loss": 0.7869,
+      "step": 3348
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3573424906689258e-05,
+      "loss": 0.7048,
+      "step": 3349
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3532723019341375e-05,
+      "loss": 0.8082,
+      "step": 3350
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3492077819591187e-05,
+      "loss": 0.6617,
+      "step": 3351
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3451489334085554e-05,
+      "loss": 0.5602,
+      "step": 3352
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3410957589434082e-05,
+      "loss": 0.7496,
+      "step": 3353
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3370482612209223e-05,
+      "loss": 0.791,
+      "step": 3354
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3330064428946254e-05,
+      "loss": 0.8307,
+      "step": 3355
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3289703066143111e-05,
+      "loss": 0.7203,
+      "step": 3356
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3249398550260561e-05,
+      "loss": 0.734,
+      "step": 3357
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3209150907722123e-05,
+      "loss": 0.7205,
+      "step": 3358
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3168960164913924e-05,
+      "loss": 0.6422,
+      "step": 3359
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3128826348184887e-05,
+      "loss": 0.595,
+      "step": 3360
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3088749483846596e-05,
+      "loss": 0.5669,
+      "step": 3361
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3048729598173248e-05,
+      "loss": 0.7438,
+      "step": 3362
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3008766717401743e-05,
+      "loss": 0.72,
+      "step": 3363
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2968860867731569e-05,
+      "loss": 0.8365,
+      "step": 3364
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2929012075324831e-05,
+      "loss": 0.7655,
+      "step": 3365
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2889220366306276e-05,
+      "loss": 0.7364,
+      "step": 3366
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2849485766763147e-05,
+      "loss": 0.596,
+      "step": 3367
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2809808302745297e-05,
+      "loss": 0.6937,
+      "step": 3368
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2770188000265137e-05,
+      "loss": 0.695,
+      "step": 3369
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2730624885297537e-05,
+      "loss": 0.6485,
+      "step": 3370
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2691118983779937e-05,
+      "loss": 0.6978,
+      "step": 3371
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2651670321612263e-05,
+      "loss": 0.7086,
+      "step": 3372
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2612278924656862e-05,
+      "loss": 0.7564,
+      "step": 3373
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2572944818738586e-05,
+      "loss": 0.6358,
+      "step": 3374
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2533668029644751e-05,
+      "loss": 0.6349,
+      "step": 3375
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2494448583125018e-05,
+      "loss": 0.7699,
+      "step": 3376
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2455286504891528e-05,
+      "loss": 0.7081,
+      "step": 3377
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2416181820618744e-05,
+      "loss": 0.6856,
+      "step": 3378
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.237713455594357e-05,
+      "loss": 0.6728,
+      "step": 3379
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.233814473646524e-05,
+      "loss": 0.6956,
+      "step": 3380
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2299212387745295e-05,
+      "loss": 0.5776,
+      "step": 3381
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2260337535307631e-05,
+      "loss": 0.621,
+      "step": 3382
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.222152020463848e-05,
+      "loss": 0.7125,
+      "step": 3383
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.218276042118629e-05,
+      "loss": 0.8168,
+      "step": 3384
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.214405821036182e-05,
+      "loss": 0.7213,
+      "step": 3385
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2105413597538107e-05,
+      "loss": 0.5724,
+      "step": 3386
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2066826608050374e-05,
+      "loss": 0.78,
+      "step": 3387
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.202829726719611e-05,
+      "loss": 0.617,
+      "step": 3388
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1989825600235016e-05,
+      "loss": 0.6821,
+      "step": 3389
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.195141163238892e-05,
+      "loss": 0.6974,
+      "step": 3390
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1913055388841909e-05,
+      "loss": 0.6424,
+      "step": 3391
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1874756894740135e-05,
+      "loss": 0.6534,
+      "step": 3392
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1836516175191948e-05,
+      "loss": 0.7664,
+      "step": 3393
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1798333255267857e-05,
+      "loss": 0.7151,
+      "step": 3394
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1760208160000363e-05,
+      "loss": 0.6757,
+      "step": 3395
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.172214091438416e-05,
+      "loss": 0.6363,
+      "step": 3396
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1684131543375986e-05,
+      "loss": 0.7267,
+      "step": 3397
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1646180071894607e-05,
+      "loss": 0.7248,
+      "step": 3398
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1608286524820878e-05,
+      "loss": 0.6677,
+      "step": 3399
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1570450926997655e-05,
+      "loss": 0.7026,
+      "step": 3400
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1532673303229769e-05,
+      "loss": 0.7734,
+      "step": 3401
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1494953678284103e-05,
+      "loss": 0.5277,
+      "step": 3402
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1457292076889525e-05,
+      "loss": 0.6706,
+      "step": 3403
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.141968852373676e-05,
+      "loss": 0.6472,
+      "step": 3404
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.13821430434786e-05,
+      "loss": 0.7212,
+      "step": 3405
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1344655660729675e-05,
+      "loss": 0.7058,
+      "step": 3406
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1307226400066573e-05,
+      "loss": 0.7681,
+      "step": 3407
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1269855286027797e-05,
+      "loss": 0.593,
+      "step": 3408
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.123254234311364e-05,
+      "loss": 0.6433,
+      "step": 3409
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1195287595786352e-05,
+      "loss": 0.6224,
+      "step": 3410
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1158091068470011e-05,
+      "loss": 0.6322,
+      "step": 3411
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1120952785550476e-05,
+      "loss": 0.787,
+      "step": 3412
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1083872771375459e-05,
+      "loss": 0.6334,
+      "step": 3413
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1046851050254502e-05,
+      "loss": 0.6295,
+      "step": 3414
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1009887646458861e-05,
+      "loss": 0.6515,
+      "step": 3415
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0972982584221592e-05,
+      "loss": 0.6799,
+      "step": 3416
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0936135887737542e-05,
+      "loss": 0.6979,
+      "step": 3417
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0899347581163221e-05,
+      "loss": 0.7052,
+      "step": 3418
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0862617688616917e-05,
+      "loss": 0.6454,
+      "step": 3419
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0825946234178574e-05,
+      "loss": 0.6757,
+      "step": 3420
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0789333241889854e-05,
+      "loss": 0.7214,
+      "step": 3421
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.075277873575412e-05,
+      "loss": 0.634,
+      "step": 3422
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0716282739736316e-05,
+      "loss": 0.5729,
+      "step": 3423
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.067984527776309e-05,
+      "loss": 0.663,
+      "step": 3424
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0643466373722711e-05,
+      "loss": 0.6547,
+      "step": 3425
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0607146051465012e-05,
+      "loss": 0.6794,
+      "step": 3426
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0570884334801457e-05,
+      "loss": 0.6314,
+      "step": 3427
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0534681247505106e-05,
+      "loss": 0.7596,
+      "step": 3428
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.049853681331051e-05,
+      "loss": 0.6964,
+      "step": 3429
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0462451055913847e-05,
+      "loss": 0.785,
+      "step": 3430
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0426423998972811e-05,
+      "loss": 0.7619,
+      "step": 3431
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0390455666106547e-05,
+      "loss": 0.6658,
+      "step": 3432
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.03545460808958e-05,
+      "loss": 0.7013,
+      "step": 3433
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0318695266882693e-05,
+      "loss": 0.8046,
+      "step": 3434
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0282903247570908e-05,
+      "loss": 0.6932,
+      "step": 3435
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.024717004642557e-05,
+      "loss": 0.6654,
+      "step": 3436
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0211495686873174e-05,
+      "loss": 0.618,
+      "step": 3437
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0175880192301713e-05,
+      "loss": 0.7176,
+      "step": 3438
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0140323586060563e-05,
+      "loss": 0.6271,
+      "step": 3439
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.010482589146048e-05,
+      "loss": 0.6717,
+      "step": 3440
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0069387131773612e-05,
+      "loss": 0.7172,
+      "step": 3441
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0034007330233486e-05,
+      "loss": 0.6903,
+      "step": 3442
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.998686510034916e-06,
+      "loss": 0.6868,
+      "step": 3443
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.963424694334122e-06,
+      "loss": 0.7248,
+      "step": 3444
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.928221906248614e-06,
+      "loss": 0.5985,
+      "step": 3445
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.893078168857173e-06,
+      "loss": 0.6184,
+      "step": 3446
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.857993505199913e-06,
+      "loss": 0.698,
+      "step": 3447
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.822967938278171e-06,
+      "loss": 0.5248,
+      "step": 3448
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.788001491054577e-06,
+      "loss": 0.6832,
+      "step": 3449
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.753094186453026e-06,
+      "loss": 0.7532,
+      "step": 3450
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.71824604735856e-06,
+      "loss": 0.6092,
+      "step": 3451
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.683457096617488e-06,
+      "loss": 0.679,
+      "step": 3452
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.648727357037335e-06,
+      "loss": 0.6772,
+      "step": 3453
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.614056851386744e-06,
+      "loss": 0.6717,
+      "step": 3454
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.579445602395576e-06,
+      "loss": 0.6338,
+      "step": 3455
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.544893632754814e-06,
+      "loss": 0.742,
+      "step": 3456
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.510400965116595e-06,
+      "loss": 0.6677,
+      "step": 3457
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.475967622094205e-06,
+      "loss": 0.6977,
+      "step": 3458
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.441593626261947e-06,
+      "loss": 0.7612,
+      "step": 3459
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.407279000155312e-06,
+      "loss": 0.8282,
+      "step": 3460
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.373023766270827e-06,
+      "loss": 0.7906,
+      "step": 3461
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.338827947066076e-06,
+      "loss": 0.7513,
+      "step": 3462
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.30469156495969e-06,
+      "loss": 0.8023,
+      "step": 3463
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.270614642331376e-06,
+      "loss": 0.6327,
+      "step": 3464
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.23659720152179e-06,
+      "loss": 0.6193,
+      "step": 3465
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.202639264832668e-06,
+      "loss": 0.6558,
+      "step": 3466
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.168740854526648e-06,
+      "loss": 0.6934,
+      "step": 3467
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.134901992827427e-06,
+      "loss": 0.7386,
+      "step": 3468
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.101122701919629e-06,
+      "loss": 0.6215,
+      "step": 3469
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.067403003948782e-06,
+      "loss": 0.7285,
+      "step": 3470
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.033742921021415e-06,
+      "loss": 0.71,
+      "step": 3471
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.000142475204964e-06,
+      "loss": 0.7302,
+      "step": 3472
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.966601688527686e-06,
+      "loss": 0.6238,
+      "step": 3473
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.933120582978827e-06,
+      "loss": 0.602,
+      "step": 3474
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.89969918050847e-06,
+      "loss": 0.6694,
+      "step": 3475
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.866337503027522e-06,
+      "loss": 0.6838,
+      "step": 3476
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.833035572407777e-06,
+      "loss": 0.7906,
+      "step": 3477
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.79979341048187e-06,
+      "loss": 0.6009,
+      "step": 3478
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.766611039043193e-06,
+      "loss": 0.5806,
+      "step": 3479
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.733488479845997e-06,
+      "loss": 0.8257,
+      "step": 3480
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.700425754605268e-06,
+      "loss": 0.6316,
+      "step": 3481
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.667422884996823e-06,
+      "loss": 0.6162,
+      "step": 3482
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.634479892657221e-06,
+      "loss": 0.6765,
+      "step": 3483
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.60159679918372e-06,
+      "loss": 0.7833,
+      "step": 3484
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.568773626134364e-06,
+      "loss": 0.765,
+      "step": 3485
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.536010395027904e-06,
+      "loss": 0.7669,
+      "step": 3486
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.503307127343751e-06,
+      "loss": 0.682,
+      "step": 3487
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.470663844522052e-06,
+      "loss": 0.6625,
+      "step": 3488
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.438080567963625e-06,
+      "loss": 0.715,
+      "step": 3489
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.405557319029912e-06,
+      "loss": 0.7023,
+      "step": 3490
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.373094119043023e-06,
+      "loss": 0.6978,
+      "step": 3491
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.340690989285726e-06,
+      "loss": 0.6732,
+      "step": 3492
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.308347951001361e-06,
+      "loss": 0.6345,
+      "step": 3493
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.276065025393908e-06,
+      "loss": 0.5816,
+      "step": 3494
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.243842233627896e-06,
+      "loss": 0.6632,
+      "step": 3495
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.21167959682848e-06,
+      "loss": 0.589,
+      "step": 3496
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.179577136081361e-06,
+      "loss": 0.5358,
+      "step": 3497
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.14753487243276e-06,
+      "loss": 0.6908,
+      "step": 3498
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.115552826889461e-06,
+      "loss": 0.7534,
+      "step": 3499
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.083631020418791e-06,
+      "loss": 0.672,
+      "step": 3500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.051769473948512e-06,
+      "loss": 0.5886,
+      "step": 3501
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.019968208366958e-06,
+      "loss": 0.732,
+      "step": 3502
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.988227244522916e-06,
+      "loss": 0.727,
+      "step": 3503
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.956546603225601e-06,
+      "loss": 0.6834,
+      "step": 3504
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.924926305244728e-06,
+      "loss": 0.6765,
+      "step": 3505
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.893366371310462e-06,
+      "loss": 0.6227,
+      "step": 3506
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.86186682211334e-06,
+      "loss": 0.6952,
+      "step": 3507
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.830427678304353e-06,
+      "loss": 0.7067,
+      "step": 3508
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.799048960494848e-06,
+      "loss": 0.7066,
+      "step": 3509
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.767730689256614e-06,
+      "loss": 0.7127,
+      "step": 3510
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.736472885121792e-06,
+      "loss": 0.5671,
+      "step": 3511
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.705275568582848e-06,
+      "loss": 0.6785,
+      "step": 3512
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.674138760092619e-06,
+      "loss": 0.6866,
+      "step": 3513
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.6430624800643e-06,
+      "loss": 0.6382,
+      "step": 3514
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.612046748871327e-06,
+      "loss": 0.6426,
+      "step": 3515
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.581091586847522e-06,
+      "loss": 0.644,
+      "step": 3516
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.5501970142869634e-06,
+      "loss": 0.6609,
+      "step": 3517
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.519363051443995e-06,
+      "loss": 0.6774,
+      "step": 3518
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.488589718533234e-06,
+      "loss": 0.7174,
+      "step": 3519
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.457877035729588e-06,
+      "loss": 0.6395,
+      "step": 3520
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.427225023168127e-06,
+      "loss": 0.7698,
+      "step": 3521
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.3966337009442e-06,
+      "loss": 0.6063,
+      "step": 3522
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.366103089113374e-06,
+      "loss": 0.5887,
+      "step": 3523
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.335633207691361e-06,
+      "loss": 0.6803,
+      "step": 3524
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.305224076654127e-06,
+      "loss": 0.6974,
+      "step": 3525
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.274875715937746e-06,
+      "loss": 0.6216,
+      "step": 3526
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.24458814543848e-06,
+      "loss": 0.7976,
+      "step": 3527
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.21436138501278e-06,
+      "loss": 0.736,
+      "step": 3528
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.184195454477138e-06,
+      "loss": 0.6963,
+      "step": 3529
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.154090373608235e-06,
+      "loss": 0.6817,
+      "step": 3530
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.124046162142861e-06,
+      "loss": 0.7671,
+      "step": 3531
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.094062839777837e-06,
+      "loss": 0.5841,
+      "step": 3532
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.064140426170141e-06,
+      "loss": 0.6577,
+      "step": 3533
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.03427894093679e-06,
+      "loss": 0.5979,
+      "step": 3534
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.004478403654835e-06,
+      "loss": 0.7418,
+      "step": 3535
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.974738833861383e-06,
+      "loss": 0.6404,
+      "step": 3536
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.945060251053603e-06,
+      "loss": 0.6472,
+      "step": 3537
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.915442674688632e-06,
+      "loss": 0.6831,
+      "step": 3538
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.885886124183649e-06,
+      "loss": 0.5881,
+      "step": 3539
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.856390618915775e-06,
+      "loss": 0.655,
+      "step": 3540
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.826956178222177e-06,
+      "loss": 0.7027,
+      "step": 3541
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.7975828213999725e-06,
+      "loss": 0.6806,
+      "step": 3542
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.768270567706158e-06,
+      "loss": 0.6387,
+      "step": 3543
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.739019436357774e-06,
+      "loss": 0.7274,
+      "step": 3544
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.709829446531734e-06,
+      "loss": 0.6556,
+      "step": 3545
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.680700617364877e-06,
+      "loss": 0.7245,
+      "step": 3546
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.651632967953925e-06,
+      "loss": 0.7239,
+      "step": 3547
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.622626517355557e-06,
+      "loss": 0.714,
+      "step": 3548
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.5936812845862395e-06,
+      "loss": 0.7507,
+      "step": 3549
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.564797288622371e-06,
+      "loss": 0.8123,
+      "step": 3550
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.535974548400192e-06,
+      "loss": 0.6595,
+      "step": 3551
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.507213082815744e-06,
+      "loss": 0.6654,
+      "step": 3552
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.478512910724966e-06,
+      "loss": 0.6483,
+      "step": 3553
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.449874050943549e-06,
+      "loss": 0.7007,
+      "step": 3554
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.4212965222470115e-06,
+      "loss": 0.6482,
+      "step": 3555
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.392780343370686e-06,
+      "loss": 0.6774,
+      "step": 3556
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.3643255330096515e-06,
+      "loss": 0.627,
+      "step": 3557
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.335932109818754e-06,
+      "loss": 0.6942,
+      "step": 3558
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.307600092412647e-06,
+      "loss": 0.7169,
+      "step": 3559
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2793294993656494e-06,
+      "loss": 0.7734,
+      "step": 3560
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.251120349211847e-06,
+      "loss": 0.6412,
+      "step": 3561
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.222972660445081e-06,
+      "loss": 0.826,
+      "step": 3562
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.194886451518822e-06,
+      "loss": 0.8978,
+      "step": 3563
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.166861740846297e-06,
+      "loss": 0.6954,
+      "step": 3564
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.138898546800398e-06,
+      "loss": 0.7732,
+      "step": 3565
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.110996887713661e-06,
+      "loss": 0.6706,
+      "step": 3566
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.083156781878318e-06,
+      "loss": 0.7175,
+      "step": 3567
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.055378247546218e-06,
+      "loss": 0.5223,
+      "step": 3568
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.027661302928867e-06,
+      "loss": 0.6253,
+      "step": 3569
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.000005966197387e-06,
+      "loss": 0.568,
+      "step": 3570
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.972412255482473e-06,
+      "loss": 0.7352,
+      "step": 3571
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.9448801888744795e-06,
+      "loss": 0.6638,
+      "step": 3572
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.917409784423334e-06,
+      "loss": 0.7207,
+      "step": 3573
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.8900010601384835e-06,
+      "loss": 0.6272,
+      "step": 3574
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.86265403398899e-06,
+      "loss": 0.793,
+      "step": 3575
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.835368723903456e-06,
+      "loss": 0.7517,
+      "step": 3576
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.808145147770016e-06,
+      "loss": 0.6889,
+      "step": 3577
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.780983323436373e-06,
+      "loss": 0.7372,
+      "step": 3578
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7538832687096586e-06,
+      "loss": 0.6594,
+      "step": 3579
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.726845001356573e-06,
+      "loss": 0.5621,
+      "step": 3580
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.699868539103326e-06,
+      "loss": 0.6722,
+      "step": 3581
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.672953899635525e-06,
+      "loss": 0.8223,
+      "step": 3582
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.646101100598344e-06,
+      "loss": 0.6327,
+      "step": 3583
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6193101595963585e-06,
+      "loss": 0.7267,
+      "step": 3584
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.592581094193583e-06,
+      "loss": 0.7317,
+      "step": 3585
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.565913921913513e-06,
+      "loss": 0.6914,
+      "step": 3586
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5393086602390066e-06,
+      "loss": 0.689,
+      "step": 3587
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.512765326612379e-06,
+      "loss": 0.5784,
+      "step": 3588
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.486283938435355e-06,
+      "loss": 0.7296,
+      "step": 3589
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.45986451306899e-06,
+      "loss": 0.7332,
+      "step": 3590
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.433507067833776e-06,
+      "loss": 0.7716,
+      "step": 3591
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.407211620009544e-06,
+      "loss": 0.7337,
+      "step": 3592
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.380978186835484e-06,
+      "loss": 0.7159,
+      "step": 3593
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.354806785510113e-06,
+      "loss": 0.6871,
+      "step": 3594
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.328697433191321e-06,
+      "loss": 0.82,
+      "step": 3595
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.30265014699628e-06,
+      "loss": 0.6564,
+      "step": 3596
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.276664944001475e-06,
+      "loss": 0.6183,
+      "step": 3597
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.250741841242734e-06,
+      "loss": 0.7437,
+      "step": 3598
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.224880855715098e-06,
+      "loss": 0.7949,
+      "step": 3599
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.199082004372957e-06,
+      "loss": 0.6514,
+      "step": 3600
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.173345304129917e-06,
+      "loss": 0.7038,
+      "step": 3601
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.147670771858848e-06,
+      "loss": 0.7311,
+      "step": 3602
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1220584243918955e-06,
+      "loss": 0.6651,
+      "step": 3603
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.096508278520384e-06,
+      "loss": 0.7588,
+      "step": 3604
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.0710203509948924e-06,
+      "loss": 0.652,
+      "step": 3605
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.045594658525232e-06,
+      "loss": 0.6412,
+      "step": 3606
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.020231217780336e-06,
+      "loss": 0.7101,
+      "step": 3607
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.994930045388413e-06,
+      "loss": 0.7717,
+      "step": 3608
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.969691157936796e-06,
+      "loss": 0.7117,
+      "step": 3609
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.944514571971981e-06,
+      "loss": 0.5564,
+      "step": 3610
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.919400303999655e-06,
+      "loss": 0.7413,
+      "step": 3611
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.8943483704846475e-06,
+      "loss": 0.7776,
+      "step": 3612
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.869358787850864e-06,
+      "loss": 0.6269,
+      "step": 3613
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.8444315724814115e-06,
+      "loss": 0.5717,
+      "step": 3614
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.819566740718439e-06,
+      "loss": 0.6604,
+      "step": 3615
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.794764308863242e-06,
+      "loss": 0.7864,
+      "step": 3616
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.770024293176223e-06,
+      "loss": 0.763,
+      "step": 3617
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.745346709876786e-06,
+      "loss": 0.7784,
+      "step": 3618
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7207315751434885e-06,
+      "loss": 0.7537,
+      "step": 3619
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.6961789051139124e-06,
+      "loss": 0.6485,
+      "step": 3620
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.671688715884681e-06,
+      "loss": 0.6689,
+      "step": 3621
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.647261023511451e-06,
+      "loss": 0.7189,
+      "step": 3622
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6228958440089565e-06,
+      "loss": 0.6563,
+      "step": 3623
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5985931933508754e-06,
+      "loss": 0.5758,
+      "step": 3624
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.574353087469929e-06,
+      "loss": 0.6609,
+      "step": 3625
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.550175542257862e-06,
+      "loss": 0.5894,
+      "step": 3626
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.526060573565338e-06,
+      "loss": 0.6835,
+      "step": 3627
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.502008197202068e-06,
+      "loss": 0.7067,
+      "step": 3628
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.47801842893667e-06,
+      "loss": 0.7825,
+      "step": 3629
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.454091284496731e-06,
+      "loss": 0.6789,
+      "step": 3630
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.430226779568835e-06,
+      "loss": 0.6477,
+      "step": 3631
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.406424929798403e-06,
+      "loss": 0.6441,
+      "step": 3632
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.38268575078985e-06,
+      "loss": 0.7222,
+      "step": 3633
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3590092581065055e-06,
+      "loss": 0.709,
+      "step": 3634
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.335395467270553e-06,
+      "loss": 0.7128,
+      "step": 3635
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.311844393763109e-06,
+      "loss": 0.6408,
+      "step": 3636
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.28835605302419e-06,
+      "loss": 0.6736,
+      "step": 3637
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.26493046045261e-06,
+      "loss": 0.7049,
+      "step": 3638
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.241567631406118e-06,
+      "loss": 0.6938,
+      "step": 3639
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2182675812012965e-06,
+      "loss": 0.5635,
+      "step": 3640
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.195030325113547e-06,
+      "loss": 0.6491,
+      "step": 3641
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.17185587837714e-06,
+      "loss": 0.7619,
+      "step": 3642
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.148744256185122e-06,
+      "loss": 0.6213,
+      "step": 3643
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.125695473689406e-06,
+      "loss": 0.6675,
+      "step": 3644
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.102709546000671e-06,
+      "loss": 0.6653,
+      "step": 3645
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.0797864881883975e-06,
+      "loss": 0.7361,
+      "step": 3646
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0569263152808405e-06,
+      "loss": 0.6684,
+      "step": 3647
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.034129042265066e-06,
+      "loss": 0.6394,
+      "step": 3648
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.011394684086845e-06,
+      "loss": 0.5542,
+      "step": 3649
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.988723255650728e-06,
+      "loss": 0.7499,
+      "step": 3650
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.966114771820051e-06,
+      "loss": 0.6807,
+      "step": 3651
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.943569247416801e-06,
+      "loss": 0.6941,
+      "step": 3652
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.921086697221743e-06,
+      "loss": 0.717,
+      "step": 3653
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.898667135974376e-06,
+      "loss": 0.7237,
+      "step": 3654
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.876310578372832e-06,
+      "loss": 0.6331,
+      "step": 3655
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.854017039074009e-06,
+      "loss": 0.619,
+      "step": 3656
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.831786532693449e-06,
+      "loss": 0.6666,
+      "step": 3657
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8096190738053818e-06,
+      "loss": 0.8119,
+      "step": 3658
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.787514676942716e-06,
+      "loss": 0.738,
+      "step": 3659
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7654733565969826e-06,
+      "loss": 0.8414,
+      "step": 3660
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7434951272183994e-06,
+      "loss": 0.6529,
+      "step": 3661
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7215800032158075e-06,
+      "loss": 0.6005,
+      "step": 3662
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.699727998956659e-06,
+      "loss": 0.7136,
+      "step": 3663
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6779391287670494e-06,
+      "loss": 0.7898,
+      "step": 3664
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6562134069316854e-06,
+      "loss": 0.6859,
+      "step": 3665
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6345508476938297e-06,
+      "loss": 0.6669,
+      "step": 3666
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6129514652553986e-06,
+      "loss": 0.6928,
+      "step": 3667
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.591415273776855e-06,
+      "loss": 0.7278,
+      "step": 3668
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.569942287377226e-06,
+      "loss": 0.8252,
+      "step": 3669
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.548532520134129e-06,
+      "loss": 0.6575,
+      "step": 3670
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.527185986083703e-06,
+      "loss": 0.7622,
+      "step": 3671
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5059026992206647e-06,
+      "loss": 0.6192,
+      "step": 3672
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.484682673498263e-06,
+      "loss": 0.7076,
+      "step": 3673
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4635259228282256e-06,
+      "loss": 0.6673,
+      "step": 3674
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.442432461080858e-06,
+      "loss": 0.6555,
+      "step": 3675
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.421402302084953e-06,
+      "loss": 0.6705,
+      "step": 3676
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.400435459627782e-06,
+      "loss": 0.7799,
+      "step": 3677
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.379531947455128e-06,
+      "loss": 0.7502,
+      "step": 3678
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3586917792712615e-06,
+      "loss": 0.6775,
+      "step": 3679
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3379149687388867e-06,
+      "loss": 0.7785,
+      "step": 3680
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.317201529479208e-06,
+      "loss": 0.6853,
+      "step": 3681
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2965514750718963e-06,
+      "loss": 0.6687,
+      "step": 3682
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2759648190549996e-06,
+      "loss": 0.5891,
+      "step": 3683
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2554415749250888e-06,
+      "loss": 0.7912,
+      "step": 3684
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.23498175613709e-06,
+      "loss": 0.7934,
+      "step": 3685
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.214585376104384e-06,
+      "loss": 0.6664,
+      "step": 3686
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1942524481987757e-06,
+      "loss": 0.7818,
+      "step": 3687
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1739829857504234e-06,
+      "loss": 0.5997,
+      "step": 3688
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1537770020479196e-06,
+      "loss": 0.6982,
+      "step": 3689
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1336345103382346e-06,
+      "loss": 0.6711,
+      "step": 3690
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1135555238266722e-06,
+      "loss": 0.6664,
+      "step": 3691
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.093540055676958e-06,
+      "loss": 0.7396,
+      "step": 3692
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0735881190111617e-06,
+      "loss": 0.7169,
+      "step": 3693
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.053699726909676e-06,
+      "loss": 0.657,
+      "step": 3694
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0338748924112483e-06,
+      "loss": 0.6877,
+      "step": 3695
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.014113628512982e-06,
+      "loss": 0.652,
+      "step": 3696
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.994415948170237e-06,
+      "loss": 0.6903,
+      "step": 3697
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9747818642967827e-06,
+      "loss": 0.639,
+      "step": 3698
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.955211389764623e-06,
+      "loss": 0.6287,
+      "step": 3699
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9357045374040825e-06,
+      "loss": 0.649,
+      "step": 3700
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.916261320003777e-06,
+      "loss": 0.802,
+      "step": 3701
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.8968817503105983e-06,
+      "loss": 0.6164,
+      "step": 3702
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8775658410297277e-06,
+      "loss": 0.6589,
+      "step": 3703
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8583136048245697e-06,
+      "loss": 0.6475,
+      "step": 3704
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.839125054316838e-06,
+      "loss": 0.6043,
+      "step": 3705
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.820000202086459e-06,
+      "loss": 0.7104,
+      "step": 3706
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8009390606716035e-06,
+      "loss": 0.7408,
+      "step": 3707
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.781941642568686e-06,
+      "loss": 0.6798,
+      "step": 3708
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7630079602323442e-06,
+      "loss": 0.6394,
+      "step": 3709
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7441380260754048e-06,
+      "loss": 0.6268,
+      "step": 3710
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7253318524689262e-06,
+      "loss": 0.76,
+      "step": 3711
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.706589451742181e-06,
+      "loss": 0.7018,
+      "step": 3712
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6879108361825835e-06,
+      "loss": 0.6948,
+      "step": 3713
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6692960180357716e-06,
+      "loss": 0.6427,
+      "step": 3714
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6507450095055618e-06,
+      "loss": 0.7047,
+      "step": 3715
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.632257822753881e-06,
+      "loss": 0.6712,
+      "step": 3716
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.613834469900889e-06,
+      "loss": 0.896,
+      "step": 3717
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5954749630248353e-06,
+      "loss": 0.6832,
+      "step": 3718
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.577179314162148e-06,
+      "loss": 0.7184,
+      "step": 3719
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5589475353073988e-06,
+      "loss": 0.7956,
+      "step": 3720
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5407796384132487e-06,
+      "loss": 0.7651,
+      "step": 3721
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.522675635390492e-06,
+      "loss": 0.746,
+      "step": 3722
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5046355381080578e-06,
+      "loss": 0.6629,
+      "step": 3723
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.486659358392951e-06,
+      "loss": 0.7833,
+      "step": 3724
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4687471080302894e-06,
+      "loss": 0.7059,
+      "step": 3725
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.450898798763268e-06,
+      "loss": 0.7012,
+      "step": 3726
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4331144422931608e-06,
+      "loss": 0.75,
+      "step": 3727
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.415394050279318e-06,
+      "loss": 0.6988,
+      "step": 3728
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.397737634339181e-06,
+      "loss": 0.7264,
+      "step": 3729
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.380145206048201e-06,
+      "loss": 0.7137,
+      "step": 3730
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.362616776939919e-06,
+      "loss": 0.6452,
+      "step": 3731
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3451523585058754e-06,
+      "loss": 0.6698,
+      "step": 3732
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.32775196219569e-06,
+      "loss": 0.6319,
+      "step": 3733
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.310415599417004e-06,
+      "loss": 0.6905,
+      "step": 3734
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2931432815354594e-06,
+      "loss": 0.6685,
+      "step": 3735
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2759350198746976e-06,
+      "loss": 0.7455,
+      "step": 3736
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.258790825716417e-06,
+      "loss": 0.6826,
+      "step": 3737
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.24171071030026e-06,
+      "loss": 0.68,
+      "step": 3738
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2246946848238804e-06,
+      "loss": 0.6061,
+      "step": 3739
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2077427604429433e-06,
+      "loss": 0.6656,
+      "step": 3740
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.190854948271026e-06,
+      "loss": 0.5905,
+      "step": 3741
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.1740312593797273e-06,
+      "loss": 0.7855,
+      "step": 3742
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1572717047986025e-06,
+      "loss": 0.7919,
+      "step": 3743
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1405762955151176e-06,
+      "loss": 0.721,
+      "step": 3744
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1239450424747508e-06,
+      "loss": 0.5418,
+      "step": 3745
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.107377956580847e-06,
+      "loss": 0.7518,
+      "step": 3746
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0908750486947405e-06,
+      "loss": 0.765,
+      "step": 3747
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.074436329635687e-06,
+      "loss": 0.7374,
+      "step": 3748
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0580618101808115e-06,
+      "loss": 0.8086,
+      "step": 3749
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.041751501065203e-06,
+      "loss": 0.7438,
+      "step": 3750
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.025505412981843e-06,
+      "loss": 0.6736,
+      "step": 3751
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.009323556581566e-06,
+      "loss": 0.5604,
+      "step": 3752
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.993205942473164e-06,
+      "loss": 0.6589,
+      "step": 3753
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.977152581223274e-06,
+      "loss": 0.6693,
+      "step": 3754
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9611634833564096e-06,
+      "loss": 0.666,
+      "step": 3755
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.945238659354953e-06,
+      "loss": 0.6172,
+      "step": 3756
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9293781196591755e-06,
+      "loss": 0.6339,
+      "step": 3757
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9135818746671586e-06,
+      "loss": 0.6054,
+      "step": 3758
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8978499347348743e-06,
+      "loss": 0.7465,
+      "step": 3759
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.882182310176095e-06,
+      "loss": 0.7345,
+      "step": 3760
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8665790112624814e-06,
+      "loss": 0.701,
+      "step": 3761
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8510400482234847e-06,
+      "loss": 0.7386,
+      "step": 3762
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8355654312463666e-06,
+      "loss": 0.6527,
+      "step": 3763
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8201551704762453e-06,
+      "loss": 0.6804,
+      "step": 3764
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8048092760160285e-06,
+      "loss": 0.6119,
+      "step": 3765
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.7895277579264014e-06,
+      "loss": 0.617,
+      "step": 3766
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7743106262258835e-06,
+      "loss": 0.6905,
+      "step": 3767
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7591578908907724e-06,
+      "loss": 0.6577,
+      "step": 3768
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7440695618551106e-06,
+      "loss": 0.7777,
+      "step": 3769
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.729045649010752e-06,
+      "loss": 0.5481,
+      "step": 3770
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7140861622073511e-06,
+      "loss": 0.6004,
+      "step": 3771
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6991911112522407e-06,
+      "loss": 0.8025,
+      "step": 3772
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.684360505910576e-06,
+      "loss": 0.7217,
+      "step": 3773
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6695943559052462e-06,
+      "loss": 0.661,
+      "step": 3774
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6548926709168633e-06,
+      "loss": 0.5793,
+      "step": 3775
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6402554605838172e-06,
+      "loss": 0.6152,
+      "step": 3776
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6256827345021764e-06,
+      "loss": 0.6295,
+      "step": 3777
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6111745022257874e-06,
+      "loss": 0.7411,
+      "step": 3778
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.596730773266175e-06,
+      "loss": 0.6035,
+      "step": 3779
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5823515570925763e-06,
+      "loss": 0.7556,
+      "step": 3780
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.568036863131972e-06,
+      "loss": 0.601,
+      "step": 3781
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.553786700769011e-06,
+      "loss": 0.6925,
+      "step": 3782
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5396010793460314e-06,
+      "loss": 0.7466,
+      "step": 3783
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5254800081630826e-06,
+      "loss": 0.6275,
+      "step": 3784
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5114234964778707e-06,
+      "loss": 0.6086,
+      "step": 3785
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4974315535058015e-06,
+      "loss": 0.7127,
+      "step": 3786
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4835041884199263e-06,
+      "loss": 0.5898,
+      "step": 3787
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4696414103509636e-06,
+      "loss": 0.6461,
+      "step": 3788
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.455843228387299e-06,
+      "loss": 0.7701,
+      "step": 3789
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4421096515749855e-06,
+      "loss": 0.7585,
+      "step": 3790
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4284406889176649e-06,
+      "loss": 0.7133,
+      "step": 3791
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4148363493766802e-06,
+      "loss": 0.6809,
+      "step": 3792
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.401296641870975e-06,
+      "loss": 0.7032,
+      "step": 3793
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3878215752771262e-06,
+      "loss": 0.6774,
+      "step": 3794
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3744111584293228e-06,
+      "loss": 0.6861,
+      "step": 3795
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.361065400119399e-06,
+      "loss": 0.8521,
+      "step": 3796
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3477843090967667e-06,
+      "loss": 0.742,
+      "step": 3797
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3345678940684613e-06,
+      "loss": 0.5918,
+      "step": 3798
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3214161636991184e-06,
+      "loss": 0.7435,
+      "step": 3799
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.30832912661093e-06,
+      "loss": 0.6431,
+      "step": 3800
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2953067913837435e-06,
+      "loss": 0.763,
+      "step": 3801
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2823491665549193e-06,
+      "loss": 0.741,
+      "step": 3802
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2694562606194392e-06,
+      "loss": 0.6664,
+      "step": 3803
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2566280820298426e-06,
+      "loss": 0.7967,
+      "step": 3804
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2438646391962129e-06,
+      "loss": 0.6042,
+      "step": 3805
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.231165940486234e-06,
+      "loss": 0.6135,
+      "step": 3806
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2185319942251027e-06,
+      "loss": 0.7164,
+      "step": 3807
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2059628086956044e-06,
+      "loss": 0.5975,
+      "step": 3808
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.193458392138025e-06,
+      "loss": 0.7094,
+      "step": 3809
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1810187527502182e-06,
+      "loss": 0.7137,
+      "step": 3810
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1686438986875715e-06,
+      "loss": 0.8664,
+      "step": 3811
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1563338380629618e-06,
+      "loss": 0.615,
+      "step": 3812
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1440885789468558e-06,
+      "loss": 0.7204,
+      "step": 3813
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1319081293671541e-06,
+      "loss": 0.6456,
+      "step": 3814
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1197924973093464e-06,
+      "loss": 0.7285,
+      "step": 3815
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1077416907163574e-06,
+      "loss": 0.7233,
+      "step": 3816
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0957557174886778e-06,
+      "loss": 0.6879,
+      "step": 3817
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0838345854842446e-06,
+      "loss": 0.7347,
+      "step": 3818
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0719783025185058e-06,
+      "loss": 0.636,
+      "step": 3819
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0601868763643996e-06,
+      "loss": 0.6901,
+      "step": 3820
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0484603147523309e-06,
+      "loss": 0.6211,
+      "step": 3821
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0367986253701944e-06,
+      "loss": 0.6789,
+      "step": 3822
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0252018158633526e-06,
+      "loss": 0.7161,
+      "step": 3823
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0136698938346011e-06,
+      "loss": 0.6204,
+      "step": 3824
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0022028668442375e-06,
+      "loss": 0.7776,
+      "step": 3825
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.90800742410003e-07,
+      "loss": 0.5942,
+      "step": 3826
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.794635280070742e-07,
+      "loss": 0.7151,
+      "step": 3827
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.68191231068083e-07,
+      "loss": 0.6033,
+      "step": 3828
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.569838589831181e-07,
+      "loss": 0.5889,
+      "step": 3829
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.458414190996689e-07,
+      "loss": 0.6318,
+      "step": 3830
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.347639187226809e-07,
+      "loss": 0.6779,
+      "step": 3831
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.237513651145225e-07,
+      "loss": 0.6256,
+      "step": 3832
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.128037654949628e-07,
+      "loss": 0.6866,
+      "step": 3833
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.019211270412275e-07,
+      "loss": 0.6584,
+      "step": 3834
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.911034568879207e-07,
+      "loss": 0.6675,
+      "step": 3835
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.803507621270579e-07,
+      "loss": 0.7179,
+      "step": 3836
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.696630498080671e-07,
+      "loss": 0.5699,
+      "step": 3837
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.590403269377655e-07,
+      "loss": 0.7092,
+      "step": 3838
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.484826004803714e-07,
+      "loss": 0.6439,
+      "step": 3839
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.379898773574924e-07,
+      "loss": 0.7267,
+      "step": 3840
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.275621644481035e-07,
+      "loss": 0.7105,
+      "step": 3841
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.171994685885698e-07,
+      "loss": 0.8542,
+      "step": 3842
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.069017965726566e-07,
+      "loss": 0.7392,
+      "step": 3843
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.966691551514527e-07,
+      "loss": 0.5934,
+      "step": 3844
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.865015510334472e-07,
+      "loss": 0.7733,
+      "step": 3845
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.763989908844749e-07,
+      "loss": 0.7374,
+      "step": 3846
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.663614813277265e-07,
+      "loss": 0.7646,
+      "step": 3847
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.563890289437825e-07,
+      "loss": 0.6627,
+      "step": 3848
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.464816402705022e-07,
+      "loss": 0.8193,
+      "step": 3849
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.366393218031564e-07,
+      "loss": 0.7067,
+      "step": 3850
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.268620799943171e-07,
+      "loss": 0.6517,
+      "step": 3851
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.171499212539123e-07,
+      "loss": 0.7392,
+      "step": 3852
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.075028519491933e-07,
+      "loss": 0.7555,
+      "step": 3853
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.979208784047453e-07,
+      "loss": 0.6058,
+      "step": 3854
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.884040069024434e-07,
+      "loss": 0.5249,
+      "step": 3855
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.78952243681541e-07,
+      "loss": 0.6706,
+      "step": 3856
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.695655949385705e-07,
+      "loss": 0.6055,
+      "step": 3857
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.602440668273758e-07,
+      "loss": 0.722,
+      "step": 3858
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.509876654591018e-07,
+      "loss": 0.6995,
+      "step": 3859
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.41796396902239e-07,
+      "loss": 0.6126,
+      "step": 3860
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.326702671825113e-07,
+      "loss": 0.6928,
+      "step": 3861
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.236092822829887e-07,
+      "loss": 0.7388,
+      "step": 3862
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.146134481440191e-07,
+      "loss": 0.6398,
+      "step": 3863
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.056827706632185e-07,
+      "loss": 0.5682,
+      "step": 3864
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.968172556955365e-07,
+      "loss": 0.6497,
+      "step": 3865
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.880169090531351e-07,
+      "loss": 0.6825,
+      "step": 3866
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.792817365054992e-07,
+      "loss": 0.6574,
+      "step": 3867
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.706117437793701e-07,
+      "loss": 0.7725,
+      "step": 3868
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.620069365587677e-07,
+      "loss": 0.741,
+      "step": 3869
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.534673204849572e-07,
+      "loss": 0.587,
+      "step": 3870
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.449929011564936e-07,
+      "loss": 0.6444,
+      "step": 3871
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.365836841291438e-07,
+      "loss": 0.5979,
+      "step": 3872
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.282396749159868e-07,
+      "loss": 0.6551,
+      "step": 3873
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.199608789873134e-07,
+      "loss": 0.8386,
+      "step": 3874
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.117473017706486e-07,
+      "loss": 0.6598,
+      "step": 3875
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.035989486508075e-07,
+      "loss": 0.6402,
+      "step": 3876
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.955158249697945e-07,
+      "loss": 0.6729,
+      "step": 3877
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.874979360268928e-07,
+      "loss": 0.7487,
+      "step": 3878
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.795452870785866e-07,
+      "loss": 0.5709,
+      "step": 3879
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7165788333860536e-07,
+      "loss": 0.8205,
+      "step": 3880
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.638357299778906e-07,
+      "loss": 0.6635,
+      "step": 3881
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.56078832124629e-07,
+      "loss": 0.6603,
+      "step": 3882
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.483871948641971e-07,
+      "loss": 0.5385,
+      "step": 3883
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4076082323920576e-07,
+      "loss": 0.7146,
+      "step": 3884
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.331997222494777e-07,
+      "loss": 0.8316,
+      "step": 3885
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.257038968520366e-07,
+      "loss": 0.6555,
+      "step": 3886
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1827335196111815e-07,
+      "loss": 0.765,
+      "step": 3887
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1090809244814785e-07,
+      "loss": 0.864,
+      "step": 3888
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0360812314177433e-07,
+      "loss": 0.743,
+      "step": 3889
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.963734488278248e-07,
+      "loss": 0.6622,
+      "step": 3890
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.892040742493164e-07,
+      "loss": 0.8674,
+      "step": 3891
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.82100004106456e-07,
+      "loss": 0.6649,
+      "step": 3892
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.7506124305666246e-07,
+      "loss": 0.649,
+      "step": 3893
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.6808779571451126e-07,
+      "loss": 0.7281,
+      "step": 3894
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.611796666517564e-07,
+      "loss": 0.6875,
+      "step": 3895
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.543368603973529e-07,
+      "loss": 0.8206,
+      "step": 3896
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.475593814374123e-07,
+      "loss": 0.6412,
+      "step": 3897
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4084723421521356e-07,
+      "loss": 0.6884,
+      "step": 3898
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.342004231312257e-07,
+      "loss": 0.6955,
+      "step": 3899
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.2761895254306287e-07,
+      "loss": 0.6637,
+      "step": 3900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.2110282676551806e-07,
+      "loss": 0.7734,
+      "step": 3901
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.146520500705297e-07,
+      "loss": 0.6223,
+      "step": 3902
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0826662668720364e-07,
+      "loss": 0.5382,
+      "step": 3903
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.019465608018024e-07,
+      "loss": 0.6362,
+      "step": 3904
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.956918565577338e-07,
+      "loss": 0.7063,
+      "step": 3905
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8950251805553997e-07,
+      "loss": 0.5882,
+      "step": 3906
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.833785493529528e-07,
+      "loss": 0.6206,
+      "step": 3907
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.773199544648164e-07,
+      "loss": 0.67,
+      "step": 3908
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.713267373630979e-07,
+      "loss": 0.8264,
+      "step": 3909
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.6539890197695427e-07,
+      "loss": 0.7322,
+      "step": 3910
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5953645219263245e-07,
+      "loss": 0.5665,
+      "step": 3911
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.537393918535358e-07,
+      "loss": 0.6743,
+      "step": 3912
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4800772476020197e-07,
+      "loss": 0.6483,
+      "step": 3913
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.423414546702807e-07,
+      "loss": 0.5782,
+      "step": 3914
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3674058529855602e-07,
+      "loss": 0.7205,
+      "step": 3915
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.312051203169352e-07,
+      "loss": 0.665,
+      "step": 3916
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.257350633544486e-07,
+      "loss": 0.6761,
+      "step": 3917
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.2033041799723875e-07,
+      "loss": 0.51,
+      "step": 3918
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.149911877885713e-07,
+      "loss": 0.8748,
+      "step": 3919
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0971737622883515e-07,
+      "loss": 0.6093,
+      "step": 3920
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.04508986775509e-07,
+      "loss": 0.6656,
+      "step": 3921
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9936602284318373e-07,
+      "loss": 0.81,
+      "step": 3922
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9428848780358444e-07,
+      "loss": 0.725,
+      "step": 3923
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.8927638498551502e-07,
+      "loss": 0.7288,
+      "step": 3924
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.843297176748804e-07,
+      "loss": 0.6541,
+      "step": 3925
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7944848911470857e-07,
+      "loss": 0.711,
+      "step": 3926
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7463270250510645e-07,
+      "loss": 0.6442,
+      "step": 3927
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6988236100329292e-07,
+      "loss": 0.6213,
+      "step": 3928
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.651974677235546e-07,
+      "loss": 0.5859,
+      "step": 3929
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.605780257373124e-07,
+      "loss": 0.8115,
+      "step": 3930
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5602403807303266e-07,
+      "loss": 0.6165,
+      "step": 3931
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5153550771630498e-07,
+      "loss": 0.7601,
+      "step": 3932
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.4711243760978653e-07,
+      "loss": 0.6696,
+      "step": 3933
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.427548306532134e-07,
+      "loss": 0.7484,
+      "step": 3934
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3846268970344466e-07,
+      "loss": 0.7123,
+      "step": 3935
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3423601757436287e-07,
+      "loss": 0.6541,
+      "step": 3936
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3007481703696256e-07,
+      "loss": 0.6967,
+      "step": 3937
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2597909081931702e-07,
+      "loss": 0.5944,
+      "step": 3938
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.219488416065673e-07,
+      "loss": 0.66,
+      "step": 3939
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.179840720409331e-07,
+      "loss": 0.6786,
+      "step": 3940
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1408478472167971e-07,
+      "loss": 0.5755,
+      "step": 3941
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1025098220518448e-07,
+      "loss": 0.7488,
+      "step": 3942
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.064826670048702e-07,
+      "loss": 0.7736,
+      "step": 3943
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0277984159122733e-07,
+      "loss": 0.7101,
+      "step": 3944
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.914250839180294e-08,
+      "loss": 0.6806,
+      "step": 3945
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.557066979123397e-08,
+      "loss": 0.7184,
+      "step": 3946
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.20643281312028e-08,
+      "loss": 0.612,
+      "step": 3947
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.862348571043733e-08,
+      "loss": 0.7672,
+      "step": 3948
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.524814478474419e-08,
+      "loss": 0.6917,
+      "step": 3949
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.193830756699772e-08,
+      "loss": 0.7284,
+      "step": 3950
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.869397622710661e-08,
+      "loss": 0.7737,
+      "step": 3951
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.551515289203615e-08,
+      "loss": 0.6833,
+      "step": 3952
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.240183964580816e-08,
+      "loss": 0.6684,
+      "step": 3953
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.935403852950107e-08,
+      "loss": 0.6261,
+      "step": 3954
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.637175154124986e-08,
+      "loss": 0.6541,
+      "step": 3955
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.34549806362239e-08,
+      "loss": 0.657,
+      "step": 3956
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.060372772663802e-08,
+      "loss": 0.5731,
+      "step": 3957
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.7817994681774735e-08,
+      "loss": 0.7464,
+      "step": 3958
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.50977833279509e-08,
+      "loss": 0.8479,
+      "step": 3959
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.2443095448506674e-08,
+      "loss": 0.6177,
+      "step": 3960
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.985393278386097e-08,
+      "loss": 0.7827,
+      "step": 3961
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.7330297031467075e-08,
+      "loss": 0.6983,
+      "step": 3962
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.487218984577934e-08,
+      "loss": 0.7159,
+      "step": 3963
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.247961283835311e-08,
+      "loss": 0.6281,
+      "step": 3964
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.015256757774477e-08,
+      "loss": 0.72,
+      "step": 3965
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.789105558954509e-08,
+      "loss": 0.6323,
+      "step": 3966
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.56950783564014e-08,
+      "loss": 0.6879,
+      "step": 3967
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.356463731798432e-08,
+      "loss": 0.6421,
+      "step": 3968
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1499733870998807e-08,
+      "loss": 0.7646,
+      "step": 3969
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.9500369369195312e-08,
+      "loss": 0.7669,
+      "step": 3970
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7566545123347553e-08,
+      "loss": 0.707,
+      "step": 3971
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5698262401263605e-08,
+      "loss": 0.6269,
+      "step": 3972
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.389552242777482e-08,
+      "loss": 0.7364,
+      "step": 3973
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.215832638474691e-08,
+      "loss": 0.7645,
+      "step": 3974
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0486675411102163e-08,
+      "loss": 0.7415,
+      "step": 3975
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.888057060274173e-08,
+      "loss": 0.7344,
+      "step": 3976
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.734001301263444e-08,
+      "loss": 0.7766,
+      "step": 3977
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5865003650761266e-08,
+      "loss": 0.7109,
+      "step": 3978
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.445554348413758e-08,
+      "loss": 0.664,
+      "step": 3979
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3111633436779791e-08,
+      "loss": 0.669,
+      "step": 3980
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.18332743897831e-08,
+      "loss": 0.6344,
+      "step": 3981
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0620467181210459e-08,
+      "loss": 0.7265,
+      "step": 3982
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.473212606170289e-09,
+      "loss": 0.6268,
+      "step": 3983
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.391511416816489e-09,
+      "loss": 0.6244,
+      "step": 3984
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.375364322292911e-09,
+      "loss": 0.7479,
+      "step": 3985
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.42477198878888e-09,
+      "loss": 0.684,
+      "step": 3986
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.539735039505889e-09,
+      "loss": 0.6448,
+      "step": 3987
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.720254054679796e-09,
+      "loss": 0.858,
+      "step": 3988
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.96632957153642e-09,
+      "loss": 0.7484,
+      "step": 3989
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.2779620843692572e-09,
+      "loss": 0.6821,
+      "step": 3990
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.6551520444617616e-09,
+      "loss": 0.7201,
+      "step": 3991
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0978998601206556e-09,
+      "loss": 0.6425,
+      "step": 3992
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6062058966870297e-09,
+      "loss": 0.7212,
+      "step": 3993
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1800704765030368e-09,
+      "loss": 0.7377,
+      "step": 3994
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.194938789451989e-10,
+      "loss": 0.6774,
+      "step": 3995
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.244763404133046e-10,
+      "loss": 0.6424,
+      "step": 3996
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.950180543082048e-10,
+      "loss": 0.7422,
+      "step": 3997
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.311191710651194e-10,
+      "loss": 0.6551,
+      "step": 3998
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.277979814253485e-11,
+      "loss": 0.7711,
+      "step": 3999
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.6624,
+      "step": 4000
+    },
+    {
+      "epoch": 5.0,
+      "step": 4000,
+      "total_flos": 56180610244608.0,
+      "train_loss": 0.7672133717834949,
+      "train_runtime": 10232.3403,
+      "train_samples_per_second": 6.256,
+      "train_steps_per_second": 0.391
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 50000,
+  "total_flos": 56180610244608.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}