diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9319 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9444880037635253,
+  "eval_steps": 500,
+  "global_step": 1550,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.4131,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.0602,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.962406251802891e-06,
+      "loss": 1.0468,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.9349,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.804820237218406e-06,
+      "loss": 0.8473,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.462406251802891e-06,
+      "loss": 0.801,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.018387305144011e-06,
+      "loss": 0.759,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.499999999999999e-06,
+      "loss": 0.7606,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.924812503605782e-06,
+      "loss": 0.6934,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.304820237218408e-06,
+      "loss": 0.6736,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.648579046593244e-06,
+      "loss": 0.6872,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.96240625180289e-06,
+      "loss": 0.6925,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.25109929535273e-06,
+      "loss": 0.6246,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.51838730514401e-06,
+      "loss": 0.6142,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.767226489021297e-06,
+      "loss": 0.6569,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.6093,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.5945,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.5672,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.5773,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5756,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5813,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5734,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5704,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5841,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5685,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.6075,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5396,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 0.5303,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 0.5325,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 0.5838,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 0.5663,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 0.5584,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 0.554,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 0.5374,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 0.5642,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.5294,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.5163,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.5464,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.5321,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.5236,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.552,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.5455,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.4956,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.5388,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.5523,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.5352,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.5225,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.5134,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.5058,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.5187,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.4732,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.5154,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.4733,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.4969,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.5198,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.4703,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.5203,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.4452,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.4923,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4909,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4906,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4775,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4883,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4951,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4821,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4866,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4498,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 0.4648,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 0.4492,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 0.4354,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 0.4374,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 0.4744,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 0.4603,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 0.4814,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 0.4786,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.4874,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.4413,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.4679,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.4551,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.4407,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.4367,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.4708,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.4518,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 0.4484,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 0.4465,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 0.4292,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 0.4528,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 0.462,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 0.4487,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 0.4593,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 0.4365,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.4775,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.438,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.4476,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.4617,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.4196,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.4281,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.4262,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.4222,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 0.4691,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 0.4281,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 0.4185,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 0.4566,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 0.4061,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 0.4161,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 0.4577,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 0.4393,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 0.4262,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 0.4063,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 0.4203,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 0.4377,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 0.4075,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 0.443,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 0.4103,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 0.4401,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 0.3772,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 0.4577,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 0.467,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 0.4007,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 0.4201,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 0.4139,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 0.4367,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 0.4018,
+      "step": 123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 0.4078,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 0.4107,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 0.4291,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 0.4336,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 0.4301,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 0.4174,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 0.4167,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 0.4047,
+      "step": 131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 0.4073,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 0.4038,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 0.3888,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 0.3988,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 0.3749,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 0.4132,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 0.4041,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 0.4302,
+      "step": 139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 0.429,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 0.3916,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 0.4019,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 0.3886,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 0.3769,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 0.382,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 0.4278,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 0.3999,
+      "step": 147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 0.3869,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 0.3717,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 0.4117,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 0.3936,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 0.3935,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 0.3948,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 0.3932,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 0.402,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 0.3583,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 0.4149,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 0.3876,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 0.3909,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 0.3768,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 0.3712,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 0.3635,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 0.4109,
+      "step": 163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 0.3977,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 0.385,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 0.3829,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 0.3636,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 0.3546,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 0.3649,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 0.3494,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 0.3783,
+      "step": 171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 0.4177,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 0.3754,
+      "step": 173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 0.3823,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 0.3776,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 0.3672,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 0.3663,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 0.3905,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 0.3604,
+      "step": 179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.3837,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.3728,
+      "step": 181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.3501,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.3428,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.3293,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.3902,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.3768,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.3918,
+      "step": 187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 0.3633,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 0.3813,
+      "step": 189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 0.3749,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 0.3546,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 0.3642,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 0.3598,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 0.3704,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 0.3639,
+      "step": 195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 0.3623,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 0.3508,
+      "step": 197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 0.346,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 0.3268,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 0.3258,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 0.3605,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 0.3709,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 0.3828,
+      "step": 203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 0.3575,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 0.352,
+      "step": 205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 0.3637,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 0.3427,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 0.3529,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 0.346,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 0.3361,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 0.3681,
+      "step": 211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 0.3577,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 0.3603,
+      "step": 213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 0.332,
+      "step": 214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 0.3293,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 0.337,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 0.3614,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 0.3189,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 0.3725,
+      "step": 219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 0.3281,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 0.3626,
+      "step": 221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 0.3648,
+      "step": 222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 0.3666,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 0.3674,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 0.3359,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 0.3303,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 0.3472,
+      "step": 227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 0.3569,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 0.3454,
+      "step": 229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 0.3324,
+      "step": 230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 0.3489,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 0.3451,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 0.3263,
+      "step": 233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 0.3386,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 0.3521,
+      "step": 235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 0.349,
+      "step": 236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 0.3581,
+      "step": 237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 0.3448,
+      "step": 238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 0.3115,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 0.338,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 0.3427,
+      "step": 241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 0.3261,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 0.3266,
+      "step": 243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 0.3632,
+      "step": 244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 0.3167,
+      "step": 245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 0.3288,
+      "step": 246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 0.3334,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 0.3447,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 0.3148,
+      "step": 249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 0.3273,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 0.3265,
+      "step": 251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 0.3316,
+      "step": 252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 0.327,
+      "step": 253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 0.3281,
+      "step": 254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 0.3115,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 0.3216,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 0.3242,
+      "step": 257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 0.3224,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 0.3398,
+      "step": 259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 0.3397,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 0.3287,
+      "step": 261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 0.3158,
+      "step": 262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 0.3241,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 0.3434,
+      "step": 264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 0.3353,
+      "step": 265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 0.3446,
+      "step": 266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 0.3061,
+      "step": 267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 0.3071,
+      "step": 268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 0.3191,
+      "step": 269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 0.316,
+      "step": 270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 0.3083,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 0.3259,
+      "step": 272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 0.3177,
+      "step": 273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 0.3167,
+      "step": 274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 0.3268,
+      "step": 275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 0.334,
+      "step": 276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 0.3313,
+      "step": 277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 0.3275,
+      "step": 278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 0.295,
+      "step": 279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 0.3157,
+      "step": 280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 0.3191,
+      "step": 281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 0.3068,
+      "step": 282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 0.3121,
+      "step": 283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 0.301,
+      "step": 284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 0.3009,
+      "step": 285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 0.3245,
+      "step": 286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 0.3339,
+      "step": 287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 0.292,
+      "step": 288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 0.3175,
+      "step": 289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 0.3037,
+      "step": 290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 0.2939,
+      "step": 291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 0.3065,
+      "step": 292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 0.2838,
+      "step": 293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 0.2955,
+      "step": 294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 0.2738,
+      "step": 295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 0.2935,
+      "step": 296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 0.3105,
+      "step": 297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 0.3226,
+      "step": 298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 0.2813,
+      "step": 299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 0.2952,
+      "step": 300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 0.3144,
+      "step": 301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 0.352,
+      "step": 302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 0.3194,
+      "step": 303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 0.317,
+      "step": 304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 0.3094,
+      "step": 305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 0.2962,
+      "step": 306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 0.2917,
+      "step": 307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 0.3334,
+      "step": 308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 0.3009,
+      "step": 309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 0.2777,
+      "step": 310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 0.3102,
+      "step": 311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 0.3084,
+      "step": 312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 0.3067,
+      "step": 313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 0.3313,
+      "step": 314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 0.3115,
+      "step": 315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 0.2766,
+      "step": 316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 0.3271,
+      "step": 317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 0.2968,
+      "step": 318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 0.3102,
+      "step": 319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 0.3173,
+      "step": 320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 0.2888,
+      "step": 321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 0.3088,
+      "step": 322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 0.2759,
+      "step": 323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 0.2929,
+      "step": 324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 0.3062,
+      "step": 325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 0.2729,
+      "step": 326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 0.2759,
+      "step": 327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 0.2889,
+      "step": 328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 0.3015,
+      "step": 329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 0.314,
+      "step": 330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 0.2998,
+      "step": 331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 0.3129,
+      "step": 332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 0.294,
+      "step": 333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 0.3079,
+      "step": 334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 0.2971,
+      "step": 335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 0.2964,
+      "step": 336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 0.3056,
+      "step": 337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 0.2926,
+      "step": 338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 0.2886,
+      "step": 339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 0.2969,
+      "step": 340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 0.2813,
+      "step": 341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 0.2878,
+      "step": 342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 0.2708,
+      "step": 343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 0.2807,
+      "step": 344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 0.3149,
+      "step": 345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 0.281,
+      "step": 346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 0.2837,
+      "step": 347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 0.3274,
+      "step": 348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 0.2797,
+      "step": 349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 0.2915,
+      "step": 350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 0.2737,
+      "step": 351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 0.2838,
+      "step": 352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 0.2774,
+      "step": 353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 0.2986,
+      "step": 354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 0.254,
+      "step": 355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 0.292,
+      "step": 356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 0.3026,
+      "step": 357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 0.3035,
+      "step": 358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 0.2844,
+      "step": 359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 0.2996,
+      "step": 360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 0.2917,
+      "step": 361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 0.3196,
+      "step": 362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 0.2631,
+      "step": 363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 0.2686,
+      "step": 364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 0.2784,
+      "step": 365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 0.2748,
+      "step": 366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 0.2775,
+      "step": 367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 0.2734,
+      "step": 368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 0.2899,
+      "step": 369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 0.266,
+      "step": 370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 0.2995,
+      "step": 371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 0.2793,
+      "step": 372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 0.2909,
+      "step": 373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 0.278,
+      "step": 374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 0.2843,
+      "step": 375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 0.2709,
+      "step": 376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 0.2726,
+      "step": 377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 0.2811,
+      "step": 378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 0.2735,
+      "step": 379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 0.2876,
+      "step": 380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 0.2903,
+      "step": 381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 0.2976,
+      "step": 382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 0.2813,
+      "step": 383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 0.2605,
+      "step": 384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 0.2604,
+      "step": 385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 0.2543,
+      "step": 386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 0.2528,
+      "step": 387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 0.2769,
+      "step": 388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 0.2931,
+      "step": 389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 0.3047,
+      "step": 390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 0.2642,
+      "step": 391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 0.2856,
+      "step": 392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 0.2929,
+      "step": 393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 0.2795,
+      "step": 394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2723,
+      "step": 395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2692,
+      "step": 396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2759,
+      "step": 397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2589,
+      "step": 398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2693,
+      "step": 399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2682,
+      "step": 400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2733,
+      "step": 401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2996,
+      "step": 402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.277,
+      "step": 403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.2544,
+      "step": 404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.2285,
+      "step": 405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.2826,
+      "step": 406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.2731,
+      "step": 407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.2618,
+      "step": 408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.2766,
+      "step": 409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.2511,
+      "step": 410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2605,
+      "step": 411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2715,
+      "step": 412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2611,
+      "step": 413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2418,
+      "step": 414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2844,
+      "step": 415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2843,
+      "step": 416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2765,
+      "step": 417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2545,
+      "step": 418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 0.2752,
+      "step": 419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 0.2509,
+      "step": 420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 0.2832,
+      "step": 421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 0.2608,
+      "step": 422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 0.275,
+      "step": 423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 0.2649,
+      "step": 424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 0.3044,
+      "step": 425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 0.2753,
+      "step": 426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 0.2649,
+      "step": 427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 0.2473,
+      "step": 428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 0.252,
+      "step": 429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 0.2668,
+      "step": 430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 0.253,
+      "step": 431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 0.2721,
+      "step": 432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 0.2864,
+      "step": 433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 0.2689,
+      "step": 434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2493,
+      "step": 435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2634,
+      "step": 436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2616,
+      "step": 437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2825,
+      "step": 438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2778,
+      "step": 439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2788,
+      "step": 440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2624,
+      "step": 441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2622,
+      "step": 442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2901,
+      "step": 443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2629,
+      "step": 444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2527,
+      "step": 445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2542,
+      "step": 446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2737,
+      "step": 447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2349,
+      "step": 448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2631,
+      "step": 449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 0.289,
+      "step": 450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 0.2767,
+      "step": 451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 0.2708,
+      "step": 452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 0.2778,
+      "step": 453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 0.252,
+      "step": 454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 0.2602,
+      "step": 455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 0.249,
+      "step": 456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 0.2745,
+      "step": 457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 0.2435,
+      "step": 458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2593,
+      "step": 459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2946,
+      "step": 460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2679,
+      "step": 461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2707,
+      "step": 462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2512,
+      "step": 463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2777,
+      "step": 464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2602,
+      "step": 465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2551,
+      "step": 466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 0.2431,
+      "step": 467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 0.2727,
+      "step": 468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 0.2618,
+      "step": 469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 0.2712,
+      "step": 470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 0.242,
+      "step": 471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 0.268,
+      "step": 472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 0.2401,
+      "step": 473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 0.2886,
+      "step": 474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 0.2713,
+      "step": 475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 0.266,
+      "step": 476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 0.2455,
+      "step": 477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 0.2546,
+      "step": 478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 0.2644,
+      "step": 479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 0.2563,
+      "step": 480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 0.2675,
+      "step": 481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 0.264,
+      "step": 482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 0.2692,
+      "step": 483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 0.2768,
+      "step": 484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 0.2564,
+      "step": 485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 0.2833,
+      "step": 486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 0.2568,
+      "step": 487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 0.2543,
+      "step": 488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 0.2516,
+      "step": 489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 0.2802,
+      "step": 490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 0.2569,
+      "step": 491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 0.2396,
+      "step": 492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 0.2678,
+      "step": 493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 0.2519,
+      "step": 494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 0.2558,
+      "step": 495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 0.2641,
+      "step": 496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 0.2653,
+      "step": 497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 0.2518,
+      "step": 498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 0.2581,
+      "step": 499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 0.2518,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 0.2459,
+      "step": 501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 0.2782,
+      "step": 502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 0.248,
+      "step": 503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 0.2533,
+      "step": 504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 0.2546,
+      "step": 505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 0.2396,
+      "step": 506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 0.2393,
+      "step": 507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 0.25,
+      "step": 508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 0.2427,
+      "step": 509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 0.2689,
+      "step": 510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 0.2488,
+      "step": 511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 0.2698,
+      "step": 512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 0.2494,
+      "step": 513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 0.2569,
+      "step": 514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2358,
+      "step": 515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2461,
+      "step": 516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2406,
+      "step": 517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2523,
+      "step": 518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2336,
+      "step": 519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2403,
+      "step": 520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2571,
+      "step": 521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2445,
+      "step": 522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 0.2476,
+      "step": 523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 0.2763,
+      "step": 524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 0.2298,
+      "step": 525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 0.2541,
+      "step": 526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 0.2681,
+      "step": 527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 0.2405,
+      "step": 528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 0.259,
+      "step": 529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 0.2507,
+      "step": 530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 0.2335,
+      "step": 531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 0.2567,
+      "step": 532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 0.2498,
+      "step": 533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 0.2748,
+      "step": 534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 0.2615,
+      "step": 535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 0.2622,
+      "step": 536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 0.2764,
+      "step": 537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 0.2778,
+      "step": 538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 0.2479,
+      "step": 539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 0.2408,
+      "step": 540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 0.2388,
+      "step": 541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 0.2882,
+      "step": 542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 0.2615,
+      "step": 543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 0.2554,
+      "step": 544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 0.253,
+      "step": 545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 0.2263,
+      "step": 546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 0.2589,
+      "step": 547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 0.2536,
+      "step": 548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 0.2463,
+      "step": 549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 0.2316,
+      "step": 550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 0.2528,
+      "step": 551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 0.2575,
+      "step": 552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 0.2523,
+      "step": 553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 0.2655,
+      "step": 554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 0.2379,
+      "step": 555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 0.2382,
+      "step": 556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 0.2383,
+      "step": 557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 0.2442,
+      "step": 558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 0.2496,
+      "step": 559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 0.2601,
+      "step": 560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 0.2504,
+      "step": 561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 0.2628,
+      "step": 562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 0.2284,
+      "step": 563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 0.2208,
+      "step": 564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 0.2495,
+      "step": 565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 0.2316,
+      "step": 566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 0.252,
+      "step": 567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 0.2615,
+      "step": 568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 0.2766,
+      "step": 569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 0.2502,
+      "step": 570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 0.2528,
+      "step": 571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 0.2405,
+      "step": 572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 0.2584,
+      "step": 573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 0.2524,
+      "step": 574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 0.2317,
+      "step": 575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 0.2415,
+      "step": 576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 0.246,
+      "step": 577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 0.2648,
+      "step": 578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 0.2403,
+      "step": 579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 0.2431,
+      "step": 580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 0.2632,
+      "step": 581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 0.246,
+      "step": 582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 0.2538,
+      "step": 583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 0.2529,
+      "step": 584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 0.2374,
+      "step": 585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2502,
+      "step": 586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2209,
+      "step": 587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2298,
+      "step": 588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2211,
+      "step": 589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2379,
+      "step": 590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2444,
+      "step": 591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2598,
+      "step": 592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2616,
+      "step": 593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 0.2851,
+      "step": 594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 0.2323,
+      "step": 595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 0.2326,
+      "step": 596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 0.2532,
+      "step": 597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 0.2548,
+      "step": 598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 0.2598,
+      "step": 599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 0.2556,
+      "step": 600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 0.2174,
+      "step": 601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 0.2437,
+      "step": 602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 0.241,
+      "step": 603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 0.2149,
+      "step": 604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 0.2261,
+      "step": 605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 0.2286,
+      "step": 606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 0.2443,
+      "step": 607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 0.2472,
+      "step": 608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 0.2506,
+      "step": 609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 0.2375,
+      "step": 610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 0.2301,
+      "step": 611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 0.2341,
+      "step": 612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 0.2393,
+      "step": 613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 0.2245,
+      "step": 614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 0.2346,
+      "step": 615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 0.2472,
+      "step": 616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 0.2313,
+      "step": 617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 0.2536,
+      "step": 618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 0.23,
+      "step": 619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 0.2405,
+      "step": 620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 0.2205,
+      "step": 621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 0.2355,
+      "step": 622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 0.2205,
+      "step": 623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 0.228,
+      "step": 624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 0.2407,
+      "step": 625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 0.2158,
+      "step": 626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 0.2237,
+      "step": 627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 0.2345,
+      "step": 628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 0.2237,
+      "step": 629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 0.2549,
+      "step": 630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 0.2584,
+      "step": 631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 0.2735,
+      "step": 632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 0.2197,
+      "step": 633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.2267,
+      "step": 634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.2252,
+      "step": 635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.2505,
+      "step": 636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.2519,
+      "step": 637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.2439,
+      "step": 638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.2432,
+      "step": 639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.2504,
+      "step": 640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.2191,
+      "step": 641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 0.2489,
+      "step": 642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 0.2306,
+      "step": 643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 0.2151,
+      "step": 644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 0.2263,
+      "step": 645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 0.2418,
+      "step": 646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 0.2377,
+      "step": 647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 0.2463,
+      "step": 648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 0.2279,
+      "step": 649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 0.235,
+      "step": 650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 0.2399,
+      "step": 651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 0.2112,
+      "step": 652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 0.2465,
+      "step": 653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 0.2495,
+      "step": 654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 0.2723,
+      "step": 655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 0.2391,
+      "step": 656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 0.224,
+      "step": 657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.2364,
+      "step": 658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.239,
+      "step": 659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.2046,
+      "step": 660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.2467,
+      "step": 661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.2189,
+      "step": 662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.2502,
+      "step": 663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.2295,
+      "step": 664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.2505,
+      "step": 665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 0.2019,
+      "step": 666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 0.2047,
+      "step": 667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 0.2467,
+      "step": 668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 0.2276,
+      "step": 669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 0.2286,
+      "step": 670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 0.2399,
+      "step": 671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 0.2183,
+      "step": 672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 0.2505,
+      "step": 673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 0.2287,
+      "step": 674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 0.2561,
+      "step": 675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 0.2395,
+      "step": 676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 0.2446,
+      "step": 677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 0.2465,
+      "step": 678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 0.2313,
+      "step": 679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 0.2508,
+      "step": 680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 0.2317,
+      "step": 681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 0.2376,
+      "step": 682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 0.2051,
+      "step": 683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 0.2443,
+      "step": 684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 0.2154,
+      "step": 685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 0.2369,
+      "step": 686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 0.2512,
+      "step": 687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 0.2652,
+      "step": 688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 0.2388,
+      "step": 689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 0.2542,
+      "step": 690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 0.2427,
+      "step": 691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 0.2232,
+      "step": 692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 0.226,
+      "step": 693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 0.2361,
+      "step": 694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 0.2441,
+      "step": 695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 0.2363,
+      "step": 696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 0.2297,
+      "step": 697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 0.2252,
+      "step": 698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 0.2269,
+      "step": 699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 0.2351,
+      "step": 700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 0.2206,
+      "step": 701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 0.2264,
+      "step": 702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 0.229,
+      "step": 703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 0.2515,
+      "step": 704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 0.2425,
+      "step": 705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 0.2574,
+      "step": 706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 0.2295,
+      "step": 707
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 0.2634,
+      "step": 708
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 0.2413,
+      "step": 709
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 0.2225,
+      "step": 710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 0.2337,
+      "step": 711
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 0.2406,
+      "step": 712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 0.2329,
+      "step": 713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 0.2446,
+      "step": 714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 0.2125,
+      "step": 715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 0.244,
+      "step": 716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 0.2348,
+      "step": 717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 0.2339,
+      "step": 718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 0.2196,
+      "step": 719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 0.2127,
+      "step": 720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 0.2404,
+      "step": 721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 0.2472,
+      "step": 722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 0.2193,
+      "step": 723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 0.2034,
+      "step": 724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 0.2205,
+      "step": 725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 0.2289,
+      "step": 726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 0.2197,
+      "step": 727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 0.2151,
+      "step": 728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 0.2389,
+      "step": 729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 0.2213,
+      "step": 730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 0.2439,
+      "step": 731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 0.2054,
+      "step": 732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 0.2197,
+      "step": 733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 0.2469,
+      "step": 734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 0.246,
+      "step": 735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 0.2211,
+      "step": 736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 0.2278,
+      "step": 737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 0.221,
+      "step": 738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 0.2088,
+      "step": 739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 0.2066,
+      "step": 740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 0.2385,
+      "step": 741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 0.2312,
+      "step": 742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 0.2381,
+      "step": 743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 0.2102,
+      "step": 744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 0.2494,
+      "step": 745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 0.2576,
+      "step": 746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 0.2266,
+      "step": 747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 0.2458,
+      "step": 748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 0.2355,
+      "step": 749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 0.2098,
+      "step": 750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 0.2311,
+      "step": 751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 0.2431,
+      "step": 752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 0.2294,
+      "step": 753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 0.2435,
+      "step": 754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 0.2124,
+      "step": 755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 0.2197,
+      "step": 756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 0.2336,
+      "step": 757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 0.222,
+      "step": 758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 0.217,
+      "step": 759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 0.2329,
+      "step": 760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 0.2353,
+      "step": 761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 0.215,
+      "step": 762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 0.2359,
+      "step": 763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 0.2248,
+      "step": 764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 0.2523,
+      "step": 765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 0.238,
+      "step": 766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 0.2338,
+      "step": 767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 0.2377,
+      "step": 768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 0.2067,
+      "step": 769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 0.2218,
+      "step": 770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 0.2338,
+      "step": 771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 0.2101,
+      "step": 772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 0.242,
+      "step": 773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 0.2247,
+      "step": 774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 0.2059,
+      "step": 775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 0.2295,
+      "step": 776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 0.2423,
+      "step": 777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 0.2246,
+      "step": 778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 0.2182,
+      "step": 779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 0.2232,
+      "step": 780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 0.2254,
+      "step": 781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 0.2089,
+      "step": 782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 0.2211,
+      "step": 783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 0.2175,
+      "step": 784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 0.2285,
+      "step": 785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 0.2228,
+      "step": 786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 0.2167,
+      "step": 787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 0.2134,
+      "step": 788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 0.2098,
+      "step": 789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 0.2325,
+      "step": 790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 0.2373,
+      "step": 791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 0.2072,
+      "step": 792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 0.2427,
+      "step": 793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.2176,
+      "step": 794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.2204,
+      "step": 795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.2168,
+      "step": 796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.2169,
+      "step": 797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.2005,
+      "step": 798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.1831,
+      "step": 799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.1969,
+      "step": 800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.1795,
+      "step": 801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 0.1932,
+      "step": 802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 0.192,
+      "step": 803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 0.1906,
+      "step": 804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 0.1946,
+      "step": 805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 0.1767,
+      "step": 806
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 0.189,
+      "step": 807
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 0.1746,
+      "step": 808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 0.2015,
+      "step": 809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 0.1758,
+      "step": 810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 0.1607,
+      "step": 811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 0.188,
+      "step": 812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 0.1668,
+      "step": 813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 0.179,
+      "step": 814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 0.1554,
+      "step": 815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 0.1597,
+      "step": 816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 0.1715,
+      "step": 817
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1811,
+      "step": 818
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1847,
+      "step": 819
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1807,
+      "step": 820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1916,
+      "step": 821
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1719,
+      "step": 822
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1877,
+      "step": 823
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1777,
+      "step": 824
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1879,
+      "step": 825
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 0.1866,
+      "step": 826
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 0.1835,
+      "step": 827
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 0.1841,
+      "step": 828
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 0.1935,
+      "step": 829
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 0.1633,
+      "step": 830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 0.1671,
+      "step": 831
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 0.1848,
+      "step": 832
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 0.1742,
+      "step": 833
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 0.1954,
+      "step": 834
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 0.1686,
+      "step": 835
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 0.1698,
+      "step": 836
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 0.1696,
+      "step": 837
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 0.1804,
+      "step": 838
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 0.192,
+      "step": 839
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 0.1675,
+      "step": 840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 0.1945,
+      "step": 841
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 0.1592,
+      "step": 842
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 0.1812,
+      "step": 843
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 0.1645,
+      "step": 844
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 0.1812,
+      "step": 845
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 0.1955,
+      "step": 846
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 0.1933,
+      "step": 847
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 0.1793,
+      "step": 848
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 0.1833,
+      "step": 849
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 0.1763,
+      "step": 850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 0.1641,
+      "step": 851
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 0.1672,
+      "step": 852
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 0.1787,
+      "step": 853
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 0.1853,
+      "step": 854
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 0.1885,
+      "step": 855
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 0.1927,
+      "step": 856
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.1804,
+      "step": 857
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.1817,
+      "step": 858
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.1619,
+      "step": 859
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.1793,
+      "step": 860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.1813,
+      "step": 861
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.1617,
+      "step": 862
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.1971,
+      "step": 863
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.1979,
+      "step": 864
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 0.1586,
+      "step": 865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 0.187,
+      "step": 866
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 0.1766,
+      "step": 867
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 0.1749,
+      "step": 868
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 0.1871,
+      "step": 869
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 0.1818,
+      "step": 870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 0.1901,
+      "step": 871
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 0.1593,
+      "step": 872
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.1807,
+      "step": 873
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.1821,
+      "step": 874
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.1838,
+      "step": 875
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.1881,
+      "step": 876
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.1786,
+      "step": 877
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.1906,
+      "step": 878
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.1849,
+      "step": 879
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.2034,
+      "step": 880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 0.1872,
+      "step": 881
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 0.1795,
+      "step": 882
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 0.1774,
+      "step": 883
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 0.1809,
+      "step": 884
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 0.17,
+      "step": 885
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 0.1701,
+      "step": 886
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 0.1741,
+      "step": 887
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 0.2074,
+      "step": 888
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 0.187,
+      "step": 889
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 0.1757,
+      "step": 890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 0.1934,
+      "step": 891
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 0.1936,
+      "step": 892
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 0.1764,
+      "step": 893
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 0.1885,
+      "step": 894
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 0.1785,
+      "step": 895
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 0.1784,
+      "step": 896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 0.168,
+      "step": 897
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 0.1769,
+      "step": 898
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 0.1949,
+      "step": 899
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 0.1769,
+      "step": 900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 0.1818,
+      "step": 901
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 0.1818,
+      "step": 902
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 0.1874,
+      "step": 903
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 0.1748,
+      "step": 904
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 0.156,
+      "step": 905
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 0.1848,
+      "step": 906
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 0.1896,
+      "step": 907
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 0.1724,
+      "step": 908
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 0.1544,
+      "step": 909
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 0.1775,
+      "step": 910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 0.1747,
+      "step": 911
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 0.1563,
+      "step": 912
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 0.1901,
+      "step": 913
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 0.1608,
+      "step": 914
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 0.1727,
+      "step": 915
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 0.1784,
+      "step": 916
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 0.1868,
+      "step": 917
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 0.1908,
+      "step": 918
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 0.1687,
+      "step": 919
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 0.1627,
+      "step": 920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 0.1846,
+      "step": 921
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 0.1965,
+      "step": 922
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 0.1915,
+      "step": 923
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 0.189,
+      "step": 924
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 0.1814,
+      "step": 925
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 0.1711,
+      "step": 926
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 0.1783,
+      "step": 927
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 0.1832,
+      "step": 928
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 0.2172,
+      "step": 929
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 0.1852,
+      "step": 930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 0.2017,
+      "step": 931
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 0.1694,
+      "step": 932
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 0.1828,
+      "step": 933
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 0.1622,
+      "step": 934
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 0.1883,
+      "step": 935
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 0.1745,
+      "step": 936
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 0.1568,
+      "step": 937
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 0.1749,
+      "step": 938
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 0.188,
+      "step": 939
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 0.1764,
+      "step": 940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 0.1756,
+      "step": 941
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 0.1892,
+      "step": 942
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 0.1815,
+      "step": 943
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 0.1694,
+      "step": 944
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 0.1616,
+      "step": 945
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 0.1941,
+      "step": 946
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 0.1848,
+      "step": 947
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 0.1458,
+      "step": 948
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 0.1767,
+      "step": 949
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 0.1615,
+      "step": 950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 0.1936,
+      "step": 951
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 0.1756,
+      "step": 952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 0.1907,
+      "step": 953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 0.178,
+      "step": 954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 0.1733,
+      "step": 955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 0.1839,
+      "step": 956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 0.1801,
+      "step": 957
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 0.1903,
+      "step": 958
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 0.184,
+      "step": 959
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 0.1797,
+      "step": 960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 0.1771,
+      "step": 961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 0.199,
+      "step": 962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 0.1698,
+      "step": 963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 0.1535,
+      "step": 964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 0.1878,
+      "step": 965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 0.172,
+      "step": 966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 0.1688,
+      "step": 967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 0.1768,
+      "step": 968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 0.1712,
+      "step": 969
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 0.1786,
+      "step": 970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 0.1651,
+      "step": 971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 0.1768,
+      "step": 972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 0.1741,
+      "step": 973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 0.1656,
+      "step": 974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 0.2045,
+      "step": 975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 0.1745,
+      "step": 976
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 0.1783,
+      "step": 977
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 0.1809,
+      "step": 978
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 0.1582,
+      "step": 979
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 0.1933,
+      "step": 980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 0.1739,
+      "step": 981
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 0.1793,
+      "step": 982
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 0.1745,
+      "step": 983
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 0.1641,
+      "step": 984
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 0.1795,
+      "step": 985
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 0.1715,
+      "step": 986
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 0.171,
+      "step": 987
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 0.1879,
+      "step": 988
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 0.1715,
+      "step": 989
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 0.1774,
+      "step": 990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 0.204,
+      "step": 991
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 0.1879,
+      "step": 992
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 0.1904,
+      "step": 993
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 0.1714,
+      "step": 994
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 0.1527,
+      "step": 995
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 0.1625,
+      "step": 996
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 0.1646,
+      "step": 997
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 0.1631,
+      "step": 998
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 0.1842,
+      "step": 999
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 0.1758,
+      "step": 1000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 0.1738,
+      "step": 1001
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 0.1569,
+      "step": 1002
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 0.1732,
+      "step": 1003
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 0.1807,
+      "step": 1004
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 0.1682,
+      "step": 1005
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 0.1721,
+      "step": 1006
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 0.1764,
+      "step": 1007
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 0.1921,
+      "step": 1008
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 0.1641,
+      "step": 1009
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 0.1852,
+      "step": 1010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 0.1769,
+      "step": 1011
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 0.1758,
+      "step": 1012
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 0.1722,
+      "step": 1013
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 0.1806,
+      "step": 1014
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 0.1886,
+      "step": 1015
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 0.1688,
+      "step": 1016
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 0.175,
+      "step": 1017
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 0.1753,
+      "step": 1018
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 0.1855,
+      "step": 1019
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 0.1859,
+      "step": 1020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 0.1653,
+      "step": 1021
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 0.1682,
+      "step": 1022
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 0.1537,
+      "step": 1023
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 0.1752,
+      "step": 1024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 0.1887,
+      "step": 1025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 0.193,
+      "step": 1026
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 0.1482,
+      "step": 1027
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 0.1699,
+      "step": 1028
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 0.1896,
+      "step": 1029
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 0.1597,
+      "step": 1030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 0.2192,
+      "step": 1031
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 0.1752,
+      "step": 1032
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 0.1644,
+      "step": 1033
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 0.1802,
+      "step": 1034
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 0.1873,
+      "step": 1035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 0.1839,
+      "step": 1036
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 0.1736,
+      "step": 1037
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 0.1666,
+      "step": 1038
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 0.1758,
+      "step": 1039
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 0.1833,
+      "step": 1040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 0.2001,
+      "step": 1041
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 0.156,
+      "step": 1042
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 0.1628,
+      "step": 1043
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 0.2069,
+      "step": 1044
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 0.1626,
+      "step": 1045
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 0.182,
+      "step": 1046
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 0.1867,
+      "step": 1047
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 0.1686,
+      "step": 1048
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 0.1695,
+      "step": 1049
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 0.183,
+      "step": 1050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 0.1741,
+      "step": 1051
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 0.1699,
+      "step": 1052
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 0.1888,
+      "step": 1053
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 0.1728,
+      "step": 1054
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 0.1692,
+      "step": 1055
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 0.194,
+      "step": 1056
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 0.1805,
+      "step": 1057
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 0.1741,
+      "step": 1058
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 0.1724,
+      "step": 1059
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 0.1881,
+      "step": 1060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 0.1653,
+      "step": 1061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 0.1643,
+      "step": 1062
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 0.1613,
+      "step": 1063
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 0.1791,
+      "step": 1064
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 0.182,
+      "step": 1065
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 0.1904,
+      "step": 1066
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 0.1743,
+      "step": 1067
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 0.1861,
+      "step": 1068
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 0.1806,
+      "step": 1069
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 0.1764,
+      "step": 1070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 0.1613,
+      "step": 1071
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 0.1688,
+      "step": 1072
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 0.1831,
+      "step": 1073
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 0.1708,
+      "step": 1074
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 0.1859,
+      "step": 1075
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 0.1907,
+      "step": 1076
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 0.1701,
+      "step": 1077
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 0.1703,
+      "step": 1078
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 0.1682,
+      "step": 1079
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 0.1638,
+      "step": 1080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 0.1624,
+      "step": 1081
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 0.1767,
+      "step": 1082
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 0.1865,
+      "step": 1083
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 0.2082,
+      "step": 1084
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 0.1817,
+      "step": 1085
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 0.1666,
+      "step": 1086
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 0.194,
+      "step": 1087
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 0.1622,
+      "step": 1088
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 0.1518,
+      "step": 1089
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 0.178,
+      "step": 1090
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 0.1565,
+      "step": 1091
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 0.1797,
+      "step": 1092
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 0.1718,
+      "step": 1093
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 0.1716,
+      "step": 1094
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 0.2004,
+      "step": 1095
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 0.1836,
+      "step": 1096
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 0.1675,
+      "step": 1097
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 0.1698,
+      "step": 1098
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 0.1775,
+      "step": 1099
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 0.1802,
+      "step": 1100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 0.192,
+      "step": 1101
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 0.1508,
+      "step": 1102
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 0.1872,
+      "step": 1103
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 0.1984,
+      "step": 1104
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 0.1679,
+      "step": 1105
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 0.1774,
+      "step": 1106
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 0.1885,
+      "step": 1107
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 0.1522,
+      "step": 1108
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 0.1848,
+      "step": 1109
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 0.168,
+      "step": 1110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 0.1581,
+      "step": 1111
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 0.1629,
+      "step": 1112
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 0.1731,
+      "step": 1113
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 0.1592,
+      "step": 1114
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 0.1625,
+      "step": 1115
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 0.1701,
+      "step": 1116
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 0.1537,
+      "step": 1117
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 0.1707,
+      "step": 1118
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 0.186,
+      "step": 1119
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 0.1813,
+      "step": 1120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 0.1965,
+      "step": 1121
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 0.1694,
+      "step": 1122
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 0.1853,
+      "step": 1123
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 0.1814,
+      "step": 1124
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 0.1585,
+      "step": 1125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 0.2027,
+      "step": 1126
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 0.1683,
+      "step": 1127
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 0.1607,
+      "step": 1128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 0.1602,
+      "step": 1129
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 0.1829,
+      "step": 1130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 0.1973,
+      "step": 1131
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 0.1701,
+      "step": 1132
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 0.166,
+      "step": 1133
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 0.1672,
+      "step": 1134
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 0.1814,
+      "step": 1135
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 0.1952,
+      "step": 1136
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 0.1847,
+      "step": 1137
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 0.1777,
+      "step": 1138
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 0.1689,
+      "step": 1139
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 0.1527,
+      "step": 1140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 0.1792,
+      "step": 1141
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 0.1786,
+      "step": 1142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 0.1722,
+      "step": 1143
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 0.178,
+      "step": 1144
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 0.1742,
+      "step": 1145
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 0.1787,
+      "step": 1146
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 0.164,
+      "step": 1147
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 0.1869,
+      "step": 1148
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 0.2028,
+      "step": 1149
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 0.1514,
+      "step": 1150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 0.1918,
+      "step": 1151
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 0.1685,
+      "step": 1152
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 0.1931,
+      "step": 1153
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 0.1637,
+      "step": 1154
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 0.1761,
+      "step": 1155
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 0.1824,
+      "step": 1156
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 0.1675,
+      "step": 1157
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 0.2011,
+      "step": 1158
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 0.1742,
+      "step": 1159
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 0.1737,
+      "step": 1160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 0.1654,
+      "step": 1161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 0.1776,
+      "step": 1162
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 0.172,
+      "step": 1163
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 0.1738,
+      "step": 1164
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 0.163,
+      "step": 1165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 0.177,
+      "step": 1166
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 0.1891,
+      "step": 1167
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 0.1864,
+      "step": 1168
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 0.1814,
+      "step": 1169
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 0.182,
+      "step": 1170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 0.1717,
+      "step": 1171
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 0.1717,
+      "step": 1172
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 0.1647,
+      "step": 1173
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 0.1882,
+      "step": 1174
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 0.1816,
+      "step": 1175
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 0.1636,
+      "step": 1176
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 0.1797,
+      "step": 1177
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 0.1559,
+      "step": 1178
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 0.1647,
+      "step": 1179
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 0.1665,
+      "step": 1180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 0.1564,
+      "step": 1181
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 0.1828,
+      "step": 1182
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 0.1768,
+      "step": 1183
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 0.1694,
+      "step": 1184
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 0.1664,
+      "step": 1185
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 0.1846,
+      "step": 1186
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 0.1567,
+      "step": 1187
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 0.1837,
+      "step": 1188
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 0.1747,
+      "step": 1189
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 0.1726,
+      "step": 1190
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 0.1475,
+      "step": 1191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.176,
+      "step": 1192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.1688,
+      "step": 1193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.1846,
+      "step": 1194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.163,
+      "step": 1195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.1735,
+      "step": 1196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.1718,
+      "step": 1197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.1973,
+      "step": 1198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.1733,
+      "step": 1199
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.1727,
+      "step": 1200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.1764,
+      "step": 1201
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.1651,
+      "step": 1202
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.1771,
+      "step": 1203
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.1809,
+      "step": 1204
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.1866,
+      "step": 1205
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.1646,
+      "step": 1206
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.1531,
+      "step": 1207
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 0.1753,
+      "step": 1208
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 0.1767,
+      "step": 1209
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 0.1633,
+      "step": 1210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 0.1569,
+      "step": 1211
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 0.1769,
+      "step": 1212
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 0.1627,
+      "step": 1213
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 0.1663,
+      "step": 1214
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 0.1789,
+      "step": 1215
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 0.1789,
+      "step": 1216
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 0.1741,
+      "step": 1217
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 0.1829,
+      "step": 1218
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 0.1655,
+      "step": 1219
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 0.1818,
+      "step": 1220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 0.1625,
+      "step": 1221
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 0.1632,
+      "step": 1222
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 0.1962,
+      "step": 1223
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.1884,
+      "step": 1224
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.1679,
+      "step": 1225
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.1804,
+      "step": 1226
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.1736,
+      "step": 1227
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.17,
+      "step": 1228
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.1725,
+      "step": 1229
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.1683,
+      "step": 1230
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.168,
+      "step": 1231
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.1821,
+      "step": 1232
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.1528,
+      "step": 1233
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.1758,
+      "step": 1234
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.1789,
+      "step": 1235
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.1654,
+      "step": 1236
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.1856,
+      "step": 1237
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.1777,
+      "step": 1238
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.1805,
+      "step": 1239
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.1492,
+      "step": 1240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2136,
+      "step": 1241
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.1546,
+      "step": 1242
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.1795,
+      "step": 1243
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.1751,
+      "step": 1244
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.1922,
+      "step": 1245
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.1687,
+      "step": 1246
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.1744,
+      "step": 1247
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 0.1735,
+      "step": 1248
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 0.1856,
+      "step": 1249
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 0.1551,
+      "step": 1250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 0.168,
+      "step": 1251
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 0.145,
+      "step": 1252
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 0.1719,
+      "step": 1253
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 0.1619,
+      "step": 1254
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 0.1857,
+      "step": 1255
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 0.1718,
+      "step": 1256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 0.1747,
+      "step": 1257
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 0.1632,
+      "step": 1258
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 0.1695,
+      "step": 1259
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 0.1809,
+      "step": 1260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 0.1681,
+      "step": 1261
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 0.1613,
+      "step": 1262
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 0.1531,
+      "step": 1263
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 0.1792,
+      "step": 1264
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 0.1735,
+      "step": 1265
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 0.1674,
+      "step": 1266
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 0.18,
+      "step": 1267
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 0.1808,
+      "step": 1268
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 0.1457,
+      "step": 1269
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 0.181,
+      "step": 1270
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 0.1661,
+      "step": 1271
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 0.1962,
+      "step": 1272
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 0.1756,
+      "step": 1273
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 0.1647,
+      "step": 1274
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 0.203,
+      "step": 1275
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 0.1879,
+      "step": 1276
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 0.164,
+      "step": 1277
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 0.1837,
+      "step": 1278
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 0.1803,
+      "step": 1279
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 0.1727,
+      "step": 1280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 0.1768,
+      "step": 1281
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 0.1665,
+      "step": 1282
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 0.1656,
+      "step": 1283
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 0.1631,
+      "step": 1284
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 0.1498,
+      "step": 1285
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 0.1688,
+      "step": 1286
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 0.1846,
+      "step": 1287
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 0.1772,
+      "step": 1288
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 0.1666,
+      "step": 1289
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 0.1516,
+      "step": 1290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 0.1748,
+      "step": 1291
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 0.1527,
+      "step": 1292
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 0.1674,
+      "step": 1293
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 0.1514,
+      "step": 1294
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 0.149,
+      "step": 1295
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 0.1783,
+      "step": 1296
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 0.1502,
+      "step": 1297
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 0.1757,
+      "step": 1298
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 0.1739,
+      "step": 1299
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 0.1766,
+      "step": 1300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 0.179,
+      "step": 1301
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 0.1808,
+      "step": 1302
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 0.1601,
+      "step": 1303
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 0.1781,
+      "step": 1304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 0.1899,
+      "step": 1305
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 0.1654,
+      "step": 1306
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 0.1797,
+      "step": 1307
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 0.1741,
+      "step": 1308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 0.16,
+      "step": 1309
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 0.1744,
+      "step": 1310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 0.1667,
+      "step": 1311
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 0.1672,
+      "step": 1312
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 0.1782,
+      "step": 1313
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 0.174,
+      "step": 1314
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 0.1559,
+      "step": 1315
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 0.1666,
+      "step": 1316
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 0.188,
+      "step": 1317
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 0.1686,
+      "step": 1318
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 0.1941,
+      "step": 1319
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 0.1746,
+      "step": 1320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 0.1751,
+      "step": 1321
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 0.1645,
+      "step": 1322
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 0.1703,
+      "step": 1323
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 0.1743,
+      "step": 1324
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 0.1664,
+      "step": 1325
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 0.1652,
+      "step": 1326
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 0.1673,
+      "step": 1327
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 0.1771,
+      "step": 1328
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 0.158,
+      "step": 1329
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 0.17,
+      "step": 1330
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 0.1563,
+      "step": 1331
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 0.1726,
+      "step": 1332
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 0.1884,
+      "step": 1333
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 0.1637,
+      "step": 1334
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 0.1727,
+      "step": 1335
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 0.1673,
+      "step": 1336
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 0.1875,
+      "step": 1337
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 0.1711,
+      "step": 1338
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 0.1864,
+      "step": 1339
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 0.1828,
+      "step": 1340
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 0.1599,
+      "step": 1341
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 0.1788,
+      "step": 1342
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 0.178,
+      "step": 1343
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 0.1639,
+      "step": 1344
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 0.1777,
+      "step": 1345
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 0.1762,
+      "step": 1346
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 0.1665,
+      "step": 1347
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 0.154,
+      "step": 1348
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 0.1744,
+      "step": 1349
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 0.1642,
+      "step": 1350
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 0.1729,
+      "step": 1351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 0.1833,
+      "step": 1352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 0.1621,
+      "step": 1353
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 0.1691,
+      "step": 1354
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 0.1582,
+      "step": 1355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 0.1675,
+      "step": 1356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 0.1502,
+      "step": 1357
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 0.1721,
+      "step": 1358
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 0.1757,
+      "step": 1359
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 0.1537,
+      "step": 1360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 0.1706,
+      "step": 1361
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 0.1826,
+      "step": 1362
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 0.1809,
+      "step": 1363
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 0.1644,
+      "step": 1364
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 0.1699,
+      "step": 1365
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 0.169,
+      "step": 1366
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 0.1705,
+      "step": 1367
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 0.1858,
+      "step": 1368
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 0.1773,
+      "step": 1369
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 0.1778,
+      "step": 1370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 0.1674,
+      "step": 1371
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 0.1791,
+      "step": 1372
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 0.1597,
+      "step": 1373
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 0.1427,
+      "step": 1374
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 0.1621,
+      "step": 1375
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 0.1637,
+      "step": 1376
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 0.1569,
+      "step": 1377
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 0.1656,
+      "step": 1378
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 0.1782,
+      "step": 1379
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 0.1725,
+      "step": 1380
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 0.1883,
+      "step": 1381
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 0.1795,
+      "step": 1382
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 0.1696,
+      "step": 1383
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 0.174,
+      "step": 1384
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 0.1779,
+      "step": 1385
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 0.1831,
+      "step": 1386
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 0.179,
+      "step": 1387
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 0.1588,
+      "step": 1388
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 0.1619,
+      "step": 1389
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 0.1872,
+      "step": 1390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 0.1717,
+      "step": 1391
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 0.1677,
+      "step": 1392
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 0.1749,
+      "step": 1393
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 0.1807,
+      "step": 1394
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 0.1685,
+      "step": 1395
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 0.1971,
+      "step": 1396
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 0.1617,
+      "step": 1397
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 0.1572,
+      "step": 1398
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 0.143,
+      "step": 1399
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 0.175,
+      "step": 1400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 0.1619,
+      "step": 1401
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 0.1709,
+      "step": 1402
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 0.1662,
+      "step": 1403
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 0.1724,
+      "step": 1404
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 0.1689,
+      "step": 1405
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 0.174,
+      "step": 1406
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 0.1699,
+      "step": 1407
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 0.1671,
+      "step": 1408
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 0.1602,
+      "step": 1409
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 0.1739,
+      "step": 1410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 0.1661,
+      "step": 1411
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 0.1753,
+      "step": 1412
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 0.1712,
+      "step": 1413
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 0.1681,
+      "step": 1414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 0.1669,
+      "step": 1415
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 0.1638,
+      "step": 1416
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 0.1546,
+      "step": 1417
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 0.1843,
+      "step": 1418
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 0.1672,
+      "step": 1419
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 0.1823,
+      "step": 1420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 0.1894,
+      "step": 1421
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 0.1579,
+      "step": 1422
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 0.1746,
+      "step": 1423
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 0.173,
+      "step": 1424
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 0.1624,
+      "step": 1425
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 0.1569,
+      "step": 1426
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 0.1768,
+      "step": 1427
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 0.1554,
+      "step": 1428
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 0.1854,
+      "step": 1429
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 0.1598,
+      "step": 1430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 0.1709,
+      "step": 1431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 0.1844,
+      "step": 1432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 0.1706,
+      "step": 1433
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 0.1506,
+      "step": 1434
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 0.1752,
+      "step": 1435
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 0.1792,
+      "step": 1436
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 0.1804,
+      "step": 1437
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 0.1632,
+      "step": 1438
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 0.1768,
+      "step": 1439
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 0.1795,
+      "step": 1440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 0.1672,
+      "step": 1441
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 0.1663,
+      "step": 1442
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 0.174,
+      "step": 1443
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 0.1687,
+      "step": 1444
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 0.1674,
+      "step": 1445
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 0.1748,
+      "step": 1446
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 0.1542,
+      "step": 1447
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 0.1622,
+      "step": 1448
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 0.1617,
+      "step": 1449
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 0.1583,
+      "step": 1450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 0.1793,
+      "step": 1451
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 0.1625,
+      "step": 1452
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 0.184,
+      "step": 1453
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 0.169,
+      "step": 1454
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 0.167,
+      "step": 1455
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 0.1753,
+      "step": 1456
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 0.1468,
+      "step": 1457
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 0.1677,
+      "step": 1458
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 0.1588,
+      "step": 1459
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 0.1665,
+      "step": 1460
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 0.1723,
+      "step": 1461
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 0.1679,
+      "step": 1462
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 0.146,
+      "step": 1463
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 0.1802,
+      "step": 1464
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 0.1516,
+      "step": 1465
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 0.1698,
+      "step": 1466
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 0.1899,
+      "step": 1467
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 0.1534,
+      "step": 1468
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 0.1743,
+      "step": 1469
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 0.1727,
+      "step": 1470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 0.1938,
+      "step": 1471
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 0.1723,
+      "step": 1472
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 0.1712,
+      "step": 1473
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 0.1543,
+      "step": 1474
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 0.182,
+      "step": 1475
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 0.1736,
+      "step": 1476
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 0.1767,
+      "step": 1477
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 0.1633,
+      "step": 1478
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 0.1672,
+      "step": 1479
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 0.1733,
+      "step": 1480
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 0.1753,
+      "step": 1481
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 0.1732,
+      "step": 1482
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 0.1881,
+      "step": 1483
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 0.1764,
+      "step": 1484
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 0.1585,
+      "step": 1485
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 0.1588,
+      "step": 1486
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 0.1486,
+      "step": 1487
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 0.1508,
+      "step": 1488
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 0.1711,
+      "step": 1489
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 0.1626,
+      "step": 1490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 0.1668,
+      "step": 1491
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 0.1602,
+      "step": 1492
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 0.1533,
+      "step": 1493
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 0.1607,
+      "step": 1494
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 0.1665,
+      "step": 1495
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 0.17,
+      "step": 1496
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 0.1634,
+      "step": 1497
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 0.1602,
+      "step": 1498
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 0.1461,
+      "step": 1499
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 0.1702,
+      "step": 1500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 0.1694,
+      "step": 1501
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 0.1879,
+      "step": 1502
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 0.1715,
+      "step": 1503
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 0.1556,
+      "step": 1504
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 0.1444,
+      "step": 1505
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 0.148,
+      "step": 1506
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 0.1649,
+      "step": 1507
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 0.1785,
+      "step": 1508
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 0.1744,
+      "step": 1509
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 0.1396,
+      "step": 1510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 0.1823,
+      "step": 1511
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 0.1559,
+      "step": 1512
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 0.1657,
+      "step": 1513
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 0.1959,
+      "step": 1514
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 0.1614,
+      "step": 1515
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 0.1551,
+      "step": 1516
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 0.1743,
+      "step": 1517
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 0.1714,
+      "step": 1518
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 0.155,
+      "step": 1519
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 0.1651,
+      "step": 1520
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 0.1603,
+      "step": 1521
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 0.16,
+      "step": 1522
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 0.1682,
+      "step": 1523
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 0.1604,
+      "step": 1524
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 0.1823,
+      "step": 1525
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 0.1784,
+      "step": 1526
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 0.1559,
+      "step": 1527
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 0.1605,
+      "step": 1528
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 0.1914,
+      "step": 1529
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 0.1573,
+      "step": 1530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 0.1807,
+      "step": 1531
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 0.1521,
+      "step": 1532
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 0.17,
+      "step": 1533
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 0.1742,
+      "step": 1534
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 0.1657,
+      "step": 1535
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 0.1676,
+      "step": 1536
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 0.1712,
+      "step": 1537
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 0.1657,
+      "step": 1538
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 0.1765,
+      "step": 1539
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 0.1723,
+      "step": 1540
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 0.1683,
+      "step": 1541
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 0.1711,
+      "step": 1542
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 0.1854,
+      "step": 1543
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 0.1691,
+      "step": 1544
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 0.1841,
+      "step": 1545
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 0.19,
+      "step": 1546
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 0.1604,
+      "step": 1547
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 0.1471,
+      "step": 1548
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 0.1701,
+      "step": 1549
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 0.1765,
+      "step": 1550
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1594,
+  "num_train_epochs": 2,
+  "save_steps": 50,
+  "total_flos": 1.9054348612927488e+17,
+  "trial_name": null,
+  "trial_params": null
+}