diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,33628 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4944976102961467,
+  "eval_steps": 500,
+  "global_step": 5600,
+  "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": 3.4968,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 3.9314,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.2664,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.88235294117647e-10,
+      "loss": 4.1062,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.176470588235294e-09,
+      "loss": 4.138,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7647058823529412e-09,
+      "loss": 3.8248,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.352941176470588e-09,
+      "loss": 4.0112,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.941176470588235e-09,
+      "loss": 3.7751,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5294117647058823e-09,
+      "loss": 4.3358,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.117647058823529e-09,
+      "loss": 4.2649,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.705882352941176e-09,
+      "loss": 3.8253,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.294117647058823e-09,
+      "loss": 3.5716,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.88235294117647e-09,
+      "loss": 3.6635,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.470588235294118e-09,
+      "loss": 4.1001,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.058823529411765e-09,
+      "loss": 3.7826,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.647058823529412e-09,
+      "loss": 4.3889,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.235294117647058e-09,
+      "loss": 4.003,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.823529411764706e-09,
+      "loss": 4.153,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.411764705882352e-09,
+      "loss": 3.9403,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.411764705882352e-09,
+      "loss": 3.8152,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-08,
+      "loss": 3.8604,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0588235294117647e-08,
+      "loss": 3.932,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1176470588235294e-08,
+      "loss": 3.8453,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.176470588235294e-08,
+      "loss": 4.0039,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2352941176470589e-08,
+      "loss": 4.0908,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2941176470588237e-08,
+      "loss": 3.9794,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3529411764705881e-08,
+      "loss": 3.6753,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.411764705882353e-08,
+      "loss": 4.113,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4705882352941176e-08,
+      "loss": 3.5571,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5294117647058823e-08,
+      "loss": 3.8468,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.588235294117647e-08,
+      "loss": 4.1301,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6470588235294116e-08,
+      "loss": 4.166,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7058823529411766e-08,
+      "loss": 3.9074,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7647058823529412e-08,
+      "loss": 4.2301,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8235294117647058e-08,
+      "loss": 4.3683,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8823529411764705e-08,
+      "loss": 3.5436,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9411764705882354e-08,
+      "loss": 4.3156,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 3.7362,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0588235294117647e-08,
+      "loss": 4.334,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1176470588235293e-08,
+      "loss": 3.9852,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.176470588235294e-08,
+      "loss": 4.2008,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.235294117647059e-08,
+      "loss": 4.283,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2941176470588232e-08,
+      "loss": 4.0906,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.352941176470588e-08,
+      "loss": 4.2326,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4117647058823528e-08,
+      "loss": 3.6496,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4117647058823528e-08,
+      "loss": 4.3855,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4705882352941177e-08,
+      "loss": 3.9715,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.529411764705882e-08,
+      "loss": 3.8754,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5882352941176473e-08,
+      "loss": 3.9501,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6470588235294116e-08,
+      "loss": 4.0956,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7058823529411763e-08,
+      "loss": 4.1817,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7647058823529412e-08,
+      "loss": 4.2062,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.823529411764706e-08,
+      "loss": 4.1388,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.88235294117647e-08,
+      "loss": 4.2405,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.941176470588235e-08,
+      "loss": 4.2903,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-08,
+      "loss": 4.2744,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.058823529411765e-08,
+      "loss": 4.0363,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.117647058823529e-08,
+      "loss": 4.1183,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.176470588235294e-08,
+      "loss": 3.9969,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.235294117647059e-08,
+      "loss": 3.8671,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.294117647058823e-08,
+      "loss": 4.2264,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3529411764705875e-08,
+      "loss": 3.9817,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.411764705882353e-08,
+      "loss": 3.824,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4705882352941174e-08,
+      "loss": 4.2127,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5294117647058824e-08,
+      "loss": 3.7903,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.588235294117647e-08,
+      "loss": 4.2467,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6470588235294116e-08,
+      "loss": 3.7463,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7058823529411766e-08,
+      "loss": 4.0796,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.764705882352941e-08,
+      "loss": 4.2687,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.823529411764705e-08,
+      "loss": 4.1286,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.882352941176471e-08,
+      "loss": 4.204,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.941176470588235e-08,
+      "loss": 4.0505,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-08,
+      "loss": 3.6725,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0588235294117644e-08,
+      "loss": 4.3486,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1176470588235293e-08,
+      "loss": 3.7032,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.176470588235294e-08,
+      "loss": 3.8587,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2352941176470586e-08,
+      "loss": 3.8928,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.294117647058823e-08,
+      "loss": 4.266,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.352941176470588e-08,
+      "loss": 3.9543,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.411764705882353e-08,
+      "loss": 4.0747,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.470588235294118e-08,
+      "loss": 3.6934,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.529411764705882e-08,
+      "loss": 3.8232,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5882352941176464e-08,
+      "loss": 4.0279,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.647058823529412e-08,
+      "loss": 4.562,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705882352941176e-08,
+      "loss": 4.1042,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7647058823529406e-08,
+      "loss": 3.9492,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8235294117647056e-08,
+      "loss": 3.847,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8823529411764705e-08,
+      "loss": 3.5875,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9411764705882355e-08,
+      "loss": 3.909,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-08,
+      "loss": 3.8583,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.058823529411764e-08,
+      "loss": 3.8217,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.117647058823529e-08,
+      "loss": 3.8383,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1764705882352947e-08,
+      "loss": 3.5649,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.235294117647059e-08,
+      "loss": 3.7961,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.294117647058823e-08,
+      "loss": 3.903,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3529411764705875e-08,
+      "loss": 4.266,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4117647058823525e-08,
+      "loss": 4.0882,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.470588235294118e-08,
+      "loss": 4.1547,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5294117647058824e-08,
+      "loss": 4.2945,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.588235294117647e-08,
+      "loss": 4.0975,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.647058823529412e-08,
+      "loss": 4.0827,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.705882352941176e-08,
+      "loss": 4.1912,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.76470588235294e-08,
+      "loss": 4.1283,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.823529411764706e-08,
+      "loss": 3.9852,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.88235294117647e-08,
+      "loss": 4.5667,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.941176470588235e-08,
+      "loss": 3.9092,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-08,
+      "loss": 4.0079,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.058823529411764e-08,
+      "loss": 3.7822,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.11764705882353e-08,
+      "loss": 4.2453,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.176470588235294e-08,
+      "loss": 3.8502,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.235294117647058e-08,
+      "loss": 4.4316,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.294117647058824e-08,
+      "loss": 3.9346,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.352941176470588e-08,
+      "loss": 4.0909,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.411764705882354e-08,
+      "loss": 4.4895,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.470588235294118e-08,
+      "loss": 3.6209,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.529411764705882e-08,
+      "loss": 3.829,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.588235294117646e-08,
+      "loss": 4.5784,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.647058823529411e-08,
+      "loss": 4.1743,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.705882352941175e-08,
+      "loss": 4.0971,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.76470588235294e-08,
+      "loss": 3.672,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.76470588235294e-08,
+      "loss": 3.814,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.823529411764706e-08,
+      "loss": 3.8764,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.88235294117647e-08,
+      "loss": 4.2151,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.941176470588235e-08,
+      "loss": 3.8224,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.999999999999999e-08,
+      "loss": 4.3235,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.058823529411765e-08,
+      "loss": 3.7266,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.117647058823529e-08,
+      "loss": 4.0726,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.176470588235293e-08,
+      "loss": 3.9926,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.235294117647058e-08,
+      "loss": 3.8424,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.294117647058823e-08,
+      "loss": 4.0516,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.352941176470589e-08,
+      "loss": 3.9491,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.411764705882353e-08,
+      "loss": 3.7362,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.470588235294118e-08,
+      "loss": 3.6055,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.529411764705882e-08,
+      "loss": 4.2161,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.588235294117646e-08,
+      "loss": 3.9211,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.64705882352941e-08,
+      "loss": 3.7963,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.705882352941176e-08,
+      "loss": 3.7986,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.764705882352942e-08,
+      "loss": 3.7207,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.823529411764706e-08,
+      "loss": 4.5592,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.88235294117647e-08,
+      "loss": 3.5639,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.941176470588235e-08,
+      "loss": 3.9825,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-08,
+      "loss": 4.4569,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.058823529411764e-08,
+      "loss": 4.0141,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.117647058823529e-08,
+      "loss": 4.0371,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.176470588235293e-08,
+      "loss": 3.6879,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.235294117647059e-08,
+      "loss": 4.013,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.294117647058824e-08,
+      "loss": 4.3003,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.352941176470589e-08,
+      "loss": 4.2041,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.411764705882353e-08,
+      "loss": 3.8291,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.470588235294117e-08,
+      "loss": 4.2425,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.529411764705881e-08,
+      "loss": 4.0633,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.588235294117646e-08,
+      "loss": 3.929,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.647058823529411e-08,
+      "loss": 4.1978,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.705882352941176e-08,
+      "loss": 4.2409,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.764705882352941e-08,
+      "loss": 4.141,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.823529411764706e-08,
+      "loss": 4.0297,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88235294117647e-08,
+      "loss": 3.5695,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.941176470588236e-08,
+      "loss": 3.8155,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-08,
+      "loss": 3.837,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.058823529411764e-08,
+      "loss": 4.305,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.117647058823528e-08,
+      "loss": 4.2308,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.176470588235293e-08,
+      "loss": 3.8413,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.235294117647058e-08,
+      "loss": 4.3042,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.294117647058824e-08,
+      "loss": 3.6695,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.352941176470588e-08,
+      "loss": 4.1596,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.411764705882353e-08,
+      "loss": 3.661,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.470588235294117e-08,
+      "loss": 4.08,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.529411764705881e-08,
+      "loss": 3.951,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.588235294117647e-08,
+      "loss": 3.9023,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.647058823529411e-08,
+      "loss": 3.8232,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.705882352941175e-08,
+      "loss": 3.3938,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.764705882352941e-08,
+      "loss": 3.7266,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.823529411764705e-08,
+      "loss": 3.7854,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.882352941176471e-08,
+      "loss": 4.1842,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.941176470588235e-08,
+      "loss": 4.6157,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-07,
+      "loss": 3.7525,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0058823529411764e-07,
+      "loss": 3.9674,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0117647058823528e-07,
+      "loss": 3.7851,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0176470588235292e-07,
+      "loss": 4.1855,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0235294117647058e-07,
+      "loss": 3.7209,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0294117647058822e-07,
+      "loss": 4.1895,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0352941176470589e-07,
+      "loss": 4.0992,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0411764705882354e-07,
+      "loss": 3.8811,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0470588235294118e-07,
+      "loss": 3.9257,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0529411764705882e-07,
+      "loss": 3.7107,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0588235294117647e-07,
+      "loss": 4.1159,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0647058823529411e-07,
+      "loss": 3.7231,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0705882352941175e-07,
+      "loss": 3.7767,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0764705882352941e-07,
+      "loss": 4.0418,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0823529411764705e-07,
+      "loss": 4.3042,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0882352941176469e-07,
+      "loss": 4.0845,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0941176470588236e-07,
+      "loss": 3.9898,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1e-07,
+      "loss": 4.0573,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1058823529411765e-07,
+      "loss": 4.1943,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1117647058823529e-07,
+      "loss": 3.9987,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1176470588235293e-07,
+      "loss": 3.5603,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1235294117647058e-07,
+      "loss": 4.0747,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1294117647058823e-07,
+      "loss": 4.7593,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1352941176470588e-07,
+      "loss": 4.5341,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1411764705882352e-07,
+      "loss": 4.3179,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1470588235294116e-07,
+      "loss": 4.5047,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.152941176470588e-07,
+      "loss": 3.5233,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1588235294117648e-07,
+      "loss": 4.2172,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1647058823529412e-07,
+      "loss": 3.4896,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1705882352941176e-07,
+      "loss": 4.2644,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.176470588235294e-07,
+      "loss": 4.1391,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1823529411764706e-07,
+      "loss": 4.0631,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.188235294117647e-07,
+      "loss": 4.1248,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1941176470588235e-07,
+      "loss": 3.9975,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-07,
+      "loss": 3.876,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2058823529411763e-07,
+      "loss": 3.736,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2117647058823528e-07,
+      "loss": 3.5656,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2176470588235292e-07,
+      "loss": 3.4691,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.223529411764706e-07,
+      "loss": 3.8895,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2294117647058823e-07,
+      "loss": 3.867,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2352941176470587e-07,
+      "loss": 4.1319,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2411764705882352e-07,
+      "loss": 3.7253,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2470588235294116e-07,
+      "loss": 4.2204,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2529411764705883e-07,
+      "loss": 3.9613,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2588235294117647e-07,
+      "loss": 3.7732,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2647058823529412e-07,
+      "loss": 3.399,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2705882352941176e-07,
+      "loss": 4.0948,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.276470588235294e-07,
+      "loss": 3.8423,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2823529411764707e-07,
+      "loss": 4.1212,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2882352941176471e-07,
+      "loss": 4.3819,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2941176470588236e-07,
+      "loss": 3.8446,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3e-07,
+      "loss": 3.6567,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3058823529411764e-07,
+      "loss": 3.7542,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3117647058823529e-07,
+      "loss": 3.9142,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3176470588235293e-07,
+      "loss": 3.9693,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3235294117647057e-07,
+      "loss": 3.8922,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3294117647058821e-07,
+      "loss": 3.545,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3352941176470586e-07,
+      "loss": 4.2549,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.341176470588235e-07,
+      "loss": 3.8443,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3470588235294117e-07,
+      "loss": 3.8136,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.352941176470588e-07,
+      "loss": 4.1243,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3588235294117648e-07,
+      "loss": 3.2729,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3647058823529413e-07,
+      "loss": 3.8121,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3705882352941177e-07,
+      "loss": 4.0457,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.376470588235294e-07,
+      "loss": 3.9258,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3823529411764705e-07,
+      "loss": 3.8571,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.388235294117647e-07,
+      "loss": 3.7367,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3941176470588234e-07,
+      "loss": 4.0287,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3999999999999998e-07,
+      "loss": 4.0312,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4058823529411763e-07,
+      "loss": 3.9622,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.411764705882353e-07,
+      "loss": 3.7508,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4176470588235294e-07,
+      "loss": 4.0502,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4235294117647058e-07,
+      "loss": 4.2795,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4294117647058822e-07,
+      "loss": 4.0181,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4352941176470587e-07,
+      "loss": 4.1683,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.441176470588235e-07,
+      "loss": 3.7848,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4470588235294115e-07,
+      "loss": 3.9384,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4529411764705882e-07,
+      "loss": 3.719,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4588235294117647e-07,
+      "loss": 3.7315,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.464705882352941e-07,
+      "loss": 3.9142,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4705882352941178e-07,
+      "loss": 3.5427,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4764705882352942e-07,
+      "loss": 3.7726,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4823529411764706e-07,
+      "loss": 3.8847,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.488235294117647e-07,
+      "loss": 3.7307,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4941176470588235e-07,
+      "loss": 3.9543,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-07,
+      "loss": 3.9337,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5058823529411764e-07,
+      "loss": 3.6955,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5117647058823528e-07,
+      "loss": 3.6089,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5176470588235292e-07,
+      "loss": 3.734,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5235294117647057e-07,
+      "loss": 3.763,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.529411764705882e-07,
+      "loss": 4.0927,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5352941176470588e-07,
+      "loss": 3.6117,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5411764705882352e-07,
+      "loss": 3.9501,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5470588235294116e-07,
+      "loss": 4.1775,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5529411764705883e-07,
+      "loss": 4.2473,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5588235294117648e-07,
+      "loss": 3.9555,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5647058823529412e-07,
+      "loss": 4.4574,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5705882352941176e-07,
+      "loss": 4.0441,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576470588235294e-07,
+      "loss": 3.6609,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5823529411764705e-07,
+      "loss": 4.1008,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.588235294117647e-07,
+      "loss": 3.3907,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5941176470588236e-07,
+      "loss": 4.4435,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6e-07,
+      "loss": 3.7527,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6058823529411765e-07,
+      "loss": 3.9326,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.611764705882353e-07,
+      "loss": 3.9428,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6176470588235293e-07,
+      "loss": 3.5153,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6235294117647058e-07,
+      "loss": 3.8369,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6294117647058822e-07,
+      "loss": 3.9229,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6352941176470586e-07,
+      "loss": 3.8843,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.641176470588235e-07,
+      "loss": 4.0581,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6470588235294117e-07,
+      "loss": 3.4472,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6529411764705882e-07,
+      "loss": 3.8908,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6588235294117649e-07,
+      "loss": 3.7528,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6647058823529413e-07,
+      "loss": 3.9742,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6705882352941177e-07,
+      "loss": 3.8558,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6764705882352942e-07,
+      "loss": 3.8831,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6823529411764706e-07,
+      "loss": 3.6252,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.688235294117647e-07,
+      "loss": 3.9213,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6941176470588234e-07,
+      "loss": 3.6309,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7e-07,
+      "loss": 3.9982,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7058823529411763e-07,
+      "loss": 4.1579,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7117647058823527e-07,
+      "loss": 4.0634,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7176470588235292e-07,
+      "loss": 3.7456,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7235294117647059e-07,
+      "loss": 3.9717,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7294117647058823e-07,
+      "loss": 3.7447,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7352941176470587e-07,
+      "loss": 3.8189,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7411764705882351e-07,
+      "loss": 3.7616,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7470588235294116e-07,
+      "loss": 3.5575,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7529411764705883e-07,
+      "loss": 3.6672,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7588235294117647e-07,
+      "loss": 3.6312,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.764705882352941e-07,
+      "loss": 4.0583,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7705882352941176e-07,
+      "loss": 3.7304,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.776470588235294e-07,
+      "loss": 3.4982,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7823529411764707e-07,
+      "loss": 3.8842,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.788235294117647e-07,
+      "loss": 4.2327,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7941176470588235e-07,
+      "loss": 3.536,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-07,
+      "loss": 3.9274,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8058823529411764e-07,
+      "loss": 3.8402,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8117647058823528e-07,
+      "loss": 3.6044,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8176470588235293e-07,
+      "loss": 3.804,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8235294117647057e-07,
+      "loss": 3.7063,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.829411764705882e-07,
+      "loss": 3.7362,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8352941176470586e-07,
+      "loss": 4.1281,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.841176470588235e-07,
+      "loss": 4.0401,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8470588235294117e-07,
+      "loss": 3.8292,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.852941176470588e-07,
+      "loss": 3.9999,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8588235294117648e-07,
+      "loss": 3.3449,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8647058823529412e-07,
+      "loss": 3.7358,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8705882352941177e-07,
+      "loss": 3.8039,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.876470588235294e-07,
+      "loss": 3.6286,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8823529411764705e-07,
+      "loss": 4.2325,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.888235294117647e-07,
+      "loss": 3.653,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8941176470588234e-07,
+      "loss": 3.8422,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8999999999999998e-07,
+      "loss": 4.1077,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9058823529411762e-07,
+      "loss": 4.042,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.911764705882353e-07,
+      "loss": 3.7463,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9176470588235294e-07,
+      "loss": 4.427,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9235294117647058e-07,
+      "loss": 3.8059,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9294117647058822e-07,
+      "loss": 3.9901,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9352941176470587e-07,
+      "loss": 3.7525,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.941176470588235e-07,
+      "loss": 4.0174,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9470588235294115e-07,
+      "loss": 3.6346,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9529411764705882e-07,
+      "loss": 3.7727,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9588235294117646e-07,
+      "loss": 3.8317,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.964705882352941e-07,
+      "loss": 3.662,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9705882352941178e-07,
+      "loss": 3.9145,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9764705882352942e-07,
+      "loss": 3.8994,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9823529411764706e-07,
+      "loss": 3.5725,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988235294117647e-07,
+      "loss": 3.8417,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941176470588235e-07,
+      "loss": 3.454,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-07,
+      "loss": 3.8123,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999590976087e-07,
+      "loss": 3.4982,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998363904386e-07,
+      "loss": 3.8234,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999631878499e-07,
+      "loss": 3.7841,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993455618075e-07,
+      "loss": 3.9991,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999989774403868e-07,
+      "loss": 3.8562,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999985275142673e-07,
+      "loss": 3.7438,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997995783486e-07,
+      "loss": 3.5935,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997382248086e-07,
+      "loss": 3.7366,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996686908118e-07,
+      "loss": 3.8108,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999959097636383e-07,
+      "loss": 3.4692,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995050814711e-07,
+      "loss": 3.9025,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999941100614062e-07,
+      "loss": 3.7405,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999930875038005e-07,
+      "loss": 3.8846,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999919831419779e-07,
+      "loss": 3.9166,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999907969760285e-07,
+      "loss": 3.705,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999895290060498e-07,
+      "loss": 4.0468,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999881792321452e-07,
+      "loss": 3.6286,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999867476544252e-07,
+      "loss": 4.0123,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999852342730068e-07,
+      "loss": 3.6986,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999836390880138e-07,
+      "loss": 4.0427,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981962099577e-07,
+      "loss": 3.774,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980203307833e-07,
+      "loss": 4.1432,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999783627129264e-07,
+      "loss": 3.9978,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999976440315007e-07,
+      "loss": 4.264,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999744361142325e-07,
+      "loss": 3.6811,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972350110767e-07,
+      "loss": 3.7251,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972350110767e-07,
+      "loss": 3.8368,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999701823047811e-07,
+      "loss": 3.5845,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999967932696452e-07,
+      "loss": 3.899,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999965601285963e-07,
+      "loss": 3.4585,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999631880735062e-07,
+      "loss": 3.6125,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999606930592784e-07,
+      "loss": 3.6257,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999581162434832e-07,
+      "loss": 4.0693,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999955457626332e-07,
+      "loss": 3.7065,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999527172080425e-07,
+      "loss": 3.7486,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999498949888382e-07,
+      "loss": 3.893,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999469909689504e-07,
+      "loss": 4.058,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999440051486168e-07,
+      "loss": 3.6273,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999409375280808e-07,
+      "loss": 3.6819,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999377881075942e-07,
+      "loss": 3.6333,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999345568874146e-07,
+      "loss": 3.6516,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999931243867806e-07,
+      "loss": 3.546,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999278490490395e-07,
+      "loss": 3.3878,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999243724313928e-07,
+      "loss": 3.7691,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999208140151507e-07,
+      "loss": 3.2397,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999171738006036e-07,
+      "loss": 3.7403,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99991345178805e-07,
+      "loss": 3.4515,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999096479777931e-07,
+      "loss": 3.5929,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999057623701462e-07,
+      "loss": 3.3663,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999901794965425e-07,
+      "loss": 3.3285,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998977457639558e-07,
+      "loss": 3.415,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998936147660682e-07,
+      "loss": 3.5152,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998894019721015e-07,
+      "loss": 3.3469,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998851073823996e-07,
+      "loss": 3.4214,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998807309973144e-07,
+      "loss": 3.1945,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998762728172033e-07,
+      "loss": 2.9608,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998717328424313e-07,
+      "loss": 3.1719,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998671110733695e-07,
+      "loss": 3.1507,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998624075103963e-07,
+      "loss": 3.1695,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998576221538964e-07,
+      "loss": 3.22,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998527550042613e-07,
+      "loss": 2.81,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998478060618892e-07,
+      "loss": 3.1095,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998427753271847e-07,
+      "loss": 2.9698,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998376628005596e-07,
+      "loss": 3.1656,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999832468482432e-07,
+      "loss": 2.9181,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998271923732268e-07,
+      "loss": 2.9359,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998218344733755e-07,
+      "loss": 2.9984,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998163947833167e-07,
+      "loss": 2.722,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998108733034954e-07,
+      "loss": 2.9629,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998052700343628e-07,
+      "loss": 2.8923,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999799584976378e-07,
+      "loss": 2.8492,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997938181300054e-07,
+      "loss": 2.9741,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997879694957168e-07,
+      "loss": 2.6408,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997820390739912e-07,
+      "loss": 2.786,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997760268653133e-07,
+      "loss": 2.7375,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999769932870175e-07,
+      "loss": 2.7395,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999763757089075e-07,
+      "loss": 2.7604,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997574995225184e-07,
+      "loss": 2.8015,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999751160171017e-07,
+      "loss": 2.7287,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997447390350893e-07,
+      "loss": 2.8068,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997382361152608e-07,
+      "loss": 2.8497,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997316514120635e-07,
+      "loss": 2.4128,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997249849260357e-07,
+      "loss": 2.573,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997182366577232e-07,
+      "loss": 2.5065,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999711406607678e-07,
+      "loss": 2.6225,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997044947764582e-07,
+      "loss": 2.5938,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996975011646304e-07,
+      "loss": 2.4246,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996904257727657e-07,
+      "loss": 2.3548,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996832686014428e-07,
+      "loss": 2.457,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999676029651248e-07,
+      "loss": 2.3492,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999668708922773e-07,
+      "loss": 2.4699,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999661306416617e-07,
+      "loss": 2.5194,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999653822133385e-07,
+      "loss": 2.4514,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996462560736897e-07,
+      "loss": 2.3588,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996386082381498e-07,
+      "loss": 2.3944,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999630878627391e-07,
+      "loss": 2.3084,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996230672420457e-07,
+      "loss": 2.5286,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999615174082753e-07,
+      "loss": 2.3598,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996071991501585e-07,
+      "loss": 2.3571,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995991424449143e-07,
+      "loss": 2.4826,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99959100396768e-07,
+      "loss": 2.171,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995827837191208e-07,
+      "loss": 2.3149,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995744816999097e-07,
+      "loss": 2.451,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995660979107253e-07,
+      "loss": 2.1898,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995576323522542e-07,
+      "loss": 2.1907,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995490850251882e-07,
+      "loss": 2.4017,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995404559302264e-07,
+      "loss": 2.2633,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995317450680757e-07,
+      "loss": 2.263,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995229524394477e-07,
+      "loss": 2.4532,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995140780450618e-07,
+      "loss": 2.3403,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995051218856448e-07,
+      "loss": 2.2413,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994960839619283e-07,
+      "loss": 2.3032,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999486964274652e-07,
+      "loss": 2.2277,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994777628245626e-07,
+      "loss": 2.2978,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994684796124116e-07,
+      "loss": 2.3325,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994591146389597e-07,
+      "loss": 2.2018,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999449667904972e-07,
+      "loss": 2.4376,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994401394112216e-07,
+      "loss": 2.3082,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994305291584882e-07,
+      "loss": 2.1065,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994208371475577e-07,
+      "loss": 2.2085,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994110633792233e-07,
+      "loss": 2.3053,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994012078542841e-07,
+      "loss": 2.0261,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993912705735465e-07,
+      "loss": 2.1622,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993812515378236e-07,
+      "loss": 2.2939,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993711507479345e-07,
+      "loss": 2.0927,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999360968204706e-07,
+      "loss": 2.1123,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999350703908971e-07,
+      "loss": 2.0177,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993403578615692e-07,
+      "loss": 2.3314,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993299300633466e-07,
+      "loss": 2.0571,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993194205151566e-07,
+      "loss": 2.1681,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993088292178587e-07,
+      "loss": 2.3628,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992981561723194e-07,
+      "loss": 2.2766,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992874013794122e-07,
+      "loss": 2.1332,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999276564840016e-07,
+      "loss": 2.4022,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992656465550182e-07,
+      "loss": 2.1707,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992546465253115e-07,
+      "loss": 2.0751,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999243564751796e-07,
+      "loss": 2.2211,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992324012353778e-07,
+      "loss": 2.1463,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992211559769708e-07,
+      "loss": 2.1774,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999209828977494e-07,
+      "loss": 2.1297,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991984202378748e-07,
+      "loss": 2.1865,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991869297590462e-07,
+      "loss": 2.1475,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999175357541948e-07,
+      "loss": 2.2598,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991637035875275e-07,
+      "loss": 2.2223,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991519678967373e-07,
+      "loss": 2.3596,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999140150470538e-07,
+      "loss": 2.3503,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991282513098957e-07,
+      "loss": 2.1513,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991162704157843e-07,
+      "loss": 2.2112,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999104207789184e-07,
+      "loss": 2.3276,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999092063431081e-07,
+      "loss": 2.2048,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990798373424692e-07,
+      "loss": 2.5556,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990675295243489e-07,
+      "loss": 2.0046,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990551399777265e-07,
+      "loss": 2.0475,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990426687036156e-07,
+      "loss": 2.164,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990301157030368e-07,
+      "loss": 2.1842,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990174809770165e-07,
+      "loss": 2.1348,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999004764526589e-07,
+      "loss": 2.0868,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989919663527937e-07,
+      "loss": 2.2901,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998979086456678e-07,
+      "loss": 2.0219,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989661248392954e-07,
+      "loss": 2.1396,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989530815017064e-07,
+      "loss": 2.3333,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989399564449777e-07,
+      "loss": 2.1894,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989267496701832e-07,
+      "loss": 2.2118,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989134611784035e-07,
+      "loss": 2.13,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989000909707255e-07,
+      "loss": 2.0294,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988866390482425e-07,
+      "loss": 2.2527,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988731054120557e-07,
+      "loss": 1.9564,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988594900632717e-07,
+      "loss": 2.078,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988457930030042e-07,
+      "loss": 2.0749,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998832014232374e-07,
+      "loss": 2.1187,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988181537525085e-07,
+      "loss": 2.0159,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988042115645408e-07,
+      "loss": 2.0392,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987901876696123e-07,
+      "loss": 2.115,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987760820688694e-07,
+      "loss": 2.1423,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987618947634667e-07,
+      "loss": 2.1477,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987476257545644e-07,
+      "loss": 2.1475,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99873327504333e-07,
+      "loss": 2.0204,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987188426309368e-07,
+      "loss": 2.1187,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987043285185665e-07,
+      "loss": 2.1218,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986897327074054e-07,
+      "loss": 2.2679,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986750551986483e-07,
+      "loss": 2.2096,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986602959934958e-07,
+      "loss": 1.9791,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986454550931547e-07,
+      "loss": 2.3815,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986305324988396e-07,
+      "loss": 1.9666,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998615528211771e-07,
+      "loss": 2.1127,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986004422331763e-07,
+      "loss": 2.1968,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985852745642897e-07,
+      "loss": 2.2003,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998570025206352e-07,
+      "loss": 2.0765,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985546941606107e-07,
+      "loss": 2.051,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985392814283198e-07,
+      "loss": 1.8822,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985237870107403e-07,
+      "loss": 2.167,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985082109091395e-07,
+      "loss": 2.1536,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998492553124792e-07,
+      "loss": 2.1892,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998476813658978e-07,
+      "loss": 1.9155,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998460992512986e-07,
+      "loss": 2.2406,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984450896881094e-07,
+      "loss": 2.0965,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984291051856497e-07,
+      "loss": 2.1192,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984130390069138e-07,
+      "loss": 1.9529,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998396891153217e-07,
+      "loss": 2.1266,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983806616258794e-07,
+      "loss": 2.0018,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983643504262292e-07,
+      "loss": 1.9539,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983479575556005e-07,
+      "loss": 2.1674,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983314830153343e-07,
+      "loss": 2.0012,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983149268067782e-07,
+      "loss": 2.3001,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982982889312866e-07,
+      "loss": 2.2651,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998281569390221e-07,
+      "loss": 2.0781,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982647681849486e-07,
+      "loss": 2.2342,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982478853168438e-07,
+      "loss": 2.038,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982309207872882e-07,
+      "loss": 2.0456,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982138745976694e-07,
+      "loss": 2.0642,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981967467493817e-07,
+      "loss": 1.9998,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981795372438261e-07,
+      "loss": 2.0406,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981622460824108e-07,
+      "loss": 1.9361,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99814487326655e-07,
+      "loss": 2.0561,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998127418797665e-07,
+      "loss": 2.1263,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998109882677184e-07,
+      "loss": 1.9618,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998092264906541e-07,
+      "loss": 1.8651,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980745654871773e-07,
+      "loss": 2.1562,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980567844205409e-07,
+      "loss": 1.9445,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980389217080865e-07,
+      "loss": 2.2188,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980209773512752e-07,
+      "loss": 2.0562,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998002951351575e-07,
+      "loss": 2.131,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979848437104602e-07,
+      "loss": 2.0584,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979666544294127e-07,
+      "loss": 1.9959,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979483835099202e-07,
+      "loss": 2.0904,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979300309534772e-07,
+      "loss": 2.3918,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979115967615852e-07,
+      "loss": 1.97,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997893080935752e-07,
+      "loss": 2.0938,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978744834774924e-07,
+      "loss": 1.8647,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978558043883279e-07,
+      "loss": 1.9616,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978370436697863e-07,
+      "loss": 1.8581,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978182013234026e-07,
+      "loss": 2.0703,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997799277350718e-07,
+      "loss": 2.1125,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977802717532803e-07,
+      "loss": 2.0307,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977611845326446e-07,
+      "loss": 2.1216,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977420156903725e-07,
+      "loss": 1.9232,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997722765228032e-07,
+      "loss": 2.0422,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977034331471975e-07,
+      "loss": 2.0217,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976840194494508e-07,
+      "loss": 1.9801,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976645241363802e-07,
+      "loss": 2.141,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976449472095798e-07,
+      "loss": 1.828,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976252886706517e-07,
+      "loss": 2.1371,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997605548521204e-07,
+      "loss": 2.2307,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975857267628513e-07,
+      "loss": 2.0327,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975658233972154e-07,
+      "loss": 2.0997,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975458384259243e-07,
+      "loss": 2.1978,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997525771850613e-07,
+      "loss": 2.1544,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975056236729228e-07,
+      "loss": 1.9377,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997485393894502e-07,
+      "loss": 1.9667,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974650825170056e-07,
+      "loss": 1.9598,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974446895420953e-07,
+      "loss": 2.0457,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974242149714392e-07,
+      "loss": 2.0581,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997403658806712e-07,
+      "loss": 2.0441,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973830210495955e-07,
+      "loss": 1.8937,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997362301701778e-07,
+      "loss": 2.1183,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973415007649546e-07,
+      "loss": 2.0288,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973206182408266e-07,
+      "loss": 2.0416,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972996541311023e-07,
+      "loss": 2.2232,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972786084374968e-07,
+      "loss": 2.0761,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972574811617315e-07,
+      "loss": 2.1222,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972362723055355e-07,
+      "loss": 2.043,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972149818706426e-07,
+      "loss": 2.1165,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971936098587954e-07,
+      "loss": 1.9758,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971721562717416e-07,
+      "loss": 2.0046,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971506211112368e-07,
+      "loss": 1.8654,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971290043790422e-07,
+      "loss": 2.0925,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971073060769265e-07,
+      "loss": 1.816,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970855262066646e-07,
+      "loss": 2.1748,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970636647700378e-07,
+      "loss": 2.0981,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970417217688352e-07,
+      "loss": 2.1133,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970196972048513e-07,
+      "loss": 2.1404,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996997591079888e-07,
+      "loss": 2.0114,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969754033957534e-07,
+      "loss": 2.0934,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969531341542632e-07,
+      "loss": 2.2276,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996930783357239e-07,
+      "loss": 1.8102,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969083510065083e-07,
+      "loss": 2.1611,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968858371039074e-07,
+      "loss": 2.189,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968632416512769e-07,
+      "loss": 2.068,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968405646504663e-07,
+      "loss": 2.1093,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968178061033298e-07,
+      "loss": 1.9606,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967949660117298e-07,
+      "loss": 2.0796,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967720443775345e-07,
+      "loss": 2.0356,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967490412026189e-07,
+      "loss": 2.1125,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967259564888645e-07,
+      "loss": 1.806,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967027902381605e-07,
+      "loss": 2.2048,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966795424524016e-07,
+      "loss": 2.0796,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966562131334894e-07,
+      "loss": 2.0021,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966328022833326e-07,
+      "loss": 1.9103,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996609309903846e-07,
+      "loss": 2.0121,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996585735996952e-07,
+      "loss": 1.9918,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965620805645785e-07,
+      "loss": 1.9991,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965383436086608e-07,
+      "loss": 2.077,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965145251311405e-07,
+      "loss": 2.1753,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964906251339668e-07,
+      "loss": 1.9777,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964666436190937e-07,
+      "loss": 2.0007,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996442580588484e-07,
+      "loss": 1.9876,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996418436044106e-07,
+      "loss": 2.1437,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996394209987934e-07,
+      "loss": 2.0268,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996369902421951e-07,
+      "loss": 1.9671,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963455133481446e-07,
+      "loss": 1.9428,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963210427685106e-07,
+      "loss": 2.2476,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99629649068505e-07,
+      "loss": 1.9274,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962718570997717e-07,
+      "loss": 2.1176,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996247142014691e-07,
+      "loss": 2.048,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962223454318297e-07,
+      "loss": 2.0808,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996197467353216e-07,
+      "loss": 1.8451,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961725077808856e-07,
+      "loss": 1.9793,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961474667168795e-07,
+      "loss": 2.2374,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996122344163247e-07,
+      "loss": 1.9138,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960971401220426e-07,
+      "loss": 2.069,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960718545953282e-07,
+      "loss": 2.003,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960464875851728e-07,
+      "loss": 2.1368,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996021039093651e-07,
+      "loss": 1.9374,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995995509122845e-07,
+      "loss": 2.0364,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959698976748424e-07,
+      "loss": 1.9852,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959442047517396e-07,
+      "loss": 2.0204,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959184303556378e-07,
+      "loss": 1.9465,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958925744886455e-07,
+      "loss": 2.1091,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958666371528774e-07,
+      "loss": 1.9849,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995840618350456e-07,
+      "loss": 1.9764,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958145180835095e-07,
+      "loss": 1.9395,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995788336354173e-07,
+      "loss": 1.9859,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995762073164588e-07,
+      "loss": 1.9465,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957357285169038e-07,
+      "loss": 2.15,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957093024132746e-07,
+      "loss": 1.9698,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956827948558626e-07,
+      "loss": 1.9639,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956562058468363e-07,
+      "loss": 2.1391,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956295353883706e-07,
+      "loss": 2.1025,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956027834826471e-07,
+      "loss": 2.0821,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995575950131855e-07,
+      "loss": 2.0573,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955490353381883e-07,
+      "loss": 1.8958,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955220391038498e-07,
+      "loss": 1.966,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954949614310474e-07,
+      "loss": 2.0766,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954678023219963e-07,
+      "loss": 2.2377,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954405617789182e-07,
+      "loss": 1.9757,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954132398040414e-07,
+      "loss": 2.1842,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953858363996012e-07,
+      "loss": 1.9828,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953583515678392e-07,
+      "loss": 2.0575,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953307853110036e-07,
+      "loss": 2.1322,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953031376313495e-07,
+      "loss": 2.1712,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952754085311394e-07,
+      "loss": 1.8022,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952475980126406e-07,
+      "loss": 1.8727,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952197060781287e-07,
+      "loss": 1.9438,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951917327298852e-07,
+      "loss": 1.9142,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951636779701987e-07,
+      "loss": 1.7922,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951355418013635e-07,
+      "loss": 2.0459,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951073242256825e-07,
+      "loss": 2.1181,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995079025245463e-07,
+      "loss": 2.0979,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950506448630203e-07,
+      "loss": 1.9916,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995022183080676e-07,
+      "loss": 2.117,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949936399007587e-07,
+      "loss": 2.1285,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994965015325603e-07,
+      "loss": 2.2083,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949363093575508e-07,
+      "loss": 2.0232,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99490752199895e-07,
+      "loss": 2.0336,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948786532521563e-07,
+      "loss": 1.9176,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948497031195306e-07,
+      "loss": 1.96,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948206716034415e-07,
+      "loss": 2.0002,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947915587062635e-07,
+      "loss": 2.0089,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947623644303787e-07,
+      "loss": 1.9669,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994733088778175e-07,
+      "loss": 2.0039,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947037317520472e-07,
+      "loss": 1.9567,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946742933543973e-07,
+      "loss": 1.7922,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946447735876332e-07,
+      "loss": 1.9348,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946151724541698e-07,
+      "loss": 1.8996,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945854899564285e-07,
+      "loss": 2.0148,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945557260968377e-07,
+      "loss": 2.0167,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994525880877832e-07,
+      "loss": 2.1915,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994495954301853e-07,
+      "loss": 1.9562,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944659463713488e-07,
+      "loss": 2.0501,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944358570887743e-07,
+      "loss": 2.0298,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994405686456591e-07,
+      "loss": 1.7929,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943754344772666e-07,
+      "loss": 1.8414,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943451011532762e-07,
+      "loss": 2.1098,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994314686487101e-07,
+      "loss": 2.0105,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942841904812292e-07,
+      "loss": 2.0887,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942536131381558e-07,
+      "loss": 2.2097,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994222954460382e-07,
+      "loss": 1.9257,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941922144504151e-07,
+      "loss": 2.0935,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941613931107708e-07,
+      "loss": 2.0491,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99413049044397e-07,
+      "loss": 2.1365,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940995064525404e-07,
+      "loss": 1.7997,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940684411390176e-07,
+      "loss": 2.0554,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940372945059418e-07,
+      "loss": 1.8915,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940060665558614e-07,
+      "loss": 1.8736,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993974757291331e-07,
+      "loss": 2.0731,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939433667149119e-07,
+      "loss": 2.1761,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939118948291719e-07,
+      "loss": 2.1037,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938803416366853e-07,
+      "loss": 2.0351,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938487071400342e-07,
+      "loss": 2.0628,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938169913418053e-07,
+      "loss": 2.0682,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993785194244594e-07,
+      "loss": 2.1079,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937533158510006e-07,
+      "loss": 1.9638,
+      "step": 738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937213561636338e-07,
+      "loss": 2.0132,
+      "step": 739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936893151851077e-07,
+      "loss": 2.0779,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936571929180433e-07,
+      "loss": 2.1761,
+      "step": 741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936249893650683e-07,
+      "loss": 1.9428,
+      "step": 742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935927045288173e-07,
+      "loss": 1.9735,
+      "step": 743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935603384119312e-07,
+      "loss": 2.0988,
+      "step": 744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935278910170576e-07,
+      "loss": 1.9703,
+      "step": 745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934953623468512e-07,
+      "loss": 1.8234,
+      "step": 746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993462752403973e-07,
+      "loss": 1.9303,
+      "step": 747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934300611910904e-07,
+      "loss": 2.2346,
+      "step": 748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933972887108776e-07,
+      "loss": 1.9744,
+      "step": 749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933644349660162e-07,
+      "loss": 2.0995,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933314999591925e-07,
+      "loss": 1.9549,
+      "step": 751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993298483693102e-07,
+      "loss": 2.2063,
+      "step": 752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932653861704454e-07,
+      "loss": 1.9567,
+      "step": 753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932322073939297e-07,
+      "loss": 2.0855,
+      "step": 754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931989473662694e-07,
+      "loss": 2.0717,
+      "step": 755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931656060901854e-07,
+      "loss": 1.988,
+      "step": 756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993132183568405e-07,
+      "loss": 2.0987,
+      "step": 757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930986798036623e-07,
+      "loss": 2.0281,
+      "step": 758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930650947986983e-07,
+      "loss": 2.1791,
+      "step": 759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99303142855626e-07,
+      "loss": 1.8802,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992997681079102e-07,
+      "loss": 2.0146,
+      "step": 761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929638523699842e-07,
+      "loss": 1.8485,
+      "step": 762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929299424316753e-07,
+      "loss": 2.0983,
+      "step": 763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928959512669481e-07,
+      "loss": 1.883,
+      "step": 764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928618788785837e-07,
+      "loss": 2.037,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992827725269369e-07,
+      "loss": 2.0556,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927934904420987e-07,
+      "loss": 2.1148,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927591743995723e-07,
+      "loss": 1.9575,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927247771445982e-07,
+      "loss": 1.9996,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926902986799895e-07,
+      "loss": 1.9509,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926557390085667e-07,
+      "loss": 2.1249,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926210981331574e-07,
+      "loss": 1.8738,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992586376056595e-07,
+      "loss": 1.8422,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99255157278172e-07,
+      "loss": 1.9223,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925166883113798e-07,
+      "loss": 2.0373,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924817226484277e-07,
+      "loss": 1.8726,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924466757957239e-07,
+      "loss": 1.7855,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992411547756136e-07,
+      "loss": 2.0362,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923763385325374e-07,
+      "loss": 1.776,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923410481278082e-07,
+      "loss": 2.0999,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923056765448355e-07,
+      "loss": 2.0666,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922702237865127e-07,
+      "loss": 1.9974,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922346898557404e-07,
+      "loss": 2.0868,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992199074755425e-07,
+      "loss": 1.7643,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921633784884801e-07,
+      "loss": 1.8832,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921276010578258e-07,
+      "loss": 2.0442,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992091742466389e-07,
+      "loss": 2.0564,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920558027171033e-07,
+      "loss": 2.0817,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920197818129082e-07,
+      "loss": 2.0147,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919836797567509e-07,
+      "loss": 1.8131,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919474965515843e-07,
+      "loss": 1.8501,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919112322003687e-07,
+      "loss": 1.8737,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918748867060704e-07,
+      "loss": 2.1535,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918384600716629e-07,
+      "loss": 2.0985,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991801952300126e-07,
+      "loss": 2.0196,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991765363394446e-07,
+      "loss": 1.898,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917286933576162e-07,
+      "loss": 2.0605,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916919421926365e-07,
+      "loss": 2.1986,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991655109902513e-07,
+      "loss": 1.8766,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916181964902595e-07,
+      "loss": 1.9842,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991581201958895e-07,
+      "loss": 1.8618,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915441263114457e-07,
+      "loss": 2.0763,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991506969550945e-07,
+      "loss": 1.9765,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914697316804325e-07,
+      "loss": 2.0527,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991432412702954e-07,
+      "loss": 2.0871,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991395012621563e-07,
+      "loss": 1.9736,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913575314393186e-07,
+      "loss": 1.8748,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991319969159287e-07,
+      "loss": 2.0093,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991282325784541e-07,
+      "loss": 2.0302,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99124460131816e-07,
+      "loss": 1.943,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99120679576323e-07,
+      "loss": 2.0741,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911689091228436e-07,
+      "loss": 1.7161,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911309414001006e-07,
+      "loss": 2.037,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910928925981063e-07,
+      "loss": 1.855,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910547627199733e-07,
+      "loss": 2.064,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910165517688218e-07,
+      "loss": 2.0134,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909782597477761e-07,
+      "loss": 1.9398,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909398866599697e-07,
+      "loss": 2.045,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990901432508542e-07,
+      "loss": 1.9583,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908628972966378e-07,
+      "loss": 2.1484,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908242810274097e-07,
+      "loss": 1.9606,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907855837040167e-07,
+      "loss": 2.0077,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907468053296248e-07,
+      "loss": 2.0784,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907079459074062e-07,
+      "loss": 1.9675,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906690054405395e-07,
+      "loss": 1.99,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906299839322105e-07,
+      "loss": 2.0721,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990590881385611e-07,
+      "loss": 2.0333,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99055169780394e-07,
+      "loss": 1.8831,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990512433190403e-07,
+      "loss": 1.9308,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904730875482113e-07,
+      "loss": 1.8449,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904336608805847e-07,
+      "loss": 2.0064,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990394153190748e-07,
+      "loss": 2.0428,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903545644819332e-07,
+      "loss": 1.749,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990314894757378e-07,
+      "loss": 1.9373,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990275144020329e-07,
+      "loss": 2.1588,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990235312274037e-07,
+      "loss": 1.9184,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901953995217606e-07,
+      "loss": 1.8945,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990155405766765e-07,
+      "loss": 2.1088,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901153310123223e-07,
+      "loss": 1.963,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99007517526171e-07,
+      "loss": 2.1221,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900349385182134e-07,
+      "loss": 2.0653,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899946207851239e-07,
+      "loss": 1.8861,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899542220657398e-07,
+      "loss": 2.007,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899137423633664e-07,
+      "loss": 1.908,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898731816813143e-07,
+      "loss": 2.0624,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898325400229022e-07,
+      "loss": 2.0519,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897918173914543e-07,
+      "loss": 1.8678,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989751013790302e-07,
+      "loss": 1.9097,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897101292227836e-07,
+      "loss": 1.7432,
+      "step": 849
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9896691636922433e-07,
+      "loss": 2.0403,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9896281172020322e-07,
+      "loss": 1.8302,
+      "step": 851
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9895869897555087e-07,
+      "loss": 2.0273,
+      "step": 852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9895457813560362e-07,
+      "loss": 1.9692,
+      "step": 853
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9895044920069867e-07,
+      "loss": 1.7768,
+      "step": 854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894631217117377e-07,
+      "loss": 1.7702,
+      "step": 855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989421670473673e-07,
+      "loss": 1.9507,
+      "step": 856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989380138296184e-07,
+      "loss": 1.9785,
+      "step": 857
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989338525182668e-07,
+      "loss": 1.948,
+      "step": 858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989296831136529e-07,
+      "loss": 1.8614,
+      "step": 859
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892550561611784e-07,
+      "loss": 1.8685,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892132002600326e-07,
+      "loss": 1.7887,
+      "step": 861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891712634365163e-07,
+      "loss": 1.8351,
+      "step": 862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891292456940603e-07,
+      "loss": 1.9966,
+      "step": 863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890871470361012e-07,
+      "loss": 1.9802,
+      "step": 864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890449674660834e-07,
+      "loss": 2.1552,
+      "step": 865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989002706987457e-07,
+      "loss": 2.0129,
+      "step": 866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889603656036797e-07,
+      "loss": 1.8237,
+      "step": 867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889179433182148e-07,
+      "loss": 2.0095,
+      "step": 868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888754401345325e-07,
+      "loss": 1.9629,
+      "step": 869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98883285605611e-07,
+      "loss": 2.107,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988790191086431e-07,
+      "loss": 1.9828,
+      "step": 871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887474452289852e-07,
+      "loss": 1.7803,
+      "step": 872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98870461848727e-07,
+      "loss": 1.7492,
+      "step": 873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886617108647887e-07,
+      "loss": 1.9275,
+      "step": 874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988618722365051e-07,
+      "loss": 1.8172,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885756529915738e-07,
+      "loss": 2.0175,
+      "step": 876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885325027478804e-07,
+      "loss": 1.9701,
+      "step": 877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884892716375009e-07,
+      "loss": 1.8121,
+      "step": 878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884459596639714e-07,
+      "loss": 2.0341,
+      "step": 879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884025668308353e-07,
+      "loss": 2.2532,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883590931416424e-07,
+      "loss": 2.0103,
+      "step": 881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883155385999486e-07,
+      "loss": 2.1588,
+      "step": 882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882719032093172e-07,
+      "loss": 2.1301,
+      "step": 883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988228186973318e-07,
+      "loss": 1.9026,
+      "step": 884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881843898955267e-07,
+      "loss": 2.0146,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881405119795265e-07,
+      "loss": 2.1298,
+      "step": 886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880965532289066e-07,
+      "loss": 2.0369,
+      "step": 887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880525136472635e-07,
+      "loss": 1.6002,
+      "step": 888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988008393238199e-07,
+      "loss": 1.896,
+      "step": 889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987964192005323e-07,
+      "loss": 1.9979,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879199099522513e-07,
+      "loss": 2.0448,
+      "step": 891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878755470826064e-07,
+      "loss": 2.0142,
+      "step": 892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878311034000171e-07,
+      "loss": 2.1048,
+      "step": 893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877865789081193e-07,
+      "loss": 1.9032,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877419736105555e-07,
+      "loss": 2.1556,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876972875109743e-07,
+      "loss": 1.8651,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876525206130314e-07,
+      "loss": 1.9092,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987607672920389e-07,
+      "loss": 1.9512,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875627444367157e-07,
+      "loss": 2.0059,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875177351656868e-07,
+      "loss": 1.9125,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987472645110985e-07,
+      "loss": 1.966,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874274742762975e-07,
+      "loss": 1.9101,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987382222665321e-07,
+      "loss": 2.0729,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873368902817562e-07,
+      "loss": 1.8941,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872914771293118e-07,
+      "loss": 1.9251,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872459832117033e-07,
+      "loss": 1.9824,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872004085326516e-07,
+      "loss": 2.0279,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871547530958852e-07,
+      "loss": 1.8949,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871090169051393e-07,
+      "loss": 1.9598,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870631999641546e-07,
+      "loss": 1.8748,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870173022766798e-07,
+      "loss": 1.8601,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869713238464695e-07,
+      "loss": 1.9766,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869252646772846e-07,
+      "loss": 1.9739,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868791247728934e-07,
+      "loss": 1.9369,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868329041370698e-07,
+      "loss": 1.9251,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867866027735952e-07,
+      "loss": 2.0108,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867402206862573e-07,
+      "loss": 1.9588,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866937578788506e-07,
+      "loss": 1.9939,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866472143551753e-07,
+      "loss": 1.9192,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866005901190395e-07,
+      "loss": 2.0165,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865538851742567e-07,
+      "loss": 1.8928,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865070995246486e-07,
+      "loss": 1.9171,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864602331740412e-07,
+      "loss": 1.7968,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864132861262694e-07,
+      "loss": 1.8122,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863662583851732e-07,
+      "loss": 1.8952,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863191499546e-07,
+      "loss": 2.0076,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862719608384032e-07,
+      "loss": 1.8167,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986224691040443e-07,
+      "loss": 2.1258,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861773405645866e-07,
+      "loss": 2.056,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986129909414708e-07,
+      "loss": 1.7433,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986082397594686e-07,
+      "loss": 1.7649,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860348051084085e-07,
+      "loss": 1.8901,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859871319597678e-07,
+      "loss": 1.8887,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859393781526648e-07,
+      "loss": 1.9107,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858915436910052e-07,
+      "loss": 2.2182,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858436285787022e-07,
+      "loss": 2.0295,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857956328196758e-07,
+      "loss": 1.8832,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857475564178522e-07,
+      "loss": 1.8261,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985699399377164e-07,
+      "loss": 2.0451,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985651161701551e-07,
+      "loss": 2.0913,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985602843394959e-07,
+      "loss": 2.139,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855544444613412e-07,
+      "loss": 1.9566,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855059649046565e-07,
+      "loss": 1.8659,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854574047288703e-07,
+      "loss": 1.8776,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854087639379557e-07,
+      "loss": 1.9549,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853600425358916e-07,
+      "loss": 1.9921,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853112405266636e-07,
+      "loss": 1.8791,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985262357914264e-07,
+      "loss": 1.7945,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852133947026914e-07,
+      "loss": 1.9272,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851643508959514e-07,
+      "loss": 1.9448,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851152264980562e-07,
+      "loss": 1.9055,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985066021513024e-07,
+      "loss": 1.9685,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850167359448803e-07,
+      "loss": 2.0452,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984967369797657e-07,
+      "loss": 1.9935,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984917923075392e-07,
+      "loss": 1.931,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848683957821306e-07,
+      "loss": 1.7065,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848187879219246e-07,
+      "loss": 1.8036,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847690994988318e-07,
+      "loss": 2.0339,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984719330516917e-07,
+      "loss": 2.0722,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984669480980252e-07,
+      "loss": 2.1021,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846195508929137e-07,
+      "loss": 2.039,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845695402589875e-07,
+      "loss": 1.8151,
+      "step": 962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9845194490825643e-07,
+      "loss": 1.7174,
+      "step": 963
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984469277367742e-07,
+      "loss": 2.1357,
+      "step": 964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9844190251186242e-07,
+      "loss": 1.9686,
+      "step": 965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9843686923393223e-07,
+      "loss": 1.7817,
+      "step": 966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984318279033954e-07,
+      "loss": 1.8839,
+      "step": 967
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842677852066427e-07,
+      "loss": 1.9006,
+      "step": 968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842172108615193e-07,
+      "loss": 1.9925,
+      "step": 969
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841665560027217e-07,
+      "loss": 1.7197,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841158206343923e-07,
+      "loss": 1.8632,
+      "step": 971
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984065004760683e-07,
+      "loss": 1.9068,
+      "step": 972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98401410838575e-07,
+      "loss": 2.0575,
+      "step": 973
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839631315137564e-07,
+      "loss": 1.9483,
+      "step": 974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839120741488733e-07,
+      "loss": 1.8561,
+      "step": 975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838609362952768e-07,
+      "loss": 2.0549,
+      "step": 976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838097179571507e-07,
+      "loss": 2.0712,
+      "step": 977
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837584191386843e-07,
+      "loss": 2.0153,
+      "step": 978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837070398440748e-07,
+      "loss": 1.9671,
+      "step": 979
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983655580077525e-07,
+      "loss": 2.0452,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836040398432443e-07,
+      "loss": 2.0344,
+      "step": 981
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835524191454492e-07,
+      "loss": 1.9208,
+      "step": 982
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835007179883624e-07,
+      "loss": 2.1596,
+      "step": 983
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983448936376213e-07,
+      "loss": 1.847,
+      "step": 984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833970743132377e-07,
+      "loss": 1.8601,
+      "step": 985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983345131803679e-07,
+      "loss": 1.9483,
+      "step": 986
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983293108851785e-07,
+      "loss": 1.8388,
+      "step": 987
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832410054618129e-07,
+      "loss": 1.8939,
+      "step": 988
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831888216380234e-07,
+      "loss": 2.0435,
+      "step": 989
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831365573846869e-07,
+      "loss": 1.8303,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830842127060778e-07,
+      "loss": 2.0251,
+      "step": 991
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983031787606479e-07,
+      "loss": 1.8736,
+      "step": 992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829792820901784e-07,
+      "loss": 1.8942,
+      "step": 993
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829266961614715e-07,
+      "loss": 1.7597,
+      "step": 994
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828740298246603e-07,
+      "loss": 2.0472,
+      "step": 995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828212830840524e-07,
+      "loss": 2.0213,
+      "step": 996
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982768455943964e-07,
+      "loss": 2.0177,
+      "step": 997
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827155484087154e-07,
+      "loss": 2.0315,
+      "step": 998
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982662560482635e-07,
+      "loss": 2.0234,
+      "step": 999
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982609492170058e-07,
+      "loss": 1.9457,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825563434753253e-07,
+      "loss": 1.9645,
+      "step": 1001
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825031144027844e-07,
+      "loss": 2.0036,
+      "step": 1002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98244980495679e-07,
+      "loss": 1.7961,
+      "step": 1003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982396415141703e-07,
+      "loss": 1.8238,
+      "step": 1004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823429449618913e-07,
+      "loss": 1.8792,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822893944217284e-07,
+      "loss": 1.7909,
+      "step": 1006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822357635255956e-07,
+      "loss": 2.0498,
+      "step": 1007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821820522778795e-07,
+      "loss": 1.9786,
+      "step": 1008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821282606829746e-07,
+      "loss": 1.6736,
+      "step": 1009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820743887452808e-07,
+      "loss": 2.0559,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820204364692055e-07,
+      "loss": 1.8967,
+      "step": 1011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819664038591618e-07,
+      "loss": 2.0758,
+      "step": 1012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819122909195703e-07,
+      "loss": 1.9372,
+      "step": 1013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818580976548574e-07,
+      "loss": 1.9424,
+      "step": 1014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818038240694563e-07,
+      "loss": 2.0424,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981749470167807e-07,
+      "loss": 1.9549,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981695035954356e-07,
+      "loss": 1.8575,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816405214335562e-07,
+      "loss": 1.9748,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981585926609867e-07,
+      "loss": 1.8041,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815312514877548e-07,
+      "loss": 1.9109,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814764960716918e-07,
+      "loss": 1.9672,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814216603661577e-07,
+      "loss": 1.8782,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813667443756383e-07,
+      "loss": 2.1748,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981311748104626e-07,
+      "loss": 1.9118,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812566715576194e-07,
+      "loss": 2.1963,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812015147391241e-07,
+      "loss": 2.1552,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811462776536528e-07,
+      "loss": 1.8698,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810909603057233e-07,
+      "loss": 2.0302,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810355626998616e-07,
+      "loss": 2.1429,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980980084840599e-07,
+      "loss": 1.9509,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809245267324738e-07,
+      "loss": 1.7505,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808688883800314e-07,
+      "loss": 1.8933,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808131697878227e-07,
+      "loss": 1.9834,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807573709604062e-07,
+      "loss": 1.9149,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807014919023464e-07,
+      "loss": 2.0273,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806455326182144e-07,
+      "loss": 1.9833,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805894931125876e-07,
+      "loss": 1.9167,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805333733900512e-07,
+      "loss": 1.9433,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804771734551952e-07,
+      "loss": 2.0633,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804208933126172e-07,
+      "loss": 1.9708,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803645329669215e-07,
+      "loss": 1.889,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803080924227186e-07,
+      "loss": 2.0516,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802515716846253e-07,
+      "loss": 1.9637,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801949707572657e-07,
+      "loss": 1.9207,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801382896452693e-07,
+      "loss": 1.767,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800815283532738e-07,
+      "loss": 1.9277,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800246868859218e-07,
+      "loss": 1.8599,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799677652478633e-07,
+      "loss": 1.9555,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799107634437554e-07,
+      "loss": 1.7847,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798536814782606e-07,
+      "loss": 1.8565,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797965193560486e-07,
+      "loss": 1.8527,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797392770817955e-07,
+      "loss": 1.8692,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979681954660184e-07,
+      "loss": 1.9668,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796245520959033e-07,
+      "loss": 1.7095,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795670693936493e-07,
+      "loss": 1.9751,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979509506558124e-07,
+      "loss": 1.9917,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794518635940368e-07,
+      "loss": 1.803,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793941405061033e-07,
+      "loss": 1.9869,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793363372990446e-07,
+      "loss": 1.968,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792784539775904e-07,
+      "loss": 2.0119,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979220490546475e-07,
+      "loss": 1.8177,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791624470104406e-07,
+      "loss": 1.9596,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791043233742352e-07,
+      "loss": 1.8548,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790461196426137e-07,
+      "loss": 1.9339,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789878358203374e-07,
+      "loss": 1.8875,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978929471912174e-07,
+      "loss": 1.927,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788710279228985e-07,
+      "loss": 2.0306,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978812503857291e-07,
+      "loss": 1.8755,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787538997201398e-07,
+      "loss": 1.8304,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786952155162387e-07,
+      "loss": 1.9711,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786364512503888e-07,
+      "loss": 2.1136,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785776069273965e-07,
+      "loss": 1.7651,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978518682552076e-07,
+      "loss": 1.9997,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784596781292478e-07,
+      "loss": 2.1913,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784005936637384e-07,
+      "loss": 1.9555,
+      "step": 1075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9783414291603814e-07,
+      "loss": 1.9486,
+      "step": 1076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9782821846240163e-07,
+      "loss": 1.7755,
+      "step": 1077
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9782228600594902e-07,
+      "loss": 1.9721,
+      "step": 1078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9781634554716559e-07,
+      "loss": 1.9498,
+      "step": 1079
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9781039708653726e-07,
+      "loss": 1.9144,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9780444062455071e-07,
+      "loss": 2.0666,
+      "step": 1081
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779847616169315e-07,
+      "loss": 1.6283,
+      "step": 1082
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779250369845254e-07,
+      "loss": 1.8816,
+      "step": 1083
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778652323531744e-07,
+      "loss": 1.9252,
+      "step": 1084
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778053477277708e-07,
+      "loss": 1.8853,
+      "step": 1085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777453831132135e-07,
+      "loss": 1.9388,
+      "step": 1086
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776853385144078e-07,
+      "loss": 1.9728,
+      "step": 1087
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776252139362654e-07,
+      "loss": 1.8661,
+      "step": 1088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775650093837055e-07,
+      "loss": 2.0905,
+      "step": 1089
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775047248616525e-07,
+      "loss": 1.9634,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977444360375038e-07,
+      "loss": 1.8288,
+      "step": 1091
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773839159288006e-07,
+      "loss": 1.8976,
+      "step": 1092
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773233915278844e-07,
+      "loss": 1.9679,
+      "step": 1093
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772627871772406e-07,
+      "loss": 2.0507,
+      "step": 1094
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772021028818276e-07,
+      "loss": 1.979,
+      "step": 1095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977141338646609e-07,
+      "loss": 1.7169,
+      "step": 1096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770804944765557e-07,
+      "loss": 1.8602,
+      "step": 1097
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770195703766447e-07,
+      "loss": 1.8287,
+      "step": 1098
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976958566351861e-07,
+      "loss": 1.9077,
+      "step": 1099
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768974824071938e-07,
+      "loss": 1.8444,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976836318547641e-07,
+      "loss": 1.8328,
+      "step": 1101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767750747782054e-07,
+      "loss": 1.7031,
+      "step": 1102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767137511038974e-07,
+      "loss": 1.9797,
+      "step": 1103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766523475297335e-07,
+      "loss": 2.0639,
+      "step": 1104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976590864060737e-07,
+      "loss": 1.7705,
+      "step": 1105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976529300701937e-07,
+      "loss": 1.7484,
+      "step": 1106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97646765745837e-07,
+      "loss": 2.0765,
+      "step": 1107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764059343350788e-07,
+      "loss": 1.8656,
+      "step": 1108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763441313371127e-07,
+      "loss": 1.9168,
+      "step": 1109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762822484695274e-07,
+      "loss": 1.8277,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976220285737385e-07,
+      "loss": 1.8551,
+      "step": 1111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761582431457545e-07,
+      "loss": 1.9705,
+      "step": 1112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760961206997114e-07,
+      "loss": 1.94,
+      "step": 1113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760339184043375e-07,
+      "loss": 1.8268,
+      "step": 1114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975971636264721e-07,
+      "loss": 1.9576,
+      "step": 1115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759092742859575e-07,
+      "loss": 1.904,
+      "step": 1116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975846832473148e-07,
+      "loss": 1.9797,
+      "step": 1117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757843108314004e-07,
+      "loss": 1.8627,
+      "step": 1118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757217093658302e-07,
+      "loss": 2.1779,
+      "step": 1119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756590280815572e-07,
+      "loss": 2.1336,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755962669837099e-07,
+      "loss": 1.8853,
+      "step": 1121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755334260774221e-07,
+      "loss": 1.792,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754705053678348e-07,
+      "loss": 1.7067,
+      "step": 1123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975407504860095e-07,
+      "loss": 1.8764,
+      "step": 1124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753444245593564e-07,
+      "loss": 1.8508,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752812644707794e-07,
+      "loss": 1.9724,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752180245995306e-07,
+      "loss": 1.8926,
+      "step": 1127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751547049507839e-07,
+      "loss": 1.9216,
+      "step": 1128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750913055297182e-07,
+      "loss": 1.7578,
+      "step": 1129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750278263415207e-07,
+      "loss": 1.9893,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974964267391384e-07,
+      "loss": 1.8722,
+      "step": 1131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749006286845075e-07,
+      "loss": 1.9768,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748369102260968e-07,
+      "loss": 1.9824,
+      "step": 1133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747731120213653e-07,
+      "loss": 1.9874,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747092340755313e-07,
+      "loss": 1.6646,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97464527639382e-07,
+      "loss": 2.0093,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745812389814645e-07,
+      "loss": 1.7783,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745171218437025e-07,
+      "loss": 1.8666,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744529249857792e-07,
+      "loss": 1.7781,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743886484129465e-07,
+      "loss": 2.0597,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743242921304627e-07,
+      "loss": 1.6979,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974259856143592e-07,
+      "loss": 1.7743,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741953404576054e-07,
+      "loss": 1.8096,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741307450777812e-07,
+      "loss": 2.024,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740660700094036e-07,
+      "loss": 1.8479,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740013152577625e-07,
+      "loss": 1.8704,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973936480828156e-07,
+      "loss": 1.7397,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738715667258876e-07,
+      "loss": 1.8961,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738065729562677e-07,
+      "loss": 1.8792,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737414995246127e-07,
+      "loss": 1.8024,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736763464362462e-07,
+      "loss": 1.8769,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973611113696498e-07,
+      "loss": 2.0001,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735458013107046e-07,
+      "loss": 1.7891,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734804092842086e-07,
+      "loss": 2.0263,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734149376223596e-07,
+      "loss": 1.8208,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733493863305135e-07,
+      "loss": 1.8826,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732837554140325e-07,
+      "loss": 1.8263,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732180448782856e-07,
+      "loss": 2.0317,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973152254728648e-07,
+      "loss": 1.9502,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730863849705023e-07,
+      "loss": 1.9454,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730204356092367e-07,
+      "loss": 1.9419,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972954406650246e-07,
+      "loss": 2.0254,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728882980989315e-07,
+      "loss": 1.8354,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728221099607017e-07,
+      "loss": 1.9393,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727558422409704e-07,
+      "loss": 2.05,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726894949451593e-07,
+      "loss": 1.972,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972623068078696e-07,
+      "loss": 1.806,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725565616470137e-07,
+      "loss": 1.8768,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724899756555539e-07,
+      "loss": 1.8566,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972423310109763e-07,
+      "loss": 2.1425,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723565650150947e-07,
+      "loss": 1.9131,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722897403770094e-07,
+      "loss": 1.9435,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722228362009732e-07,
+      "loss": 1.8026,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721558524924592e-07,
+      "loss": 1.8617,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720887892569476e-07,
+      "loss": 1.9002,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972021646499924e-07,
+      "loss": 2.0652,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719544242268808e-07,
+      "loss": 1.9472,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718871224433178e-07,
+      "loss": 1.9899,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97181974115474e-07,
+      "loss": 2.0571,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717522803666596e-07,
+      "loss": 1.8423,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716847400845956e-07,
+      "loss": 1.9628,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716171203140726e-07,
+      "loss": 2.0837,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715494210606226e-07,
+      "loss": 1.8585,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714816423297834e-07,
+      "loss": 1.9354,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714137841270997e-07,
+      "loss": 1.9752,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971345846458123e-07,
+      "loss": 1.8909,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712778293284106e-07,
+      "loss": 1.8411,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712097327435266e-07,
+      "loss": 1.8633,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711415567090418e-07,
+      "loss": 1.8656,
+      "step": 1189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9710733012305335e-07,
+      "loss": 2.0105,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9710049663135846e-07,
+      "loss": 2.0941,
+      "step": 1191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9709365519637858e-07,
+      "loss": 1.927,
+      "step": 1192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970868058186734e-07,
+      "loss": 1.786,
+      "step": 1193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9707994849880313e-07,
+      "loss": 1.8723,
+      "step": 1194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9707308323732883e-07,
+      "loss": 1.8945,
+      "step": 1195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706621003481207e-07,
+      "loss": 1.9355,
+      "step": 1196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705932889181512e-07,
+      "loss": 1.8003,
+      "step": 1197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705243980890086e-07,
+      "loss": 1.9227,
+      "step": 1198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970455427866329e-07,
+      "loss": 1.9233,
+      "step": 1199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970386378255754e-07,
+      "loss": 1.8618,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970317249262933e-07,
+      "loss": 1.8052,
+      "step": 1201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97024804089352e-07,
+      "loss": 1.9745,
+      "step": 1202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701787531531772e-07,
+      "loss": 1.8947,
+      "step": 1203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701093860475728e-07,
+      "loss": 1.7958,
+      "step": 1204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970039939582381e-07,
+      "loss": 2.0361,
+      "step": 1205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969970413763283e-07,
+      "loss": 1.8176,
+      "step": 1206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699008085959663e-07,
+      "loss": 1.8404,
+      "step": 1207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969831124086125e-07,
+      "loss": 1.8958,
+      "step": 1208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697613602394596e-07,
+      "loss": 2.0067,
+      "step": 1209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969691517061677e-07,
+      "loss": 1.8523,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696215945584908e-07,
+      "loss": 1.7676,
+      "step": 1211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969551592735621e-07,
+      "loss": 2.0352,
+      "step": 1212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694815115987944e-07,
+      "loss": 1.8935,
+      "step": 1213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694113511537433e-07,
+      "loss": 1.9792,
+      "step": 1214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693411114062077e-07,
+      "loss": 1.8237,
+      "step": 1215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692707923619333e-07,
+      "loss": 2.0793,
+      "step": 1216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969200394026673e-07,
+      "loss": 1.8064,
+      "step": 1217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969129916406185e-07,
+      "loss": 1.8709,
+      "step": 1218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690593595062347e-07,
+      "loss": 2.1747,
+      "step": 1219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689887233325946e-07,
+      "loss": 1.7799,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968918007891043e-07,
+      "loss": 1.7677,
+      "step": 1221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688472131873643e-07,
+      "loss": 1.9692,
+      "step": 1222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687763392273502e-07,
+      "loss": 1.9857,
+      "step": 1223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687053860167983e-07,
+      "loss": 1.8846,
+      "step": 1224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968634353561513e-07,
+      "loss": 1.7713,
+      "step": 1225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685632418673057e-07,
+      "loss": 1.9622,
+      "step": 1226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684920509399924e-07,
+      "loss": 1.8543,
+      "step": 1227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968420780785398e-07,
+      "loss": 1.9731,
+      "step": 1228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683494314093525e-07,
+      "loss": 2.0344,
+      "step": 1229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968278002817692e-07,
+      "loss": 1.9653,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682064950162602e-07,
+      "loss": 1.9766,
+      "step": 1231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681349080109068e-07,
+      "loss": 1.8473,
+      "step": 1232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680632418074882e-07,
+      "loss": 1.9313,
+      "step": 1233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679914964118663e-07,
+      "loss": 2.0946,
+      "step": 1234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679196718299108e-07,
+      "loss": 2.0642,
+      "step": 1235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967847768067497e-07,
+      "loss": 1.9509,
+      "step": 1236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677757851305074e-07,
+      "loss": 1.8988,
+      "step": 1237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96770372302483e-07,
+      "loss": 1.8938,
+      "step": 1238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96763158175636e-07,
+      "loss": 1.7786,
+      "step": 1239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675593613309992e-07,
+      "loss": 1.8088,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674870617546552e-07,
+      "loss": 1.9747,
+      "step": 1241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674146830332425e-07,
+      "loss": 1.8649,
+      "step": 1242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673422251726823e-07,
+      "loss": 2.0778,
+      "step": 1243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672696881789017e-07,
+      "loss": 2.0144,
+      "step": 1244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671970720578348e-07,
+      "loss": 1.9007,
+      "step": 1245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967124376815422e-07,
+      "loss": 1.974,
+      "step": 1246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670516024576096e-07,
+      "loss": 1.887,
+      "step": 1247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669787489903516e-07,
+      "loss": 2.0737,
+      "step": 1248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669058164196072e-07,
+      "loss": 1.8764,
+      "step": 1249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966832804751343e-07,
+      "loss": 1.9416,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667597139915314e-07,
+      "loss": 1.8607,
+      "step": 1251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666865441461518e-07,
+      "loss": 1.9846,
+      "step": 1252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96661329522119e-07,
+      "loss": 1.6323,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665399672226374e-07,
+      "loss": 1.8589,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664665601564933e-07,
+      "loss": 1.9102,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663930740287623e-07,
+      "loss": 1.9544,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663195088454564e-07,
+      "loss": 1.7448,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662458646125932e-07,
+      "loss": 2.1772,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661721413361977e-07,
+      "loss": 2.1147,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660983390222997e-07,
+      "loss": 1.7168,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660244576769376e-07,
+      "loss": 1.9957,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659504973061548e-07,
+      "loss": 1.8834,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658764579160017e-07,
+      "loss": 1.8918,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658023395125355e-07,
+      "loss": 1.7009,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657281421018187e-07,
+      "loss": 1.8921,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656538656899212e-07,
+      "loss": 1.8604,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655795102829195e-07,
+      "loss": 1.7677,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655050758868958e-07,
+      "loss": 1.6281,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654305625079397e-07,
+      "loss": 1.8499,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653559701521462e-07,
+      "loss": 1.8695,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652812988256179e-07,
+      "loss": 1.8957,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965206548534463e-07,
+      "loss": 1.7912,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965131719284796e-07,
+      "loss": 1.9202,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650568110827387e-07,
+      "loss": 1.8297,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649818239344195e-07,
+      "loss": 1.7306,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649067578459716e-07,
+      "loss": 1.9256,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648316128235364e-07,
+      "loss": 1.7275,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964756388873261e-07,
+      "loss": 1.7971,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646810860012991e-07,
+      "loss": 1.9936,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964605704213811e-07,
+      "loss": 1.8491,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964530243516963e-07,
+      "loss": 1.7942,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644547039169286e-07,
+      "loss": 1.9288,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643790854198865e-07,
+      "loss": 1.9964,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643033880320236e-07,
+      "loss": 1.9279,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964227611759531e-07,
+      "loss": 1.9637,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641517566086093e-07,
+      "loss": 1.8341,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640758225854626e-07,
+      "loss": 2.1403,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639998096963026e-07,
+      "loss": 1.767,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639237179473486e-07,
+      "loss": 1.9839,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963847547344824e-07,
+      "loss": 1.7519,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637712978949607e-07,
+      "loss": 1.9178,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963694969603996e-07,
+      "loss": 2.029,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963618562478174e-07,
+      "loss": 1.9911,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635420765237453e-07,
+      "loss": 1.9451,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634655117469662e-07,
+      "loss": 1.9366,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633888681541012e-07,
+      "loss": 1.7832,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963312145751419e-07,
+      "loss": 2.0509,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632353445451964e-07,
+      "loss": 2.0064,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631584645417157e-07,
+      "loss": 1.9736,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630815057472668e-07,
+      "loss": 1.8974,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630044681681445e-07,
+      "loss": 1.9212,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629273518106514e-07,
+      "loss": 1.9382,
+      "step": 1302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962850156681096e-07,
+      "loss": 1.8411,
+      "step": 1303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9627728827857925e-07,
+      "loss": 1.8238,
+      "step": 1304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9626955301310632e-07,
+      "loss": 1.9299,
+      "step": 1305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9626180987232353e-07,
+      "loss": 1.8289,
+      "step": 1306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9625405885686433e-07,
+      "loss": 1.9017,
+      "step": 1307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9624629996736276e-07,
+      "loss": 1.7874,
+      "step": 1308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962385332044536e-07,
+      "loss": 1.8436,
+      "step": 1309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623075856877217e-07,
+      "loss": 1.7346,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622297606095444e-07,
+      "loss": 1.9594,
+      "step": 1311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962151856816371e-07,
+      "loss": 1.9692,
+      "step": 1312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620738743145744e-07,
+      "loss": 1.982,
+      "step": 1313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619958131105338e-07,
+      "loss": 1.9436,
+      "step": 1314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961917673210635e-07,
+      "loss": 1.9011,
+      "step": 1315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96183945462127e-07,
+      "loss": 1.9244,
+      "step": 1316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961761157348838e-07,
+      "loss": 1.876,
+      "step": 1317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616827813997435e-07,
+      "loss": 2.0048,
+      "step": 1318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616043267803983e-07,
+      "loss": 1.9583,
+      "step": 1319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615257934972204e-07,
+      "loss": 1.7694,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614471815566342e-07,
+      "loss": 2.0141,
+      "step": 1321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613684909650706e-07,
+      "loss": 1.8047,
+      "step": 1322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612897217289666e-07,
+      "loss": 2.0137,
+      "step": 1323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961210873854766e-07,
+      "loss": 1.9281,
+      "step": 1324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611319473489192e-07,
+      "loss": 1.7636,
+      "step": 1325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610529422178822e-07,
+      "loss": 1.8176,
+      "step": 1326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609738584681188e-07,
+      "loss": 2.0636,
+      "step": 1327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608946961060978e-07,
+      "loss": 1.7671,
+      "step": 1328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608154551382952e-07,
+      "loss": 1.8471,
+      "step": 1329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960736135571193e-07,
+      "loss": 1.8985,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606567374112808e-07,
+      "loss": 1.7681,
+      "step": 1331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605772606650527e-07,
+      "loss": 2.1343,
+      "step": 1332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604977053390112e-07,
+      "loss": 1.9136,
+      "step": 1333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604180714396637e-07,
+      "loss": 1.9881,
+      "step": 1334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960338358973525e-07,
+      "loss": 1.8322,
+      "step": 1335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602585679471155e-07,
+      "loss": 1.9659,
+      "step": 1336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960178698366963e-07,
+      "loss": 1.9467,
+      "step": 1337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600987502396006e-07,
+      "loss": 1.9405,
+      "step": 1338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600187235715692e-07,
+      "loss": 1.7809,
+      "step": 1339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599386183694148e-07,
+      "loss": 1.8679,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598584346396908e-07,
+      "loss": 1.9374,
+      "step": 1341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959778172388956e-07,
+      "loss": 1.8805,
+      "step": 1342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596978316237768e-07,
+      "loss": 1.9032,
+      "step": 1343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596174123507253e-07,
+      "loss": 1.8476,
+      "step": 1344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595369145763803e-07,
+      "loss": 1.8543,
+      "step": 1345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594563383073264e-07,
+      "loss": 1.7314,
+      "step": 1346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593756835501558e-07,
+      "loss": 1.8133,
+      "step": 1347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592949503114662e-07,
+      "loss": 1.9127,
+      "step": 1348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592141385978615e-07,
+      "loss": 2.0369,
+      "step": 1349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959133248415953e-07,
+      "loss": 1.9303,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959052279772358e-07,
+      "loss": 1.8609,
+      "step": 1351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589712326736997e-07,
+      "loss": 2.0114,
+      "step": 1352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588901071266086e-07,
+      "loss": 1.8325,
+      "step": 1353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588089031377206e-07,
+      "loss": 1.7951,
+      "step": 1354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958727620713679e-07,
+      "loss": 1.7612,
+      "step": 1355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586462598611327e-07,
+      "loss": 1.9578,
+      "step": 1356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585648205867382e-07,
+      "loss": 1.7774,
+      "step": 1357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584833028971568e-07,
+      "loss": 1.8989,
+      "step": 1358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584017067990574e-07,
+      "loss": 1.8453,
+      "step": 1359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958320032299115e-07,
+      "loss": 1.9325,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582382794040107e-07,
+      "loss": 1.8479,
+      "step": 1361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581564481204325e-07,
+      "loss": 1.7757,
+      "step": 1362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580745384550744e-07,
+      "loss": 1.7787,
+      "step": 1363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579925504146374e-07,
+      "loss": 1.8587,
+      "step": 1364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957910484005828e-07,
+      "loss": 1.8186,
+      "step": 1365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578283392353596e-07,
+      "loss": 1.9002,
+      "step": 1366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577461161099526e-07,
+      "loss": 1.8377,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957663814636333e-07,
+      "loss": 1.8755,
+      "step": 1368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575814348212333e-07,
+      "loss": 1.8008,
+      "step": 1369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574989766713928e-07,
+      "loss": 1.9177,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574164401935564e-07,
+      "loss": 1.8756,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573338253944768e-07,
+      "loss": 1.797,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572511322809118e-07,
+      "loss": 1.783,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957168360859626e-07,
+      "loss": 1.9794,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957085511137391e-07,
+      "loss": 1.9808,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957002583120984e-07,
+      "loss": 1.7994,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569195768171886e-07,
+      "loss": 2.0092,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568364922327954e-07,
+      "loss": 1.7842,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567533293746012e-07,
+      "loss": 2.076,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566700882494093e-07,
+      "loss": 1.8171,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565867688640286e-07,
+      "loss": 1.8785,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565033712252756e-07,
+      "loss": 1.9229,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564198953399724e-07,
+      "loss": 1.8135,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563363412149477e-07,
+      "loss": 1.766,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562527088570366e-07,
+      "loss": 1.689,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956168998273081e-07,
+      "loss": 1.9522,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560852094699283e-07,
+      "loss": 1.8788,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956001342454433e-07,
+      "loss": 1.943,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559173972334557e-07,
+      "loss": 1.8915,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955833373813864e-07,
+      "loss": 2.0585,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955749272202531e-07,
+      "loss": 1.801,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556650924063368e-07,
+      "loss": 1.9378,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555808344321676e-07,
+      "loss": 1.965,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955496498286916e-07,
+      "loss": 1.6912,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554120839774812e-07,
+      "loss": 1.8645,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955327591510769e-07,
+      "loss": 1.907,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552430208936908e-07,
+      "loss": 1.8973,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955158372133165e-07,
+      "loss": 1.7503,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550736452361168e-07,
+      "loss": 1.9335,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549888402094765e-07,
+      "loss": 1.804,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954903957060182e-07,
+      "loss": 1.7798,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954818995795177e-07,
+      "loss": 1.9803,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547339564214118e-07,
+      "loss": 1.8323,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954648838945843e-07,
+      "loss": 1.8545,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545636433754337e-07,
+      "loss": 1.944,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544783697171533e-07,
+      "loss": 1.9045,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543930179779774e-07,
+      "loss": 1.8657,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954307588164888e-07,
+      "loss": 1.9181,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542220802848744e-07,
+      "loss": 1.9131,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541364943449307e-07,
+      "loss": 1.7733,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954050830352059e-07,
+      "loss": 1.9705,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539650883132665e-07,
+      "loss": 1.8012,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953879268235568e-07,
+      "loss": 1.8037,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953793370125983e-07,
+      "loss": 1.8754,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953707393991539e-07,
+      "loss": 1.8675,
+      "step": 1415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9536213398392692e-07,
+      "loss": 1.8606,
+      "step": 1416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9535352076762128e-07,
+      "loss": 1.9611,
+      "step": 1417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9534489975094165e-07,
+      "loss": 1.8838,
+      "step": 1418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9533627093459323e-07,
+      "loss": 1.819,
+      "step": 1419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9532763431928193e-07,
+      "loss": 1.846,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9531898990571422e-07,
+      "loss": 1.8626,
+      "step": 1421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9531033769459727e-07,
+      "loss": 2.0259,
+      "step": 1422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953016776866389e-07,
+      "loss": 1.7961,
+      "step": 1423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952930098825475e-07,
+      "loss": 1.868,
+      "step": 1424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952843342830322e-07,
+      "loss": 1.7329,
+      "step": 1425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952756508888026e-07,
+      "loss": 1.7996,
+      "step": 1426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526695970056918e-07,
+      "loss": 1.8791,
+      "step": 1427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952582607190428e-07,
+      "loss": 1.939,
+      "step": 1428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524955394493513e-07,
+      "loss": 1.8916,
+      "step": 1429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524083937895846e-07,
+      "loss": 1.8249,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952321170218256e-07,
+      "loss": 1.9271,
+      "step": 1431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522338687425015e-07,
+      "loss": 1.9216,
+      "step": 1432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521464893694625e-07,
+      "loss": 2.0455,
+      "step": 1433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952059032106287e-07,
+      "loss": 1.842,
+      "step": 1434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519714969601294e-07,
+      "loss": 1.8919,
+      "step": 1435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518838839381507e-07,
+      "loss": 2.0277,
+      "step": 1436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951796193047518e-07,
+      "loss": 2.029,
+      "step": 1437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951708424295405e-07,
+      "loss": 1.8711,
+      "step": 1438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516205776889911e-07,
+      "loss": 1.7991,
+      "step": 1439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951532653235463e-07,
+      "loss": 1.8672,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514446509420132e-07,
+      "loss": 1.7521,
+      "step": 1441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513565708158408e-07,
+      "loss": 1.7323,
+      "step": 1442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951268412864151e-07,
+      "loss": 2.0248,
+      "step": 1443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511801770941555e-07,
+      "loss": 1.7496,
+      "step": 1444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510918635130728e-07,
+      "loss": 1.8557,
+      "step": 1445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510034721281272e-07,
+      "loss": 1.9156,
+      "step": 1446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509150029465494e-07,
+      "loss": 1.7602,
+      "step": 1447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508264559755764e-07,
+      "loss": 1.9943,
+      "step": 1448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507378312224522e-07,
+      "loss": 1.9332,
+      "step": 1449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950649128694427e-07,
+      "loss": 1.9221,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950560348398756e-07,
+      "loss": 1.7888,
+      "step": 1451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504714903427026e-07,
+      "loss": 2.0943,
+      "step": 1452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950382554533536e-07,
+      "loss": 1.8928,
+      "step": 1453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502935409785312e-07,
+      "loss": 1.9685,
+      "step": 1454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95020444968497e-07,
+      "loss": 1.871,
+      "step": 1455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501152806601408e-07,
+      "loss": 1.904,
+      "step": 1456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950026033911337e-07,
+      "loss": 1.8022,
+      "step": 1457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499367094458608e-07,
+      "loss": 1.866,
+      "step": 1458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498473072710187e-07,
+      "loss": 2.0519,
+      "step": 1459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497578273941242e-07,
+      "loss": 1.9254,
+      "step": 1460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496682698224972e-07,
+      "loss": 1.8274,
+      "step": 1461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949578634563464e-07,
+      "loss": 2.2036,
+      "step": 1462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494889216243573e-07,
+      "loss": 1.892,
+      "step": 1463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949399131012516e-07,
+      "loss": 1.9718,
+      "step": 1464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949309262735285e-07,
+      "loss": 1.914,
+      "step": 1465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492193168000167e-07,
+      "loss": 1.9325,
+      "step": 1466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491292932140684e-07,
+      "loss": 1.9142,
+      "step": 1467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490391919848047e-07,
+      "loss": 1.8182,
+      "step": 1468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489490131195968e-07,
+      "loss": 1.762,
+      "step": 1469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948858756625821e-07,
+      "loss": 1.8435,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487684225108612e-07,
+      "loss": 1.7329,
+      "step": 1471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486780107821067e-07,
+      "loss": 1.8324,
+      "step": 1472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485875214469542e-07,
+      "loss": 1.8045,
+      "step": 1473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948496954512806e-07,
+      "loss": 1.8603,
+      "step": 1474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484063099870703e-07,
+      "loss": 1.8814,
+      "step": 1475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483155878771633e-07,
+      "loss": 1.9182,
+      "step": 1476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948224788190506e-07,
+      "loss": 2.014,
+      "step": 1477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481339109345256e-07,
+      "loss": 1.7716,
+      "step": 1478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480429561166574e-07,
+      "loss": 1.8724,
+      "step": 1479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479519237443412e-07,
+      "loss": 1.7494,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947860813825024e-07,
+      "loss": 1.965,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477696263661593e-07,
+      "loss": 1.8268,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476783613752067e-07,
+      "loss": 1.8692,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475870188596314e-07,
+      "loss": 1.8711,
+      "step": 1484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474955988269066e-07,
+      "loss": 1.8425,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474041012845105e-07,
+      "loss": 1.7905,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473125262399283e-07,
+      "loss": 1.653,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94722087370065e-07,
+      "loss": 1.9375,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947129143674175e-07,
+      "loss": 1.8168,
+      "step": 1489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470373361680064e-07,
+      "loss": 2.179,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469454511896543e-07,
+      "loss": 1.933,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946853488746636e-07,
+      "loss": 1.8656,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467614488464736e-07,
+      "loss": 1.8932,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946669331496697e-07,
+      "loss": 1.8493,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465771367048416e-07,
+      "loss": 1.9999,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464848644784495e-07,
+      "loss": 1.7524,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946392514825069e-07,
+      "loss": 2.0171,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463000877522548e-07,
+      "loss": 1.6722,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462075832675678e-07,
+      "loss": 1.7691,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461150013785753e-07,
+      "loss": 1.7664,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946022342092851e-07,
+      "loss": 1.9709,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945929605417975e-07,
+      "loss": 1.8565,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945836791361533e-07,
+      "loss": 1.8376,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457438999311181e-07,
+      "loss": 2.131,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456509311343295e-07,
+      "loss": 1.8408,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945557884978772e-07,
+      "loss": 1.8805,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454647614720576e-07,
+      "loss": 1.8153,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453715606218037e-07,
+      "loss": 1.8434,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452782824356352e-07,
+      "loss": 1.8749,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451849269211825e-07,
+      "loss": 1.9475,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450914940860827e-07,
+      "loss": 1.8383,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449979839379786e-07,
+      "loss": 1.8702,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94490439648452e-07,
+      "loss": 2.015,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944810731733363e-07,
+      "loss": 1.8447,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447169896921696e-07,
+      "loss": 2.0285,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446231703686083e-07,
+      "loss": 1.8779,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944529273770354e-07,
+      "loss": 1.8849,
+      "step": 1517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444352999050882e-07,
+      "loss": 1.8581,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443412487804976e-07,
+      "loss": 1.833,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442471204042773e-07,
+      "loss": 1.8115,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441529147841263e-07,
+      "loss": 2.0083,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440586319277518e-07,
+      "loss": 1.9765,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943964271842866e-07,
+      "loss": 1.7966,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438698345371883e-07,
+      "loss": 1.7332,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437753200184443e-07,
+      "loss": 2.1539,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436807282943656e-07,
+      "loss": 1.9331,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435860593726903e-07,
+      "loss": 1.9864,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434913132611625e-07,
+      "loss": 2.0315,
+      "step": 1528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9433964899675333e-07,
+      "loss": 1.9525,
+      "step": 1529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9433015894995593e-07,
+      "loss": 1.747,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.943206611865004e-07,
+      "loss": 2.0202,
+      "step": 1531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9431115570716373e-07,
+      "loss": 1.8826,
+      "step": 1532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9430164251272343e-07,
+      "loss": 1.9958,
+      "step": 1533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9429212160395783e-07,
+      "loss": 1.8908,
+      "step": 1534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9428259298164572e-07,
+      "loss": 1.6626,
+      "step": 1535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427305664656659e-07,
+      "loss": 1.7654,
+      "step": 1536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942635125995006e-07,
+      "loss": 1.8238,
+      "step": 1537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425396084122844e-07,
+      "loss": 1.8915,
+      "step": 1538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424440137253155e-07,
+      "loss": 1.9199,
+      "step": 1539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423483419419188e-07,
+      "loss": 1.8771,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422525930699213e-07,
+      "loss": 1.8486,
+      "step": 1541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421567671171553e-07,
+      "loss": 1.9772,
+      "step": 1542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420608640914598e-07,
+      "loss": 1.7878,
+      "step": 1543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419648840006803e-07,
+      "loss": 1.8579,
+      "step": 1544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418688268526687e-07,
+      "loss": 1.7309,
+      "step": 1545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941772692655282e-07,
+      "loss": 1.7666,
+      "step": 1546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416764814163856e-07,
+      "loss": 1.7869,
+      "step": 1547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415801931438496e-07,
+      "loss": 1.7303,
+      "step": 1548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414838278455506e-07,
+      "loss": 1.5816,
+      "step": 1549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413873855293717e-07,
+      "loss": 2.016,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941290866203203e-07,
+      "loss": 2.086,
+      "step": 1551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941194269874939e-07,
+      "loss": 1.9957,
+      "step": 1552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410975965524829e-07,
+      "loss": 1.9162,
+      "step": 1553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410008462437428e-07,
+      "loss": 2.0202,
+      "step": 1554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409040189566332e-07,
+      "loss": 1.7495,
+      "step": 1555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940807114699075e-07,
+      "loss": 1.8986,
+      "step": 1556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407101334789955e-07,
+      "loss": 1.7756,
+      "step": 1557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940613075304328e-07,
+      "loss": 1.8254,
+      "step": 1558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405159401830126e-07,
+      "loss": 1.6518,
+      "step": 1559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404187281229954e-07,
+      "loss": 2.0081,
+      "step": 1560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403214391322287e-07,
+      "loss": 1.9134,
+      "step": 1561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402240732186713e-07,
+      "loss": 1.8987,
+      "step": 1562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940126630390288e-07,
+      "loss": 1.7937,
+      "step": 1563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400291106550504e-07,
+      "loss": 1.8242,
+      "step": 1564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399315140209356e-07,
+      "loss": 1.8744,
+      "step": 1565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398338404959281e-07,
+      "loss": 1.8229,
+      "step": 1566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397360900880178e-07,
+      "loss": 1.9621,
+      "step": 1567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939638262805201e-07,
+      "loss": 1.9786,
+      "step": 1568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395403586554804e-07,
+      "loss": 1.9046,
+      "step": 1569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394423776468654e-07,
+      "loss": 1.6017,
+      "step": 1570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939344319787371e-07,
+      "loss": 1.8152,
+      "step": 1571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392461850850186e-07,
+      "loss": 1.9437,
+      "step": 1572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391479735478367e-07,
+      "loss": 1.7033,
+      "step": 1573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390496851838592e-07,
+      "loss": 1.9177,
+      "step": 1574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389513200011263e-07,
+      "loss": 2.0796,
+      "step": 1575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388528780076849e-07,
+      "loss": 1.9457,
+      "step": 1576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387543592115884e-07,
+      "loss": 1.8388,
+      "step": 1577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386557636208956e-07,
+      "loss": 1.8451,
+      "step": 1578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938557091243672e-07,
+      "loss": 1.719,
+      "step": 1579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93845834208799e-07,
+      "loss": 1.8468,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383595161619275e-07,
+      "loss": 1.996,
+      "step": 1581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938260613473569e-07,
+      "loss": 1.8969,
+      "step": 1582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938161634031005e-07,
+      "loss": 1.7657,
+      "step": 1583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380625778423325e-07,
+      "loss": 1.9924,
+      "step": 1584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379634449156551e-07,
+      "loss": 1.778,
+      "step": 1585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937864235259082e-07,
+      "loss": 1.8908,
+      "step": 1586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377649488807296e-07,
+      "loss": 1.882,
+      "step": 1587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376655857887193e-07,
+      "loss": 1.8279,
+      "step": 1588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375661459911798e-07,
+      "loss": 1.834,
+      "step": 1589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374666294962456e-07,
+      "loss": 1.7294,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373670363120577e-07,
+      "loss": 1.8031,
+      "step": 1591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372673664467635e-07,
+      "loss": 1.7814,
+      "step": 1592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371676199085162e-07,
+      "loss": 1.8313,
+      "step": 1593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370677967054757e-07,
+      "loss": 1.9345,
+      "step": 1594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369678968458078e-07,
+      "loss": 1.9599,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936867920337685e-07,
+      "loss": 1.7184,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936767867189286e-07,
+      "loss": 2.0143,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366677374087953e-07,
+      "loss": 1.8529,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936567531004404e-07,
+      "loss": 1.9087,
+      "step": 1599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364672479843098e-07,
+      "loss": 1.8305,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363668883567162e-07,
+      "loss": 1.7891,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362664521298326e-07,
+      "loss": 1.9193,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936165939311876e-07,
+      "loss": 1.9965,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360653499110682e-07,
+      "loss": 1.8764,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935964683935638e-07,
+      "loss": 1.9572,
+      "step": 1605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358639413938205e-07,
+      "loss": 1.9216,
+      "step": 1606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935763122293857e-07,
+      "loss": 1.7296,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356622266439947e-07,
+      "loss": 1.7243,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355612544524876e-07,
+      "loss": 1.8483,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354602057275957e-07,
+      "loss": 1.8512,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353590804775853e-07,
+      "loss": 1.7233,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352578787107286e-07,
+      "loss": 1.9612,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351566004353046e-07,
+      "loss": 1.8576,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350552456595987e-07,
+      "loss": 1.6818,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349538143919016e-07,
+      "loss": 1.9276,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934852306640511e-07,
+      "loss": 1.9825,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934750722413731e-07,
+      "loss": 2.0237,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346490617198715e-07,
+      "loss": 1.9232,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934547324567249e-07,
+      "loss": 1.97,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344455109641854e-07,
+      "loss": 1.905,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343436209190104e-07,
+      "loss": 1.9844,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342416544400587e-07,
+      "loss": 1.8267,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341396115356718e-07,
+      "loss": 1.8304,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340374922141972e-07,
+      "loss": 1.831,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339352964839888e-07,
+      "loss": 1.9922,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338330243534067e-07,
+      "loss": 2.0544,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337306758308173e-07,
+      "loss": 1.786,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336282509245927e-07,
+      "loss": 1.741,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335257496431126e-07,
+      "loss": 1.8878,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334231719947612e-07,
+      "loss": 1.8605,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333205179879307e-07,
+      "loss": 1.8644,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933217787631018e-07,
+      "loss": 1.8759,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331149809324275e-07,
+      "loss": 1.871,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933012097900569e-07,
+      "loss": 1.913,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329091385438586e-07,
+      "loss": 1.8812,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932806102870719e-07,
+      "loss": 1.7422,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327029908895792e-07,
+      "loss": 1.8717,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325998026088742e-07,
+      "loss": 2.01,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932496538037045e-07,
+      "loss": 2.0551,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323931971825397e-07,
+      "loss": 1.9604,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322897800538116e-07,
+      "loss": 1.8662,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932186286659321e-07,
+      "loss": 1.8264,
+      "step": 1642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.932082717007534e-07,
+      "loss": 1.9498,
+      "step": 1643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931979071106923e-07,
+      "loss": 1.9432,
+      "step": 1644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9318753489659668e-07,
+      "loss": 1.8867,
+      "step": 1645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9317715505931509e-07,
+      "loss": 1.8233,
+      "step": 1646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9316676759969655e-07,
+      "loss": 1.8877,
+      "step": 1647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9315637251859088e-07,
+      "loss": 1.776,
+      "step": 1648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314596981684842e-07,
+      "loss": 1.759,
+      "step": 1649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313555949532015e-07,
+      "loss": 1.6459,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931251415548577e-07,
+      "loss": 1.925,
+      "step": 1651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931147159963133e-07,
+      "loss": 1.8774,
+      "step": 1652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310428282053984e-07,
+      "loss": 1.8825,
+      "step": 1653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309384202839078e-07,
+      "loss": 1.5974,
+      "step": 1654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930833936207202e-07,
+      "loss": 1.7802,
+      "step": 1655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307293759838289e-07,
+      "loss": 1.9505,
+      "step": 1656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306247396223415e-07,
+      "loss": 1.5608,
+      "step": 1657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305200271312996e-07,
+      "loss": 1.7362,
+      "step": 1658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304152385192695e-07,
+      "loss": 1.9641,
+      "step": 1659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303103737948233e-07,
+      "loss": 2.0606,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302054329665392e-07,
+      "loss": 1.8106,
+      "step": 1661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301004160430022e-07,
+      "loss": 1.9008,
+      "step": 1662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299953230328027e-07,
+      "loss": 1.7816,
+      "step": 1663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298901539445386e-07,
+      "loss": 1.8818,
+      "step": 1664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297849087868122e-07,
+      "loss": 2.0049,
+      "step": 1665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929679587568234e-07,
+      "loss": 1.7142,
+      "step": 1666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295741902974193e-07,
+      "loss": 1.7283,
+      "step": 1667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294687169829901e-07,
+      "loss": 1.9408,
+      "step": 1668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929363167633575e-07,
+      "loss": 1.8941,
+      "step": 1669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929257542257808e-07,
+      "loss": 1.6092,
+      "step": 1670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291518408643303e-07,
+      "loss": 2.0746,
+      "step": 1671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290460634617878e-07,
+      "loss": 1.7988,
+      "step": 1672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289402100588347e-07,
+      "loss": 1.6268,
+      "step": 1673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288342806641293e-07,
+      "loss": 1.9959,
+      "step": 1674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928728275286338e-07,
+      "loss": 1.8863,
+      "step": 1675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928622193934132e-07,
+      "loss": 1.935,
+      "step": 1676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285160366161897e-07,
+      "loss": 1.7647,
+      "step": 1677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284098033411946e-07,
+      "loss": 1.8102,
+      "step": 1678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928303494117838e-07,
+      "loss": 1.9595,
+      "step": 1679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928197108954816e-07,
+      "loss": 1.8056,
+      "step": 1680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280906478608311e-07,
+      "loss": 1.8132,
+      "step": 1681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279841108445928e-07,
+      "loss": 1.9132,
+      "step": 1682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278774979148165e-07,
+      "loss": 1.7936,
+      "step": 1683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927770809080223e-07,
+      "loss": 1.7918,
+      "step": 1684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276640443495404e-07,
+      "loss": 1.7382,
+      "step": 1685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275572037315025e-07,
+      "loss": 1.6831,
+      "step": 1686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274502872348495e-07,
+      "loss": 1.6567,
+      "step": 1687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273432948683273e-07,
+      "loss": 1.8462,
+      "step": 1688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272362266406886e-07,
+      "loss": 1.9027,
+      "step": 1689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271290825606923e-07,
+      "loss": 2.0702,
+      "step": 1690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270218626371028e-07,
+      "loss": 1.8216,
+      "step": 1691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269145668786918e-07,
+      "loss": 1.7959,
+      "step": 1692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268071952942364e-07,
+      "loss": 2.0256,
+      "step": 1693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92669974789252e-07,
+      "loss": 2.052,
+      "step": 1694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926592224682332e-07,
+      "loss": 1.9321,
+      "step": 1695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264846256724687e-07,
+      "loss": 1.8809,
+      "step": 1696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263769508717326e-07,
+      "loss": 1.7603,
+      "step": 1697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926269200288931e-07,
+      "loss": 1.8495,
+      "step": 1698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261613739328794e-07,
+      "loss": 1.8338,
+      "step": 1699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926053471812398e-07,
+      "loss": 1.7838,
+      "step": 1700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259454939363137e-07,
+      "loss": 1.919,
+      "step": 1701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258374403134599e-07,
+      "loss": 1.7823,
+      "step": 1702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257293109526754e-07,
+      "loss": 1.8319,
+      "step": 1703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256211058628063e-07,
+      "loss": 1.7518,
+      "step": 1704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925512825052704e-07,
+      "loss": 1.9264,
+      "step": 1705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254044685312264e-07,
+      "loss": 1.8245,
+      "step": 1706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252960363072374e-07,
+      "loss": 1.8909,
+      "step": 1707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251875283896078e-07,
+      "loss": 2.0109,
+      "step": 1708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250789447872135e-07,
+      "loss": 1.7047,
+      "step": 1709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249702855089374e-07,
+      "loss": 1.9275,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248615505636684e-07,
+      "loss": 1.8603,
+      "step": 1711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247527399603014e-07,
+      "loss": 1.8706,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246438537077379e-07,
+      "loss": 1.9396,
+      "step": 1713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245348918148847e-07,
+      "loss": 1.8874,
+      "step": 1714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244258542906563e-07,
+      "loss": 1.8468,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243167411439717e-07,
+      "loss": 1.8629,
+      "step": 1716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242075523837568e-07,
+      "loss": 1.9025,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240982880189448e-07,
+      "loss": 1.9432,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239889480584733e-07,
+      "loss": 1.8924,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238795325112868e-07,
+      "loss": 1.727,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923770041386336e-07,
+      "loss": 1.865,
+      "step": 1721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923660474692578e-07,
+      "loss": 1.9334,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923550832438976e-07,
+      "loss": 1.8414,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234411146344992e-07,
+      "loss": 1.9033,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233313212881223e-07,
+      "loss": 2.0449,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232214524088282e-07,
+      "loss": 1.8501,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231115080056037e-07,
+      "loss": 1.7695,
+      "step": 1727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230014880874433e-07,
+      "loss": 1.8065,
+      "step": 1728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228913926633467e-07,
+      "loss": 1.6926,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922781221742321e-07,
+      "loss": 2.0407,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226709753333775e-07,
+      "loss": 1.7673,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922560653445536e-07,
+      "loss": 1.9307,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922450256087821e-07,
+      "loss": 1.9766,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223397832692637e-07,
+      "loss": 1.7119,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922229234998901e-07,
+      "loss": 1.8422,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922118611285776e-07,
+      "loss": 1.7465,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220079121389387e-07,
+      "loss": 1.758,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921897137567445e-07,
+      "loss": 1.801,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217862875803564e-07,
+      "loss": 1.8342,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921675362186741e-07,
+      "loss": 1.9248,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215643613956732e-07,
+      "loss": 1.9693,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214532852162335e-07,
+      "loss": 1.8566,
+      "step": 1742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921342133657508e-07,
+      "loss": 1.8297,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212309067285898e-07,
+      "loss": 1.8518,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921119604438578e-07,
+      "loss": 2.0706,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921008226796577e-07,
+      "loss": 1.739,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208967738116984e-07,
+      "loss": 1.7624,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207852454930595e-07,
+      "loss": 1.8934,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920673641849784e-07,
+      "loss": 1.8999,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205619628910016e-07,
+      "loss": 1.9322,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920450208625848e-07,
+      "loss": 1.8306,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203383790634656e-07,
+      "loss": 1.9399,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920226474213002e-07,
+      "loss": 1.7811,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920114494083612e-07,
+      "loss": 1.8744,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920002438684456e-07,
+      "loss": 2.1072,
+      "step": 1755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9198903080247006e-07,
+      "loss": 1.7557,
+      "step": 1756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919778102113519e-07,
+      "loss": 1.8242,
+      "step": 1757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9196658209600894e-07,
+      "loss": 1.8514,
+      "step": 1758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9195534645735977e-07,
+      "loss": 1.8324,
+      "step": 1759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9194410329632346e-07,
+      "loss": 1.7069,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919328526138198e-07,
+      "loss": 1.672,
+      "step": 1761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9192159441076916e-07,
+      "loss": 1.685,
+      "step": 1762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191032868809246e-07,
+      "loss": 1.87,
+      "step": 1763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918990554467113e-07,
+      "loss": 1.7259,
+      "step": 1764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188777468754792e-07,
+      "loss": 1.9437,
+      "step": 1765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187648641152517e-07,
+      "loss": 1.8091,
+      "step": 1766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186519061956638e-07,
+      "loss": 1.7057,
+      "step": 1767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185388731259566e-07,
+      "loss": 1.8372,
+      "step": 1768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184257649153767e-07,
+      "loss": 1.6972,
+      "step": 1769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183125815731774e-07,
+      "loss": 1.9574,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181993231086167e-07,
+      "loss": 1.7104,
+      "step": 1771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180859895309602e-07,
+      "loss": 1.9356,
+      "step": 1772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179725808494794e-07,
+      "loss": 1.8947,
+      "step": 1773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917859097073451e-07,
+      "loss": 1.8816,
+      "step": 1774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917745538212159e-07,
+      "loss": 1.8432,
+      "step": 1775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176319042748924e-07,
+      "loss": 1.662,
+      "step": 1776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917518195270948e-07,
+      "loss": 1.6878,
+      "step": 1777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174044112096272e-07,
+      "loss": 1.7724,
+      "step": 1778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172905521002383e-07,
+      "loss": 1.6294,
+      "step": 1779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171766179520952e-07,
+      "loss": 1.8232,
+      "step": 1780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170626087745183e-07,
+      "loss": 1.8189,
+      "step": 1781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169485245768342e-07,
+      "loss": 1.9285,
+      "step": 1782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168343653683757e-07,
+      "loss": 1.6723,
+      "step": 1783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916720131158481e-07,
+      "loss": 1.9776,
+      "step": 1784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916605821956496e-07,
+      "loss": 1.9755,
+      "step": 1785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916491437771771e-07,
+      "loss": 1.9167,
+      "step": 1786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916376978613663e-07,
+      "loss": 1.8089,
+      "step": 1787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916262444491536e-07,
+      "loss": 1.7966,
+      "step": 1788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916147835414759e-07,
+      "loss": 1.8947,
+      "step": 1789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160331513927072e-07,
+      "loss": 2.0256,
+      "step": 1790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159183924347632e-07,
+      "loss": 1.9084,
+      "step": 1791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158035585503142e-07,
+      "loss": 1.7535,
+      "step": 1792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156886497487547e-07,
+      "loss": 1.7944,
+      "step": 1793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155736660394842e-07,
+      "loss": 1.7224,
+      "step": 1794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154586074319087e-07,
+      "loss": 1.9045,
+      "step": 1795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153434739354415e-07,
+      "loss": 1.895,
+      "step": 1796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152282655595004e-07,
+      "loss": 1.9342,
+      "step": 1797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91511298231351e-07,
+      "loss": 1.9821,
+      "step": 1798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149976242069016e-07,
+      "loss": 1.6472,
+      "step": 1799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148821912491116e-07,
+      "loss": 1.8301,
+      "step": 1800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147666834495827e-07,
+      "loss": 1.7837,
+      "step": 1801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146511008177643e-07,
+      "loss": 1.9152,
+      "step": 1802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145354433631115e-07,
+      "loss": 1.8208,
+      "step": 1803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914419711095086e-07,
+      "loss": 1.8675,
+      "step": 1804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143039040231548e-07,
+      "loss": 1.8167,
+      "step": 1805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141880221567918e-07,
+      "loss": 1.7745,
+      "step": 1806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140720655054761e-07,
+      "loss": 1.7015,
+      "step": 1807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913956034078694e-07,
+      "loss": 1.8389,
+      "step": 1808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913839927885938e-07,
+      "loss": 1.925,
+      "step": 1809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913723746936705e-07,
+      "loss": 1.8978,
+      "step": 1810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136074912405e-07,
+      "loss": 1.6916,
+      "step": 1811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134911608068328e-07,
+      "loss": 1.9095,
+      "step": 1812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133747556452199e-07,
+      "loss": 1.8876,
+      "step": 1813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132582757651837e-07,
+      "loss": 1.9292,
+      "step": 1814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913141721176253e-07,
+      "loss": 2.0147,
+      "step": 1815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130250918879628e-07,
+      "loss": 1.7437,
+      "step": 1816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912908387909853e-07,
+      "loss": 1.8052,
+      "step": 1817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127916092514715e-07,
+      "loss": 1.9133,
+      "step": 1818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912674755922371e-07,
+      "loss": 1.8965,
+      "step": 1819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125578279321105e-07,
+      "loss": 1.9185,
+      "step": 1820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124408252902554e-07,
+      "loss": 1.7649,
+      "step": 1821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123237480063773e-07,
+      "loss": 1.9893,
+      "step": 1822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122065960900536e-07,
+      "loss": 1.9017,
+      "step": 1823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120893695508675e-07,
+      "loss": 1.8491,
+      "step": 1824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119720683984092e-07,
+      "loss": 1.5233,
+      "step": 1825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118546926422738e-07,
+      "loss": 2.0061,
+      "step": 1826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117372422920642e-07,
+      "loss": 1.6321,
+      "step": 1827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116197173573875e-07,
+      "loss": 1.891,
+      "step": 1828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911502117847858e-07,
+      "loss": 1.8684,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113844437730963e-07,
+      "loss": 1.9089,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112666951427286e-07,
+      "loss": 2.0067,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911148871966387e-07,
+      "loss": 1.9734,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91103097425371e-07,
+      "loss": 1.7724,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109130020143426e-07,
+      "loss": 1.7796,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107949552579351e-07,
+      "loss": 1.934,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106768339941446e-07,
+      "loss": 1.8249,
+      "step": 1836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105586382326337e-07,
+      "loss": 1.798,
+      "step": 1837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104403679830714e-07,
+      "loss": 1.7692,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910322023255133e-07,
+      "loss": 1.7515,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102036040584997e-07,
+      "loss": 1.9876,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100851104028583e-07,
+      "loss": 1.7954,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099665422979029e-07,
+      "loss": 1.8513,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909847899753332e-07,
+      "loss": 1.844,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097291827788517e-07,
+      "loss": 1.889,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096103913841738e-07,
+      "loss": 1.9641,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094915255790153e-07,
+      "loss": 1.7867,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093725853731008e-07,
+      "loss": 1.749,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092535707761598e-07,
+      "loss": 1.8661,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091344817979277e-07,
+      "loss": 1.7197,
+      "step": 1849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090153184481474e-07,
+      "loss": 1.767,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908896080736567e-07,
+      "loss": 1.992,
+      "step": 1851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087767686729404e-07,
+      "loss": 1.9977,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086573822670277e-07,
+      "loss": 1.7856,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908537921528596e-07,
+      "loss": 2.0115,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908418386467417e-07,
+      "loss": 1.9782,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082987770932694e-07,
+      "loss": 1.8354,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908179093415938e-07,
+      "loss": 1.867,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080593354452134e-07,
+      "loss": 1.7047,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079395031908926e-07,
+      "loss": 1.7564,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078195966627781e-07,
+      "loss": 1.7872,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076996158706792e-07,
+      "loss": 1.7367,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075795608244105e-07,
+      "loss": 1.8613,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074594315337933e-07,
+      "loss": 2.0169,
+      "step": 1863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907339228008655e-07,
+      "loss": 1.8313,
+      "step": 1864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072189502588283e-07,
+      "loss": 1.8115,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070985982941528e-07,
+      "loss": 1.8523,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069781721244738e-07,
+      "loss": 1.8205,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068576717596428e-07,
+      "loss": 1.8656,
+      "step": 1868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9067370972095173e-07,
+      "loss": 1.8246,
+      "step": 1869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9066164484839606e-07,
+      "loss": 1.9369,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.906495725592843e-07,
+      "loss": 2.0265,
+      "step": 1871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9063749285460397e-07,
+      "loss": 1.7505,
+      "step": 1872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.906254057353432e-07,
+      "loss": 1.6812,
+      "step": 1873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9061331120249088e-07,
+      "loss": 1.8353,
+      "step": 1874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9060120925703632e-07,
+      "loss": 1.7681,
+      "step": 1875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905890998999696e-07,
+      "loss": 1.7647,
+      "step": 1876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905769831322812e-07,
+      "loss": 1.6795,
+      "step": 1877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905648589549625e-07,
+      "loss": 1.9263,
+      "step": 1878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055272736900513e-07,
+      "loss": 1.8097,
+      "step": 1879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054058837540161e-07,
+      "loss": 1.8819,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90528441975145e-07,
+      "loss": 1.9834,
+      "step": 1881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051628816922884e-07,
+      "loss": 1.8969,
+      "step": 1882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050412695864744e-07,
+      "loss": 1.9779,
+      "step": 1883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049195834439562e-07,
+      "loss": 1.721,
+      "step": 1884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047978232746884e-07,
+      "loss": 1.7294,
+      "step": 1885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046759890886316e-07,
+      "loss": 1.7619,
+      "step": 1886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045540808957523e-07,
+      "loss": 1.9992,
+      "step": 1887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044320987060232e-07,
+      "loss": 1.9261,
+      "step": 1888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043100425294228e-07,
+      "loss": 1.8444,
+      "step": 1889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041879123759363e-07,
+      "loss": 1.8749,
+      "step": 1890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040657082555542e-07,
+      "loss": 1.6623,
+      "step": 1891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039434301782737e-07,
+      "loss": 1.8617,
+      "step": 1892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038210781540975e-07,
+      "loss": 1.8916,
+      "step": 1893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903698652193035e-07,
+      "loss": 1.6943,
+      "step": 1894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035761523051003e-07,
+      "loss": 1.8247,
+      "step": 1895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034535785003155e-07,
+      "loss": 1.7489,
+      "step": 1896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033309307887069e-07,
+      "loss": 1.6707,
+      "step": 1897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903208209180308e-07,
+      "loss": 1.8228,
+      "step": 1898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030854136851583e-07,
+      "loss": 1.6867,
+      "step": 1899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902962544313303e-07,
+      "loss": 1.8814,
+      "step": 1900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028396010747927e-07,
+      "loss": 2.0263,
+      "step": 1901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027165839796856e-07,
+      "loss": 1.8601,
+      "step": 1902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025934930380444e-07,
+      "loss": 1.8661,
+      "step": 1903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024703282599393e-07,
+      "loss": 1.8336,
+      "step": 1904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902347089655445e-07,
+      "loss": 1.9128,
+      "step": 1905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022237772346435e-07,
+      "loss": 1.9037,
+      "step": 1906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021003910076224e-07,
+      "loss": 1.6385,
+      "step": 1907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019769309844745e-07,
+      "loss": 1.8877,
+      "step": 1908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018533971753007e-07,
+      "loss": 1.8347,
+      "step": 1909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017297895902057e-07,
+      "loss": 1.8199,
+      "step": 1910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016061082393012e-07,
+      "loss": 1.7298,
+      "step": 1911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014823531327053e-07,
+      "loss": 1.7504,
+      "step": 1912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013585242805416e-07,
+      "loss": 1.9408,
+      "step": 1913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012346216929398e-07,
+      "loss": 1.904,
+      "step": 1914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901110645380036e-07,
+      "loss": 1.9766,
+      "step": 1915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900986595351972e-07,
+      "loss": 1.7929,
+      "step": 1916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008624716188954e-07,
+      "loss": 1.7862,
+      "step": 1917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007382741909607e-07,
+      "loss": 1.7998,
+      "step": 1918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900614003078327e-07,
+      "loss": 1.9097,
+      "step": 1919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900489658291161e-07,
+      "loss": 1.7714,
+      "step": 1920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900365239839634e-07,
+      "loss": 1.8161,
+      "step": 1921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900240747733925e-07,
+      "loss": 1.7068,
+      "step": 1922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001161819842171e-07,
+      "loss": 1.752,
+      "step": 1923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999915426007007e-07,
+      "loss": 2.0682,
+      "step": 1924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998668295935722e-07,
+      "loss": 1.8522,
+      "step": 1925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997420429730335e-07,
+      "loss": 1.7947,
+      "step": 1926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996171827492927e-07,
+      "loss": 1.9067,
+      "step": 1927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994922489325638e-07,
+      "loss": 1.924,
+      "step": 1928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993672415330674e-07,
+      "loss": 1.9449,
+      "step": 1929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992421605610293e-07,
+      "loss": 1.8979,
+      "step": 1930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899117006026682e-07,
+      "loss": 1.8542,
+      "step": 1931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989917779402635e-07,
+      "loss": 1.901,
+      "step": 1932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988664763120183e-07,
+      "loss": 1.8715,
+      "step": 1933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987411011521967e-07,
+      "loss": 1.9159,
+      "step": 1934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986156524710547e-07,
+      "loss": 1.7408,
+      "step": 1935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984901302788545e-07,
+      "loss": 1.823,
+      "step": 1936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898364534585865e-07,
+      "loss": 1.8946,
+      "step": 1937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89823886540236e-07,
+      "loss": 1.9527,
+      "step": 1938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981131227386203e-07,
+      "loss": 1.9489,
+      "step": 1939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979873066049318e-07,
+      "loss": 1.783,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978614170115872e-07,
+      "loss": 1.8988,
+      "step": 1941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977354539688846e-07,
+      "loss": 1.6869,
+      "step": 1942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976094174871286e-07,
+      "loss": 1.8096,
+      "step": 1943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974833075766292e-07,
+      "loss": 1.8322,
+      "step": 1944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897357124247703e-07,
+      "loss": 1.9583,
+      "step": 1945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897230867510673e-07,
+      "loss": 1.7657,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897104537375867e-07,
+      "loss": 1.9102,
+      "step": 1947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896978133853619e-07,
+      "loss": 1.9085,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968516569542703e-07,
+      "loss": 1.9159,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967251066881672e-07,
+      "loss": 1.6702,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965984830656613e-07,
+      "loss": 1.9541,
+      "step": 1951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896471786097112e-07,
+      "loss": 1.902,
+      "step": 1952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963450157928827e-07,
+      "loss": 1.9016,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896218172163345e-07,
+      "loss": 1.8883,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896091255218874e-07,
+      "loss": 1.9081,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959642649698534e-07,
+      "loss": 1.5907,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895837201426671e-07,
+      "loss": 1.8981,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895710064599721e-07,
+      "loss": 2.0165,
+      "step": 1958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955828544994042e-07,
+      "loss": 1.7691,
+      "step": 1959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895455571136127e-07,
+      "loss": 1.908,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895328214520301e-07,
+      "loss": 2.0222,
+      "step": 1961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952007846623458e-07,
+      "loss": 1.9409,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895073281572685e-07,
+      "loss": 1.882,
+      "step": 1963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949457052617493e-07,
+      "loss": 1.9855,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948180557399747e-07,
+      "loss": 1.7105,
+      "step": 1965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946903330178037e-07,
+      "loss": 1.8548,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945625371056844e-07,
+      "loss": 1.9622,
+      "step": 1967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944346680140718e-07,
+      "loss": 1.727,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943067257534255e-07,
+      "loss": 1.6283,
+      "step": 1969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941787103342124e-07,
+      "loss": 1.9274,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940506217669042e-07,
+      "loss": 1.8355,
+      "step": 1971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939224600619795e-07,
+      "loss": 1.9559,
+      "step": 1972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937942252299222e-07,
+      "loss": 1.8763,
+      "step": 1973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936659172812228e-07,
+      "loss": 1.6801,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935375362263777e-07,
+      "loss": 1.8512,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893409082075889e-07,
+      "loss": 1.8613,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932805548402642e-07,
+      "loss": 1.9393,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931519545300181e-07,
+      "loss": 1.9546,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893023281155671e-07,
+      "loss": 1.8024,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928945347277482e-07,
+      "loss": 1.8659,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927657152567824e-07,
+      "loss": 1.8307,
+      "step": 1981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8926368227533116e-07,
+      "loss": 1.7614,
+      "step": 1982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8925078572278798e-07,
+      "loss": 1.6503,
+      "step": 1983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.892378818691037e-07,
+      "loss": 1.7027,
+      "step": 1984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8922497071533388e-07,
+      "loss": 1.9345,
+      "step": 1985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8921205226253476e-07,
+      "loss": 1.6629,
+      "step": 1986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8919912651176307e-07,
+      "loss": 1.8927,
+      "step": 1987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891861934640763e-07,
+      "loss": 1.7825,
+      "step": 1988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917325312053235e-07,
+      "loss": 1.7912,
+      "step": 1989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916030548218984e-07,
+      "loss": 1.7447,
+      "step": 1990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914735055010792e-07,
+      "loss": 1.9253,
+      "step": 1991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891343883253464e-07,
+      "loss": 1.8725,
+      "step": 1992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912141880896564e-07,
+      "loss": 1.7993,
+      "step": 1993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910844200202657e-07,
+      "loss": 1.663,
+      "step": 1994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909545790559083e-07,
+      "loss": 1.8606,
+      "step": 1995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908246652072054e-07,
+      "loss": 1.9217,
+      "step": 1996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906946784847842e-07,
+      "loss": 1.7827,
+      "step": 1997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890564618899279e-07,
+      "loss": 1.9513,
+      "step": 1998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904344864613284e-07,
+      "loss": 1.9141,
+      "step": 1999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903042811815788e-07,
+      "loss": 1.8969,
+      "step": 2000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890174003070681e-07,
+      "loss": 1.8923,
+      "step": 2001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900436521392925e-07,
+      "loss": 1.8893,
+      "step": 2002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899132283980768e-07,
+      "loss": 1.8082,
+      "step": 2003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889782731857703e-07,
+      "loss": 1.8166,
+      "step": 2004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896521625288467e-07,
+      "loss": 1.927,
+      "step": 2005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895215204221883e-07,
+      "loss": 1.7018,
+      "step": 2006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893908055484157e-07,
+      "loss": 1.8634,
+      "step": 2007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892600179182216e-07,
+      "loss": 1.9026,
+      "step": 2008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891291575423054e-07,
+      "loss": 1.845,
+      "step": 2009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889982244313716e-07,
+      "loss": 1.6787,
+      "step": 2010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888867218596132e-07,
+      "loss": 1.8424,
+      "step": 2011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887361400473023e-07,
+      "loss": 1.9604,
+      "step": 2012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886049887956067e-07,
+      "loss": 1.7482,
+      "step": 2013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884737648517728e-07,
+      "loss": 1.9332,
+      "step": 2014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888342468226536e-07,
+      "loss": 1.736,
+      "step": 2015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888211098930637e-07,
+      "loss": 1.7234,
+      "step": 2016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888079656974822e-07,
+      "loss": 1.7414,
+      "step": 2017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887948142369844e-07,
+      "loss": 1.8086,
+      "step": 2018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878165551264615e-07,
+      "loss": 1.7519,
+      "step": 2019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876848952554388e-07,
+      "loss": 1.7504,
+      "step": 2020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875531627675465e-07,
+      "loss": 1.8744,
+      "step": 2021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874213576735608e-07,
+      "loss": 1.7702,
+      "step": 2022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887289479984264e-07,
+      "loss": 1.8961,
+      "step": 2023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887157529710444e-07,
+      "loss": 1.898,
+      "step": 2024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870255068628955e-07,
+      "loss": 1.7307,
+      "step": 2025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868934114524185e-07,
+      "loss": 2.0081,
+      "step": 2026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867612434898188e-07,
+      "loss": 1.8363,
+      "step": 2027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866290029859085e-07,
+      "loss": 1.8627,
+      "step": 2028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864966899515055e-07,
+      "loss": 1.9988,
+      "step": 2029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863643043974337e-07,
+      "loss": 1.748,
+      "step": 2030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862318463345228e-07,
+      "loss": 1.8581,
+      "step": 2031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860993157736088e-07,
+      "loss": 1.8489,
+      "step": 2032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859667127255326e-07,
+      "loss": 1.7993,
+      "step": 2033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858340372011424e-07,
+      "loss": 1.7278,
+      "step": 2034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857012892112913e-07,
+      "loss": 1.7147,
+      "step": 2035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855684687668393e-07,
+      "loss": 1.8789,
+      "step": 2036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854355758786512e-07,
+      "loss": 1.7409,
+      "step": 2037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853026105575984e-07,
+      "loss": 1.733,
+      "step": 2038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851695728145584e-07,
+      "loss": 1.8168,
+      "step": 2039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885036462660414e-07,
+      "loss": 1.6687,
+      "step": 2040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849032801060537e-07,
+      "loss": 1.8952,
+      "step": 2041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884770025162374e-07,
+      "loss": 1.8428,
+      "step": 2042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846366978402744e-07,
+      "loss": 1.7203,
+      "step": 2043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845032981506624e-07,
+      "loss": 1.6885,
+      "step": 2044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843698261044502e-07,
+      "loss": 1.8149,
+      "step": 2045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884236281712557e-07,
+      "loss": 1.8292,
+      "step": 2046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841026649859071e-07,
+      "loss": 1.918,
+      "step": 2047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883968975935431e-07,
+      "loss": 1.7667,
+      "step": 2048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838352145720653e-07,
+      "loss": 1.8062,
+      "step": 2049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883701380906752e-07,
+      "loss": 1.9718,
+      "step": 2050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835674749504396e-07,
+      "loss": 1.6826,
+      "step": 2051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883433496714082e-07,
+      "loss": 1.8877,
+      "step": 2052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832994462086398e-07,
+      "loss": 1.8067,
+      "step": 2053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883165323445078e-07,
+      "loss": 1.8283,
+      "step": 2054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830311284343697e-07,
+      "loss": 1.9121,
+      "step": 2055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828968611874915e-07,
+      "loss": 1.7692,
+      "step": 2056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827625217154278e-07,
+      "loss": 1.5656,
+      "step": 2057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882628110029168e-07,
+      "loss": 1.7005,
+      "step": 2058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882493626139708e-07,
+      "loss": 1.8703,
+      "step": 2059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823590700580487e-07,
+      "loss": 1.8172,
+      "step": 2060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822244417951976e-07,
+      "loss": 1.8432,
+      "step": 2061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882089741362168e-07,
+      "loss": 1.8344,
+      "step": 2062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881954968769979e-07,
+      "loss": 1.8597,
+      "step": 2063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818201240296555e-07,
+      "loss": 1.8513,
+      "step": 2064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881685207152229e-07,
+      "loss": 1.7598,
+      "step": 2065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815502181487357e-07,
+      "loss": 1.8201,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814151570302187e-07,
+      "loss": 1.9293,
+      "step": 2067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812800238077263e-07,
+      "loss": 1.8335,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811448184923135e-07,
+      "loss": 1.6929,
+      "step": 2069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810095410950407e-07,
+      "loss": 1.8686,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808741916269737e-07,
+      "loss": 1.6342,
+      "step": 2071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807387700991852e-07,
+      "loss": 1.7375,
+      "step": 2072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806032765227532e-07,
+      "loss": 1.8776,
+      "step": 2073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804677109087619e-07,
+      "loss": 1.7401,
+      "step": 2074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880332073268301e-07,
+      "loss": 1.8214,
+      "step": 2075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801963636124665e-07,
+      "loss": 1.8813,
+      "step": 2076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800605819523598e-07,
+      "loss": 1.8169,
+      "step": 2077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799247282990885e-07,
+      "loss": 1.7928,
+      "step": 2078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797888026637665e-07,
+      "loss": 1.6262,
+      "step": 2079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796528050575126e-07,
+      "loss": 1.8181,
+      "step": 2080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795167354914527e-07,
+      "loss": 2.0107,
+      "step": 2081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793805939767175e-07,
+      "loss": 1.9995,
+      "step": 2082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792443805244443e-07,
+      "loss": 1.8221,
+      "step": 2083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791080951457756e-07,
+      "loss": 2.0412,
+      "step": 2084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789717378518604e-07,
+      "loss": 1.8644,
+      "step": 2085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788353086538535e-07,
+      "loss": 1.7789,
+      "step": 2086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786988075629157e-07,
+      "loss": 1.9449,
+      "step": 2087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785622345902126e-07,
+      "loss": 1.813,
+      "step": 2088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784255897469173e-07,
+      "loss": 1.9018,
+      "step": 2089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782888730442075e-07,
+      "loss": 1.9476,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878152084493268e-07,
+      "loss": 1.9521,
+      "step": 2091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780152241052877e-07,
+      "loss": 1.7915,
+      "step": 2092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778782918914632e-07,
+      "loss": 1.8838,
+      "step": 2093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777412878629964e-07,
+      "loss": 1.867,
+      "step": 2094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776042120310942e-07,
+      "loss": 1.975,
+      "step": 2095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8774670644069704e-07,
+      "loss": 1.8991,
+      "step": 2096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8773298450018442e-07,
+      "loss": 1.8637,
+      "step": 2097
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.877192553826941e-07,
+      "loss": 1.931,
+      "step": 2098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8770551908934917e-07,
+      "loss": 1.7158,
+      "step": 2099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8769177562127335e-07,
+      "loss": 1.8865,
+      "step": 2100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8767802497959088e-07,
+      "loss": 1.9696,
+      "step": 2101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766426716542667e-07,
+      "loss": 1.6726,
+      "step": 2102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765050217990616e-07,
+      "loss": 1.8605,
+      "step": 2103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763673002415537e-07,
+      "loss": 1.6958,
+      "step": 2104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762295069930097e-07,
+      "loss": 1.7444,
+      "step": 2105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760916420647014e-07,
+      "loss": 1.9836,
+      "step": 2106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759537054679066e-07,
+      "loss": 1.9126,
+      "step": 2107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758156972139098e-07,
+      "loss": 1.816,
+      "step": 2108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756776173140007e-07,
+      "loss": 1.761,
+      "step": 2109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755394657794744e-07,
+      "loss": 1.8748,
+      "step": 2110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754012426216326e-07,
+      "loss": 1.9637,
+      "step": 2111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752629478517822e-07,
+      "loss": 1.8547,
+      "step": 2112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751245814812373e-07,
+      "loss": 1.8496,
+      "step": 2113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749861435213161e-07,
+      "loss": 1.8643,
+      "step": 2114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874847633983344e-07,
+      "loss": 1.807,
+      "step": 2115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747090528786512e-07,
+      "loss": 1.8395,
+      "step": 2116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874570400218575e-07,
+      "loss": 1.7777,
+      "step": 2117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744316760144574e-07,
+      "loss": 2.0868,
+      "step": 2118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742928802776467e-07,
+      "loss": 1.8357,
+      "step": 2119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741540130194974e-07,
+      "loss": 2.0744,
+      "step": 2120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740150742513689e-07,
+      "loss": 1.8634,
+      "step": 2121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738760639846273e-07,
+      "loss": 1.7612,
+      "step": 2122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737369822306447e-07,
+      "loss": 1.7302,
+      "step": 2123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735978290007982e-07,
+      "loss": 1.7616,
+      "step": 2124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734586043064718e-07,
+      "loss": 1.8472,
+      "step": 2125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733193081590537e-07,
+      "loss": 1.7922,
+      "step": 2126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87317994056994e-07,
+      "loss": 1.8518,
+      "step": 2127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873040501550531e-07,
+      "loss": 1.7829,
+      "step": 2128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872900991112234e-07,
+      "loss": 1.7208,
+      "step": 2129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727614092664612e-07,
+      "loss": 1.7913,
+      "step": 2130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872621756024631e-07,
+      "loss": 1.7977,
+      "step": 2131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872482031398168e-07,
+      "loss": 1.8805,
+      "step": 2132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872342235398502e-07,
+      "loss": 1.9661,
+      "step": 2133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87220236803707e-07,
+      "loss": 1.8949,
+      "step": 2134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720624293253123e-07,
+      "loss": 1.8683,
+      "step": 2135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719224192746775e-07,
+      "loss": 1.7774,
+      "step": 2136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717823378966188e-07,
+      "loss": 1.7491,
+      "step": 2137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871642185202596e-07,
+      "loss": 1.736,
+      "step": 2138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715019612040735e-07,
+      "loss": 1.9178,
+      "step": 2139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713616659125227e-07,
+      "loss": 1.8685,
+      "step": 2140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712212993394204e-07,
+      "loss": 1.8034,
+      "step": 2141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710808614962493e-07,
+      "loss": 1.844,
+      "step": 2142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709403523944977e-07,
+      "loss": 1.8156,
+      "step": 2143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87079977204566e-07,
+      "loss": 1.9613,
+      "step": 2144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706591204612368e-07,
+      "loss": 1.7649,
+      "step": 2145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705183976527335e-07,
+      "loss": 2.0239,
+      "step": 2146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870377603631662e-07,
+      "loss": 1.7094,
+      "step": 2147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702367384095398e-07,
+      "loss": 2.0019,
+      "step": 2148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700958019978912e-07,
+      "loss": 1.7925,
+      "step": 2149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699547944082443e-07,
+      "loss": 1.7108,
+      "step": 2150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698137156521349e-07,
+      "loss": 1.8243,
+      "step": 2151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696725657411034e-07,
+      "loss": 1.8466,
+      "step": 2152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869531344686697e-07,
+      "loss": 1.7325,
+      "step": 2153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693900525004685e-07,
+      "loss": 1.8203,
+      "step": 2154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692486891939753e-07,
+      "loss": 1.8751,
+      "step": 2155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691072547787826e-07,
+      "loss": 2.0274,
+      "step": 2156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689657492664602e-07,
+      "loss": 1.8784,
+      "step": 2157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688241726685833e-07,
+      "loss": 1.9149,
+      "step": 2158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868682524996734e-07,
+      "loss": 1.8699,
+      "step": 2159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685408062624998e-07,
+      "loss": 1.957,
+      "step": 2160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868399016477474e-07,
+      "loss": 1.6066,
+      "step": 2161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682571556532557e-07,
+      "loss": 1.8716,
+      "step": 2162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681152238014495e-07,
+      "loss": 1.6705,
+      "step": 2163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679732209336662e-07,
+      "loss": 1.7565,
+      "step": 2164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678311470615226e-07,
+      "loss": 1.8535,
+      "step": 2165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676890021966408e-07,
+      "loss": 1.8887,
+      "step": 2166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867546786350649e-07,
+      "loss": 1.9557,
+      "step": 2167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674044995351807e-07,
+      "loss": 1.6607,
+      "step": 2168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867262141761876e-07,
+      "loss": 1.7705,
+      "step": 2169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671197130423813e-07,
+      "loss": 1.5703,
+      "step": 2170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669772133883463e-07,
+      "loss": 1.8341,
+      "step": 2171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668346428114293e-07,
+      "loss": 1.8065,
+      "step": 2172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666920013232928e-07,
+      "loss": 2.0248,
+      "step": 2173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866549288935606e-07,
+      "loss": 1.7972,
+      "step": 2174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664065056600431e-07,
+      "loss": 1.9149,
+      "step": 2175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662636515082842e-07,
+      "loss": 1.7838,
+      "step": 2176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661207264920157e-07,
+      "loss": 1.7404,
+      "step": 2177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86597773062293e-07,
+      "loss": 1.8248,
+      "step": 2178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658346639127244e-07,
+      "loss": 1.7775,
+      "step": 2179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656915263731021e-07,
+      "loss": 1.9197,
+      "step": 2180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865548318015773e-07,
+      "loss": 1.796,
+      "step": 2181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654050388524522e-07,
+      "loss": 1.952,
+      "step": 2182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652616888948604e-07,
+      "loss": 1.7028,
+      "step": 2183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651182681547245e-07,
+      "loss": 1.8688,
+      "step": 2184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649747766437767e-07,
+      "loss": 1.9091,
+      "step": 2185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648312143737555e-07,
+      "loss": 1.8668,
+      "step": 2186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646875813564052e-07,
+      "loss": 1.747,
+      "step": 2187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864543877603475e-07,
+      "loss": 1.8066,
+      "step": 2188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864400103126721e-07,
+      "loss": 1.7766,
+      "step": 2189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642562579379052e-07,
+      "loss": 1.789,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641123420487937e-07,
+      "loss": 1.6928,
+      "step": 2191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639683554711603e-07,
+      "loss": 1.8112,
+      "step": 2192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638242982167837e-07,
+      "loss": 1.7822,
+      "step": 2193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863680170297448e-07,
+      "loss": 1.7345,
+      "step": 2194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863535971724944e-07,
+      "loss": 1.6143,
+      "step": 2195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633917025110675e-07,
+      "loss": 1.6997,
+      "step": 2196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632473626676209e-07,
+      "loss": 1.9811,
+      "step": 2197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631029522064114e-07,
+      "loss": 1.8352,
+      "step": 2198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629584711392529e-07,
+      "loss": 2.1007,
+      "step": 2199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628139194779645e-07,
+      "loss": 1.7778,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626692972343707e-07,
+      "loss": 1.8779,
+      "step": 2201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625246044203028e-07,
+      "loss": 1.9419,
+      "step": 2202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623798410475974e-07,
+      "loss": 1.8528,
+      "step": 2203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622350071280967e-07,
+      "loss": 1.747,
+      "step": 2204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862090102673649e-07,
+      "loss": 1.7341,
+      "step": 2205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861945127696108e-07,
+      "loss": 1.8887,
+      "step": 2206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861800082207333e-07,
+      "loss": 1.756,
+      "step": 2207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616549662191904e-07,
+      "loss": 1.9616,
+      "step": 2208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8615097797435503e-07,
+      "loss": 1.9253,
+      "step": 2209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.86136452279229e-07,
+      "loss": 1.7724,
+      "step": 2210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8612191953772926e-07,
+      "loss": 1.8061,
+      "step": 2211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8610737975104462e-07,
+      "loss": 1.7772,
+      "step": 2212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860928329203645e-07,
+      "loss": 1.6384,
+      "step": 2213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8607827904687895e-07,
+      "loss": 1.7542,
+      "step": 2214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8606371813177845e-07,
+      "loss": 1.8991,
+      "step": 2215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604915017625426e-07,
+      "loss": 1.7255,
+      "step": 2216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603457518149804e-07,
+      "loss": 1.7951,
+      "step": 2217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860199931487021e-07,
+      "loss": 1.8372,
+      "step": 2218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600540407905937e-07,
+      "loss": 1.9479,
+      "step": 2219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599080797376327e-07,
+      "loss": 1.9078,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859762048340078e-07,
+      "loss": 1.8996,
+      "step": 2221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596159466098763e-07,
+      "loss": 1.7773,
+      "step": 2222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859469774558979e-07,
+      "loss": 1.8794,
+      "step": 2223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593235321993436e-07,
+      "loss": 1.7869,
+      "step": 2224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859177219542934e-07,
+      "loss": 1.7984,
+      "step": 2225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590308366017184e-07,
+      "loss": 1.8158,
+      "step": 2226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588843833876723e-07,
+      "loss": 1.6711,
+      "step": 2227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858737859912776e-07,
+      "loss": 1.7143,
+      "step": 2228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585912661890162e-07,
+      "loss": 1.8429,
+      "step": 2229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584446022283844e-07,
+      "loss": 1.8706,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582978680428786e-07,
+      "loss": 1.8632,
+      "step": 2231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581510636445029e-07,
+      "loss": 1.8198,
+      "step": 2232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580041890452657e-07,
+      "loss": 1.8762,
+      "step": 2233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578572442571825e-07,
+      "loss": 1.8298,
+      "step": 2234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577102292922744e-07,
+      "loss": 1.798,
+      "step": 2235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575631441625675e-07,
+      "loss": 1.7575,
+      "step": 2236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574159888800938e-07,
+      "loss": 1.8066,
+      "step": 2237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857268763456892e-07,
+      "loss": 1.8925,
+      "step": 2238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571214679050053e-07,
+      "loss": 1.7682,
+      "step": 2239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569741022364836e-07,
+      "loss": 1.8304,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856826666463382e-07,
+      "loss": 1.8939,
+      "step": 2241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856679160597761e-07,
+      "loss": 1.8097,
+      "step": 2242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565315846516878e-07,
+      "loss": 1.8385,
+      "step": 2243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563839386372347e-07,
+      "loss": 1.8088,
+      "step": 2244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85623622256648e-07,
+      "loss": 1.7989,
+      "step": 2245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856088436451507e-07,
+      "loss": 2.0478,
+      "step": 2246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855940580304406e-07,
+      "loss": 1.9029,
+      "step": 2247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855792654137272e-07,
+      "loss": 1.8039,
+      "step": 2248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855644657962206e-07,
+      "loss": 1.7947,
+      "step": 2249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855496591791315e-07,
+      "loss": 1.7545,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553484556367114e-07,
+      "loss": 1.8198,
+      "step": 2251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552002495105134e-07,
+      "loss": 1.9145,
+      "step": 2252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855051973424845e-07,
+      "loss": 1.8904,
+      "step": 2253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549036273918365e-07,
+      "loss": 1.8563,
+      "step": 2254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854755211423622e-07,
+      "loss": 1.8644,
+      "step": 2255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546067255323438e-07,
+      "loss": 2.096,
+      "step": 2256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544581697301483e-07,
+      "loss": 1.7951,
+      "step": 2257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854309544029188e-07,
+      "loss": 1.8857,
+      "step": 2258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541608484416215e-07,
+      "loss": 2.0191,
+      "step": 2259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540120829796125e-07,
+      "loss": 1.7863,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538632476553308e-07,
+      "loss": 1.9423,
+      "step": 2261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853714342480952e-07,
+      "loss": 1.716,
+      "step": 2262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535653674686574e-07,
+      "loss": 1.8536,
+      "step": 2263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534163226306332e-07,
+      "loss": 1.8615,
+      "step": 2264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532672079790725e-07,
+      "loss": 1.8188,
+      "step": 2265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531180235261736e-07,
+      "loss": 2.0087,
+      "step": 2266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529687692841402e-07,
+      "loss": 1.7662,
+      "step": 2267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528194452651823e-07,
+      "loss": 1.7439,
+      "step": 2268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526700514815152e-07,
+      "loss": 1.6804,
+      "step": 2269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85252058794536e-07,
+      "loss": 1.8395,
+      "step": 2270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523710546689437e-07,
+      "loss": 1.8879,
+      "step": 2271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522214516644987e-07,
+      "loss": 1.6786,
+      "step": 2272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520717789442633e-07,
+      "loss": 1.91,
+      "step": 2273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519220365204814e-07,
+      "loss": 1.9318,
+      "step": 2274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517722244054025e-07,
+      "loss": 1.6051,
+      "step": 2275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516223426112822e-07,
+      "loss": 1.8264,
+      "step": 2276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514723911503817e-07,
+      "loss": 1.6951,
+      "step": 2277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851322370034967e-07,
+      "loss": 1.8073,
+      "step": 2278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851172279277311e-07,
+      "loss": 1.6999,
+      "step": 2279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510221188896921e-07,
+      "loss": 2.0547,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508718888843943e-07,
+      "loss": 2.0091,
+      "step": 2281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507215892737063e-07,
+      "loss": 1.8462,
+      "step": 2282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505712200699237e-07,
+      "loss": 1.8536,
+      "step": 2283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504207812853475e-07,
+      "loss": 1.8981,
+      "step": 2284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502702729322846e-07,
+      "loss": 1.7098,
+      "step": 2285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501196950230468e-07,
+      "loss": 1.7495,
+      "step": 2286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499690475699522e-07,
+      "loss": 1.8906,
+      "step": 2287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498183305853248e-07,
+      "loss": 1.7151,
+      "step": 2288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496675440814936e-07,
+      "loss": 2.0128,
+      "step": 2289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849516688070794e-07,
+      "loss": 1.9058,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493657625655665e-07,
+      "loss": 1.9917,
+      "step": 2291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492147675781576e-07,
+      "loss": 1.8686,
+      "step": 2292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490637031209193e-07,
+      "loss": 1.9441,
+      "step": 2293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489125692062096e-07,
+      "loss": 1.7717,
+      "step": 2294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487613658463917e-07,
+      "loss": 1.7851,
+      "step": 2295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848610093053835e-07,
+      "loss": 1.7972,
+      "step": 2296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484587508409143e-07,
+      "loss": 1.6149,
+      "step": 2297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483073392200098e-07,
+      "loss": 1.8782,
+      "step": 2298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481558582035084e-07,
+      "loss": 1.8571,
+      "step": 2299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480043078038015e-07,
+      "loss": 1.732,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478526880332863e-07,
+      "loss": 1.8279,
+      "step": 2301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477009989043668e-07,
+      "loss": 1.9742,
+      "step": 2302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475492404294513e-07,
+      "loss": 1.8096,
+      "step": 2303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473974126209546e-07,
+      "loss": 1.8732,
+      "step": 2304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847245515491297e-07,
+      "loss": 1.7284,
+      "step": 2305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470935490529045e-07,
+      "loss": 1.9097,
+      "step": 2306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469415133182083e-07,
+      "loss": 1.9533,
+      "step": 2307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467894082996457e-07,
+      "loss": 1.9918,
+      "step": 2308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466372340096601e-07,
+      "loss": 1.7344,
+      "step": 2309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464849904606997e-07,
+      "loss": 1.8302,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463326776652186e-07,
+      "loss": 1.9472,
+      "step": 2311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461802956356773e-07,
+      "loss": 1.7843,
+      "step": 2312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460278443845407e-07,
+      "loss": 1.9635,
+      "step": 2313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458753239242802e-07,
+      "loss": 1.8214,
+      "step": 2314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457227342673735e-07,
+      "loss": 1.6278,
+      "step": 2315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845570075426302e-07,
+      "loss": 1.8514,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454173474135545e-07,
+      "loss": 1.733,
+      "step": 2317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845264550241625e-07,
+      "loss": 1.9101,
+      "step": 2318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451116839230125e-07,
+      "loss": 1.5834,
+      "step": 2319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449587484702229e-07,
+      "loss": 1.7387,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448057438957667e-07,
+      "loss": 1.7119,
+      "step": 2321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8446526702121602e-07,
+      "loss": 1.6632,
+      "step": 2322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8444995274319258e-07,
+      "loss": 1.7813,
+      "step": 2323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8443463155675916e-07,
+      "loss": 1.8872,
+      "step": 2324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8441930346316904e-07,
+      "loss": 1.9127,
+      "step": 2325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.844039684636762e-07,
+      "loss": 1.7501,
+      "step": 2326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8438862655953505e-07,
+      "loss": 1.852,
+      "step": 2327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8437327775200067e-07,
+      "loss": 1.8696,
+      "step": 2328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435792204232867e-07,
+      "loss": 1.7992,
+      "step": 2329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434255943177522e-07,
+      "loss": 1.8139,
+      "step": 2330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432718992159702e-07,
+      "loss": 1.8003,
+      "step": 2331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431181351305143e-07,
+      "loss": 1.7934,
+      "step": 2332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429643020739625e-07,
+      "loss": 1.7997,
+      "step": 2333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428104000588994e-07,
+      "loss": 1.8458,
+      "step": 2334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426564290979149e-07,
+      "loss": 1.7941,
+      "step": 2335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425023892036045e-07,
+      "loss": 1.7721,
+      "step": 2336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423482803885697e-07,
+      "loss": 1.82,
+      "step": 2337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421941026654168e-07,
+      "loss": 1.7147,
+      "step": 2338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420398560467588e-07,
+      "loss": 1.6818,
+      "step": 2339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418855405452135e-07,
+      "loss": 1.7865,
+      "step": 2340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417311561734047e-07,
+      "loss": 1.824,
+      "step": 2341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415767029439614e-07,
+      "loss": 1.8181,
+      "step": 2342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414221808695193e-07,
+      "loss": 1.8356,
+      "step": 2343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412675899627188e-07,
+      "loss": 1.592,
+      "step": 2344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411129302362063e-07,
+      "loss": 1.8023,
+      "step": 2345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409582017026333e-07,
+      "loss": 1.9014,
+      "step": 2346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408034043746576e-07,
+      "loss": 1.8846,
+      "step": 2347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406485382649425e-07,
+      "loss": 1.8686,
+      "step": 2348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404936033861565e-07,
+      "loss": 1.6841,
+      "step": 2349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840338599750974e-07,
+      "loss": 1.7253,
+      "step": 2350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401835273720754e-07,
+      "loss": 2.0635,
+      "step": 2351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400283862621459e-07,
+      "loss": 1.8844,
+      "step": 2352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839873176433877e-07,
+      "loss": 1.7714,
+      "step": 2353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397178978999658e-07,
+      "loss": 1.7586,
+      "step": 2354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395625506731147e-07,
+      "loss": 1.7342,
+      "step": 2355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839407134766032e-07,
+      "loss": 1.8606,
+      "step": 2356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392516501914307e-07,
+      "loss": 1.7157,
+      "step": 2357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839096096962031e-07,
+      "loss": 1.6125,
+      "step": 2358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389404750905577e-07,
+      "loss": 1.8158,
+      "step": 2359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387847845897416e-07,
+      "loss": 1.7366,
+      "step": 2360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838629025472318e-07,
+      "loss": 1.8678,
+      "step": 2361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384731977510303e-07,
+      "loss": 1.7644,
+      "step": 2362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838317301438624e-07,
+      "loss": 1.6878,
+      "step": 2363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381613365478544e-07,
+      "loss": 1.9062,
+      "step": 2364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380053030914782e-07,
+      "loss": 1.8294,
+      "step": 2365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837849201082261e-07,
+      "loss": 1.8112,
+      "step": 2366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837693030532972e-07,
+      "loss": 1.7319,
+      "step": 2367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375367914563865e-07,
+      "loss": 1.8238,
+      "step": 2368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373804838652865e-07,
+      "loss": 1.8626,
+      "step": 2369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372241077724578e-07,
+      "loss": 2.0149,
+      "step": 2370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370676631906933e-07,
+      "loss": 1.709,
+      "step": 2371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369111501327906e-07,
+      "loss": 1.8876,
+      "step": 2372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367545686115536e-07,
+      "loss": 1.8433,
+      "step": 2373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836597918639791e-07,
+      "loss": 1.9234,
+      "step": 2374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836441200230318e-07,
+      "loss": 1.7283,
+      "step": 2375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362844133959543e-07,
+      "loss": 1.7775,
+      "step": 2376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836127558149526e-07,
+      "loss": 1.8599,
+      "step": 2377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835970634503865e-07,
+      "loss": 1.8691,
+      "step": 2378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358136424718082e-07,
+      "loss": 1.8282,
+      "step": 2379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356565820661976e-07,
+      "loss": 1.97,
+      "step": 2380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354994532998825e-07,
+      "loss": 1.8163,
+      "step": 2381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353422561857165e-07,
+      "loss": 1.8714,
+      "step": 2382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835184990736559e-07,
+      "loss": 1.9156,
+      "step": 2383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350276569652747e-07,
+      "loss": 1.7482,
+      "step": 2384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348702548847346e-07,
+      "loss": 1.9445,
+      "step": 2385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347127845078153e-07,
+      "loss": 1.9828,
+      "step": 2386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834555245847398e-07,
+      "loss": 1.8873,
+      "step": 2387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343976389163703e-07,
+      "loss": 1.9192,
+      "step": 2388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342399637276255e-07,
+      "loss": 1.5987,
+      "step": 2389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834082220294062e-07,
+      "loss": 1.749,
+      "step": 2390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833924408628584e-07,
+      "loss": 1.8102,
+      "step": 2391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337665287441008e-07,
+      "loss": 1.7425,
+      "step": 2392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336085806535284e-07,
+      "loss": 1.8519,
+      "step": 2393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334505643697874e-07,
+      "loss": 1.7148,
+      "step": 2394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833292479905804e-07,
+      "loss": 1.7927,
+      "step": 2395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833134327274511e-07,
+      "loss": 1.8563,
+      "step": 2396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329761064888456e-07,
+      "loss": 1.879,
+      "step": 2397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328178175617507e-07,
+      "loss": 1.9162,
+      "step": 2398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326594605061756e-07,
+      "loss": 1.7445,
+      "step": 2399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325010353350744e-07,
+      "loss": 1.8485,
+      "step": 2400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323425420614075e-07,
+      "loss": 1.8354,
+      "step": 2401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321839806981397e-07,
+      "loss": 1.8622,
+      "step": 2402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832025351258242e-07,
+      "loss": 1.6758,
+      "step": 2403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318666537546923e-07,
+      "loss": 1.7753,
+      "step": 2404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317078882004716e-07,
+      "loss": 1.6862,
+      "step": 2405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315490546085676e-07,
+      "loss": 1.8261,
+      "step": 2406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313901529919749e-07,
+      "loss": 1.7723,
+      "step": 2407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312311833636911e-07,
+      "loss": 1.707,
+      "step": 2408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310721457367213e-07,
+      "loss": 1.8965,
+      "step": 2409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309130401240752e-07,
+      "loss": 2.1182,
+      "step": 2410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309130401240752e-07,
+      "loss": 1.8853,
+      "step": 2411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307538665387692e-07,
+      "loss": 1.8262,
+      "step": 2412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305946249938234e-07,
+      "loss": 1.6475,
+      "step": 2413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304353155022651e-07,
+      "loss": 2.0269,
+      "step": 2414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302759380771267e-07,
+      "loss": 1.823,
+      "step": 2415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301164927314456e-07,
+      "loss": 1.8693,
+      "step": 2416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299569794782658e-07,
+      "loss": 1.8846,
+      "step": 2417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297973983306352e-07,
+      "loss": 1.8666,
+      "step": 2418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296377493016094e-07,
+      "loss": 1.7451,
+      "step": 2419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829478032404248e-07,
+      "loss": 1.6507,
+      "step": 2420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293182476516164e-07,
+      "loss": 1.6608,
+      "step": 2421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829158395056786e-07,
+      "loss": 1.8617,
+      "step": 2422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289984746328338e-07,
+      "loss": 1.9424,
+      "step": 2423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288384863928416e-07,
+      "loss": 1.7953,
+      "step": 2424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286784303498972e-07,
+      "loss": 1.584,
+      "step": 2425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285183065170942e-07,
+      "loss": 1.7834,
+      "step": 2426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283581149075314e-07,
+      "loss": 1.6395,
+      "step": 2427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281978555343132e-07,
+      "loss": 1.7109,
+      "step": 2428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280375284105496e-07,
+      "loss": 1.8115,
+      "step": 2429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827877133549356e-07,
+      "loss": 1.6195,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827716670963854e-07,
+      "loss": 1.9552,
+      "step": 2431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275561406671692e-07,
+      "loss": 1.6674,
+      "step": 2432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827395542672435e-07,
+      "loss": 2.0574,
+      "step": 2433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272348769927878e-07,
+      "loss": 1.823,
+      "step": 2434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.827074143641372e-07,
+      "loss": 1.918,
+      "step": 2435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8269133426313356e-07,
+      "loss": 1.9392,
+      "step": 2436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826752473975833e-07,
+      "loss": 2.1244,
+      "step": 2437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8265915376880246e-07,
+      "loss": 1.7894,
+      "step": 2438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8264305337810747e-07,
+      "loss": 1.8124,
+      "step": 2439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8262694622681552e-07,
+      "loss": 1.757,
+      "step": 2440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8261083231624422e-07,
+      "loss": 1.8447,
+      "step": 2441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825947116477117e-07,
+      "loss": 1.736,
+      "step": 2442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825785842225368e-07,
+      "loss": 2.0186,
+      "step": 2443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256245004203878e-07,
+      "loss": 1.8544,
+      "step": 2444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254630910753748e-07,
+      "loss": 1.8207,
+      "step": 2445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253016142035334e-07,
+      "loss": 1.7443,
+      "step": 2446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825140069818073e-07,
+      "loss": 1.6647,
+      "step": 2447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249784579322088e-07,
+      "loss": 1.7184,
+      "step": 2448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248167785591612e-07,
+      "loss": 1.8099,
+      "step": 2449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246550317121562e-07,
+      "loss": 1.8891,
+      "step": 2450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244932174044263e-07,
+      "loss": 1.82,
+      "step": 2451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243313356492077e-07,
+      "loss": 1.8092,
+      "step": 2452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241693864597435e-07,
+      "loss": 2.0775,
+      "step": 2453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240073698492823e-07,
+      "loss": 1.8601,
+      "step": 2454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238452858310773e-07,
+      "loss": 1.8025,
+      "step": 2455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236831344183878e-07,
+      "loss": 1.7097,
+      "step": 2456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823520915624479e-07,
+      "loss": 1.8455,
+      "step": 2457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233586294626206e-07,
+      "loss": 1.7834,
+      "step": 2458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231962759460887e-07,
+      "loss": 1.9304,
+      "step": 2459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230338550881648e-07,
+      "loss": 1.906,
+      "step": 2460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822871366902135e-07,
+      "loss": 1.7574,
+      "step": 2461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227088114012924e-07,
+      "loss": 1.7964,
+      "step": 2462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225461885989343e-07,
+      "loss": 1.9225,
+      "step": 2463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822383498508364e-07,
+      "loss": 1.8551,
+      "step": 2464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822220741142891e-07,
+      "loss": 1.7603,
+      "step": 2465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822057916515829e-07,
+      "loss": 1.9778,
+      "step": 2466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821895024640498e-07,
+      "loss": 1.7948,
+      "step": 2467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821732065530223e-07,
+      "loss": 1.8728,
+      "step": 2468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215690391983355e-07,
+      "loss": 1.7512,
+      "step": 2469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821405945658171e-07,
+      "loss": 1.8619,
+      "step": 2470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212427849230725e-07,
+      "loss": 1.6626,
+      "step": 2471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210795570063863e-07,
+      "loss": 1.9054,
+      "step": 2472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209162619214657e-07,
+      "loss": 1.7206,
+      "step": 2473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207528996816687e-07,
+      "loss": 1.8372,
+      "step": 2474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205894703003594e-07,
+      "loss": 1.6998,
+      "step": 2475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820425973790907e-07,
+      "loss": 1.6703,
+      "step": 2476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820425973790907e-07,
+      "loss": 1.7267,
+      "step": 2477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202624101666865e-07,
+      "loss": 1.7396,
+      "step": 2478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200987794410777e-07,
+      "loss": 1.5693,
+      "step": 2479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199350816274668e-07,
+      "loss": 1.734,
+      "step": 2480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197713167392447e-07,
+      "loss": 1.8157,
+      "step": 2481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819607484789809e-07,
+      "loss": 1.7926,
+      "step": 2482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194435857925609e-07,
+      "loss": 1.6979,
+      "step": 2483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192796197609086e-07,
+      "loss": 1.7106,
+      "step": 2484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191155867082652e-07,
+      "loss": 1.594,
+      "step": 2485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189514866480494e-07,
+      "loss": 1.737,
+      "step": 2486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187873195936854e-07,
+      "loss": 1.9703,
+      "step": 2487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186230855586029e-07,
+      "loss": 1.8469,
+      "step": 2488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818458784556237e-07,
+      "loss": 1.7471,
+      "step": 2489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182944166000282e-07,
+      "loss": 1.8462,
+      "step": 2490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181299817034228e-07,
+      "loss": 1.8972,
+      "step": 2491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179654798798722e-07,
+      "loss": 1.9377,
+      "step": 2492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178009111428334e-07,
+      "loss": 1.8837,
+      "step": 2493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817636275505769e-07,
+      "loss": 1.9302,
+      "step": 2494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174715729821474e-07,
+      "loss": 1.7962,
+      "step": 2495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817306803585441e-07,
+      "loss": 1.9524,
+      "step": 2496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171419673291294e-07,
+      "loss": 1.6924,
+      "step": 2497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169770642266971e-07,
+      "loss": 1.8881,
+      "step": 2498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168120942916342e-07,
+      "loss": 1.8343,
+      "step": 2499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8166470575374354e-07,
+      "loss": 1.8664,
+      "step": 2500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8164819539776014e-07,
+      "loss": 1.8269,
+      "step": 2501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816316783625639e-07,
+      "loss": 1.6025,
+      "step": 2502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161515464950598e-07,
+      "loss": 1.6968,
+      "step": 2503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159862425993805e-07,
+      "loss": 1.8095,
+      "step": 2504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158208719521247e-07,
+      "loss": 1.6837,
+      "step": 2505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156554345668198e-07,
+      "loss": 1.8199,
+      "step": 2506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154899304569992e-07,
+      "loss": 1.9081,
+      "step": 2507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153243596362026e-07,
+      "loss": 1.7813,
+      "step": 2508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815158722117974e-07,
+      "loss": 1.859,
+      "step": 2509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149930179158637e-07,
+      "loss": 1.8529,
+      "step": 2510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148272470434267e-07,
+      "loss": 1.8776,
+      "step": 2511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814661409514224e-07,
+      "loss": 1.7735,
+      "step": 2512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814495505341822e-07,
+      "loss": 1.7284,
+      "step": 2513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143295345397924e-07,
+      "loss": 1.728,
+      "step": 2514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141634971217123e-07,
+      "loss": 1.7734,
+      "step": 2515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139973931011645e-07,
+      "loss": 1.7918,
+      "step": 2516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813831222491737e-07,
+      "loss": 1.8553,
+      "step": 2517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136649853070237e-07,
+      "loss": 1.5971,
+      "step": 2518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813498681560623e-07,
+      "loss": 1.6946,
+      "step": 2519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133323112661397e-07,
+      "loss": 1.7948,
+      "step": 2520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131658744371836e-07,
+      "loss": 1.6001,
+      "step": 2521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81299937108737e-07,
+      "loss": 1.9464,
+      "step": 2522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81283280123032e-07,
+      "loss": 1.7225,
+      "step": 2523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126661648796593e-07,
+      "loss": 1.7925,
+      "step": 2524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124994620490198e-07,
+      "loss": 1.894,
+      "step": 2525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123326927520383e-07,
+      "loss": 1.8509,
+      "step": 2526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121658570023582e-07,
+      "loss": 2.0687,
+      "step": 2527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119989548136263e-07,
+      "loss": 1.8044,
+      "step": 2528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118319861994968e-07,
+      "loss": 1.8384,
+      "step": 2529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116649511736284e-07,
+      "loss": 1.7874,
+      "step": 2530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114978497496853e-07,
+      "loss": 1.7843,
+      "step": 2531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811330681941337e-07,
+      "loss": 1.6449,
+      "step": 2532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111634477622589e-07,
+      "loss": 1.9643,
+      "step": 2533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810996147226131e-07,
+      "loss": 1.8077,
+      "step": 2534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108287803466402e-07,
+      "loss": 1.859,
+      "step": 2535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106613471374773e-07,
+      "loss": 1.6495,
+      "step": 2536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104938476123392e-07,
+      "loss": 1.8873,
+      "step": 2537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103262817849282e-07,
+      "loss": 1.8784,
+      "step": 2538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101586496689522e-07,
+      "loss": 1.8692,
+      "step": 2539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809990951278124e-07,
+      "loss": 1.8911,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809823186626162e-07,
+      "loss": 1.8607,
+      "step": 2541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809655355726791e-07,
+      "loss": 1.8065,
+      "step": 2542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094874585937392e-07,
+      "loss": 1.9763,
+      "step": 2543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093194952407424e-07,
+      "loss": 1.7821,
+      "step": 2544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80915146568154e-07,
+      "loss": 1.5883,
+      "step": 2545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089833699298783e-07,
+      "loss": 2.0126,
+      "step": 2546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088152079995078e-07,
+      "loss": 1.8439,
+      "step": 2547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086469799041854e-07,
+      "loss": 1.7467,
+      "step": 2548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8084786856576724e-07,
+      "loss": 1.8508,
+      "step": 2549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8083103252737367e-07,
+      "loss": 1.9101,
+      "step": 2550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8081418987661506e-07,
+      "loss": 1.7676,
+      "step": 2551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8079734061486923e-07,
+      "loss": 1.8606,
+      "step": 2552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807804847435145e-07,
+      "loss": 1.6059,
+      "step": 2553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8076362226392985e-07,
+      "loss": 1.761,
+      "step": 2554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8074675317749462e-07,
+      "loss": 1.7734,
+      "step": 2555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807298774855888e-07,
+      "loss": 1.897,
+      "step": 2556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071299518959294e-07,
+      "loss": 1.9205,
+      "step": 2557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069610629088806e-07,
+      "loss": 1.6627,
+      "step": 2558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067921079085573e-07,
+      "loss": 1.7634,
+      "step": 2559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066230869087815e-07,
+      "loss": 1.8773,
+      "step": 2560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064539999233793e-07,
+      "loss": 1.6814,
+      "step": 2561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062848469661835e-07,
+      "loss": 1.8302,
+      "step": 2562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061156280510312e-07,
+      "loss": 1.8752,
+      "step": 2563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805946343191765e-07,
+      "loss": 1.899,
+      "step": 2564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057769924022337e-07,
+      "loss": 1.9547,
+      "step": 2565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056075756962906e-07,
+      "loss": 1.8261,
+      "step": 2566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054380930877954e-07,
+      "loss": 1.7647,
+      "step": 2567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805268544590612e-07,
+      "loss": 1.8595,
+      "step": 2568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050989302186105e-07,
+      "loss": 1.652,
+      "step": 2569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049292499856663e-07,
+      "loss": 1.7521,
+      "step": 2570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80475950390566e-07,
+      "loss": 1.7971,
+      "step": 2571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045896919924775e-07,
+      "loss": 1.8727,
+      "step": 2572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80441981426001e-07,
+      "loss": 1.7661,
+      "step": 2573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042498707221547e-07,
+      "loss": 1.9101,
+      "step": 2574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804079861392814e-07,
+      "loss": 1.7243,
+      "step": 2575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803909786285895e-07,
+      "loss": 1.9082,
+      "step": 2576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037396454153109e-07,
+      "loss": 1.8744,
+      "step": 2577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035694387949798e-07,
+      "loss": 1.7021,
+      "step": 2578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033991664388258e-07,
+      "loss": 1.7346,
+      "step": 2579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032288283607776e-07,
+      "loss": 1.8322,
+      "step": 2580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80305842457477e-07,
+      "loss": 1.7603,
+      "step": 2581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028879550947426e-07,
+      "loss": 1.9735,
+      "step": 2582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027174199346406e-07,
+      "loss": 1.7039,
+      "step": 2583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025468191084148e-07,
+      "loss": 1.8143,
+      "step": 2584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023761526300212e-07,
+      "loss": 1.7946,
+      "step": 2585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022054205134207e-07,
+      "loss": 1.7112,
+      "step": 2586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020346227725805e-07,
+      "loss": 1.8586,
+      "step": 2587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018637594214726e-07,
+      "loss": 1.9009,
+      "step": 2588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801692830474074e-07,
+      "loss": 1.8612,
+      "step": 2589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801521835944368e-07,
+      "loss": 1.8966,
+      "step": 2590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013507758463427e-07,
+      "loss": 1.6466,
+      "step": 2591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011796501939912e-07,
+      "loss": 1.8642,
+      "step": 2592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801008459001313e-07,
+      "loss": 1.9828,
+      "step": 2593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800837202282312e-07,
+      "loss": 2.0039,
+      "step": 2594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800665880050998e-07,
+      "loss": 1.9056,
+      "step": 2595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004944923213858e-07,
+      "loss": 1.8312,
+      "step": 2596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003230391074957e-07,
+      "loss": 1.993,
+      "step": 2597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001515204233538e-07,
+      "loss": 1.9222,
+      "step": 2598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999799362829907e-07,
+      "loss": 1.8621,
+      "step": 2599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998082867004426e-07,
+      "loss": 1.816,
+      "step": 2600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799636571689752e-07,
+      "loss": 1.9191,
+      "step": 2601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994647912649654e-07,
+      "loss": 1.9525,
+      "step": 2602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992929454401358e-07,
+      "loss": 1.8485,
+      "step": 2603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991210342293205e-07,
+      "loss": 2.0191,
+      "step": 2604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798949057646583e-07,
+      "loss": 1.754,
+      "step": 2605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987770157059913e-07,
+      "loss": 1.6842,
+      "step": 2606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79860490842162e-07,
+      "loss": 1.8453,
+      "step": 2607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984327358075476e-07,
+      "loss": 1.8438,
+      "step": 2608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798260497877859e-07,
+      "loss": 1.9039,
+      "step": 2609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980881946466443e-07,
+      "loss": 1.8527,
+      "step": 2610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797915826127998e-07,
+      "loss": 1.9728,
+      "step": 2611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977433923360218e-07,
+      "loss": 1.876,
+      "step": 2612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975708932848207e-07,
+      "loss": 1.6581,
+      "step": 2613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797398328988506e-07,
+      "loss": 1.739,
+      "step": 2614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972256994611944e-07,
+      "loss": 1.9104,
+      "step": 2615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797053004717008e-07,
+      "loss": 1.7717,
+      "step": 2616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968802447700742e-07,
+      "loss": 1.8672,
+      "step": 2617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967074196345253e-07,
+      "loss": 1.8601,
+      "step": 2618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965345293244992e-07,
+      "loss": 1.917,
+      "step": 2619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963615738541388e-07,
+      "loss": 1.8156,
+      "step": 2620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961885532375938e-07,
+      "loss": 2.0876,
+      "step": 2621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960154674890168e-07,
+      "loss": 1.8613,
+      "step": 2622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795842316622568e-07,
+      "loss": 1.6793,
+      "step": 2623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956691006524115e-07,
+      "loss": 1.7578,
+      "step": 2624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954958195927175e-07,
+      "loss": 1.8611,
+      "step": 2625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953224734576612e-07,
+      "loss": 1.6955,
+      "step": 2626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951490622614224e-07,
+      "loss": 1.7478,
+      "step": 2627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794975586018188e-07,
+      "loss": 1.9929,
+      "step": 2628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948020447421488e-07,
+      "loss": 1.7183,
+      "step": 2629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794628438447501e-07,
+      "loss": 1.9736,
+      "step": 2630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944547671484468e-07,
+      "loss": 1.8658,
+      "step": 2631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794281030859193e-07,
+      "loss": 1.5375,
+      "step": 2632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941072295939524e-07,
+      "loss": 1.7284,
+      "step": 2633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793933363366943e-07,
+      "loss": 1.8092,
+      "step": 2634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793759432192387e-07,
+      "loss": 1.6196,
+      "step": 2635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935854360845137e-07,
+      "loss": 1.9787,
+      "step": 2636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934113750575563e-07,
+      "loss": 1.956,
+      "step": 2637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793237249125754e-07,
+      "loss": 1.7618,
+      "step": 2638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930630583033512e-07,
+      "loss": 1.9319,
+      "step": 2639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928888026045975e-07,
+      "loss": 1.891,
+      "step": 2640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927144820437476e-07,
+      "loss": 1.8873,
+      "step": 2641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792540096635062e-07,
+      "loss": 1.7217,
+      "step": 2642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923656463928063e-07,
+      "loss": 1.826,
+      "step": 2643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921911313312515e-07,
+      "loss": 1.8557,
+      "step": 2644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920165514646733e-07,
+      "loss": 2.0229,
+      "step": 2645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918419068073534e-07,
+      "loss": 1.7918,
+      "step": 2646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916671973735784e-07,
+      "loss": 1.906,
+      "step": 2647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914924231776408e-07,
+      "loss": 1.8284,
+      "step": 2648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913175842338374e-07,
+      "loss": 1.7827,
+      "step": 2649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911426805564717e-07,
+      "loss": 1.736,
+      "step": 2650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909677121598509e-07,
+      "loss": 1.8352,
+      "step": 2651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907926790582884e-07,
+      "loss": 1.9714,
+      "step": 2652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790617581266103e-07,
+      "loss": 1.7313,
+      "step": 2653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904424187976185e-07,
+      "loss": 1.8616,
+      "step": 2654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902671916671635e-07,
+      "loss": 1.8729,
+      "step": 2655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790091899889073e-07,
+      "loss": 1.8568,
+      "step": 2656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899165434776865e-07,
+      "loss": 1.693,
+      "step": 2657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789741122447349e-07,
+      "loss": 1.8005,
+      "step": 2658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895656368124108e-07,
+      "loss": 1.7672,
+      "step": 2659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893900865872277e-07,
+      "loss": 1.825,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78921447178616e-07,
+      "loss": 1.8172,
+      "step": 2661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7890387924235744e-07,
+      "loss": 1.7336,
+      "step": 2662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7888630485138418e-07,
+      "loss": 1.7958,
+      "step": 2663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.788687240071339e-07,
+      "loss": 1.816,
+      "step": 2664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7885113671104486e-07,
+      "loss": 1.7529,
+      "step": 2665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.788335429645557e-07,
+      "loss": 1.7696,
+      "step": 2666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7881594276910574e-07,
+      "loss": 1.8388,
+      "step": 2667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787983361261347e-07,
+      "loss": 1.8683,
+      "step": 2668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7878072303708294e-07,
+      "loss": 1.7606,
+      "step": 2669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876310350339123e-07,
+      "loss": 1.7491,
+      "step": 2670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874547752650103e-07,
+      "loss": 1.811,
+      "step": 2671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787278451078541e-07,
+      "loss": 1.7326,
+      "step": 2672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78710206248893e-07,
+      "loss": 1.7387,
+      "step": 2673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869256095106057e-07,
+      "loss": 1.7456,
+      "step": 2674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867490921580032e-07,
+      "loss": 1.8376,
+      "step": 2675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865725104455623e-07,
+      "loss": 1.7461,
+      "step": 2676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863958643877283e-07,
+      "loss": 1.804,
+      "step": 2677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786219153998952e-07,
+      "loss": 1.7949,
+      "step": 2678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860423792936885e-07,
+      "loss": 1.7317,
+      "step": 2679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858655402863994e-07,
+      "loss": 1.7396,
+      "step": 2680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856886369915505e-07,
+      "loss": 1.8165,
+      "step": 2681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855116694236139e-07,
+      "loss": 1.7894,
+      "step": 2682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853346375970656e-07,
+      "loss": 1.8785,
+      "step": 2683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851575415263887e-07,
+      "loss": 1.8557,
+      "step": 2684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849803812260695e-07,
+      "loss": 1.9165,
+      "step": 2685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848031567106013e-07,
+      "loss": 1.7244,
+      "step": 2686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846258679944815e-07,
+      "loss": 2.0275,
+      "step": 2687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784448515092213e-07,
+      "loss": 1.8532,
+      "step": 2688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784271098018305e-07,
+      "loss": 1.6807,
+      "step": 2689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840936167872698e-07,
+      "loss": 1.8546,
+      "step": 2690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783916071413627e-07,
+      "loss": 2.0016,
+      "step": 2691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837384619119006e-07,
+      "loss": 1.7815,
+      "step": 2692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835607882966203e-07,
+      "loss": 1.7879,
+      "step": 2693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833830505823197e-07,
+      "loss": 1.7414,
+      "step": 2694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832052487835392e-07,
+      "loss": 1.7922,
+      "step": 2695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830273829148236e-07,
+      "loss": 1.7385,
+      "step": 2696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828494529907235e-07,
+      "loss": 1.9153,
+      "step": 2697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782671459025794e-07,
+      "loss": 1.9012,
+      "step": 2698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824934010345963e-07,
+      "loss": 1.8052,
+      "step": 2699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782315279031696e-07,
+      "loss": 1.7963,
+      "step": 2700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821370930316647e-07,
+      "loss": 1.8794,
+      "step": 2701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819588430490785e-07,
+      "loss": 1.7713,
+      "step": 2702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781780529098519e-07,
+      "loss": 1.9373,
+      "step": 2703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816021511945736e-07,
+      "loss": 1.8261,
+      "step": 2704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814237093518345e-07,
+      "loss": 1.6605,
+      "step": 2705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812452035848988e-07,
+      "loss": 1.8517,
+      "step": 2706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810666339083687e-07,
+      "loss": 1.8334,
+      "step": 2707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808880003368533e-07,
+      "loss": 1.8283,
+      "step": 2708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807093028849645e-07,
+      "loss": 1.7744,
+      "step": 2709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805305415673214e-07,
+      "loss": 2.144,
+      "step": 2710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803517163985467e-07,
+      "loss": 1.7755,
+      "step": 2711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801728273932698e-07,
+      "loss": 1.8768,
+      "step": 2712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799938745661244e-07,
+      "loss": 1.6915,
+      "step": 2713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77981485793175e-07,
+      "loss": 2.0052,
+      "step": 2714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796357775047904e-07,
+      "loss": 1.8217,
+      "step": 2715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794566332998958e-07,
+      "loss": 1.7668,
+      "step": 2716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792774253317208e-07,
+      "loss": 1.8883,
+      "step": 2717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779098153614926e-07,
+      "loss": 1.8243,
+      "step": 2718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789188181641758e-07,
+      "loss": 1.909,
+      "step": 2719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787394189941415e-07,
+      "loss": 1.5451,
+      "step": 2720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785599561194981e-07,
+      "loss": 1.7161,
+      "step": 2721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778380429554927e-07,
+      "loss": 1.9412,
+      "step": 2722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782008393151146e-07,
+      "loss": 1.9279,
+      "step": 2723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780211854147513e-07,
+      "loss": 1.7939,
+      "step": 2724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778414678685345e-07,
+      "loss": 1.7914,
+      "step": 2725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776616866911654e-07,
+      "loss": 1.8375,
+      "step": 2726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774818418973516e-07,
+      "loss": 1.9855,
+      "step": 2727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773019335018044e-07,
+      "loss": 1.8322,
+      "step": 2728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777121961519242e-07,
+      "loss": 1.7668,
+      "step": 2729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769419259643863e-07,
+      "loss": 1.9212,
+      "step": 2730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767618268519656e-07,
+      "loss": 1.8039,
+      "step": 2731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765816641967127e-07,
+      "loss": 1.6244,
+      "step": 2732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764014380133654e-07,
+      "loss": 1.5578,
+      "step": 2733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762211483166678e-07,
+      "loss": 1.9353,
+      "step": 2734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776040795121368e-07,
+      "loss": 1.7788,
+      "step": 2735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758603784422195e-07,
+      "loss": 1.8326,
+      "step": 2736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756798982939815e-07,
+      "loss": 1.7082,
+      "step": 2737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754993546914182e-07,
+      "loss": 1.6975,
+      "step": 2738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753187476492994e-07,
+      "loss": 1.7987,
+      "step": 2739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751380771823984e-07,
+      "loss": 1.7824,
+      "step": 2740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774957343305496e-07,
+      "loss": 1.5799,
+      "step": 2741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774776546033377e-07,
+      "loss": 1.8154,
+      "step": 2742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745956853808308e-07,
+      "loss": 1.7895,
+      "step": 2743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744147613626533e-07,
+      "loss": 1.8745,
+      "step": 2744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742337739936447e-07,
+      "loss": 1.6239,
+      "step": 2745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740527232886104e-07,
+      "loss": 1.826,
+      "step": 2746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773871609262362e-07,
+      "loss": 1.9137,
+      "step": 2747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736904319297145e-07,
+      "loss": 1.6243,
+      "step": 2748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735091913054896e-07,
+      "loss": 1.913,
+      "step": 2749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773327887404514e-07,
+      "loss": 1.7793,
+      "step": 2750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731465202416183e-07,
+      "loss": 1.8156,
+      "step": 2751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77296508983164e-07,
+      "loss": 1.9219,
+      "step": 2752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727835961894205e-07,
+      "loss": 1.7209,
+      "step": 2753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772602039329807e-07,
+      "loss": 1.7269,
+      "step": 2754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772420419267652e-07,
+      "loss": 1.9362,
+      "step": 2755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722387360178127e-07,
+      "loss": 1.7396,
+      "step": 2756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722387360178127e-07,
+      "loss": 1.9445,
+      "step": 2757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720569895951513e-07,
+      "loss": 1.7598,
+      "step": 2758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718751800145362e-07,
+      "loss": 1.7755,
+      "step": 2759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716933072908398e-07,
+      "loss": 1.8534,
+      "step": 2760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77151137143894e-07,
+      "loss": 1.7486,
+      "step": 2761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713293724737207e-07,
+      "loss": 1.6608,
+      "step": 2762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711473104100697e-07,
+      "loss": 1.8232,
+      "step": 2763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709651852628805e-07,
+      "loss": 1.7904,
+      "step": 2764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707829970470525e-07,
+      "loss": 1.7833,
+      "step": 2765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770600745777489e-07,
+      "loss": 1.7918,
+      "step": 2766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770418431469099e-07,
+      "loss": 1.7581,
+      "step": 2767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702360541367968e-07,
+      "loss": 1.5967,
+      "step": 2768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770053613795502e-07,
+      "loss": 1.7167,
+      "step": 2769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698711104601387e-07,
+      "loss": 1.8336,
+      "step": 2770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696885441456368e-07,
+      "loss": 1.5996,
+      "step": 2771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769505914866931e-07,
+      "loss": 1.7193,
+      "step": 2772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769323222638961e-07,
+      "loss": 1.6318,
+      "step": 2773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691404674766722e-07,
+      "loss": 1.7548,
+      "step": 2774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.768957649395015e-07,
+      "loss": 1.7944,
+      "step": 2775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7687747684089446e-07,
+      "loss": 1.9356,
+      "step": 2776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7685918245334215e-07,
+      "loss": 2.0,
+      "step": 2777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7684088177834112e-07,
+      "loss": 1.8529,
+      "step": 2778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7682257481738853e-07,
+      "loss": 1.7225,
+      "step": 2779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7680426157198186e-07,
+      "loss": 1.9125,
+      "step": 2780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767859420436193e-07,
+      "loss": 1.7483,
+      "step": 2781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7676761623379943e-07,
+      "loss": 1.8832,
+      "step": 2782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7674928414402145e-07,
+      "loss": 1.7678,
+      "step": 2783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76730945775785e-07,
+      "loss": 1.7227,
+      "step": 2784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671260113059018e-07,
+      "loss": 1.9218,
+      "step": 2785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766942502099377e-07,
+      "loss": 1.6061,
+      "step": 2786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766758930153288e-07,
+      "loss": 1.884,
+      "step": 2787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7665752954826516e-07,
+      "loss": 1.6795,
+      "step": 2788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663915981024897e-07,
+      "loss": 1.7129,
+      "step": 2789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662078380278303e-07,
+      "loss": 1.7724,
+      "step": 2790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766024015273705e-07,
+      "loss": 1.8465,
+      "step": 2791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658401298551516e-07,
+      "loss": 1.9579,
+      "step": 2792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656561817872134e-07,
+      "loss": 1.82,
+      "step": 2793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7654721710849377e-07,
+      "loss": 1.7385,
+      "step": 2794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652880977633778e-07,
+      "loss": 1.9008,
+      "step": 2795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765103961837591e-07,
+      "loss": 1.6672,
+      "step": 2796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649197633226414e-07,
+      "loss": 1.8296,
+      "step": 2797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764735502233597e-07,
+      "loss": 1.9783,
+      "step": 2798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645511785855313e-07,
+      "loss": 1.693,
+      "step": 2799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643667923935225e-07,
+      "loss": 1.7593,
+      "step": 2800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641823436726545e-07,
+      "loss": 1.6914,
+      "step": 2801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639978324380165e-07,
+      "loss": 1.839,
+      "step": 2802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763813258704702e-07,
+      "loss": 1.7932,
+      "step": 2803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76362862248781e-07,
+      "loss": 1.7865,
+      "step": 2804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634439238024445e-07,
+      "loss": 1.85,
+      "step": 2805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763259162663715e-07,
+      "loss": 1.7722,
+      "step": 2806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630743390867357e-07,
+      "loss": 1.7174,
+      "step": 2807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628894530866263e-07,
+      "loss": 1.7973,
+      "step": 2808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762704504678511e-07,
+      "loss": 1.7889,
+      "step": 2809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625194938775195e-07,
+      "loss": 1.8577,
+      "step": 2810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623344206987869e-07,
+      "loss": 1.6514,
+      "step": 2811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621492851574526e-07,
+      "loss": 1.7968,
+      "step": 2812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619640872686623e-07,
+      "loss": 1.6446,
+      "step": 2813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617788270475653e-07,
+      "loss": 1.9695,
+      "step": 2814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761593504509317e-07,
+      "loss": 1.785,
+      "step": 2815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761408119669078e-07,
+      "loss": 1.6309,
+      "step": 2816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612226725420135e-07,
+      "loss": 1.8699,
+      "step": 2817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610371631432935e-07,
+      "loss": 1.8814,
+      "step": 2818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608515914880942e-07,
+      "loss": 1.9525,
+      "step": 2819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760665957591596e-07,
+      "loss": 1.8138,
+      "step": 2820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604802614689844e-07,
+      "loss": 1.7471,
+      "step": 2821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602945031354509e-07,
+      "loss": 1.8673,
+      "step": 2822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601086826061907e-07,
+      "loss": 1.7379,
+      "step": 2823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759922799896405e-07,
+      "loss": 1.7574,
+      "step": 2824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597368550213e-07,
+      "loss": 1.9483,
+      "step": 2825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759550847996087e-07,
+      "loss": 1.8785,
+      "step": 2826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759364778835982e-07,
+      "loss": 1.8113,
+      "step": 2827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591786475562066e-07,
+      "loss": 1.95,
+      "step": 2828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589924541719871e-07,
+      "loss": 1.6198,
+      "step": 2829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588061986985551e-07,
+      "loss": 1.7481,
+      "step": 2830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758619881151147e-07,
+      "loss": 1.7926,
+      "step": 2831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584335015450044e-07,
+      "loss": 1.6416,
+      "step": 2832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582470598953742e-07,
+      "loss": 1.9088,
+      "step": 2833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580605562175087e-07,
+      "loss": 1.7835,
+      "step": 2834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7578739905266637e-07,
+      "loss": 1.5993,
+      "step": 2835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576873628381023e-07,
+      "loss": 1.634,
+      "step": 2836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575006731670906e-07,
+      "loss": 1.8612,
+      "step": 2837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573139215289015e-07,
+      "loss": 1.7725,
+      "step": 2838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571271079388117e-07,
+      "loss": 1.8424,
+      "step": 2839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569402324121036e-07,
+      "loss": 1.7092,
+      "step": 2840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567532949640646e-07,
+      "loss": 1.6117,
+      "step": 2841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565662956099866e-07,
+      "loss": 1.9205,
+      "step": 2842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563792343651672e-07,
+      "loss": 1.6589,
+      "step": 2843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561921112449095e-07,
+      "loss": 1.9713,
+      "step": 2844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560049262645206e-07,
+      "loss": 1.9156,
+      "step": 2845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755817679439313e-07,
+      "loss": 1.956,
+      "step": 2846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755630370784605e-07,
+      "loss": 1.85,
+      "step": 2847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7554430003157186e-07,
+      "loss": 2.0043,
+      "step": 2848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755255568047982e-07,
+      "loss": 1.7432,
+      "step": 2849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550680739967279e-07,
+      "loss": 1.6341,
+      "step": 2850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754880518177294e-07,
+      "loss": 1.7875,
+      "step": 2851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754692900605024e-07,
+      "loss": 1.8324,
+      "step": 2852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545052212952652e-07,
+      "loss": 1.7986,
+      "step": 2853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754317480263371e-07,
+      "loss": 1.6904,
+      "step": 2854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541296775246997e-07,
+      "loss": 1.6121,
+      "step": 2855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753941813094614e-07,
+      "loss": 1.7449,
+      "step": 2856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753753886988482e-07,
+      "loss": 1.7527,
+      "step": 2857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535658992216775e-07,
+      "loss": 1.6484,
+      "step": 2858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533778498095787e-07,
+      "loss": 1.8444,
+      "step": 2859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531897387675686e-07,
+      "loss": 1.6849,
+      "step": 2860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753001566111036e-07,
+      "loss": 1.7975,
+      "step": 2861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752813331855374e-07,
+      "loss": 1.8652,
+      "step": 2862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526250360159808e-07,
+      "loss": 1.6632,
+      "step": 2863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752436678608261e-07,
+      "loss": 1.6588,
+      "step": 2864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522482596476217e-07,
+      "loss": 2.0375,
+      "step": 2865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520597791494777e-07,
+      "loss": 1.9698,
+      "step": 2866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518712371292467e-07,
+      "loss": 1.7693,
+      "step": 2867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751682633602353e-07,
+      "loss": 1.8341,
+      "step": 2868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751493968584225e-07,
+      "loss": 1.8864,
+      "step": 2869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513052420902963e-07,
+      "loss": 1.7361,
+      "step": 2870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511164541360057e-07,
+      "loss": 1.5364,
+      "step": 2871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750927604736797e-07,
+      "loss": 1.887,
+      "step": 2872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750738693908119e-07,
+      "loss": 1.8219,
+      "step": 2873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505497216654254e-07,
+      "loss": 1.6862,
+      "step": 2874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503606880241751e-07,
+      "loss": 1.7001,
+      "step": 2875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750171592999832e-07,
+      "loss": 1.8693,
+      "step": 2876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749982436607865e-07,
+      "loss": 1.7916,
+      "step": 2877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749793218863748e-07,
+      "loss": 1.9106,
+      "step": 2878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496039397829597e-07,
+      "loss": 1.7428,
+      "step": 2879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494145993809841e-07,
+      "loss": 1.8208,
+      "step": 2880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492251976733105e-07,
+      "loss": 2.0079,
+      "step": 2881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490357346754326e-07,
+      "loss": 1.9533,
+      "step": 2882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748846210402849e-07,
+      "loss": 1.833,
+      "step": 2883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486566248710646e-07,
+      "loss": 1.903,
+      "step": 2884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484669780955877e-07,
+      "loss": 1.8531,
+      "step": 2885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482772700919324e-07,
+      "loss": 1.8367,
+      "step": 2886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480875008756177e-07,
+      "loss": 1.7748,
+      "step": 2887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7478976704621679e-07,
+      "loss": 1.7009,
+      "step": 2888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.747707778867112e-07,
+      "loss": 2.0737,
+      "step": 2889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7475178261059837e-07,
+      "loss": 1.7218,
+      "step": 2890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7473278121943226e-07,
+      "loss": 1.9292,
+      "step": 2891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.747137737147672e-07,
+      "loss": 1.7128,
+      "step": 2892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7469476009815817e-07,
+      "loss": 1.7727,
+      "step": 2893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746757403711605e-07,
+      "loss": 1.6998,
+      "step": 2894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7465671453533016e-07,
+      "loss": 1.7822,
+      "step": 2895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7463768259222356e-07,
+      "loss": 1.759,
+      "step": 2896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746186445433975e-07,
+      "loss": 2.0201,
+      "step": 2897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459960039040952e-07,
+      "loss": 1.6813,
+      "step": 2898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458055013481742e-07,
+      "loss": 2.005,
+      "step": 2899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745614937781797e-07,
+      "loss": 1.7041,
+      "step": 2900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454243132205514e-07,
+      "loss": 1.6855,
+      "step": 2901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452336276800326e-07,
+      "loss": 1.8556,
+      "step": 2902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450428811758387e-07,
+      "loss": 1.7013,
+      "step": 2903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448520737235738e-07,
+      "loss": 1.6629,
+      "step": 2904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446612053388474e-07,
+      "loss": 1.7297,
+      "step": 2905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444702760372728e-07,
+      "loss": 1.7693,
+      "step": 2906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442792858344694e-07,
+      "loss": 1.72,
+      "step": 2907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440882347460612e-07,
+      "loss": 1.9155,
+      "step": 2908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7438971227876765e-07,
+      "loss": 2.0736,
+      "step": 2909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437059499749496e-07,
+      "loss": 1.6848,
+      "step": 2910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435147163235193e-07,
+      "loss": 1.8402,
+      "step": 2911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433234218490292e-07,
+      "loss": 1.6958,
+      "step": 2912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431320665671286e-07,
+      "loss": 1.7592,
+      "step": 2913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429406504934706e-07,
+      "loss": 1.7898,
+      "step": 2914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7427491736437148e-07,
+      "loss": 1.6789,
+      "step": 2915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742557636033524e-07,
+      "loss": 1.8573,
+      "step": 2916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423660376785677e-07,
+      "loss": 1.8184,
+      "step": 2917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421743785945188e-07,
+      "loss": 1.8205,
+      "step": 2918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419826587970565e-07,
+      "loss": 1.8016,
+      "step": 2919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417908783018645e-07,
+      "loss": 1.797,
+      "step": 2920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741599037124631e-07,
+      "loss": 1.8904,
+      "step": 2921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414071352810491e-07,
+      "loss": 1.761,
+      "step": 2922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412151727868183e-07,
+      "loss": 1.95,
+      "step": 2923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410231496576416e-07,
+      "loss": 1.747,
+      "step": 2924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740831065909227e-07,
+      "loss": 1.8378,
+      "step": 2925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406389215572883e-07,
+      "loss": 1.7859,
+      "step": 2926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404467166175438e-07,
+      "loss": 1.8608,
+      "step": 2927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402544511057168e-07,
+      "loss": 1.894,
+      "step": 2928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7400621250375355e-07,
+      "loss": 1.8088,
+      "step": 2929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7398697384287332e-07,
+      "loss": 1.8537,
+      "step": 2930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396772912950478e-07,
+      "loss": 1.9003,
+      "step": 2931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394847836522224e-07,
+      "loss": 1.8401,
+      "step": 2932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392922155160054e-07,
+      "loss": 1.9165,
+      "step": 2933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390995869021492e-07,
+      "loss": 1.8537,
+      "step": 2934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389068978264123e-07,
+      "loss": 1.7254,
+      "step": 2935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387141483045572e-07,
+      "loss": 1.623,
+      "step": 2936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385213383523523e-07,
+      "loss": 1.8293,
+      "step": 2937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7383284679855698e-07,
+      "loss": 1.8482,
+      "step": 2938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381355372199872e-07,
+      "loss": 1.7041,
+      "step": 2939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737942546071388e-07,
+      "loss": 1.876,
+      "step": 2940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377494945555595e-07,
+      "loss": 1.7598,
+      "step": 2941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375563826882938e-07,
+      "loss": 1.8287,
+      "step": 2942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737363210485389e-07,
+      "loss": 1.7781,
+      "step": 2943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371699779626466e-07,
+      "loss": 1.8492,
+      "step": 2944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369766851358748e-07,
+      "loss": 1.8202,
+      "step": 2945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367833320208858e-07,
+      "loss": 1.8889,
+      "step": 2946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365899186334966e-07,
+      "loss": 1.7839,
+      "step": 2947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7363964449895293e-07,
+      "loss": 1.866,
+      "step": 2948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362029111048111e-07,
+      "loss": 1.9011,
+      "step": 2949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360093169951738e-07,
+      "loss": 1.8649,
+      "step": 2950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358156626764543e-07,
+      "loss": 1.7781,
+      "step": 2951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735621948164495e-07,
+      "loss": 1.7049,
+      "step": 2952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735428173475142e-07,
+      "loss": 1.7321,
+      "step": 2953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352343386242475e-07,
+      "loss": 1.7476,
+      "step": 2954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350404436276676e-07,
+      "loss": 1.7995,
+      "step": 2955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734846488501264e-07,
+      "loss": 1.9188,
+      "step": 2956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346524732609034e-07,
+      "loss": 1.7481,
+      "step": 2957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344583979224572e-07,
+      "loss": 1.8911,
+      "step": 2958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342642625018015e-07,
+      "loss": 1.889,
+      "step": 2959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340700670148175e-07,
+      "loss": 1.8167,
+      "step": 2960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338758114773917e-07,
+      "loss": 1.88,
+      "step": 2961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336814959054145e-07,
+      "loss": 1.8711,
+      "step": 2962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334871203147822e-07,
+      "loss": 1.9643,
+      "step": 2963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332926847213955e-07,
+      "loss": 1.7554,
+      "step": 2964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330981891411603e-07,
+      "loss": 1.7605,
+      "step": 2965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329036335899875e-07,
+      "loss": 1.7958,
+      "step": 2966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732709018083792e-07,
+      "loss": 1.7688,
+      "step": 2967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732514342638495e-07,
+      "loss": 1.7263,
+      "step": 2968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323196072700212e-07,
+      "loss": 1.8427,
+      "step": 2969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321248119943018e-07,
+      "loss": 1.7809,
+      "step": 2970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731929956827271e-07,
+      "loss": 1.7613,
+      "step": 2971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317350417848693e-07,
+      "loss": 1.7892,
+      "step": 2972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731540066883042e-07,
+      "loss": 1.7328,
+      "step": 2973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313450321377385e-07,
+      "loss": 1.8531,
+      "step": 2974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731149937564914e-07,
+      "loss": 1.9266,
+      "step": 2975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309547831805276e-07,
+      "loss": 1.845,
+      "step": 2976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307595690005444e-07,
+      "loss": 1.9555,
+      "step": 2977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305642950409335e-07,
+      "loss": 1.7839,
+      "step": 2978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303689613176692e-07,
+      "loss": 1.9247,
+      "step": 2979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301735678467313e-07,
+      "loss": 1.7522,
+      "step": 2980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7299781146441035e-07,
+      "loss": 1.5757,
+      "step": 2981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297826017257745e-07,
+      "loss": 1.7832,
+      "step": 2982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295870291077388e-07,
+      "loss": 1.8869,
+      "step": 2983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729391396805995e-07,
+      "loss": 1.828,
+      "step": 2984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291957048365467e-07,
+      "loss": 1.7324,
+      "step": 2985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728999953215402e-07,
+      "loss": 1.7405,
+      "step": 2986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728804141958575e-07,
+      "loss": 1.928,
+      "step": 2987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286082710820836e-07,
+      "loss": 1.7381,
+      "step": 2988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284123406019514e-07,
+      "loss": 1.7479,
+      "step": 2989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728216350534206e-07,
+      "loss": 1.6397,
+      "step": 2990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280203008948804e-07,
+      "loss": 1.8223,
+      "step": 2991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278241917000126e-07,
+      "loss": 1.7565,
+      "step": 2992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727628022965645e-07,
+      "loss": 2.0047,
+      "step": 2993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274317947078254e-07,
+      "loss": 1.8586,
+      "step": 2994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727235506942606e-07,
+      "loss": 1.6566,
+      "step": 2995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270391596860443e-07,
+      "loss": 1.7391,
+      "step": 2996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268427529542022e-07,
+      "loss": 1.682,
+      "step": 2997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726646286763147e-07,
+      "loss": 1.7709,
+      "step": 2998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72644976112895e-07,
+      "loss": 1.8467,
+      "step": 2999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262531760676886e-07,
+      "loss": 1.718,
+      "step": 3000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260565315954444e-07,
+      "loss": 1.6917,
+      "step": 3001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7258598277283032e-07,
+      "loss": 1.7586,
+      "step": 3002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7256630644823568e-07,
+      "loss": 1.6621,
+      "step": 3003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7254662418737015e-07,
+      "loss": 1.6229,
+      "step": 3004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.725269359918438e-07,
+      "loss": 1.8902,
+      "step": 3005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.725072418632672e-07,
+      "loss": 1.6414,
+      "step": 3006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7248754180325146e-07,
+      "loss": 1.9748,
+      "step": 3007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724678358134082e-07,
+      "loss": 1.6902,
+      "step": 3008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7244812389534933e-07,
+      "loss": 1.8254,
+      "step": 3009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7242840605068744e-07,
+      "loss": 1.8113,
+      "step": 3010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724086822810356e-07,
+      "loss": 1.7741,
+      "step": 3011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723889525880072e-07,
+      "loss": 1.7405,
+      "step": 3012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236921697321636e-07,
+      "loss": 1.6736,
+      "step": 3013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723494754382774e-07,
+      "loss": 1.7222,
+      "step": 3014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723297279848054e-07,
+      "loss": 1.775,
+      "step": 3015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723099746144157e-07,
+      "loss": 1.8041,
+      "step": 3016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229021532872428e-07,
+      "loss": 1.8804,
+      "step": 3017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722704501293475e-07,
+      "loss": 1.8278,
+      "step": 3018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225067901790228e-07,
+      "loss": 1.7724,
+      "step": 3019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72230901996006e-07,
+      "loss": 1.9004,
+      "step": 3020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221111906527645e-07,
+      "loss": 1.731,
+      "step": 3021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7219133022733204e-07,
+      "loss": 1.897,
+      "step": 3022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217153548379155e-07,
+      "loss": 1.7374,
+      "step": 3023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215173483627434e-07,
+      "loss": 2.0288,
+      "step": 3024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213192828640013e-07,
+      "loss": 1.9778,
+      "step": 3025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7211211583578924e-07,
+      "loss": 1.8288,
+      "step": 3026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720922974860624e-07,
+      "loss": 1.7903,
+      "step": 3027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207247323884085e-07,
+      "loss": 2.0006,
+      "step": 3028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720526430957463e-07,
+      "loss": 1.7431,
+      "step": 3029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203280705840094e-07,
+      "loss": 1.9698,
+      "step": 3030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201296512842748e-07,
+      "loss": 1.7878,
+      "step": 3031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719931173074491e-07,
+      "loss": 1.9005,
+      "step": 3032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197326359708938e-07,
+      "loss": 1.7226,
+      "step": 3033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719534039989725e-07,
+      "loss": 1.7086,
+      "step": 3034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719335385147231e-07,
+      "loss": 2.042,
+      "step": 3035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719136671459662e-07,
+      "loss": 1.9048,
+      "step": 3036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189378989432742e-07,
+      "loss": 1.659,
+      "step": 3037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187390676143277e-07,
+      "loss": 1.8333,
+      "step": 3038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185401774890885e-07,
+      "loss": 1.6543,
+      "step": 3039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718341228583826e-07,
+      "loss": 1.7117,
+      "step": 3040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718142220914816e-07,
+      "loss": 1.7219,
+      "step": 3041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179431544983375e-07,
+      "loss": 1.9425,
+      "step": 3042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177440293506759e-07,
+      "loss": 1.7763,
+      "step": 3043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175448454881197e-07,
+      "loss": 1.8407,
+      "step": 3044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173456029269638e-07,
+      "loss": 1.936,
+      "step": 3045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171463016835066e-07,
+      "loss": 1.7839,
+      "step": 3046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169469417740526e-07,
+      "loss": 1.8063,
+      "step": 3047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167475232149097e-07,
+      "loss": 1.7548,
+      "step": 3048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165480460223914e-07,
+      "loss": 1.7382,
+      "step": 3049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163485102128165e-07,
+      "loss": 1.7014,
+      "step": 3050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161489158025076e-07,
+      "loss": 1.7507,
+      "step": 3051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715949262807792e-07,
+      "loss": 1.9864,
+      "step": 3052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715749551245003e-07,
+      "loss": 1.8917,
+      "step": 3053
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155497811304777e-07,
+      "loss": 1.6713,
+      "step": 3054
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153499524805584e-07,
+      "loss": 1.6542,
+      "step": 3055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151500653115917e-07,
+      "loss": 1.575,
+      "step": 3056
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149501196399295e-07,
+      "loss": 1.5453,
+      "step": 3057
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147501154819285e-07,
+      "loss": 1.7828,
+      "step": 3058
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145500528539493e-07,
+      "loss": 1.8151,
+      "step": 3059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714349931772359e-07,
+      "loss": 1.7261,
+      "step": 3060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141497522535276e-07,
+      "loss": 1.7915,
+      "step": 3061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139495143138315e-07,
+      "loss": 1.6441,
+      "step": 3062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137492179696503e-07,
+      "loss": 1.752,
+      "step": 3063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135488632373697e-07,
+      "loss": 1.8113,
+      "step": 3064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133484501333798e-07,
+      "loss": 2.0013,
+      "step": 3065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131479786740748e-07,
+      "loss": 1.9084,
+      "step": 3066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129474488758547e-07,
+      "loss": 1.8407,
+      "step": 3067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127468607551235e-07,
+      "loss": 1.9973,
+      "step": 3068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125462143282908e-07,
+      "loss": 1.9337,
+      "step": 3069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7123455096117696e-07,
+      "loss": 1.9496,
+      "step": 3070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121447466219793e-07,
+      "loss": 1.8607,
+      "step": 3071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711943925375343e-07,
+      "loss": 1.806,
+      "step": 3072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117430458882884e-07,
+      "loss": 1.8019,
+      "step": 3073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711542108177249e-07,
+      "loss": 1.895,
+      "step": 3074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113411122586624e-07,
+      "loss": 1.8438,
+      "step": 3075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7111400581489707e-07,
+      "loss": 1.7949,
+      "step": 3076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109389458646213e-07,
+      "loss": 1.5873,
+      "step": 3077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710737775422066e-07,
+      "loss": 1.8788,
+      "step": 3078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105365468377617e-07,
+      "loss": 1.8032,
+      "step": 3079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103352601281698e-07,
+      "loss": 1.7371,
+      "step": 3080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101339153097564e-07,
+      "loss": 1.8674,
+      "step": 3081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099325123989928e-07,
+      "loss": 1.7008,
+      "step": 3082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097310514123543e-07,
+      "loss": 1.7147,
+      "step": 3083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095295323663215e-07,
+      "loss": 1.7756,
+      "step": 3084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7093279552773796e-07,
+      "loss": 1.7935,
+      "step": 3085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091263201620187e-07,
+      "loss": 1.7147,
+      "step": 3086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7089246270367336e-07,
+      "loss": 1.7925,
+      "step": 3087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087228759180236e-07,
+      "loss": 1.8918,
+      "step": 3088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7085210668223926e-07,
+      "loss": 1.7171,
+      "step": 3089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083191997663502e-07,
+      "loss": 1.6162,
+      "step": 3090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081172747664097e-07,
+      "loss": 1.805,
+      "step": 3091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079152918390896e-07,
+      "loss": 1.7601,
+      "step": 3092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707713251000913e-07,
+      "loss": 1.8199,
+      "step": 3093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075111522684078e-07,
+      "loss": 1.7308,
+      "step": 3094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073089956581068e-07,
+      "loss": 1.744,
+      "step": 3095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865473e-07,
+      "loss": 1.7792,
+      "step": 3096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069045088702716e-07,
+      "loss": 1.7378,
+      "step": 3097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706702178725826e-07,
+      "loss": 1.6773,
+      "step": 3098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7064997907697628e-07,
+      "loss": 1.6645,
+      "step": 3099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7062973450186377e-07,
+      "loss": 1.5947,
+      "step": 3100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706094841489012e-07,
+      "loss": 1.7801,
+      "step": 3101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058922801974513e-07,
+      "loss": 1.6699,
+      "step": 3102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056896611605263e-07,
+      "loss": 1.7774,
+      "step": 3103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705486984394812e-07,
+      "loss": 1.8436,
+      "step": 3104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052842499168887e-07,
+      "loss": 1.6941,
+      "step": 3105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050814577433405e-07,
+      "loss": 1.9508,
+      "step": 3106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704878607890757e-07,
+      "loss": 1.8516,
+      "step": 3107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046757003757323e-07,
+      "loss": 1.7258,
+      "step": 3108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044727352148656e-07,
+      "loss": 1.6929,
+      "step": 3109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70426971242476e-07,
+      "loss": 1.8616,
+      "step": 3110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040666320220235e-07,
+      "loss": 1.8568,
+      "step": 3111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7038634940232694e-07,
+      "loss": 1.963,
+      "step": 3112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036602984451155e-07,
+      "loss": 1.8206,
+      "step": 3113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034570453041838e-07,
+      "loss": 1.822,
+      "step": 3114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7032537346171018e-07,
+      "loss": 1.7722,
+      "step": 3115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7030503664005006e-07,
+      "loss": 1.7891,
+      "step": 3116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7028469406710176e-07,
+      "loss": 1.9253,
+      "step": 3117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7026434574452934e-07,
+      "loss": 1.7298,
+      "step": 3118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.702439916739974e-07,
+      "loss": 1.9048,
+      "step": 3119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.70223631857171e-07,
+      "loss": 1.656,
+      "step": 3120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7020326629571566e-07,
+      "loss": 1.8343,
+      "step": 3121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701828949912974e-07,
+      "loss": 1.9013,
+      "step": 3122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701625179455827e-07,
+      "loss": 1.8983,
+      "step": 3123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7014213516023847e-07,
+      "loss": 1.6517,
+      "step": 3124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012174663693214e-07,
+      "loss": 1.7894,
+      "step": 3125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010135237733158e-07,
+      "loss": 1.7797,
+      "step": 3126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7008095238310512e-07,
+      "loss": 1.8494,
+      "step": 3127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006054665592164e-07,
+      "loss": 1.7952,
+      "step": 3128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004013519745036e-07,
+      "loss": 1.6681,
+      "step": 3129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001971800936108e-07,
+      "loss": 1.84,
+      "step": 3130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999929509332397e-07,
+      "loss": 1.7165,
+      "step": 3131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699788664510098e-07,
+      "loss": 1.6157,
+      "step": 3132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6995843208408962e-07,
+      "loss": 1.7357,
+      "step": 3133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993799199423514e-07,
+      "loss": 2.0106,
+      "step": 3134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991754618311846e-07,
+      "loss": 1.7169,
+      "step": 3135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989709465241212e-07,
+      "loss": 1.9385,
+      "step": 3136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987663740378917e-07,
+      "loss": 1.7566,
+      "step": 3137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985617443892308e-07,
+      "loss": 1.6819,
+      "step": 3138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983570575948786e-07,
+      "loss": 2.0513,
+      "step": 3139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698152313671579e-07,
+      "loss": 1.9046,
+      "step": 3140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979475126360813e-07,
+      "loss": 1.9163,
+      "step": 3141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977426545051393e-07,
+      "loss": 1.656,
+      "step": 3142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975377392955113e-07,
+      "loss": 1.815,
+      "step": 3143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69733276702396e-07,
+      "loss": 1.7144,
+      "step": 3144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971277377072537e-07,
+      "loss": 1.7582,
+      "step": 3145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6969226513621644e-07,
+      "loss": 1.8859,
+      "step": 3146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696717508005469e-07,
+      "loss": 1.7254,
+      "step": 3147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69651230765395e-07,
+      "loss": 1.8808,
+      "step": 3148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963070503243925e-07,
+      "loss": 1.9238,
+      "step": 3149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961017360335887e-07,
+      "loss": 1.7469,
+      "step": 3150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958963647983337e-07,
+      "loss": 1.6867,
+      "step": 3151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956909366354281e-07,
+      "loss": 1.634,
+      "step": 3152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954854515616765e-07,
+      "loss": 1.675,
+      "step": 3153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952799095938893e-07,
+      "loss": 1.8837,
+      "step": 3154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69507431074888e-07,
+      "loss": 1.9387,
+      "step": 3155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694868655043468e-07,
+      "loss": 1.7725,
+      "step": 3156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946629424944768e-07,
+      "loss": 1.814,
+      "step": 3157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944571731187348e-07,
+      "loss": 1.8314,
+      "step": 3158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942513469330749e-07,
+      "loss": 1.7751,
+      "step": 3159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940454639543344e-07,
+      "loss": 1.7632,
+      "step": 3160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693839524199356e-07,
+      "loss": 1.7436,
+      "step": 3161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693633527684986e-07,
+      "loss": 1.7738,
+      "step": 3162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934274744280764e-07,
+      "loss": 1.8558,
+      "step": 3163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693221364445483e-07,
+      "loss": 1.7933,
+      "step": 3164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930151977540667e-07,
+      "loss": 1.6044,
+      "step": 3165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928089743706928e-07,
+      "loss": 1.7666,
+      "step": 3166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926026943122317e-07,
+      "loss": 1.8778,
+      "step": 3167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923963575955575e-07,
+      "loss": 1.8156,
+      "step": 3168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6921899642375506e-07,
+      "loss": 1.7771,
+      "step": 3169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919835142550936e-07,
+      "loss": 1.7553,
+      "step": 3170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6917770076650762e-07,
+      "loss": 1.8553,
+      "step": 3171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915704444843908e-07,
+      "loss": 1.8777,
+      "step": 3172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691363824729936e-07,
+      "loss": 1.7564,
+      "step": 3173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911571484186137e-07,
+      "loss": 1.7994,
+      "step": 3174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6909504155673317e-07,
+      "loss": 1.6519,
+      "step": 3175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907436261930003e-07,
+      "loss": 1.7277,
+      "step": 3176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905367803125376e-07,
+      "loss": 1.9816,
+      "step": 3177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690329877942864e-07,
+      "loss": 1.813,
+      "step": 3178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901229191009042e-07,
+      "loss": 1.8155,
+      "step": 3179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899159038035898e-07,
+      "loss": 1.8301,
+      "step": 3180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897088320678545e-07,
+      "loss": 1.9004,
+      "step": 3181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895017039106386e-07,
+      "loss": 1.833,
+      "step": 3182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892945193488857e-07,
+      "loss": 1.7274,
+      "step": 3183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890872783995446e-07,
+      "loss": 1.9224,
+      "step": 3184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888799810795685e-07,
+      "loss": 1.7741,
+      "step": 3185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886726274059155e-07,
+      "loss": 1.7557,
+      "step": 3186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688465217395548e-07,
+      "loss": 1.7451,
+      "step": 3187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882577510654333e-07,
+      "loss": 1.8049,
+      "step": 3188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880502284325432e-07,
+      "loss": 1.8207,
+      "step": 3189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878426495138536e-07,
+      "loss": 1.8308,
+      "step": 3190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876350143263455e-07,
+      "loss": 1.7105,
+      "step": 3191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687427322887005e-07,
+      "loss": 1.9788,
+      "step": 3192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872195752128218e-07,
+      "loss": 1.7699,
+      "step": 3193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870117713207907e-07,
+      "loss": 1.7399,
+      "step": 3194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686803911227911e-07,
+      "loss": 1.8886,
+      "step": 3195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686595994951187e-07,
+      "loss": 1.8646,
+      "step": 3196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686388022507627e-07,
+      "loss": 1.8669,
+      "step": 3197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686179993914244e-07,
+      "loss": 1.9044,
+      "step": 3198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685971909188056e-07,
+      "loss": 1.6814,
+      "step": 3199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685763768346085e-07,
+      "loss": 1.886,
+      "step": 3200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855555714053584e-07,
+      "loss": 1.8841,
+      "step": 3201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853473183829072e-07,
+      "loss": 1.7806,
+      "step": 3202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851390092957681e-07,
+      "loss": 1.7818,
+      "step": 3203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684930644160981e-07,
+      "loss": 1.9394,
+      "step": 3204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847222229955917e-07,
+      "loss": 1.5505,
+      "step": 3205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68451374581665e-07,
+      "loss": 1.8025,
+      "step": 3206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68430521264121e-07,
+      "loss": 1.6864,
+      "step": 3207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840966234863315e-07,
+      "loss": 1.9791,
+      "step": 3208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838879783690772e-07,
+      "loss": 1.967,
+      "step": 3209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836792773065156e-07,
+      "loss": 1.8245,
+      "step": 3210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6834705203157193e-07,
+      "loss": 1.881,
+      "step": 3211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683261707413766e-07,
+      "loss": 1.7645,
+      "step": 3212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830528386177375e-07,
+      "loss": 1.7894,
+      "step": 3213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68284391394472e-07,
+      "loss": 1.5978,
+      "step": 3214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826349334118044e-07,
+      "loss": 1.7634,
+      "step": 3215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682425897036087e-07,
+      "loss": 1.9323,
+      "step": 3216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822168048346673e-07,
+      "loss": 1.7777,
+      "step": 3217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820076568246501e-07,
+      "loss": 1.7322,
+      "step": 3218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817984530231454e-07,
+      "loss": 1.9277,
+      "step": 3219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6815891934472665e-07,
+      "loss": 1.86,
+      "step": 3220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813798781141317e-07,
+      "loss": 1.6115,
+      "step": 3221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811705070408644e-07,
+      "loss": 1.813,
+      "step": 3222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809610802445918e-07,
+      "loss": 1.7811,
+      "step": 3223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807515977424465e-07,
+      "loss": 1.7827,
+      "step": 3224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805420595515644e-07,
+      "loss": 1.8539,
+      "step": 3225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803324656890875e-07,
+      "loss": 1.6461,
+      "step": 3226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6801228161721612e-07,
+      "loss": 1.6802,
+      "step": 3227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6799131110179356e-07,
+      "loss": 1.8564,
+      "step": 3228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6797033502435666e-07,
+      "loss": 1.6622,
+      "step": 3229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.679493533866212e-07,
+      "loss": 1.8029,
+      "step": 3230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.679283661903037e-07,
+      "loss": 1.6969,
+      "step": 3231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67907373437121e-07,
+      "loss": 1.7372,
+      "step": 3232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6788637512879042e-07,
+      "loss": 1.9056,
+      "step": 3233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6786537126702965e-07,
+      "loss": 1.8043,
+      "step": 3234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6784436185355694e-07,
+      "loss": 1.6485,
+      "step": 3235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6782334689009098e-07,
+      "loss": 1.7835,
+      "step": 3236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6780232637835088e-07,
+      "loss": 1.7514,
+      "step": 3237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778130032005623e-07,
+      "loss": 1.8049,
+      "step": 3238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776026871692705e-07,
+      "loss": 1.9757,
+      "step": 3239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773923157068385e-07,
+      "loss": 1.99,
+      "step": 3240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6771818888304753e-07,
+      "loss": 1.9009,
+      "step": 3241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676971406557395e-07,
+      "loss": 1.7316,
+      "step": 3242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676760868904816e-07,
+      "loss": 1.7632,
+      "step": 3243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765502758899613e-07,
+      "loss": 1.9486,
+      "step": 3244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676339627530059e-07,
+      "loss": 1.7508,
+      "step": 3245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67612892384234e-07,
+      "loss": 1.5791,
+      "step": 3246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759181648440417e-07,
+      "loss": 1.8253,
+      "step": 3247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675707350552405e-07,
+      "loss": 1.9091,
+      "step": 3248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6754964809846754e-07,
+      "loss": 1.7642,
+      "step": 3249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752855561581033e-07,
+      "loss": 1.7461,
+      "step": 3250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750745760899433e-07,
+      "loss": 1.7532,
+      "step": 3251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674863540797454e-07,
+      "loss": 1.9885,
+      "step": 3252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746524502979003e-07,
+      "loss": 1.9737,
+      "step": 3253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674441304608549e-07,
+      "loss": 1.6468,
+      "step": 3254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742301037466738e-07,
+      "loss": 1.5688,
+      "step": 3255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740188477295516e-07,
+      "loss": 1.8148,
+      "step": 3256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673807536574464e-07,
+      "loss": 1.6528,
+      "step": 3257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673596170298698e-07,
+      "loss": 1.9153,
+      "step": 3258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6733847489195432e-07,
+      "loss": 1.6164,
+      "step": 3259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673173272454296e-07,
+      "loss": 1.8881,
+      "step": 3260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6729617409202554e-07,
+      "loss": 1.925,
+      "step": 3261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672750154334726e-07,
+      "loss": 1.7306,
+      "step": 3262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725385127150168e-07,
+      "loss": 1.6854,
+      "step": 3263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723268160784408e-07,
+      "loss": 1.818,
+      "step": 3264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6721150644423157e-07,
+      "loss": 1.6243,
+      "step": 3265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719032578239644e-07,
+      "loss": 1.7624,
+      "step": 3266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671691396240713e-07,
+      "loss": 1.763,
+      "step": 3267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671479479709893e-07,
+      "loss": 1.8261,
+      "step": 3268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6712675082488405e-07,
+      "loss": 1.8801,
+      "step": 3269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710554818748951e-07,
+      "loss": 1.8383,
+      "step": 3270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708434006054023e-07,
+      "loss": 1.8132,
+      "step": 3271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670631264457711e-07,
+      "loss": 1.6775,
+      "step": 3272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670419073449175e-07,
+      "loss": 1.8811,
+      "step": 3273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670206827597153e-07,
+      "loss": 1.8158,
+      "step": 3274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699945269190066e-07,
+      "loss": 1.8256,
+      "step": 3275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669782171432104e-07,
+      "loss": 1.7194,
+      "step": 3276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695697611538163e-07,
+      "loss": 1.5455,
+      "step": 3277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693572961015203e-07,
+      "loss": 1.8962,
+      "step": 3278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691447762925963e-07,
+      "loss": 1.605,
+      "step": 3279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689322017444293e-07,
+      "loss": 1.6781,
+      "step": 3280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687195724744095e-07,
+      "loss": 1.5248,
+      "step": 3281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685068884999298e-07,
+      "loss": 1.866,
+      "step": 3282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66829414983839e-07,
+      "loss": 1.7218,
+      "step": 3283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680813565071925e-07,
+      "loss": 1.6496,
+      "step": 3284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667868508523745e-07,
+      "loss": 1.6378,
+      "step": 3285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66765560590546e-07,
+      "loss": 1.751,
+      "step": 3286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674426486697528e-07,
+      "loss": 1.863,
+      "step": 3287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667229636834045e-07,
+      "loss": 1.7081,
+      "step": 3288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667016570415762e-07,
+      "loss": 1.6443,
+      "step": 3289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6668034494323338e-07,
+      "loss": 1.7489,
+      "step": 3290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665902739011942e-07,
+      "loss": 1.7988,
+      "step": 3291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663770438397822e-07,
+      "loss": 1.7444,
+      "step": 3292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661637592655412e-07,
+      "loss": 1.7234,
+      "step": 3293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659504201959188e-07,
+      "loss": 1.9476,
+      "step": 3294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657370266483674e-07,
+      "loss": 1.716,
+      "step": 3295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665523578640343e-07,
+      "loss": 1.8222,
+      "step": 3296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653100761893073e-07,
+      "loss": 1.8544,
+      "step": 3297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650965193127252e-07,
+      "loss": 1.7788,
+      "step": 3298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648829080280678e-07,
+      "loss": 1.6652,
+      "step": 3299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646692423528079e-07,
+      "loss": 1.8236,
+      "step": 3300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664455522304426e-07,
+      "loss": 1.6814,
+      "step": 3301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664241747900404e-07,
+      "loss": 1.8472,
+      "step": 3302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640279191582308e-07,
+      "loss": 1.9575,
+      "step": 3303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638140360953978e-07,
+      "loss": 1.7596,
+      "step": 3304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663600098729402e-07,
+      "loss": 1.8046,
+      "step": 3305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663386107077745e-07,
+      "loss": 1.8328,
+      "step": 3306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631720611579314e-07,
+      "loss": 1.6274,
+      "step": 3307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629579609874715e-07,
+      "loss": 1.7744,
+      "step": 3308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627438065838802e-07,
+      "loss": 1.6167,
+      "step": 3309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625295979646755e-07,
+      "loss": 1.7655,
+      "step": 3310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623153351473817e-07,
+      "loss": 1.7485,
+      "step": 3311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621010181495258e-07,
+      "loss": 1.8177,
+      "step": 3312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66188664698864e-07,
+      "loss": 1.876,
+      "step": 3313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616722216822614e-07,
+      "loss": 1.6722,
+      "step": 3314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614577422479303e-07,
+      "loss": 1.6842,
+      "step": 3315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612432087031927e-07,
+      "loss": 1.6235,
+      "step": 3316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661028621065598e-07,
+      "loss": 1.866,
+      "step": 3317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6608139793527012e-07,
+      "loss": 1.8918,
+      "step": 3318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605992835820605e-07,
+      "loss": 1.5351,
+      "step": 3319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660384533771239e-07,
+      "loss": 1.8555,
+      "step": 3320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660169729937804e-07,
+      "loss": 1.9264,
+      "step": 3321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599548720993287e-07,
+      "loss": 1.9579,
+      "step": 3322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597399602733882e-07,
+      "loss": 1.7281,
+      "step": 3323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595249944775635e-07,
+      "loss": 1.6777,
+      "step": 3324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593099747294404e-07,
+      "loss": 1.7209,
+      "step": 3325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6590949010466083e-07,
+      "loss": 1.8048,
+      "step": 3326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588797734466616e-07,
+      "loss": 1.759,
+      "step": 3327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586645919471978e-07,
+      "loss": 1.8457,
+      "step": 3328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584493565658205e-07,
+      "loss": 1.6384,
+      "step": 3329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582340673201366e-07,
+      "loss": 1.656,
+      "step": 3330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580187242277588e-07,
+      "loss": 1.9974,
+      "step": 3331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6578033273063017e-07,
+      "loss": 1.8424,
+      "step": 3332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657587876573387e-07,
+      "loss": 1.7934,
+      "step": 3333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573723720466387e-07,
+      "loss": 1.9966,
+      "step": 3334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6571568137436866e-07,
+      "loss": 1.8523,
+      "step": 3335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569412016821644e-07,
+      "loss": 1.7617,
+      "step": 3336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567255358797101e-07,
+      "loss": 1.6879,
+      "step": 3337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565098163539665e-07,
+      "loss": 1.7741,
+      "step": 3338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562940431225797e-07,
+      "loss": 1.7596,
+      "step": 3339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560782162032018e-07,
+      "loss": 1.7048,
+      "step": 3340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.655862335613488e-07,
+      "loss": 1.7862,
+      "step": 3341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.655646401371099e-07,
+      "loss": 1.9181,
+      "step": 3342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.655430413493698e-07,
+      "loss": 1.808,
+      "step": 3343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.655214371998955e-07,
+      "loss": 1.7638,
+      "step": 3344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.654998276904543e-07,
+      "loss": 1.8393,
+      "step": 3345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6547821282281397e-07,
+      "loss": 2.0385,
+      "step": 3346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.654565925987426e-07,
+      "loss": 1.7391,
+      "step": 3347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65434967020009e-07,
+      "loss": 1.6032,
+      "step": 3348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.654133360883821e-07,
+      "loss": 1.7008,
+      "step": 3349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6539169980563148e-07,
+      "loss": 1.9303,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537005817352712e-07,
+      "loss": 1.6134,
+      "step": 3351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534841119383932e-07,
+      "loss": 1.813,
+      "step": 3352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65326758868339e-07,
+      "loss": 1.7131,
+      "step": 3353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530510119879737e-07,
+      "loss": 1.8502,
+      "step": 3354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528343818698614e-07,
+      "loss": 1.8114,
+      "step": 3355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6526176983467749e-07,
+      "loss": 1.7659,
+      "step": 3356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652400961436439e-07,
+      "loss": 1.7513,
+      "step": 3357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6521841711565848e-07,
+      "loss": 1.7802,
+      "step": 3358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6519673275249462e-07,
+      "loss": 1.6801,
+      "step": 3359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517504305592622e-07,
+      "loss": 1.771,
+      "step": 3360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651533480277276e-07,
+      "loss": 1.8828,
+      "step": 3361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513164766967353e-07,
+      "loss": 1.6795,
+      "step": 3362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651099419835392e-07,
+      "loss": 1.9906,
+      "step": 3363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508823097110018e-07,
+      "loss": 1.8727,
+      "step": 3364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506651463413263e-07,
+      "loss": 1.9374,
+      "step": 3365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504479297441304e-07,
+      "loss": 1.9211,
+      "step": 3366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502306599371825e-07,
+      "loss": 1.7851,
+      "step": 3367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650013336938257e-07,
+      "loss": 1.6765,
+      "step": 3368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497959607651323e-07,
+      "loss": 1.7697,
+      "step": 3369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495785314355902e-07,
+      "loss": 1.7607,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493610489674173e-07,
+      "loss": 1.7696,
+      "step": 3371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491435133784054e-07,
+      "loss": 1.7481,
+      "step": 3372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489259246863498e-07,
+      "loss": 1.6803,
+      "step": 3373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64870828290905e-07,
+      "loss": 1.7323,
+      "step": 3374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484905880643097e-07,
+      "loss": 1.7715,
+      "step": 3375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482728401699383e-07,
+      "loss": 1.878,
+      "step": 3376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480550392437486e-07,
+      "loss": 1.812,
+      "step": 3377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478371853035571e-07,
+      "loss": 1.7091,
+      "step": 3378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476192783671858e-07,
+      "loss": 1.6593,
+      "step": 3379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64740131845246e-07,
+      "loss": 1.8574,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6471833055772105e-07,
+      "loss": 1.8095,
+      "step": 3381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469652397592711e-07,
+      "loss": 1.7897,
+      "step": 3382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467471210164813e-07,
+      "loss": 1.8618,
+      "step": 3383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465289493666838e-07,
+      "loss": 1.858,
+      "step": 3384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463107248277266e-07,
+      "loss": 1.6414,
+      "step": 3385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646092447417461e-07,
+      "loss": 1.7779,
+      "step": 3386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645874117153743e-07,
+      "loss": 1.9259,
+      "step": 3387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456557340544336e-07,
+      "loss": 1.5464,
+      "step": 3388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454372981373975e-07,
+      "loss": 1.8487,
+      "step": 3389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452188094205036e-07,
+      "loss": 1.7161,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450002679216254e-07,
+      "loss": 2.0366,
+      "step": 3391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447816736586405e-07,
+      "loss": 1.7225,
+      "step": 3392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445630266494308e-07,
+      "loss": 1.7936,
+      "step": 3393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443443269118834e-07,
+      "loss": 1.6462,
+      "step": 3394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644125574463888e-07,
+      "loss": 1.7782,
+      "step": 3395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439067693233406e-07,
+      "loss": 1.8272,
+      "step": 3396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64368791150814e-07,
+      "loss": 1.708,
+      "step": 3397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434690010361896e-07,
+      "loss": 1.81,
+      "step": 3398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432500379253974e-07,
+      "loss": 1.9031,
+      "step": 3399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430310221936762e-07,
+      "loss": 1.8906,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642811953858942e-07,
+      "loss": 1.929,
+      "step": 3401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425928329391157e-07,
+      "loss": 1.819,
+      "step": 3402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6423736594521226e-07,
+      "loss": 1.836,
+      "step": 3403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421544334158918e-07,
+      "loss": 1.8625,
+      "step": 3404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419351548483576e-07,
+      "loss": 1.7644,
+      "step": 3405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417158237674576e-07,
+      "loss": 1.8054,
+      "step": 3406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641496440191134e-07,
+      "loss": 1.6628,
+      "step": 3407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641277004137334e-07,
+      "loss": 1.9198,
+      "step": 3408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6410575156240084e-07,
+      "loss": 1.7527,
+      "step": 3409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640837974669112e-07,
+      "loss": 1.7189,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406183812906042e-07,
+      "loss": 1.758,
+      "step": 3411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403987355064492e-07,
+      "loss": 1.7995,
+      "step": 3412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401790373346152e-07,
+      "loss": 1.7039,
+      "step": 3413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399592867930744e-07,
+      "loss": 1.7474,
+      "step": 3414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397394838998028e-07,
+      "loss": 1.8453,
+      "step": 3415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395196286727823e-07,
+      "loss": 1.845,
+      "step": 3416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392997211299977e-07,
+      "loss": 1.8348,
+      "step": 3417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390797612894386e-07,
+      "loss": 1.57,
+      "step": 3418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388597491690983e-07,
+      "loss": 1.9137,
+      "step": 3419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386396847869758e-07,
+      "loss": 1.8271,
+      "step": 3420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638419568161072e-07,
+      "loss": 1.8015,
+      "step": 3421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381993993093949e-07,
+      "loss": 1.7942,
+      "step": 3422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637979178249955e-07,
+      "loss": 1.8337,
+      "step": 3423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637758905000767e-07,
+      "loss": 1.9243,
+      "step": 3424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63753857957985e-07,
+      "loss": 1.7908,
+      "step": 3425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637318202005229e-07,
+      "loss": 1.6218,
+      "step": 3426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370977722949308e-07,
+      "loss": 1.6546,
+      "step": 3427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6368772904669878e-07,
+      "loss": 1.6799,
+      "step": 3428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366567565394368e-07,
+      "loss": 1.9211,
+      "step": 3429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364361705303183e-07,
+      "loss": 1.9103,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6362155324576775e-07,
+      "loss": 1.9216,
+      "step": 3431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6359948423395631e-07,
+      "loss": 1.7547,
+      "step": 3432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357741001940292e-07,
+      "loss": 1.8606,
+      "step": 3433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355533060391333e-07,
+      "loss": 1.7728,
+      "step": 3434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353324598929375e-07,
+      "loss": 1.8032,
+      "step": 3435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635111561773508e-07,
+      "loss": 1.7253,
+      "step": 3436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348906116989154e-07,
+      "loss": 1.7574,
+      "step": 3437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6346696096872345e-07,
+      "loss": 1.7129,
+      "step": 3438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634448555756544e-07,
+      "loss": 1.7627,
+      "step": 3439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634227449924928e-07,
+      "loss": 1.6132,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634006292210473e-07,
+      "loss": 1.9036,
+      "step": 3441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337850826312712e-07,
+      "loss": 1.8491,
+      "step": 3442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335638212054185e-07,
+      "loss": 1.9665,
+      "step": 3443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333425079510157e-07,
+      "loss": 1.9967,
+      "step": 3444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331211428861665e-07,
+      "loss": 1.7195,
+      "step": 3445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328997260289802e-07,
+      "loss": 1.7554,
+      "step": 3446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326782573975692e-07,
+      "loss": 1.7629,
+      "step": 3447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632456737010051e-07,
+      "loss": 1.7555,
+      "step": 3448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322351648845473e-07,
+      "loss": 1.7646,
+      "step": 3449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320135410391834e-07,
+      "loss": 1.7916,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6317918654920888e-07,
+      "loss": 1.9607,
+      "step": 3451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315701382613988e-07,
+      "loss": 1.958,
+      "step": 3452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313483593652507e-07,
+      "loss": 1.7518,
+      "step": 3453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311265288217872e-07,
+      "loss": 1.8037,
+      "step": 3454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6309046466491557e-07,
+      "loss": 1.8593,
+      "step": 3455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6306827128655068e-07,
+      "loss": 1.908,
+      "step": 3456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6304607274889956e-07,
+      "loss": 1.8436,
+      "step": 3457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.630238690537782e-07,
+      "loss": 2.0163,
+      "step": 3458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.630016602030029e-07,
+      "loss": 1.7307,
+      "step": 3459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6297944619839052e-07,
+      "loss": 1.8126,
+      "step": 3460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6295722704175827e-07,
+      "loss": 1.8475,
+      "step": 3461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6293500273492372e-07,
+      "loss": 1.9537,
+      "step": 3462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6291277327970502e-07,
+      "loss": 1.8358,
+      "step": 3463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628905386779205e-07,
+      "loss": 2.0533,
+      "step": 3464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628682989313892e-07,
+      "loss": 1.7998,
+      "step": 3465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628460540419304e-07,
+      "loss": 1.8985,
+      "step": 3466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628238040113638e-07,
+      "loss": 1.9033,
+      "step": 3467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280154884150958e-07,
+      "loss": 1.6548,
+      "step": 3468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627792885341883e-07,
+      "loss": 1.6746,
+      "step": 3469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62757023091221e-07,
+      "loss": 1.5568,
+      "step": 3470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273475251442908e-07,
+      "loss": 1.6599,
+      "step": 3471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627124768056344e-07,
+      "loss": 2.0211,
+      "step": 3472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269019596665915e-07,
+      "loss": 1.6729,
+      "step": 3473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626679099993261e-07,
+      "loss": 1.6742,
+      "step": 3474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264561890545832e-07,
+      "loss": 1.7493,
+      "step": 3475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262332268687928e-07,
+      "loss": 1.7322,
+      "step": 3476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6260102134541293e-07,
+      "loss": 1.7811,
+      "step": 3477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625787148828837e-07,
+      "loss": 1.8501,
+      "step": 3478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625564033011163e-07,
+      "loss": 1.6352,
+      "step": 3479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6253408660193593e-07,
+      "loss": 1.9386,
+      "step": 3480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625117647871682e-07,
+      "loss": 1.7883,
+      "step": 3481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6248943785863917e-07,
+      "loss": 1.6981,
+      "step": 3482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246710581817525e-07,
+      "loss": 1.8331,
+      "step": 3483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244476866760334e-07,
+      "loss": 1.7919,
+      "step": 3484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624224264087507e-07,
+      "loss": 1.8388,
+      "step": 3485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240007904344508e-07,
+      "loss": 1.8667,
+      "step": 3486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237772657351459e-07,
+      "loss": 1.7959,
+      "step": 3487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6235536900078768e-07,
+      "loss": 1.7258,
+      "step": 3488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233300632709344e-07,
+      "loss": 1.5798,
+      "step": 3489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623106385542611e-07,
+      "loss": 1.6108,
+      "step": 3490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622882656841206e-07,
+      "loss": 1.8186,
+      "step": 3491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226588771850204e-07,
+      "loss": 1.8045,
+      "step": 3492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224350465923612e-07,
+      "loss": 1.9689,
+      "step": 3493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622211165081538e-07,
+      "loss": 1.754,
+      "step": 3494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6219872326708662e-07,
+      "loss": 1.7793,
+      "step": 3495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217632493786637e-07,
+      "loss": 1.7523,
+      "step": 3496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621539215223254e-07,
+      "loss": 1.6342,
+      "step": 3497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213151302229641e-07,
+      "loss": 1.8342,
+      "step": 3498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621090994396125e-07,
+      "loss": 1.8561,
+      "step": 3499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208668077610724e-07,
+      "loss": 1.8604,
+      "step": 3500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620642570336146e-07,
+      "loss": 1.8431,
+      "step": 3501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620418282139689e-07,
+      "loss": 1.8293,
+      "step": 3502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6201939431900488e-07,
+      "loss": 1.9793,
+      "step": 3503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199695535055786e-07,
+      "loss": 1.721,
+      "step": 3504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197451131046338e-07,
+      "loss": 1.7415,
+      "step": 3505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619520622005575e-07,
+      "loss": 1.6765,
+      "step": 3506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619296080226766e-07,
+      "loss": 1.6821,
+      "step": 3507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190714877865764e-07,
+      "loss": 1.871,
+      "step": 3508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188468447033782e-07,
+      "loss": 1.686,
+      "step": 3509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186221509955486e-07,
+      "loss": 1.7482,
+      "step": 3510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183974066814685e-07,
+      "loss": 1.9177,
+      "step": 3511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618172611779523e-07,
+      "loss": 1.8191,
+      "step": 3512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179477663081017e-07,
+      "loss": 1.7778,
+      "step": 3513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177228702855976e-07,
+      "loss": 1.7958,
+      "step": 3514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174979237304087e-07,
+      "loss": 1.6882,
+      "step": 3515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617272926660936e-07,
+      "loss": 1.8157,
+      "step": 3516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170478790955864e-07,
+      "loss": 1.6569,
+      "step": 3517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616822781052769e-07,
+      "loss": 1.8454,
+      "step": 3518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165976325508984e-07,
+      "loss": 2.0098,
+      "step": 3519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163724336083923e-07,
+      "loss": 1.7379,
+      "step": 3520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616147184243674e-07,
+      "loss": 1.7001,
+      "step": 3521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159218844751686e-07,
+      "loss": 1.7717,
+      "step": 3522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615696534321308e-07,
+      "loss": 1.8206,
+      "step": 3523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615471133800526e-07,
+      "loss": 1.6654,
+      "step": 3524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152456829312616e-07,
+      "loss": 1.9291,
+      "step": 3525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6150201817319584e-07,
+      "loss": 1.7982,
+      "step": 3526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614794630221063e-07,
+      "loss": 1.5712,
+      "step": 3527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145690284170262e-07,
+      "loss": 1.7951,
+      "step": 3528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143433763383043e-07,
+      "loss": 1.7237,
+      "step": 3529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614117674003356e-07,
+      "loss": 1.69,
+      "step": 3530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6138919214306447e-07,
+      "loss": 1.7401,
+      "step": 3531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6136661186386387e-07,
+      "loss": 1.8364,
+      "step": 3532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613440265645809e-07,
+      "loss": 1.9075,
+      "step": 3533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132143624706322e-07,
+      "loss": 2.0137,
+      "step": 3534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129884091315877e-07,
+      "loss": 1.8484,
+      "step": 3535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6127624056471598e-07,
+      "loss": 1.5927,
+      "step": 3536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125363520358365e-07,
+      "loss": 1.9088,
+      "step": 3537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6123102483161105e-07,
+      "loss": 1.7775,
+      "step": 3538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120840945064776e-07,
+      "loss": 1.7467,
+      "step": 3539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118578906254383e-07,
+      "loss": 1.779,
+      "step": 3540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116316366914976e-07,
+      "loss": 1.8826,
+      "step": 3541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611405332723164e-07,
+      "loss": 1.8599,
+      "step": 3542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111789787389506e-07,
+      "loss": 1.8739,
+      "step": 3543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109525747573728e-07,
+      "loss": 1.7701,
+      "step": 3544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6107261207969534e-07,
+      "loss": 1.6713,
+      "step": 3545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610499616876216e-07,
+      "loss": 1.7848,
+      "step": 3546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102730630136908e-07,
+      "loss": 1.8197,
+      "step": 3547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61004645922791e-07,
+      "loss": 1.9345,
+      "step": 3548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098198055374116e-07,
+      "loss": 1.8224,
+      "step": 3549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6095931019607364e-07,
+      "loss": 1.8075,
+      "step": 3550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093663485164303e-07,
+      "loss": 1.7579,
+      "step": 3551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6091395452230426e-07,
+      "loss": 1.8916,
+      "step": 3552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608912692099127e-07,
+      "loss": 1.8881,
+      "step": 3553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608685789163241e-07,
+      "loss": 1.9802,
+      "step": 3554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084588364339464e-07,
+      "loss": 1.747,
+      "step": 3555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082318339298092e-07,
+      "loss": 1.6792,
+      "step": 3556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080047816693992e-07,
+      "loss": 1.8138,
+      "step": 3557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60777767967129e-07,
+      "loss": 2.1476,
+      "step": 3558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075505279540601e-07,
+      "loss": 1.7006,
+      "step": 3559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6073233265362915e-07,
+      "loss": 1.8205,
+      "step": 3560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070960754365702e-07,
+      "loss": 1.7207,
+      "step": 3561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068687746734867e-07,
+      "loss": 1.8117,
+      "step": 3562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066414242656352e-07,
+      "loss": 1.8163,
+      "step": 3563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064140242316135e-07,
+      "loss": 1.7421,
+      "step": 3564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6061865745900249e-07,
+      "loss": 1.7836,
+      "step": 3565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059590753594755e-07,
+      "loss": 1.9121,
+      "step": 3566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057315265585754e-07,
+      "loss": 1.7459,
+      "step": 3567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6055039282059396e-07,
+      "loss": 1.8823,
+      "step": 3568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.605276280320187e-07,
+      "loss": 1.8081,
+      "step": 3569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6050485829199397e-07,
+      "loss": 1.6915,
+      "step": 3570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6048208360238252e-07,
+      "loss": 1.744,
+      "step": 3571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6045930396504733e-07,
+      "loss": 1.7053,
+      "step": 3572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6043651938185195e-07,
+      "loss": 1.7277,
+      "step": 3573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.604137298546602e-07,
+      "loss": 1.7123,
+      "step": 3574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603909353853365e-07,
+      "loss": 1.7435,
+      "step": 3575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6036813597574544e-07,
+      "loss": 1.9547,
+      "step": 3576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6034533162775214e-07,
+      "loss": 1.693,
+      "step": 3577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032252234322213e-07,
+      "loss": 1.8422,
+      "step": 3578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602997081240213e-07,
+      "loss": 1.7834,
+      "step": 3579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027688897201593e-07,
+      "loss": 1.7917,
+      "step": 3580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025406488907283e-07,
+      "loss": 1.9029,
+      "step": 3581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023123587705905e-07,
+      "loss": 1.9057,
+      "step": 3582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020840193784208e-07,
+      "loss": 1.667,
+      "step": 3583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6018556307328993e-07,
+      "loss": 1.9307,
+      "step": 3584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016271928527087e-07,
+      "loss": 1.8466,
+      "step": 3585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013987057565366e-07,
+      "loss": 1.8642,
+      "step": 3586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601170169463074e-07,
+      "loss": 1.8615,
+      "step": 3587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009415839910164e-07,
+      "loss": 1.764,
+      "step": 3588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007129493590636e-07,
+      "loss": 1.8589,
+      "step": 3589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004842655859183e-07,
+      "loss": 1.7038,
+      "step": 3590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002555326902889e-07,
+      "loss": 1.811,
+      "step": 3591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000267506908858e-07,
+      "loss": 1.786,
+      "step": 3592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5997979196064248e-07,
+      "loss": 1.7467,
+      "step": 3593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995690394556257e-07,
+      "loss": 1.7584,
+      "step": 3594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993401102572118e-07,
+      "loss": 1.8297,
+      "step": 3595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5991111320299104e-07,
+      "loss": 1.7286,
+      "step": 3596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988821047924534e-07,
+      "loss": 1.8881,
+      "step": 3597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598653028563576e-07,
+      "loss": 1.549,
+      "step": 3598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984239033620177e-07,
+      "loss": 1.8124,
+      "step": 3599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981947292065224e-07,
+      "loss": 1.8406,
+      "step": 3600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5979655061158372e-07,
+      "loss": 1.7749,
+      "step": 3601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597736234108714e-07,
+      "loss": 1.8312,
+      "step": 3602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597506913203908e-07,
+      "loss": 1.8427,
+      "step": 3603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972775434201795e-07,
+      "loss": 1.699,
+      "step": 3604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970481247762911e-07,
+      "loss": 1.8727,
+      "step": 3605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5968186572910108e-07,
+      "loss": 1.7037,
+      "step": 3606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965891409831098e-07,
+      "loss": 1.8493,
+      "step": 3607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963595758713644e-07,
+      "loss": 1.656,
+      "step": 3608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961299619745534e-07,
+      "loss": 1.7229,
+      "step": 3609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5959002993114609e-07,
+      "loss": 1.5708,
+      "step": 3610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5956705879008736e-07,
+      "loss": 1.9694,
+      "step": 3611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954408277615836e-07,
+      "loss": 1.7943,
+      "step": 3612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952110189123863e-07,
+      "loss": 1.8602,
+      "step": 3613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949811613720813e-07,
+      "loss": 1.725,
+      "step": 3614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947512551594714e-07,
+      "loss": 1.8382,
+      "step": 3615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945213002933644e-07,
+      "loss": 1.6158,
+      "step": 3616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942912967925722e-07,
+      "loss": 1.6013,
+      "step": 3617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940612446759095e-07,
+      "loss": 1.8416,
+      "step": 3618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593831143962196e-07,
+      "loss": 1.8617,
+      "step": 3619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593600994670255e-07,
+      "loss": 1.8832,
+      "step": 3620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5933707968189137e-07,
+      "loss": 1.7687,
+      "step": 3621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931405504270034e-07,
+      "loss": 1.8396,
+      "step": 3622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929102555133595e-07,
+      "loss": 1.8115,
+      "step": 3623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592679912096821e-07,
+      "loss": 1.8775,
+      "step": 3624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924495201962316e-07,
+      "loss": 1.7873,
+      "step": 3625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922190798304378e-07,
+      "loss": 1.7654,
+      "step": 3626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5919885910182912e-07,
+      "loss": 1.7431,
+      "step": 3627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917580537786466e-07,
+      "loss": 1.7501,
+      "step": 3628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915274681303633e-07,
+      "loss": 1.6855,
+      "step": 3629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912968340923037e-07,
+      "loss": 1.787,
+      "step": 3630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910661516833356e-07,
+      "loss": 1.8989,
+      "step": 3631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908354209223297e-07,
+      "loss": 1.9221,
+      "step": 3632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906046418281603e-07,
+      "loss": 1.8976,
+      "step": 3633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903738144197073e-07,
+      "loss": 1.7856,
+      "step": 3634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901429387158523e-07,
+      "loss": 1.7122,
+      "step": 3635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899120147354829e-07,
+      "loss": 1.78,
+      "step": 3636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589681042497489e-07,
+      "loss": 1.5699,
+      "step": 3637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894500220207663e-07,
+      "loss": 1.7734,
+      "step": 3638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892189533242122e-07,
+      "loss": 1.6827,
+      "step": 3639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5889878364267305e-07,
+      "loss": 1.7813,
+      "step": 3640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588756671347226e-07,
+      "loss": 1.6727,
+      "step": 3641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885254581046113e-07,
+      "loss": 1.7512,
+      "step": 3642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882941967177988e-07,
+      "loss": 1.9162,
+      "step": 3643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880628872057075e-07,
+      "loss": 1.7632,
+      "step": 3644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58783152958726e-07,
+      "loss": 1.812,
+      "step": 3645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876001238813817e-07,
+      "loss": 1.6835,
+      "step": 3646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873686701070035e-07,
+      "loss": 1.79,
+      "step": 3647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5871371682830587e-07,
+      "loss": 1.925,
+      "step": 3648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586905618428486e-07,
+      "loss": 1.8151,
+      "step": 3649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866740205622264e-07,
+      "loss": 1.7792,
+      "step": 3650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864423747032268e-07,
+      "loss": 1.7498,
+      "step": 3651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862106808704358e-07,
+      "loss": 1.6861,
+      "step": 3652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859789390828077e-07,
+      "loss": 1.9137,
+      "step": 3653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5857471493593e-07,
+      "loss": 1.6799,
+      "step": 3654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585515311718874e-07,
+      "loss": 1.9483,
+      "step": 3655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852834261804958e-07,
+      "loss": 2.0032,
+      "step": 3656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850514927631338e-07,
+      "loss": 1.7719,
+      "step": 3657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848195114857618e-07,
+      "loss": 1.7206,
+      "step": 3658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845874823673572e-07,
+      "loss": 1.7803,
+      "step": 3659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843554054269004e-07,
+      "loss": 1.8144,
+      "step": 3660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584123280683377e-07,
+      "loss": 1.6261,
+      "step": 3661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838911081557756e-07,
+      "loss": 1.8827,
+      "step": 3662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583658887863089e-07,
+      "loss": 1.6972,
+      "step": 3663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834266198243145e-07,
+      "loss": 1.9087,
+      "step": 3664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583194304058452e-07,
+      "loss": 1.7162,
+      "step": 3665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829619405845064e-07,
+      "loss": 1.7513,
+      "step": 3666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582729529421486e-07,
+      "loss": 1.6264,
+      "step": 3667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5824970705884033e-07,
+      "loss": 1.6391,
+      "step": 3668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822645641042742e-07,
+      "loss": 1.7623,
+      "step": 3669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5820320099881195e-07,
+      "loss": 1.8365,
+      "step": 3670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817994082589627e-07,
+      "loss": 1.6566,
+      "step": 3671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581566758935832e-07,
+      "loss": 1.767,
+      "step": 3672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581334062037759e-07,
+      "loss": 1.7986,
+      "step": 3673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811013175837796e-07,
+      "loss": 1.8722,
+      "step": 3674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5808685255929334e-07,
+      "loss": 1.8896,
+      "step": 3675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806356860842636e-07,
+      "loss": 1.6693,
+      "step": 3676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580402799076818e-07,
+      "loss": 1.6825,
+      "step": 3677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5801698645896476e-07,
+      "loss": 1.5969,
+      "step": 3678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799368826418075e-07,
+      "loss": 1.7359,
+      "step": 3679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579703853252357e-07,
+      "loss": 1.7723,
+      "step": 3680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.579470776440359e-07,
+      "loss": 1.827,
+      "step": 3681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5792376522248803e-07,
+      "loss": 1.6803,
+      "step": 3682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.579004480624991e-07,
+      "loss": 1.7939,
+      "step": 3683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5787712616597667e-07,
+      "loss": 1.7208,
+      "step": 3684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5785379953482848e-07,
+      "loss": 2.0839,
+      "step": 3685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5783046817096284e-07,
+      "loss": 1.7428,
+      "step": 3686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5780713207628832e-07,
+      "loss": 1.9328,
+      "step": 3687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5778379125271396e-07,
+      "loss": 1.726,
+      "step": 3688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5776044570214908e-07,
+      "loss": 1.7247,
+      "step": 3689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5773709542650354e-07,
+      "loss": 1.7613,
+      "step": 3690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771374042768746e-07,
+      "loss": 1.743,
+      "step": 3691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769038070761144e-07,
+      "loss": 1.7731,
+      "step": 3692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576670162681863e-07,
+      "loss": 1.6065,
+      "step": 3693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764364711132352e-07,
+      "loss": 1.8019,
+      "step": 3694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762027323893467e-07,
+      "loss": 1.8276,
+      "step": 3695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759689465293192e-07,
+      "loss": 1.7529,
+      "step": 3696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757351135522773e-07,
+      "loss": 1.5869,
+      "step": 3697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755012334773498e-07,
+      "loss": 1.7752,
+      "step": 3698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752673063236693e-07,
+      "loss": 1.5587,
+      "step": 3699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750333321103714e-07,
+      "loss": 1.908,
+      "step": 3700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747993108565975e-07,
+      "loss": 1.8168,
+      "step": 3701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745652425814907e-07,
+      "loss": 1.7595,
+      "step": 3702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743311273041992e-07,
+      "loss": 1.9001,
+      "step": 3703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574096965043875e-07,
+      "loss": 1.7464,
+      "step": 3704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573862755819673e-07,
+      "loss": 1.8017,
+      "step": 3705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736284996507537e-07,
+      "loss": 1.9144,
+      "step": 3706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5733941965562795e-07,
+      "loss": 1.7303,
+      "step": 3707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731598465554177e-07,
+      "loss": 1.7318,
+      "step": 3708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729254496673394e-07,
+      "loss": 1.7867,
+      "step": 3709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726910059112195e-07,
+      "loss": 1.6235,
+      "step": 3710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572456515306236e-07,
+      "loss": 1.8163,
+      "step": 3711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722219778715725e-07,
+      "loss": 1.8206,
+      "step": 3712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719873936264142e-07,
+      "loss": 1.6568,
+      "step": 3713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5717527625899517e-07,
+      "loss": 1.8944,
+      "step": 3714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571518084781379e-07,
+      "loss": 1.8724,
+      "step": 3715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712833602198935e-07,
+      "loss": 1.8827,
+      "step": 3716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5710485889246966e-07,
+      "loss": 1.7068,
+      "step": 3717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708137709149948e-07,
+      "loss": 1.7938,
+      "step": 3718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705789062099966e-07,
+      "loss": 1.7654,
+      "step": 3719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5703439948289146e-07,
+      "loss": 1.5464,
+      "step": 3720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5701090367909667e-07,
+      "loss": 1.7949,
+      "step": 3721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569874032115373e-07,
+      "loss": 1.7011,
+      "step": 3722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5696389808213578e-07,
+      "loss": 1.7792,
+      "step": 3723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56940388292815e-07,
+      "loss": 1.6185,
+      "step": 3724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691687384549811e-07,
+      "loss": 1.8524,
+      "step": 3725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568933547421088e-07,
+      "loss": 1.8322,
+      "step": 3726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686983098457092e-07,
+      "loss": 1.951,
+      "step": 3727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568463025748089e-07,
+      "loss": 1.6954,
+      "step": 3728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568227695147475e-07,
+      "loss": 1.6816,
+      "step": 3729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5679923180631175e-07,
+      "loss": 1.7263,
+      "step": 3730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677568945142725e-07,
+      "loss": 1.84,
+      "step": 3731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567521424520198e-07,
+      "loss": 1.7022,
+      "step": 3732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672859081001567e-07,
+      "loss": 1.9589,
+      "step": 3733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5670503452734153e-07,
+      "loss": 1.8799,
+      "step": 3734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566814736059244e-07,
+      "loss": 1.7704,
+      "step": 3735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5665790804769162e-07,
+      "loss": 1.7707,
+      "step": 3736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663433785457101e-07,
+      "loss": 1.7157,
+      "step": 3737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566107630284907e-07,
+      "loss": 1.9741,
+      "step": 3738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658718357137926e-07,
+      "loss": 1.964,
+      "step": 3739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5656359948516554e-07,
+      "loss": 1.9126,
+      "step": 3740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654001077177892e-07,
+      "loss": 1.8264,
+      "step": 3741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5651641743314899e-07,
+      "loss": 1.8194,
+      "step": 3742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649281947120586e-07,
+      "loss": 1.6103,
+      "step": 3743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646921688787988e-07,
+      "loss": 1.6583,
+      "step": 3744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564456096851019e-07,
+      "loss": 1.867,
+      "step": 3745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642199786480315e-07,
+      "loss": 1.6975,
+      "step": 3746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563983814289151e-07,
+      "loss": 1.7194,
+      "step": 3747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637476037936975e-07,
+      "loss": 1.6078,
+      "step": 3748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635113471809934e-07,
+      "loss": 1.8924,
+      "step": 3749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632750444703664e-07,
+      "loss": 1.8034,
+      "step": 3750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630386956811473e-07,
+      "loss": 1.6942,
+      "step": 3751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628023008326697e-07,
+      "loss": 1.8301,
+      "step": 3752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562565859944273e-07,
+      "loss": 1.7289,
+      "step": 3753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623293730352979e-07,
+      "loss": 1.9482,
+      "step": 3754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620928401250908e-07,
+      "loss": 1.838,
+      "step": 3755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618562612330016e-07,
+      "loss": 1.6794,
+      "step": 3756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5616196363783827e-07,
+      "loss": 1.7271,
+      "step": 3757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561382965580592e-07,
+      "loss": 1.759,
+      "step": 3758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611462488589898e-07,
+      "loss": 1.6879,
+      "step": 3759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609094862329405e-07,
+      "loss": 1.8316,
+      "step": 3760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5606726777218131e-07,
+      "loss": 1.676,
+      "step": 3761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604358233449793e-07,
+      "loss": 1.7414,
+      "step": 3762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5601989231218148e-07,
+      "loss": 1.7502,
+      "step": 3763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599619770716993e-07,
+      "loss": 1.7007,
+      "step": 3764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597249852140158e-07,
+      "loss": 1.8782,
+      "step": 3765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559487947568152e-07,
+      "loss": 1.8772,
+      "step": 3766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592508641534981e-07,
+      "loss": 2.0437,
+      "step": 3767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5590137349894493e-07,
+      "loss": 1.7149,
+      "step": 3768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587765600954033e-07,
+      "loss": 1.6881,
+      "step": 3769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5585393394907626e-07,
+      "loss": 1.6402,
+      "step": 3770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583020731949325e-07,
+      "loss": 1.7593,
+      "step": 3771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580647612273229e-07,
+      "loss": 1.6885,
+      "step": 3772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578274036073468e-07,
+      "loss": 1.8423,
+      "step": 3773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575900003544215e-07,
+      "loss": 1.833,
+      "step": 3774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557352551487967e-07,
+      "loss": 1.6775,
+      "step": 3775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571150570274087e-07,
+      "loss": 1.5161,
+      "step": 3776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568775169921745e-07,
+      "loss": 1.6624,
+      "step": 3777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556639931401696e-07,
+      "loss": 1.8058,
+      "step": 3778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564023002754093e-07,
+      "loss": 1.6554,
+      "step": 3779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561646236327527e-07,
+      "loss": 1.6764,
+      "step": 3780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559269014931708e-07,
+      "loss": 1.8179,
+      "step": 3781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556891338761094e-07,
+      "loss": 1.5532,
+      "step": 3782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5554513208010196e-07,
+      "loss": 1.7028,
+      "step": 3783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5552134622873547e-07,
+      "loss": 1.7274,
+      "step": 3784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549755583545738e-07,
+      "loss": 1.8829,
+      "step": 3785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554737609022138e-07,
+      "loss": 1.7222,
+      "step": 3786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544996143095128e-07,
+      "loss": 1.6878,
+      "step": 3787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542615742361673e-07,
+      "loss": 1.625,
+      "step": 3788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540234888215741e-07,
+      "loss": 1.8402,
+      "step": 3789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5537853580852102e-07,
+      "loss": 1.7925,
+      "step": 3790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535471820465553e-07,
+      "loss": 1.7266,
+      "step": 3791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553308960725094e-07,
+      "loss": 1.6382,
+      "step": 3792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553070694140313e-07,
+      "loss": 1.7996,
+      "step": 3793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5528323823117048e-07,
+      "loss": 1.796,
+      "step": 3794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5525940252587636e-07,
+      "loss": 1.8588,
+      "step": 3795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5523556230009882e-07,
+      "loss": 1.7467,
+      "step": 3796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.552117175557881e-07,
+      "loss": 1.6546,
+      "step": 3797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5518786829489491e-07,
+      "loss": 1.7659,
+      "step": 3798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.551640145193701e-07,
+      "loss": 1.6205,
+      "step": 3799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5514015623116512e-07,
+      "loss": 1.9424,
+      "step": 3800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5511629343223165e-07,
+      "loss": 1.8541,
+      "step": 3801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550924261245218e-07,
+      "loss": 1.7477,
+      "step": 3802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5506855430998798e-07,
+      "loss": 1.7413,
+      "step": 3803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5504467799058305e-07,
+      "loss": 1.7071,
+      "step": 3804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502079716826024e-07,
+      "loss": 1.7436,
+      "step": 3805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499691184497309e-07,
+      "loss": 1.9676,
+      "step": 3806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497302202267552e-07,
+      "loss": 1.8523,
+      "step": 3807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5494912770332182e-07,
+      "loss": 1.8142,
+      "step": 3808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549252288888667e-07,
+      "loss": 1.7078,
+      "step": 3809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490132558126522e-07,
+      "loss": 1.714,
+      "step": 3810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487741778247273e-07,
+      "loss": 1.9602,
+      "step": 3811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54853505494445e-07,
+      "loss": 1.7183,
+      "step": 3812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548295887191382e-07,
+      "loss": 1.7674,
+      "step": 3813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548056674585088e-07,
+      "loss": 1.7587,
+      "step": 3814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478174171451372e-07,
+      "loss": 1.895,
+      "step": 3815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475781148911017e-07,
+      "loss": 1.7284,
+      "step": 3816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5473387678425572e-07,
+      "loss": 1.7178,
+      "step": 3817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5470993760190844e-07,
+      "loss": 1.8059,
+      "step": 3818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5468599394402662e-07,
+      "loss": 1.8463,
+      "step": 3819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466204581256897e-07,
+      "loss": 1.7265,
+      "step": 3820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463809320949452e-07,
+      "loss": 1.7961,
+      "step": 3821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461413613676276e-07,
+      "loss": 1.7285,
+      "step": 3822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545901745963335e-07,
+      "loss": 1.8701,
+      "step": 3823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456620859016681e-07,
+      "loss": 1.7602,
+      "step": 3824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454223812022335e-07,
+      "loss": 1.8857,
+      "step": 3825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451826318846397e-07,
+      "loss": 1.5626,
+      "step": 3826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5449428379684993e-07,
+      "loss": 1.8169,
+      "step": 3827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5447029994734283e-07,
+      "loss": 1.7967,
+      "step": 3828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5444631164190472e-07,
+      "loss": 1.6934,
+      "step": 3829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442231888249792e-07,
+      "loss": 1.6164,
+      "step": 3830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439832167108518e-07,
+      "loss": 1.7427,
+      "step": 3831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437432000962955e-07,
+      "loss": 1.9299,
+      "step": 3832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435031390009453e-07,
+      "loss": 1.5625,
+      "step": 3833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432630334444387e-07,
+      "loss": 1.6107,
+      "step": 3834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430228834464184e-07,
+      "loss": 1.7382,
+      "step": 3835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427826890265286e-07,
+      "loss": 1.8969,
+      "step": 3836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425424502044194e-07,
+      "loss": 1.7842,
+      "step": 3837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423021669997428e-07,
+      "loss": 1.5972,
+      "step": 3838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420618394321555e-07,
+      "loss": 1.9654,
+      "step": 3839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418214675213177e-07,
+      "loss": 1.7968,
+      "step": 3840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415810512868922e-07,
+      "loss": 1.8078,
+      "step": 3841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5413405907485468e-07,
+      "loss": 1.7413,
+      "step": 3842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5411000859259523e-07,
+      "loss": 1.6431,
+      "step": 3843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540859536838783e-07,
+      "loss": 1.9484,
+      "step": 3844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406189435067165e-07,
+      "loss": 1.8797,
+      "step": 3845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403783059494354e-07,
+      "loss": 1.6056,
+      "step": 3846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401376241866242e-07,
+      "loss": 1.6907,
+      "step": 3847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398968982379725e-07,
+      "loss": 1.5675,
+      "step": 3848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396561281231723e-07,
+      "loss": 1.7811,
+      "step": 3849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53941531386192e-07,
+      "loss": 1.6654,
+      "step": 3850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391744554739154e-07,
+      "loss": 1.7624,
+      "step": 3851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538933552978862e-07,
+      "loss": 1.7506,
+      "step": 3852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386926063964657e-07,
+      "loss": 1.9107,
+      "step": 3853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538451615746439e-07,
+      "loss": 1.715,
+      "step": 3854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382105810484943e-07,
+      "loss": 1.7532,
+      "step": 3855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5379695023223503e-07,
+      "loss": 1.718,
+      "step": 3856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377283795877282e-07,
+      "loss": 1.5741,
+      "step": 3857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374872128643528e-07,
+      "loss": 1.7491,
+      "step": 3858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372460021719533e-07,
+      "loss": 1.842,
+      "step": 3859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537004747530261e-07,
+      "loss": 1.8746,
+      "step": 3860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367634489590128e-07,
+      "loss": 1.7717,
+      "step": 3861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365221064779465e-07,
+      "loss": 1.6049,
+      "step": 3862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536280720106807e-07,
+      "loss": 1.6685,
+      "step": 3863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360392898653393e-07,
+      "loss": 1.9234,
+      "step": 3864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357978157732943e-07,
+      "loss": 1.701,
+      "step": 3865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355562978504253e-07,
+      "loss": 1.9225,
+      "step": 3866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353147361164902e-07,
+      "loss": 1.7229,
+      "step": 3867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350731305912493e-07,
+      "loss": 1.8128,
+      "step": 3868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5348314812944678e-07,
+      "loss": 1.5682,
+      "step": 3869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5345897882459128e-07,
+      "loss": 1.7888,
+      "step": 3870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343480514653567e-07,
+      "loss": 1.9291,
+      "step": 3871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341062709725745e-07,
+      "loss": 1.7983,
+      "step": 3872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533864446787345e-07,
+      "loss": 1.8016,
+      "step": 3873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336225789294507e-07,
+      "loss": 1.7424,
+      "step": 3874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333806674186774e-07,
+      "loss": 1.8954,
+      "step": 3875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331387122748145e-07,
+      "loss": 1.8153,
+      "step": 3876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328967135176552e-07,
+      "loss": 1.8255,
+      "step": 3877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326546711669966e-07,
+      "loss": 1.6586,
+      "step": 3878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5324125852426376e-07,
+      "loss": 1.6749,
+      "step": 3879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321704557643837e-07,
+      "loss": 1.7392,
+      "step": 3880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5319282827520415e-07,
+      "loss": 1.8182,
+      "step": 3881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316860662254215e-07,
+      "loss": 1.7881,
+      "step": 3882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314438062043383e-07,
+      "loss": 1.6616,
+      "step": 3883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5312015027086104e-07,
+      "loss": 1.8466,
+      "step": 3884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5309591557580595e-07,
+      "loss": 1.8603,
+      "step": 3885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5307167653725103e-07,
+      "loss": 1.7959,
+      "step": 3886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304743315717913e-07,
+      "loss": 1.6212,
+      "step": 3887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5302318543757348e-07,
+      "loss": 1.6403,
+      "step": 3888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299893338041772e-07,
+      "loss": 1.6365,
+      "step": 3889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5297467698769576e-07,
+      "loss": 1.7892,
+      "step": 3890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295041626139183e-07,
+      "loss": 1.91,
+      "step": 3891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292615120349063e-07,
+      "loss": 1.7816,
+      "step": 3892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290188181597717e-07,
+      "loss": 1.821,
+      "step": 3893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5287760810083674e-07,
+      "loss": 1.7559,
+      "step": 3894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5285333006005512e-07,
+      "loss": 1.7113,
+      "step": 3895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528290476956183e-07,
+      "loss": 1.817,
+      "step": 3896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280476100951271e-07,
+      "loss": 1.7563,
+      "step": 3897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527804700037252e-07,
+      "loss": 1.7241,
+      "step": 3898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275617468024278e-07,
+      "loss": 1.8167,
+      "step": 3899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273187504105298e-07,
+      "loss": 1.8015,
+      "step": 3900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527075710881436e-07,
+      "loss": 1.6868,
+      "step": 3901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268326282350286e-07,
+      "loss": 1.7035,
+      "step": 3902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5265895024911927e-07,
+      "loss": 1.7033,
+      "step": 3903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526346333669817e-07,
+      "loss": 1.7079,
+      "step": 3904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5261031217907937e-07,
+      "loss": 1.8477,
+      "step": 3905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258598668740193e-07,
+      "loss": 1.8917,
+      "step": 3906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5256165689393929e-07,
+      "loss": 1.559,
+      "step": 3907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5253732280068173e-07,
+      "loss": 1.5018,
+      "step": 3908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5251298440961987e-07,
+      "loss": 1.8194,
+      "step": 3909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5248864172274482e-07,
+      "loss": 1.8959,
+      "step": 3910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5246429474204783e-07,
+      "loss": 1.74,
+      "step": 3911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.524399434695206e-07,
+      "loss": 1.7132,
+      "step": 3912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5241558790715523e-07,
+      "loss": 1.7468,
+      "step": 3913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5239122805694408e-07,
+      "loss": 1.7421,
+      "step": 3914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523668639208799e-07,
+      "loss": 1.7944,
+      "step": 3915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5234249550095585e-07,
+      "loss": 1.7807,
+      "step": 3916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523181227991653e-07,
+      "loss": 1.9735,
+      "step": 3917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229374581750212e-07,
+      "loss": 1.8215,
+      "step": 3918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226936455796048e-07,
+      "loss": 1.6059,
+      "step": 3919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5224497902253476e-07,
+      "loss": 1.7921,
+      "step": 3920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222058921321997e-07,
+      "loss": 1.8197,
+      "step": 3921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521961951320112e-07,
+      "loss": 1.6988,
+      "step": 3922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217179678090405e-07,
+      "loss": 1.7598,
+      "step": 3923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214739416189443e-07,
+      "loss": 1.8199,
+      "step": 3924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212298727697855e-07,
+      "loss": 1.8791,
+      "step": 3925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209857612815306e-07,
+      "loss": 1.584,
+      "step": 3926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207416071741487e-07,
+      "loss": 1.7439,
+      "step": 3927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520497410467613e-07,
+      "loss": 1.8913,
+      "step": 3928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202531711819e-07,
+      "loss": 1.8258,
+      "step": 3929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520008889336989e-07,
+      "loss": 1.7059,
+      "step": 3930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197645649528642e-07,
+      "loss": 1.5314,
+      "step": 3931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195201980495126e-07,
+      "loss": 1.853,
+      "step": 3932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519275788646924e-07,
+      "loss": 1.6694,
+      "step": 3933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190313367650924e-07,
+      "loss": 1.6752,
+      "step": 3934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187868424240153e-07,
+      "loss": 1.9409,
+      "step": 3935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185423056436934e-07,
+      "loss": 1.716,
+      "step": 3936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518297726444131e-07,
+      "loss": 1.6694,
+      "step": 3937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518053104845336e-07,
+      "loss": 1.8219,
+      "step": 3938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517808440867319e-07,
+      "loss": 2.0209,
+      "step": 3939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175637345300955e-07,
+      "loss": 1.8815,
+      "step": 3940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173189858536835e-07,
+      "loss": 1.771,
+      "step": 3941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170741948581043e-07,
+      "loss": 1.9073,
+      "step": 3942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5168293615633828e-07,
+      "loss": 1.9655,
+      "step": 3943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5165844859895482e-07,
+      "loss": 1.8059,
+      "step": 3944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163395681566318e-07,
+      "loss": 1.8471,
+      "step": 3945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5160946080846692e-07,
+      "loss": 1.7427,
+      "step": 3946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158496057936998e-07,
+      "loss": 1.7503,
+      "step": 3947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156045613037653e-07,
+      "loss": 1.9148,
+      "step": 3948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153594746349122e-07,
+      "loss": 1.5422,
+      "step": 3949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151143458071896e-07,
+      "loss": 1.7307,
+      "step": 3950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148691748406496e-07,
+      "loss": 1.6414,
+      "step": 3951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146239617553486e-07,
+      "loss": 1.8933,
+      "step": 3952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143787065713466e-07,
+      "loss": 1.7481,
+      "step": 3953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141334093087065e-07,
+      "loss": 1.8977,
+      "step": 3954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138880699874948e-07,
+      "loss": 1.7237,
+      "step": 3955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136426886277812e-07,
+      "loss": 1.8737,
+      "step": 3956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513397265249639e-07,
+      "loss": 1.8184,
+      "step": 3957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131517998731456e-07,
+      "loss": 1.6848,
+      "step": 3958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129062925183807e-07,
+      "loss": 1.716,
+      "step": 3959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512660743205428e-07,
+      "loss": 1.6296,
+      "step": 3960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124151519543746e-07,
+      "loss": 1.5491,
+      "step": 3961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121695187853117e-07,
+      "loss": 1.8111,
+      "step": 3962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119238437183323e-07,
+      "loss": 2.0467,
+      "step": 3963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116781267735345e-07,
+      "loss": 1.7974,
+      "step": 3964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114323679710187e-07,
+      "loss": 1.655,
+      "step": 3965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111865673308892e-07,
+      "loss": 1.8012,
+      "step": 3966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109407248732538e-07,
+      "loss": 1.6139,
+      "step": 3967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510694840618224e-07,
+      "loss": 1.7652,
+      "step": 3968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104489145859133e-07,
+      "loss": 1.8371,
+      "step": 3969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102029467964406e-07,
+      "loss": 1.7892,
+      "step": 3970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509956937269927e-07,
+      "loss": 1.7926,
+      "step": 3971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5097108860264963e-07,
+      "loss": 1.7558,
+      "step": 3972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509464793086278e-07,
+      "loss": 1.5639,
+      "step": 3973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092186584694032e-07,
+      "loss": 1.814,
+      "step": 3974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089724821960068e-07,
+      "loss": 1.8233,
+      "step": 3975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508726264286227e-07,
+      "loss": 1.5444,
+      "step": 3976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084800047602058e-07,
+      "loss": 1.5835,
+      "step": 3977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082337036380888e-07,
+      "loss": 1.7165,
+      "step": 3978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507987360940024e-07,
+      "loss": 1.8857,
+      "step": 3979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077409766861637e-07,
+      "loss": 1.7571,
+      "step": 3980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074945508966633e-07,
+      "loss": 1.7991,
+      "step": 3981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072480835916816e-07,
+      "loss": 1.8314,
+      "step": 3982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070015747913804e-07,
+      "loss": 1.8363,
+      "step": 3983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506755024515926e-07,
+      "loss": 1.6828,
+      "step": 3984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5065084327854875e-07,
+      "loss": 1.8028,
+      "step": 3985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062617996202363e-07,
+      "loss": 1.5785,
+      "step": 3986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506015125040349e-07,
+      "loss": 1.8251,
+      "step": 3987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057684090660042e-07,
+      "loss": 1.6407,
+      "step": 3988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057684090660042e-07,
+      "loss": 1.8258,
+      "step": 3989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055216517173847e-07,
+      "loss": 1.9083,
+      "step": 3990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052748530146765e-07,
+      "loss": 1.799,
+      "step": 3991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505028012978069e-07,
+      "loss": 1.8057,
+      "step": 3992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047811316277548e-07,
+      "loss": 1.8183,
+      "step": 3993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045342089839302e-07,
+      "loss": 1.7416,
+      "step": 3994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504287245066794e-07,
+      "loss": 1.6174,
+      "step": 3995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040402398965496e-07,
+      "loss": 1.7066,
+      "step": 3996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037931934934028e-07,
+      "loss": 1.8693,
+      "step": 3997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503546105877564e-07,
+      "loss": 1.7691,
+      "step": 3998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032989770692445e-07,
+      "loss": 1.7812,
+      "step": 3999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030518070886627e-07,
+      "loss": 1.6982,
+      "step": 4000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028045959560367e-07,
+      "loss": 1.7264,
+      "step": 4001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5025573436915907e-07,
+      "loss": 1.788,
+      "step": 4002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50231005031555e-07,
+      "loss": 1.664,
+      "step": 4003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502062715848145e-07,
+      "loss": 1.8601,
+      "step": 4004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501815340309609e-07,
+      "loss": 1.9011,
+      "step": 4005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501567923720178e-07,
+      "loss": 1.6576,
+      "step": 4006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013204661000922e-07,
+      "loss": 1.7005,
+      "step": 4007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010729674695944e-07,
+      "loss": 1.6884,
+      "step": 4008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500825427848932e-07,
+      "loss": 1.6628,
+      "step": 4009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005778472583544e-07,
+      "loss": 1.6539,
+      "step": 4010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003302257181149e-07,
+      "loss": 1.8758,
+      "step": 4011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000825632484699e-07,
+      "loss": 1.7731,
+      "step": 4012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49983485986968e-07,
+      "loss": 1.7452,
+      "step": 4013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995871156020076e-07,
+      "loss": 1.8038,
+      "step": 4014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993393304657203e-07,
+      "loss": 1.7605,
+      "step": 4015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990915044810874e-07,
+      "loss": 1.755,
+      "step": 4016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498843637668383e-07,
+      "loss": 1.739,
+      "step": 4017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498595730047883e-07,
+      "loss": 1.8106,
+      "step": 4018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498347781639868e-07,
+      "loss": 1.8385,
+      "step": 4019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498099792464621e-07,
+      "loss": 1.8031,
+      "step": 4020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4978517625424289e-07,
+      "loss": 1.6356,
+      "step": 4021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4976036918935816e-07,
+      "loss": 1.6751,
+      "step": 4022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4973555805383725e-07,
+      "loss": 1.7234,
+      "step": 4023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4971074284970986e-07,
+      "loss": 1.7107,
+      "step": 4024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4968592357900593e-07,
+      "loss": 1.7934,
+      "step": 4025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4966110024375586e-07,
+      "loss": 1.873,
+      "step": 4026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.496362728459903e-07,
+      "loss": 1.9132,
+      "step": 4027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.496114413877402e-07,
+      "loss": 1.7438,
+      "step": 4028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4958660587103693e-07,
+      "loss": 1.7429,
+      "step": 4029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.495617662979122e-07,
+      "loss": 1.6923,
+      "step": 4030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4953692267039792e-07,
+      "loss": 1.7141,
+      "step": 4031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951207499052647e-07,
+      "loss": 1.5869,
+      "step": 4032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494872232603305e-07,
+      "loss": 1.5073,
+      "step": 4033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49462367481843e-07,
+      "loss": 1.6719,
+      "step": 4034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943750765709727e-07,
+      "loss": 1.9633,
+      "step": 4035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49412643788127e-07,
+      "loss": 1.855,
+      "step": 4036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938777587696616e-07,
+      "loss": 1.7397,
+      "step": 4037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936290392564904e-07,
+      "loss": 1.6658,
+      "step": 4038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933802793621033e-07,
+      "loss": 1.9029,
+      "step": 4039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931314791068497e-07,
+      "loss": 1.596,
+      "step": 4040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492882638511083e-07,
+      "loss": 1.6974,
+      "step": 4041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926337575951588e-07,
+      "loss": 1.7856,
+      "step": 4042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4923848363794375e-07,
+      "loss": 1.6933,
+      "step": 4043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921358748842817e-07,
+      "loss": 1.6521,
+      "step": 4044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918868731300576e-07,
+      "loss": 1.8425,
+      "step": 4045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491637831137135e-07,
+      "loss": 1.7342,
+      "step": 4046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913887489258862e-07,
+      "loss": 1.7771,
+      "step": 4047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911396265166884e-07,
+      "loss": 1.6695,
+      "step": 4048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49089046392992e-07,
+      "loss": 1.7841,
+      "step": 4049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490641261185964e-07,
+      "loss": 1.7281,
+      "step": 4050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490392018305206e-07,
+      "loss": 1.782,
+      "step": 4051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901427353080362e-07,
+      "loss": 1.7219,
+      "step": 4052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898934122148464e-07,
+      "loss": 1.7078,
+      "step": 4053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896440490460325e-07,
+      "loss": 1.7205,
+      "step": 4054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893946458219936e-07,
+      "loss": 1.638,
+      "step": 4055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891452025631327e-07,
+      "loss": 1.7417,
+      "step": 4056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888957192898543e-07,
+      "loss": 1.8172,
+      "step": 4057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488646196022568e-07,
+      "loss": 1.632,
+      "step": 4058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488396632781686e-07,
+      "loss": 1.6461,
+      "step": 4059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488147029587624e-07,
+      "loss": 1.6198,
+      "step": 4060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878973864608005e-07,
+      "loss": 1.894,
+      "step": 4061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487647703421637e-07,
+      "loss": 1.806,
+      "step": 4062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873979804905594e-07,
+      "loss": 1.7057,
+      "step": 4063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487148217687996e-07,
+      "loss": 1.7327,
+      "step": 4064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868984150343792e-07,
+      "loss": 1.8373,
+      "step": 4065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4866485725501428e-07,
+      "loss": 1.9149,
+      "step": 4066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486398690255726e-07,
+      "loss": 1.7508,
+      "step": 4067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861487681715702e-07,
+      "loss": 1.7724,
+      "step": 4068
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48589880631812e-07,
+      "loss": 1.6291,
+      "step": 4069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856488047158239e-07,
+      "loss": 1.6586,
+      "step": 4070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853987633851328e-07,
+      "loss": 1.9159,
+      "step": 4071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851486823465015e-07,
+      "loss": 1.73,
+      "step": 4072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484898561620388e-07,
+      "loss": 1.8418,
+      "step": 4073
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484648401227253e-07,
+      "loss": 1.7289,
+      "step": 4074
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4843982011875612e-07,
+      "loss": 1.5987,
+      "step": 4075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4841479615217796e-07,
+      "loss": 1.7986,
+      "step": 4076
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4838976822503794e-07,
+      "loss": 1.7511,
+      "step": 4077
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836473633938348e-07,
+      "loss": 1.8555,
+      "step": 4078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833970049726227e-07,
+      "loss": 1.7389,
+      "step": 4079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831466070072235e-07,
+      "loss": 1.7264,
+      "step": 4080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828961695181216e-07,
+      "loss": 2.0103,
+      "step": 4081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826456925258036e-07,
+      "loss": 1.8265,
+      "step": 4082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823951760507598e-07,
+      "loss": 1.7751,
+      "step": 4083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821446201134835e-07,
+      "loss": 1.7842,
+      "step": 4084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818940247344714e-07,
+      "loss": 1.9206,
+      "step": 4085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481643389934223e-07,
+      "loss": 1.8324,
+      "step": 4086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813927157332426e-07,
+      "loss": 1.8735,
+      "step": 4087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811420021520355e-07,
+      "loss": 1.7756,
+      "step": 4088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808912492111116e-07,
+      "loss": 1.6586,
+      "step": 4089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806404569309837e-07,
+      "loss": 1.8541,
+      "step": 4090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803896253321678e-07,
+      "loss": 1.8316,
+      "step": 4091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480138754435183e-07,
+      "loss": 1.6737,
+      "step": 4092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479887844260552e-07,
+      "loss": 1.7287,
+      "step": 4093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796368948288003e-07,
+      "loss": 1.8589,
+      "step": 4094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479385906160457e-07,
+      "loss": 1.7154,
+      "step": 4095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791348782760535e-07,
+      "loss": 1.7884,
+      "step": 4096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4788838111961258e-07,
+      "loss": 1.7222,
+      "step": 4097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786327049412117e-07,
+      "loss": 1.6464,
+      "step": 4098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783815595318535e-07,
+      "loss": 1.7184,
+      "step": 4099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781303749885963e-07,
+      "loss": 1.7536,
+      "step": 4100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778791513319875e-07,
+      "loss": 1.7684,
+      "step": 4101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776278885825785e-07,
+      "loss": 1.8065,
+      "step": 4102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773765867609243e-07,
+      "loss": 1.7046,
+      "step": 4103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771252458875822e-07,
+      "loss": 1.8226,
+      "step": 4104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476873865983113e-07,
+      "loss": 1.7438,
+      "step": 4105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476622447068081e-07,
+      "loss": 1.6475,
+      "step": 4106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763709891630536e-07,
+      "loss": 1.7021,
+      "step": 4107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476119492288601e-07,
+      "loss": 1.8944,
+      "step": 4108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475867956465297e-07,
+      "loss": 1.7159,
+      "step": 4109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475616381713718e-07,
+      "loss": 1.6639,
+      "step": 4110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753647680544446e-07,
+      "loss": 1.8105,
+      "step": 4111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751131155080598e-07,
+      "loss": 1.7082,
+      "step": 4112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47486142409515e-07,
+      "loss": 1.6998,
+      "step": 4113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4746096938363045e-07,
+      "loss": 1.6957,
+      "step": 4114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743579247521165e-07,
+      "loss": 1.9605,
+      "step": 4115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741061168631815e-07,
+      "loss": 1.8905,
+      "step": 4116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473854270190099e-07,
+      "loss": 1.8189,
+      "step": 4117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736023847534706e-07,
+      "loss": 1.7678,
+      "step": 4118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733504605739024e-07,
+      "loss": 1.7926,
+      "step": 4119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730984976720027e-07,
+      "loss": 1.6774,
+      "step": 4120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728464960683834e-07,
+      "loss": 1.7281,
+      "step": 4121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472594455783659e-07,
+      "loss": 1.6985,
+      "step": 4122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723423768384484e-07,
+      "loss": 1.7919,
+      "step": 4123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4720902592533724e-07,
+      "loss": 1.8191,
+      "step": 4124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718381030490553e-07,
+      "loss": 1.7647,
+      "step": 4125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471585908246125e-07,
+      "loss": 1.684,
+      "step": 4126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471333674865212e-07,
+      "loss": 1.6541,
+      "step": 4127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710814029269502e-07,
+      "loss": 1.7633,
+      "step": 4128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708290924519765e-07,
+      "loss": 1.6248,
+      "step": 4129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705767434609316e-07,
+      "loss": 1.7618,
+      "step": 4130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703243559744588e-07,
+      "loss": 1.6801,
+      "step": 4131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700719300132042e-07,
+      "loss": 1.7176,
+      "step": 4132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698194655978174e-07,
+      "loss": 1.5788,
+      "step": 4133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4695669627489517e-07,
+      "loss": 1.8225,
+      "step": 4134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4693144214872625e-07,
+      "loss": 1.7386,
+      "step": 4135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4690618418334093e-07,
+      "loss": 1.9231,
+      "step": 4136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4688092238080543e-07,
+      "loss": 1.6501,
+      "step": 4137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4685565674318626e-07,
+      "loss": 1.879,
+      "step": 4138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.468303872725503e-07,
+      "loss": 1.8007,
+      "step": 4139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4680511397096468e-07,
+      "loss": 1.7367,
+      "step": 4140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4677983684049688e-07,
+      "loss": 1.817,
+      "step": 4141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4675455588321475e-07,
+      "loss": 1.7481,
+      "step": 4142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.467292711011863e-07,
+      "loss": 1.8409,
+      "step": 4143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4670398249648003e-07,
+      "loss": 1.6015,
+      "step": 4144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466786900711646e-07,
+      "loss": 1.7198,
+      "step": 4145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665339382730913e-07,
+      "loss": 1.8003,
+      "step": 4146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466280937669829e-07,
+      "loss": 1.84,
+      "step": 4147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660278989225562e-07,
+      "loss": 1.7927,
+      "step": 4148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657748220519723e-07,
+      "loss": 1.8156,
+      "step": 4149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4655217070787804e-07,
+      "loss": 1.5738,
+      "step": 4150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652685540236866e-07,
+      "loss": 1.6702,
+      "step": 4151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650153629074002e-07,
+      "loss": 1.6709,
+      "step": 4152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647621337506325e-07,
+      "loss": 1.6308,
+      "step": 4153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645088665741004e-07,
+      "loss": 1.9644,
+      "step": 4154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642555613985212e-07,
+      "loss": 1.7351,
+      "step": 4155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640022182446168e-07,
+      "loss": 1.8498,
+      "step": 4156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637488371331118e-07,
+      "loss": 1.8107,
+      "step": 4157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634954180847343e-07,
+      "loss": 1.8082,
+      "step": 4158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632419611202146e-07,
+      "loss": 1.7623,
+      "step": 4159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629884662602877e-07,
+      "loss": 1.8304,
+      "step": 4160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462734933525689e-07,
+      "loss": 1.6933,
+      "step": 4161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624813629371605e-07,
+      "loss": 1.7161,
+      "step": 4162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622277545154448e-07,
+      "loss": 1.7989,
+      "step": 4163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461974108281288e-07,
+      "loss": 1.8904,
+      "step": 4164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617204242554397e-07,
+      "loss": 1.6361,
+      "step": 4165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614667024586525e-07,
+      "loss": 1.8117,
+      "step": 4166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461212942911682e-07,
+      "loss": 1.6089,
+      "step": 4167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4609591456352874e-07,
+      "loss": 1.8655,
+      "step": 4168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607053106502302e-07,
+      "loss": 1.8043,
+      "step": 4169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604514379772747e-07,
+      "loss": 1.7048,
+      "step": 4170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601975276371903e-07,
+      "loss": 1.5147,
+      "step": 4171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459943579650747e-07,
+      "loss": 1.875,
+      "step": 4172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459689594038719e-07,
+      "loss": 1.7382,
+      "step": 4173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594355708218838e-07,
+      "loss": 1.8002,
+      "step": 4174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459181510021022e-07,
+      "loss": 1.8546,
+      "step": 4175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589274116569162e-07,
+      "loss": 1.6833,
+      "step": 4176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586732757503536e-07,
+      "loss": 1.7787,
+      "step": 4177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584191023221232e-07,
+      "loss": 1.6611,
+      "step": 4178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581648913930183e-07,
+      "loss": 1.7925,
+      "step": 4179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457910642983834e-07,
+      "loss": 1.7363,
+      "step": 4180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457656357115369e-07,
+      "loss": 1.6955,
+      "step": 4181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574020338084256e-07,
+      "loss": 1.7497,
+      "step": 4182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457147673083808e-07,
+      "loss": 1.8284,
+      "step": 4183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568932749623245e-07,
+      "loss": 1.785,
+      "step": 4184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456638839464786e-07,
+      "loss": 1.6869,
+      "step": 4185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456384366612007e-07,
+      "loss": 1.7253,
+      "step": 4186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561298564248036e-07,
+      "loss": 1.7973,
+      "step": 4187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558753089239972e-07,
+      "loss": 1.8137,
+      "step": 4188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45562072413041e-07,
+      "loss": 1.7001,
+      "step": 4189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553661020648686e-07,
+      "loss": 1.7987,
+      "step": 4190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455111442748202e-07,
+      "loss": 1.6903,
+      "step": 4191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454856746201243e-07,
+      "loss": 1.7163,
+      "step": 4192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546020124448271e-07,
+      "loss": 1.6435,
+      "step": 4193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454347241499792e-07,
+      "loss": 1.7065,
+      "step": 4194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540924333869802e-07,
+      "loss": 1.7825,
+      "step": 4195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538375881272353e-07,
+      "loss": 1.7953,
+      "step": 4196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535827057414053e-07,
+      "loss": 1.5199,
+      "step": 4197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453327786250341e-07,
+      "loss": 1.7455,
+      "step": 4198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530728296748956e-07,
+      "loss": 1.6944,
+      "step": 4199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528178360359258e-07,
+      "loss": 1.7633,
+      "step": 4200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525628053542916e-07,
+      "loss": 1.7497,
+      "step": 4201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523077376508554e-07,
+      "loss": 1.7095,
+      "step": 4202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520526329464836e-07,
+      "loss": 1.8082,
+      "step": 4203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517974912620439e-07,
+      "loss": 1.8108,
+      "step": 4204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515423126184088e-07,
+      "loss": 1.6762,
+      "step": 4205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512870970364534e-07,
+      "loss": 1.658,
+      "step": 4206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451031844537055e-07,
+      "loss": 1.7702,
+      "step": 4207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507765551410947e-07,
+      "loss": 1.6756,
+      "step": 4208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505212288694563e-07,
+      "loss": 1.7849,
+      "step": 4209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450265865743027e-07,
+      "loss": 1.7991,
+      "step": 4210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500104657826966e-07,
+      "loss": 1.6631,
+      "step": 4211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4497550290093578e-07,
+      "loss": 1.7171,
+      "step": 4212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494995554439065e-07,
+      "loss": 1.7254,
+      "step": 4213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449244045107242e-07,
+      "loss": 1.6389,
+      "step": 4214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489884980202663e-07,
+      "loss": 1.8141,
+      "step": 4215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448732914203884e-07,
+      "loss": 1.8169,
+      "step": 4216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484772936790034e-07,
+      "loss": 1.6006,
+      "step": 4217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482216364665352e-07,
+      "loss": 1.8768,
+      "step": 4218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479659425873935e-07,
+      "loss": 1.7803,
+      "step": 4219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477102120624955e-07,
+      "loss": 1.6901,
+      "step": 4220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447454444912761e-07,
+      "loss": 1.7454,
+      "step": 4221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447198641159113e-07,
+      "loss": 1.714,
+      "step": 4222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469428008224777e-07,
+      "loss": 1.6752,
+      "step": 4223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466869239237833e-07,
+      "loss": 1.9758,
+      "step": 4224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464310104839624e-07,
+      "loss": 1.6109,
+      "step": 4225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4461750605239498e-07,
+      "loss": 1.6984,
+      "step": 4226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459190740646836e-07,
+      "loss": 1.515,
+      "step": 4227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456630511271046e-07,
+      "loss": 1.71,
+      "step": 4228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454069917321566e-07,
+      "loss": 1.6913,
+      "step": 4229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451508959007862e-07,
+      "loss": 1.759,
+      "step": 4230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448947636539442e-07,
+      "loss": 1.8206,
+      "step": 4231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446385950125825e-07,
+      "loss": 1.8052,
+      "step": 4232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443823899976572e-07,
+      "loss": 1.7832,
+      "step": 4233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441261486301274e-07,
+      "loss": 1.6241,
+      "step": 4234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438698709309545e-07,
+      "loss": 1.7415,
+      "step": 4235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4436135569211035e-07,
+      "loss": 1.7654,
+      "step": 4236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433572066215418e-07,
+      "loss": 1.7817,
+      "step": 4237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431008200532403e-07,
+      "loss": 1.6111,
+      "step": 4238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428443972371729e-07,
+      "loss": 1.7295,
+      "step": 4239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425879381943157e-07,
+      "loss": 1.8804,
+      "step": 4240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423314429456485e-07,
+      "loss": 1.8854,
+      "step": 4241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442074911512154e-07,
+      "loss": 1.7724,
+      "step": 4242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418183439148173e-07,
+      "loss": 1.8721,
+      "step": 4243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441561740174627e-07,
+      "loss": 1.7509,
+      "step": 4244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441305100312575e-07,
+      "loss": 1.9316,
+      "step": 4245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4410484243496552e-07,
+      "loss": 1.6955,
+      "step": 4246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4407917123068647e-07,
+      "loss": 1.7791,
+      "step": 4247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4405349642052044e-07,
+      "loss": 1.7196,
+      "step": 4248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.440278180065677e-07,
+      "loss": 1.7806,
+      "step": 4249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.440021359909289e-07,
+      "loss": 1.727,
+      "step": 4250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4397645037570491e-07,
+      "loss": 2.115,
+      "step": 4251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43950761162997e-07,
+      "loss": 1.7935,
+      "step": 4252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4392506835490664e-07,
+      "loss": 1.8941,
+      "step": 4253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4389937195353562e-07,
+      "loss": 1.5835,
+      "step": 4254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43873671960986e-07,
+      "loss": 1.754,
+      "step": 4255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4384796837936016e-07,
+      "loss": 1.7704,
+      "step": 4256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4382226121076085e-07,
+      "loss": 1.8158,
+      "step": 4257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379655045729102e-07,
+      "loss": 1.7209,
+      "step": 4258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377083612105388e-07,
+      "loss": 1.5612,
+      "step": 4259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43745118204153e-07,
+      "loss": 1.7618,
+      "step": 4260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4371939670869227e-07,
+      "loss": 1.6823,
+      "step": 4261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436936716367758e-07,
+      "loss": 1.8044,
+      "step": 4262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43667942990508e-07,
+      "loss": 1.6248,
+      "step": 4263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364221077199364e-07,
+      "loss": 1.7739,
+      "step": 4264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436164749833377e-07,
+      "loss": 1.8117,
+      "step": 4265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359073562664556e-07,
+      "loss": 1.7504,
+      "step": 4266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4356499270402276e-07,
+      "loss": 1.8675,
+      "step": 4267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435392462175752e-07,
+      "loss": 1.8789,
+      "step": 4268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351349616940907e-07,
+      "loss": 1.7356,
+      "step": 4269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348774256163087e-07,
+      "loss": 1.7899,
+      "step": 4270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346198539634735e-07,
+      "loss": 1.761,
+      "step": 4271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343622467566556e-07,
+      "loss": 1.9906,
+      "step": 4272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341046040169284e-07,
+      "loss": 1.7744,
+      "step": 4273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433846925765369e-07,
+      "loss": 1.8349,
+      "step": 4274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433589212023056e-07,
+      "loss": 1.7652,
+      "step": 4275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333314628110718e-07,
+      "loss": 1.6713,
+      "step": 4276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330736781505016e-07,
+      "loss": 1.6784,
+      "step": 4277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432815858062433e-07,
+      "loss": 1.6617,
+      "step": 4278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432558002567958e-07,
+      "loss": 1.7378,
+      "step": 4279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323001116881693e-07,
+      "loss": 1.7276,
+      "step": 4280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432042185444164e-07,
+      "loss": 1.8267,
+      "step": 4281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317842238570416e-07,
+      "loss": 1.8515,
+      "step": 4282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315262269479052e-07,
+      "loss": 1.8747,
+      "step": 4283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312681947378593e-07,
+      "loss": 1.6019,
+      "step": 4284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310101272480124e-07,
+      "loss": 1.5765,
+      "step": 4285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307520244994763e-07,
+      "loss": 1.9146,
+      "step": 4286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304938865133642e-07,
+      "loss": 1.6625,
+      "step": 4287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302357133107937e-07,
+      "loss": 1.5907,
+      "step": 4288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429977504912884e-07,
+      "loss": 1.6192,
+      "step": 4289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297192613407577e-07,
+      "loss": 1.7729,
+      "step": 4290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294609826155414e-07,
+      "loss": 1.8182,
+      "step": 4291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292026687583627e-07,
+      "loss": 1.6118,
+      "step": 4292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428944319790353e-07,
+      "loss": 1.7253,
+      "step": 4293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428685935732646e-07,
+      "loss": 1.7815,
+      "step": 4294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42842751660638e-07,
+      "loss": 1.8664,
+      "step": 4295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428169062432694e-07,
+      "loss": 1.798,
+      "step": 4296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427910573232731e-07,
+      "loss": 1.7895,
+      "step": 4297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276520490276363e-07,
+      "loss": 1.8832,
+      "step": 4298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273934898385594e-07,
+      "loss": 1.8977,
+      "step": 4299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271348956866506e-07,
+      "loss": 1.8222,
+      "step": 4300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426876266593065e-07,
+      "loss": 1.7171,
+      "step": 4301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266176025789587e-07,
+      "loss": 1.7239,
+      "step": 4302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263589036654925e-07,
+      "loss": 1.7099,
+      "step": 4303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426100169873829e-07,
+      "loss": 1.6723,
+      "step": 4304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425841401225134e-07,
+      "loss": 1.8404,
+      "step": 4305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255825977405755e-07,
+      "loss": 1.7646,
+      "step": 4306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4253237594413255e-07,
+      "loss": 1.6818,
+      "step": 4307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250648863485574e-07,
+      "loss": 1.7791,
+      "step": 4308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248059784834493e-07,
+      "loss": 1.9277,
+      "step": 4309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4245470358671804e-07,
+      "loss": 1.7665,
+      "step": 4310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242880585209333e-07,
+      "loss": 1.6679,
+      "step": 4311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424029046465894e-07,
+      "loss": 1.7346,
+      "step": 4312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237699997232508e-07,
+      "loss": 1.7771,
+      "step": 4313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235109183141952e-07,
+      "loss": 1.7069,
+      "step": 4314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232518022599206e-07,
+      "loss": 1.662,
+      "step": 4315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422992651581625e-07,
+      "loss": 1.7869,
+      "step": 4316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227334663005072e-07,
+      "loss": 1.8222,
+      "step": 4317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224742464377703e-07,
+      "loss": 1.794,
+      "step": 4318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222149920146194e-07,
+      "loss": 1.943,
+      "step": 4319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421955703052263e-07,
+      "loss": 1.6443,
+      "step": 4320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216963795719124e-07,
+      "loss": 1.8502,
+      "step": 4321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421437021594781e-07,
+      "loss": 1.8217,
+      "step": 4322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211776291420855e-07,
+      "loss": 1.8629,
+      "step": 4323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209182022350458e-07,
+      "loss": 1.7984,
+      "step": 4324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206587408948844e-07,
+      "loss": 1.811,
+      "step": 4325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420399245142826e-07,
+      "loss": 1.8333,
+      "step": 4326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201397150000992e-07,
+      "loss": 1.7319,
+      "step": 4327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198801504879337e-07,
+      "loss": 1.8092,
+      "step": 4328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196205516275643e-07,
+      "loss": 1.6016,
+      "step": 4329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419360918440227e-07,
+      "loss": 1.7011,
+      "step": 4330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419101250947161e-07,
+      "loss": 1.9035,
+      "step": 4331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418841549169608e-07,
+      "loss": 1.8417,
+      "step": 4332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185818131288136e-07,
+      "loss": 1.7811,
+      "step": 4333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183220428460248e-07,
+      "loss": 1.7933,
+      "step": 4334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4180622383424924e-07,
+      "loss": 1.7166,
+      "step": 4335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178023996394696e-07,
+      "loss": 1.7906,
+      "step": 4336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417542526758212e-07,
+      "loss": 1.752,
+      "step": 4337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172826197199795e-07,
+      "loss": 1.6968,
+      "step": 4338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170226785460325e-07,
+      "loss": 1.7432,
+      "step": 4339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416762703257636e-07,
+      "loss": 1.6941,
+      "step": 4340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4165026938760577e-07,
+      "loss": 1.6555,
+      "step": 4341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416242650422567e-07,
+      "loss": 1.8791,
+      "step": 4342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159825729184367e-07,
+      "loss": 1.8445,
+      "step": 4343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157224613849427e-07,
+      "loss": 1.8436,
+      "step": 4344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415462315843363e-07,
+      "loss": 1.6597,
+      "step": 4345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152021363149791e-07,
+      "loss": 1.8178,
+      "step": 4346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149419228210746e-07,
+      "loss": 1.7364,
+      "step": 4347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146816753829365e-07,
+      "loss": 1.6926,
+      "step": 4348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144213940218538e-07,
+      "loss": 1.7748,
+      "step": 4349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141610787591197e-07,
+      "loss": 1.7081,
+      "step": 4350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139007296160283e-07,
+      "loss": 1.486,
+      "step": 4351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413640346613878e-07,
+      "loss": 1.7907,
+      "step": 4352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413379929773969e-07,
+      "loss": 1.7786,
+      "step": 4353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131194791176046e-07,
+      "loss": 1.7557,
+      "step": 4354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4128589946660913e-07,
+      "loss": 1.6452,
+      "step": 4355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412598476440738e-07,
+      "loss": 1.7191,
+      "step": 4356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123379244628556e-07,
+      "loss": 1.757,
+      "step": 4357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4120773387537593e-07,
+      "loss": 1.8494,
+      "step": 4358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118167193347657e-07,
+      "loss": 1.8318,
+      "step": 4359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4115560662271954e-07,
+      "loss": 2.1129,
+      "step": 4360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4112953794523703e-07,
+      "loss": 1.9189,
+      "step": 4361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4110346590316162e-07,
+      "loss": 1.9446,
+      "step": 4362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4107739049862614e-07,
+      "loss": 1.8167,
+      "step": 4363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4105131173376365e-07,
+      "loss": 1.7255,
+      "step": 4364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4102522961070754e-07,
+      "loss": 1.784,
+      "step": 4365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4099914413159146e-07,
+      "loss": 1.6804,
+      "step": 4366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.409730552985493e-07,
+      "loss": 1.7469,
+      "step": 4367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4094696311371528e-07,
+      "loss": 1.8959,
+      "step": 4368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4092086757922383e-07,
+      "loss": 1.7522,
+      "step": 4369
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408947686972097e-07,
+      "loss": 1.5691,
+      "step": 4370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4086866646980796e-07,
+      "loss": 1.6973,
+      "step": 4371
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084256089915383e-07,
+      "loss": 1.806,
+      "step": 4372
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081645198738287e-07,
+      "loss": 1.8104,
+      "step": 4373
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079033973663096e-07,
+      "loss": 1.8265,
+      "step": 4374
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076422414903417e-07,
+      "loss": 1.8677,
+      "step": 4375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073810522672892e-07,
+      "loss": 1.6525,
+      "step": 4376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407119829718518e-07,
+      "loss": 1.9909,
+      "step": 4377
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068585738653978e-07,
+      "loss": 1.8295,
+      "step": 4378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065972847293003e-07,
+      "loss": 1.6965,
+      "step": 4379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063359623316004e-07,
+      "loss": 1.9465,
+      "step": 4380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060746066936759e-07,
+      "loss": 1.5262,
+      "step": 4381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058132178369064e-07,
+      "loss": 1.8868,
+      "step": 4382
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055517957826748e-07,
+      "loss": 1.7683,
+      "step": 4383
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052903405523665e-07,
+      "loss": 1.8532,
+      "step": 4384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050288521673707e-07,
+      "loss": 1.8141,
+      "step": 4385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047673306490774e-07,
+      "loss": 1.6637,
+      "step": 4386
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045057760188808e-07,
+      "loss": 1.6621,
+      "step": 4387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404244188298177e-07,
+      "loss": 1.9493,
+      "step": 4388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039825675083657e-07,
+      "loss": 1.7354,
+      "step": 4389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037209136708478e-07,
+      "loss": 1.7882,
+      "step": 4390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034592268070287e-07,
+      "loss": 1.8883,
+      "step": 4391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031975069383152e-07,
+      "loss": 1.8433,
+      "step": 4392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029357540861176e-07,
+      "loss": 1.564,
+      "step": 4393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026739682718482e-07,
+      "loss": 1.7479,
+      "step": 4394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4024121495169223e-07,
+      "loss": 1.7287,
+      "step": 4395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021502978427581e-07,
+      "loss": 1.9096,
+      "step": 4396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018884132707763e-07,
+      "loss": 1.8901,
+      "step": 4397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016264958224008e-07,
+      "loss": 1.6478,
+      "step": 4398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013645455190566e-07,
+      "loss": 1.7573,
+      "step": 4399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401102562382173e-07,
+      "loss": 1.6541,
+      "step": 4400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008405464331817e-07,
+      "loss": 1.7742,
+      "step": 4401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400578497693517e-07,
+      "loss": 1.9195,
+      "step": 4402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003164161846152e-07,
+      "loss": 1.8717,
+      "step": 4403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400054301927916e-07,
+      "loss": 1.6741,
+      "step": 4404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3997921549448616e-07,
+      "loss": 1.6203,
+      "step": 4405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399529975256897e-07,
+      "loss": 1.9683,
+      "step": 4406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992677628854698e-07,
+      "loss": 1.7475,
+      "step": 4407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3990055178520303e-07,
+      "loss": 1.8997,
+      "step": 4408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987432401780308e-07,
+      "loss": 1.8207,
+      "step": 4409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3984809298849273e-07,
+      "loss": 1.7722,
+      "step": 4410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982185869941785e-07,
+      "loss": 1.7772,
+      "step": 4411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979562115272445e-07,
+      "loss": 1.5348,
+      "step": 4412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976938035055892e-07,
+      "loss": 1.7754,
+      "step": 4413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397431362950679e-07,
+      "loss": 1.8239,
+      "step": 4414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971688898839827e-07,
+      "loss": 1.6686,
+      "step": 4415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3969063843269713e-07,
+      "loss": 1.7458,
+      "step": 4416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966438463011195e-07,
+      "loss": 1.7518,
+      "step": 4417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963812758279046e-07,
+      "loss": 1.7324,
+      "step": 4418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961186729288055e-07,
+      "loss": 1.8832,
+      "step": 4419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958560376253046e-07,
+      "loss": 1.5747,
+      "step": 4420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955933699388865e-07,
+      "loss": 1.879,
+      "step": 4421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395330669891039e-07,
+      "loss": 1.8163,
+      "step": 4422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950679375032518e-07,
+      "loss": 1.7944,
+      "step": 4423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394805172797018e-07,
+      "loss": 1.6852,
+      "step": 4424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394542375793833e-07,
+      "loss": 1.617,
+      "step": 4425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394279546515195e-07,
+      "loss": 1.7544,
+      "step": 4426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940166849826041e-07,
+      "loss": 1.5883,
+      "step": 4427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937537912175642e-07,
+      "loss": 1.7109,
+      "step": 4428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393490865241581e-07,
+      "loss": 1.6837,
+      "step": 4429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932279070761632e-07,
+      "loss": 1.7677,
+      "step": 4430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392964916742822e-07,
+      "loss": 1.666,
+      "step": 4431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927018942630713e-07,
+      "loss": 1.822,
+      "step": 4432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392438839658428e-07,
+      "loss": 1.9213,
+      "step": 4433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3921757529504102e-07,
+      "loss": 1.6717,
+      "step": 4434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919126341605405e-07,
+      "loss": 1.8209,
+      "step": 4435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391649483310343e-07,
+      "loss": 1.8239,
+      "step": 4436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3913863004213446e-07,
+      "loss": 1.6211,
+      "step": 4437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911230855150753e-07,
+      "loss": 1.7713,
+      "step": 4438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908598386130667e-07,
+      "loss": 1.8314,
+      "step": 4439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905965597368544e-07,
+      "loss": 1.7503,
+      "step": 4440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903332489079756e-07,
+      "loss": 1.8792,
+      "step": 4441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3900699061479699e-07,
+      "loss": 1.7948,
+      "step": 4442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898065314783804e-07,
+      "loss": 1.8576,
+      "step": 4443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895431249207525e-07,
+      "loss": 1.7303,
+      "step": 4444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892796864966342e-07,
+      "loss": 1.8376,
+      "step": 4445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890162162275759e-07,
+      "loss": 1.7422,
+      "step": 4446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38875271413513e-07,
+      "loss": 1.6989,
+      "step": 4447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884891802408535e-07,
+      "loss": 1.5419,
+      "step": 4448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388225614566304e-07,
+      "loss": 1.6705,
+      "step": 4449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387962017133043e-07,
+      "loss": 1.8413,
+      "step": 4450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387698387962633e-07,
+      "loss": 1.9177,
+      "step": 4451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874347270766408e-07,
+      "loss": 1.9193,
+      "step": 4452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3871710344966352e-07,
+      "loss": 1.7287,
+      "step": 4453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869073102441873e-07,
+      "loss": 1.8006,
+      "step": 4454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386643554340871e-07,
+      "loss": 1.6689,
+      "step": 4455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863797668082628e-07,
+      "loss": 1.8955,
+      "step": 4456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386115947667942e-07,
+      "loss": 1.7462,
+      "step": 4457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38585209694149e-07,
+      "loss": 1.7841,
+      "step": 4458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855882146504915e-07,
+      "loss": 1.9452,
+      "step": 4459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853243008165322e-07,
+      "loss": 1.7292,
+      "step": 4460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850603554612031e-07,
+      "loss": 1.736,
+      "step": 4461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384796378606095e-07,
+      "loss": 1.7722,
+      "step": 4462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384532370272803e-07,
+      "loss": 1.601,
+      "step": 4463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384268330482924e-07,
+      "loss": 1.6625,
+      "step": 4464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384004259258058e-07,
+      "loss": 1.7264,
+      "step": 4465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3837401566198067e-07,
+      "loss": 1.9028,
+      "step": 4466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834760225897762e-07,
+      "loss": 1.8561,
+      "step": 4467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832118571895722e-07,
+      "loss": 1.7415,
+      "step": 4468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382947660440806e-07,
+      "loss": 1.8826,
+      "step": 4469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826834323650897e-07,
+      "loss": 1.7471,
+      "step": 4470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824191729840385e-07,
+      "loss": 1.9403,
+      "step": 4471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38215488231927e-07,
+      "loss": 1.813,
+      "step": 4472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3818905603924045e-07,
+      "loss": 1.8185,
+      "step": 4473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.381626207225065e-07,
+      "loss": 1.854,
+      "step": 4474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3813618228388763e-07,
+      "loss": 1.8984,
+      "step": 4475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.381097407255467e-07,
+      "loss": 1.8228,
+      "step": 4476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3808329604964671e-07,
+      "loss": 1.9179,
+      "step": 4477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3805684825835096e-07,
+      "loss": 1.9072,
+      "step": 4478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3803039735382304e-07,
+      "loss": 1.7553,
+      "step": 4479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3800394333822672e-07,
+      "loss": 1.8397,
+      "step": 4480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3797748621372606e-07,
+      "loss": 1.7654,
+      "step": 4481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3795102598248545e-07,
+      "loss": 1.8232,
+      "step": 4482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3792456264666938e-07,
+      "loss": 1.6866,
+      "step": 4483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3789809620844272e-07,
+      "loss": 1.6598,
+      "step": 4484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787162666997055e-07,
+      "loss": 1.6479,
+      "step": 4485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378451540334182e-07,
+      "loss": 1.8838,
+      "step": 4486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781867830095128e-07,
+      "loss": 1.7169,
+      "step": 4487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377921994747356e-07,
+      "loss": 1.7521,
+      "step": 4488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776571755693724e-07,
+      "loss": 1.8486,
+      "step": 4489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773923254972256e-07,
+      "loss": 1.6032,
+      "step": 4490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771274445525822e-07,
+      "loss": 1.6856,
+      "step": 4491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37686253275711e-07,
+      "loss": 1.8869,
+      "step": 4492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765975901324805e-07,
+      "loss": 1.7461,
+      "step": 4493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763326167003667e-07,
+      "loss": 1.6729,
+      "step": 4494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3760676124824455e-07,
+      "loss": 1.9445,
+      "step": 4495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758025775003947e-07,
+      "loss": 1.8098,
+      "step": 4496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755375117758958e-07,
+      "loss": 1.7264,
+      "step": 4497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752724153306328e-07,
+      "loss": 1.4958,
+      "step": 4498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375007288186291e-07,
+      "loss": 1.8548,
+      "step": 4499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747421303645602e-07,
+      "loss": 1.9541,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3744769418871304e-07,
+      "loss": 1.6919,
+      "step": 4501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374211722775696e-07,
+      "loss": 1.6955,
+      "step": 4502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739464730519533e-07,
+      "loss": 1.5957,
+      "step": 4503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736811927376007e-07,
+      "loss": 1.6681,
+      "step": 4504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373415881854339e-07,
+      "loss": 1.7974,
+      "step": 4505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3731505404238727e-07,
+      "loss": 1.7571,
+      "step": 4506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728851684679075e-07,
+      "loss": 1.7081,
+      "step": 4507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726197660081523e-07,
+      "loss": 1.6557,
+      "step": 4508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372354333066318e-07,
+      "loss": 1.7693,
+      "step": 4509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720888696641187e-07,
+      "loss": 1.9784,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3718233758232702e-07,
+      "loss": 1.7185,
+      "step": 4511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715578515654913e-07,
+      "loss": 1.6761,
+      "step": 4512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712922969125035e-07,
+      "loss": 1.7213,
+      "step": 4513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3710267118860298e-07,
+      "loss": 1.6851,
+      "step": 4514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370761096507797e-07,
+      "loss": 1.7067,
+      "step": 4515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370495450799533e-07,
+      "loss": 1.8019,
+      "step": 4516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702297747829697e-07,
+      "loss": 2.0094,
+      "step": 4517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36996406847984e-07,
+      "loss": 1.7472,
+      "step": 4518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696983319118798e-07,
+      "loss": 1.7456,
+      "step": 4519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694325651008283e-07,
+      "loss": 1.8568,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691667680684263e-07,
+      "loss": 1.8506,
+      "step": 4521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689009408364172e-07,
+      "loss": 1.9249,
+      "step": 4522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686350834265465e-07,
+      "loss": 1.8554,
+      "step": 4523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683691958605632e-07,
+      "loss": 1.7197,
+      "step": 4524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368103278160218e-07,
+      "loss": 1.7552,
+      "step": 4525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678373303472645e-07,
+      "loss": 1.7645,
+      "step": 4526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367571352443458e-07,
+      "loss": 1.8166,
+      "step": 4527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673053444705566e-07,
+      "loss": 1.7012,
+      "step": 4528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670393064503217e-07,
+      "loss": 1.773,
+      "step": 4529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667732384045164e-07,
+      "loss": 1.7757,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366507140354906e-07,
+      "loss": 1.6549,
+      "step": 4531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662410123232585e-07,
+      "loss": 1.7374,
+      "step": 4532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659748543313448e-07,
+      "loss": 1.826,
+      "step": 4533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365708666400938e-07,
+      "loss": 1.7994,
+      "step": 4534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365442448553813e-07,
+      "loss": 1.8123,
+      "step": 4535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365176200811748e-07,
+      "loss": 1.7157,
+      "step": 4536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649099231965237e-07,
+      "loss": 1.6801,
+      "step": 4537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364643615729922e-07,
+      "loss": 1.7873,
+      "step": 4538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364377278433729e-07,
+      "loss": 1.6598,
+      "step": 4539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364110911329732e-07,
+      "loss": 1.8503,
+      "step": 4540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638445144397211e-07,
+      "loss": 1.701,
+      "step": 4541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635780877854885e-07,
+      "loss": 1.8908,
+      "step": 4542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36331163138883e-07,
+      "loss": 1.7493,
+      "step": 4543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3630451452715422e-07,
+      "loss": 1.7296,
+      "step": 4544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627786294554256e-07,
+      "loss": 1.7858,
+      "step": 4545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362512083962282e-07,
+      "loss": 1.723,
+      "step": 4546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622455088139165e-07,
+      "loss": 1.5628,
+      "step": 4547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3619789040321355e-07,
+      "loss": 1.7523,
+      "step": 4548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361712269638749e-07,
+      "loss": 1.648,
+      "step": 4549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361445605655569e-07,
+      "loss": 1.7316,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611789121044102e-07,
+      "loss": 1.6827,
+      "step": 4551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609121890070892e-07,
+      "loss": 1.8639,
+      "step": 4552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606454363854246e-07,
+      "loss": 1.8662,
+      "step": 4553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603786542612388e-07,
+      "loss": 1.426,
+      "step": 4554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601118426563556e-07,
+      "loss": 1.6834,
+      "step": 4555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598450015926016e-07,
+      "loss": 1.5375,
+      "step": 4556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359578131091805e-07,
+      "loss": 1.8439,
+      "step": 4557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359311231175798e-07,
+      "loss": 1.7002,
+      "step": 4558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590443018664138e-07,
+      "loss": 1.7053,
+      "step": 4559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587773431854888e-07,
+      "loss": 1.7841,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358510355154861e-07,
+      "loss": 1.7973,
+      "step": 4561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3582433377963719e-07,
+      "loss": 1.8012,
+      "step": 4562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357976291131864e-07,
+      "loss": 1.6738,
+      "step": 4563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357709215183184e-07,
+      "loss": 1.9789,
+      "step": 4564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574421099721795e-07,
+      "loss": 1.7861,
+      "step": 4565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571749755207008e-07,
+      "loss": 1.7385,
+      "step": 4566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569078118506004e-07,
+      "loss": 1.6304,
+      "step": 4567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566406189837347e-07,
+      "loss": 1.7574,
+      "step": 4568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356373396941961e-07,
+      "loss": 1.7076,
+      "step": 4569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561061457471385e-07,
+      "loss": 1.9292,
+      "step": 4570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558388654211307e-07,
+      "loss": 1.6698,
+      "step": 4571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555715559858014e-07,
+      "loss": 1.7512,
+      "step": 4572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3553042174630187e-07,
+      "loss": 1.8001,
+      "step": 4573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3550368498746517e-07,
+      "loss": 1.6736,
+      "step": 4574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547694532425724e-07,
+      "loss": 1.8081,
+      "step": 4575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545020275886553e-07,
+      "loss": 1.7213,
+      "step": 4576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542345729347767e-07,
+      "loss": 1.85,
+      "step": 4577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353967089302816e-07,
+      "loss": 1.6857,
+      "step": 4578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353699576714655e-07,
+      "loss": 1.7537,
+      "step": 4579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534320351921764e-07,
+      "loss": 1.6872,
+      "step": 4580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531644647572673e-07,
+      "loss": 1.6962,
+      "step": 4581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528968654318162e-07,
+      "loss": 1.8134,
+      "step": 4582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526292372377136e-07,
+      "loss": 1.566,
+      "step": 4583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3523615801968534e-07,
+      "loss": 1.6287,
+      "step": 4584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520938943311302e-07,
+      "loss": 1.7356,
+      "step": 4585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351826179662443e-07,
+      "loss": 1.627,
+      "step": 4586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3515584362126916e-07,
+      "loss": 1.9264,
+      "step": 4587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3512906640037786e-07,
+      "loss": 1.8459,
+      "step": 4588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3510228630576091e-07,
+      "loss": 1.721,
+      "step": 4589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.350755033396091e-07,
+      "loss": 2.0354,
+      "step": 4590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.350487175041134e-07,
+      "loss": 1.6311,
+      "step": 4591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3502192880146494e-07,
+      "loss": 1.7777,
+      "step": 4592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3499513723385523e-07,
+      "loss": 1.9695,
+      "step": 4593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3496834280347594e-07,
+      "loss": 1.9202,
+      "step": 4594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3494154551251897e-07,
+      "loss": 1.6999,
+      "step": 4595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3491474536317646e-07,
+      "loss": 1.7406,
+      "step": 4596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348879423576408e-07,
+      "loss": 1.8973,
+      "step": 4597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3486113649810463e-07,
+      "loss": 1.7158,
+      "step": 4598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3483432778676075e-07,
+      "loss": 1.904,
+      "step": 4599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480751622580227e-07,
+      "loss": 1.6371,
+      "step": 4600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478070181742252e-07,
+      "loss": 1.8152,
+      "step": 4601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475388456381498e-07,
+      "loss": 1.6886,
+      "step": 4602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347270644671735e-07,
+      "loss": 1.737,
+      "step": 4603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347002415296921e-07,
+      "loss": 1.8745,
+      "step": 4604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346734157535649e-07,
+      "loss": 1.626,
+      "step": 4605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464658714098655e-07,
+      "loss": 1.5014,
+      "step": 4606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461975569415166e-07,
+      "loss": 1.6847,
+      "step": 4607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459292141525517e-07,
+      "loss": 1.6384,
+      "step": 4608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456608430649226e-07,
+      "loss": 1.8427,
+      "step": 4609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453924437005837e-07,
+      "loss": 1.6389,
+      "step": 4610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345124016081491e-07,
+      "loss": 1.6669,
+      "step": 4611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448555602296034e-07,
+      "loss": 2.0774,
+      "step": 4612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445870761668813e-07,
+      "loss": 1.6718,
+      "step": 4613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3443185639152889e-07,
+      "loss": 1.8456,
+      "step": 4614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344050023496791e-07,
+      "loss": 1.6291,
+      "step": 4615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343781454933356e-07,
+      "loss": 1.7559,
+      "step": 4616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435128582469538e-07,
+      "loss": 1.6117,
+      "step": 4617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432442334595563e-07,
+      "loss": 1.7609,
+      "step": 4618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34297558059314e-07,
+      "loss": 1.5552,
+      "step": 4619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427068996696803e-07,
+      "loss": 1.8832,
+      "step": 4620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424381907111577e-07,
+      "loss": 1.7919,
+      "step": 4621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421694537395528e-07,
+      "loss": 1.9786,
+      "step": 4622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419006887768508e-07,
+      "loss": 1.7987,
+      "step": 4623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416318958450367e-07,
+      "loss": 1.9399,
+      "step": 4624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413630749661005e-07,
+      "loss": 1.8276,
+      "step": 4625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410942261620312e-07,
+      "loss": 1.8727,
+      "step": 4626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340825349454824e-07,
+      "loss": 1.8088,
+      "step": 4627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3405564448664727e-07,
+      "loss": 1.6012,
+      "step": 4628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402875124189756e-07,
+      "loss": 1.6072,
+      "step": 4629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340018552134332e-07,
+      "loss": 1.6953,
+      "step": 4630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3397495640345456e-07,
+      "loss": 1.7878,
+      "step": 4631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394805481416197e-07,
+      "loss": 1.7866,
+      "step": 4632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339211504477561e-07,
+      "loss": 1.8056,
+      "step": 4633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389424330643797e-07,
+      "loss": 1.6253,
+      "step": 4634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338673333924086e-07,
+      "loss": 1.9058,
+      "step": 4635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338404207078694e-07,
+      "loss": 1.7242,
+      "step": 4636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381350525502194e-07,
+      "loss": 1.5188,
+      "step": 4637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378658703606806e-07,
+      "loss": 1.701,
+      "step": 4638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375966605320974e-07,
+      "loss": 1.8474,
+      "step": 4639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373274230864932e-07,
+      "loss": 1.8354,
+      "step": 4640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370581580458925e-07,
+      "loss": 1.6533,
+      "step": 4641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367888654323224e-07,
+      "loss": 1.7526,
+      "step": 4642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365195452678123e-07,
+      "loss": 1.8484,
+      "step": 4643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362501975743938e-07,
+      "loss": 1.6323,
+      "step": 4644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335980822374102e-07,
+      "loss": 1.8916,
+      "step": 4645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3357114196889712e-07,
+      "loss": 1.8532,
+      "step": 4646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335441989541041e-07,
+      "loss": 1.759,
+      "step": 4647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335172531952352e-07,
+      "loss": 1.6969,
+      "step": 4648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349030469449468e-07,
+      "loss": 1.6713,
+      "step": 4649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346335345408705e-07,
+      "loss": 1.6705,
+      "step": 4650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343639947621709e-07,
+      "loss": 1.9775,
+      "step": 4651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3340944276308977e-07,
+      "loss": 1.5584,
+      "step": 4652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338248331691023e-07,
+      "loss": 1.7251,
+      "step": 4653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335552113988392e-07,
+      "loss": 1.8134,
+      "step": 4654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332855623421646e-07,
+      "loss": 1.7502,
+      "step": 4655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330158860211366e-07,
+      "loss": 1.8537,
+      "step": 4656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327461824578167e-07,
+      "loss": 1.8814,
+      "step": 4657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324764516742682e-07,
+      "loss": 1.836,
+      "step": 4658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3322066936925556e-07,
+      "loss": 1.8046,
+      "step": 4659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319369085347467e-07,
+      "loss": 1.7663,
+      "step": 4660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316670962229112e-07,
+      "loss": 1.712,
+      "step": 4661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331397256779121e-07,
+      "loss": 1.7843,
+      "step": 4662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311273902254504e-07,
+      "loss": 1.7789,
+      "step": 4663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308574965839757e-07,
+      "loss": 1.7438,
+      "step": 4664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3305875758767754e-07,
+      "loss": 1.6201,
+      "step": 4665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3303176281259302e-07,
+      "loss": 1.7392,
+      "step": 4666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300476533535237e-07,
+      "loss": 1.741,
+      "step": 4667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297776515816406e-07,
+      "loss": 1.764,
+      "step": 4668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295076228323683e-07,
+      "loss": 1.6871,
+      "step": 4669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292375671277965e-07,
+      "loss": 1.9273,
+      "step": 4670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289674844900175e-07,
+      "loss": 1.7782,
+      "step": 4671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286973749411248e-07,
+      "loss": 1.4801,
+      "step": 4672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328427238503215e-07,
+      "loss": 1.8885,
+      "step": 4673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328157075198386e-07,
+      "loss": 1.7985,
+      "step": 4674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3278868850487395e-07,
+      "loss": 1.5501,
+      "step": 4675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276166680763772e-07,
+      "loss": 1.7248,
+      "step": 4676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273464243034046e-07,
+      "loss": 1.6796,
+      "step": 4677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270761537519293e-07,
+      "loss": 1.7501,
+      "step": 4678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268058564440604e-07,
+      "loss": 1.6227,
+      "step": 4679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265355324019095e-07,
+      "loss": 1.6675,
+      "step": 4680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32626518164759e-07,
+      "loss": 1.7845,
+      "step": 4681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259948042032186e-07,
+      "loss": 1.7168,
+      "step": 4682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257244000909132e-07,
+      "loss": 1.7532,
+      "step": 4683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325453969332794e-07,
+      "loss": 1.6962,
+      "step": 4684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251835119509837e-07,
+      "loss": 1.667,
+      "step": 4685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324913027967607e-07,
+      "loss": 1.7395,
+      "step": 4686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246425174047909e-07,
+      "loss": 1.7117,
+      "step": 4687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243719802846644e-07,
+      "loss": 1.7662,
+      "step": 4688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324101416629358e-07,
+      "loss": 1.6895,
+      "step": 4689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323830826461006e-07,
+      "loss": 1.7269,
+      "step": 4690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235602098017437e-07,
+      "loss": 1.7567,
+      "step": 4691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323289566673709e-07,
+      "loss": 1.7989,
+      "step": 4692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230188970990415e-07,
+      "loss": 1.6209,
+      "step": 4693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227482010998833e-07,
+      "loss": 1.8411,
+      "step": 4694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3224774786983787e-07,
+      "loss": 1.8007,
+      "step": 4695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222067299166745e-07,
+      "loss": 1.6987,
+      "step": 4696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219359547769187e-07,
+      "loss": 1.7395,
+      "step": 4697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321665153301262e-07,
+      "loss": 1.6443,
+      "step": 4698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213943255118575e-07,
+      "loss": 1.7251,
+      "step": 4699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.32112347143086e-07,
+      "loss": 1.8189,
+      "step": 4700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.320852591080427e-07,
+      "loss": 1.5608,
+      "step": 4701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3205816844827173e-07,
+      "loss": 1.8333,
+      "step": 4702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3203107516598927e-07,
+      "loss": 1.6554,
+      "step": 4703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3200397926341168e-07,
+      "loss": 1.7208,
+      "step": 4704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3197688074275557e-07,
+      "loss": 1.7639,
+      "step": 4705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3194977960623766e-07,
+      "loss": 1.7131,
+      "step": 4706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3192267585607497e-07,
+      "loss": 1.8124,
+      "step": 4707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3189556949448474e-07,
+      "loss": 1.7145,
+      "step": 4708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318684605236844e-07,
+      "loss": 1.7955,
+      "step": 4709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3184134894589153e-07,
+      "loss": 1.8101,
+      "step": 4710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181423476332406e-07,
+      "loss": 1.8849,
+      "step": 4711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178711797820004e-07,
+      "loss": 1.7172,
+      "step": 4712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175999859273775e-07,
+      "loss": 1.7855,
+      "step": 4713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173287660915571e-07,
+      "loss": 1.6659,
+      "step": 4714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170575202967255e-07,
+      "loss": 1.9583,
+      "step": 4715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167862485650726e-07,
+      "loss": 1.904,
+      "step": 4716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165149509187898e-07,
+      "loss": 1.7065,
+      "step": 4717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162436273800703e-07,
+      "loss": 1.7096,
+      "step": 4718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159722779711096e-07,
+      "loss": 1.6868,
+      "step": 4719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157009027141052e-07,
+      "loss": 1.7488,
+      "step": 4720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154295016312574e-07,
+      "loss": 1.8952,
+      "step": 4721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315158074744768e-07,
+      "loss": 1.7636,
+      "step": 4722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148866220768405e-07,
+      "loss": 1.6069,
+      "step": 4723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146151436496815e-07,
+      "loss": 1.8071,
+      "step": 4724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314343639485499e-07,
+      "loss": 1.8393,
+      "step": 4725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140721096065037e-07,
+      "loss": 1.5599,
+      "step": 4726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138005540349078e-07,
+      "loss": 1.7531,
+      "step": 4727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135289727929257e-07,
+      "loss": 1.6832,
+      "step": 4728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132573659027742e-07,
+      "loss": 1.5821,
+      "step": 4729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312985733386672e-07,
+      "loss": 1.7061,
+      "step": 4730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31271407526684e-07,
+      "loss": 1.8026,
+      "step": 4731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312442391565501e-07,
+      "loss": 1.8205,
+      "step": 4732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121706823048805e-07,
+      "loss": 1.6867,
+      "step": 4733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118989475072052e-07,
+      "loss": 1.6674,
+      "step": 4734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3116271871947042e-07,
+      "loss": 1.8853,
+      "step": 4735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113554013896088e-07,
+      "loss": 1.8163,
+      "step": 4736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110835901141527e-07,
+      "loss": 1.8454,
+      "step": 4737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310811753390571e-07,
+      "loss": 1.8016,
+      "step": 4738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105398912411017e-07,
+      "loss": 1.7818,
+      "step": 4739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102680036879842e-07,
+      "loss": 1.7655,
+      "step": 4740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30999609075346e-07,
+      "loss": 1.6539,
+      "step": 4741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309724152459773e-07,
+      "loss": 1.608,
+      "step": 4742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309452188829169e-07,
+      "loss": 1.6061,
+      "step": 4743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3091801998838964e-07,
+      "loss": 1.6764,
+      "step": 4744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089081856462044e-07,
+      "loss": 1.8769,
+      "step": 4745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086361461383456e-07,
+      "loss": 1.8908,
+      "step": 4746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083640813825742e-07,
+      "loss": 1.8293,
+      "step": 4747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308091991401146e-07,
+      "loss": 1.6926,
+      "step": 4748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078198762163197e-07,
+      "loss": 1.8612,
+      "step": 4749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307547735850355e-07,
+      "loss": 1.7868,
+      "step": 4750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072755703255151e-07,
+      "loss": 1.6572,
+      "step": 4751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307003379664064e-07,
+      "loss": 1.6657,
+      "step": 4752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067311638882681e-07,
+      "loss": 1.6176,
+      "step": 4753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064589230203958e-07,
+      "loss": 1.7802,
+      "step": 4754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061866570827181e-07,
+      "loss": 1.6557,
+      "step": 4755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059143660975078e-07,
+      "loss": 1.5083,
+      "step": 4756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056420500870393e-07,
+      "loss": 1.7854,
+      "step": 4757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053697090735892e-07,
+      "loss": 1.7876,
+      "step": 4758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050973430794366e-07,
+      "loss": 1.832,
+      "step": 4759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048249521268623e-07,
+      "loss": 1.7011,
+      "step": 4760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304552536238149e-07,
+      "loss": 1.8145,
+      "step": 4761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042800954355816e-07,
+      "loss": 1.6363,
+      "step": 4762
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040076297414473e-07,
+      "loss": 1.794,
+      "step": 4763
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037351391780348e-07,
+      "loss": 1.8399,
+      "step": 4764
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037351391780348e-07,
+      "loss": 2.0364,
+      "step": 4765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034626237676356e-07,
+      "loss": 1.8032,
+      "step": 4766
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031900835325424e-07,
+      "loss": 1.7292,
+      "step": 4767
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3029175184950502e-07,
+      "loss": 1.7904,
+      "step": 4768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026449286774567e-07,
+      "loss": 1.6966,
+      "step": 4769
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023723141020603e-07,
+      "loss": 1.8911,
+      "step": 4770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302099674791163e-07,
+      "loss": 1.8473,
+      "step": 4771
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301827010767067e-07,
+      "loss": 1.765,
+      "step": 4772
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3015543220520784e-07,
+      "loss": 1.5663,
+      "step": 4773
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012816086685038e-07,
+      "loss": 1.6005,
+      "step": 4774
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301008870638653e-07,
+      "loss": 1.732,
+      "step": 4775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007361079848366e-07,
+      "loss": 1.8015,
+      "step": 4776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004633207293688e-07,
+      "loss": 1.7545,
+      "step": 4777
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001905088945645e-07,
+      "loss": 1.6645,
+      "step": 4778
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999176725027403e-07,
+      "loss": 1.6002,
+      "step": 4779
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299644811576217e-07,
+      "loss": 1.8603,
+      "step": 4780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993719261373142e-07,
+      "loss": 1.8442,
+      "step": 4781
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990990162083567e-07,
+      "loss": 1.7001,
+      "step": 4782
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298826081811669e-07,
+      "loss": 1.8519,
+      "step": 4783
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298553122969579e-07,
+      "loss": 1.6726,
+      "step": 4784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982801397044153e-07,
+      "loss": 1.7762,
+      "step": 4785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29800713203851e-07,
+      "loss": 1.7484,
+      "step": 4786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977340999941958e-07,
+      "loss": 1.5768,
+      "step": 4787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974610435938087e-07,
+      "loss": 1.8055,
+      "step": 4788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971879628596851e-07,
+      "loss": 1.7186,
+      "step": 4789
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969148578141654e-07,
+      "loss": 1.9181,
+      "step": 4790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29664172847959e-07,
+      "loss": 1.7752,
+      "step": 4791
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2963685748783025e-07,
+      "loss": 1.818,
+      "step": 4792
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960953970326486e-07,
+      "loss": 1.6917,
+      "step": 4793
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958221949649747e-07,
+      "loss": 1.8562,
+      "step": 4794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295548968697631e-07,
+      "loss": 1.7978,
+      "step": 4795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952757182529678e-07,
+      "loss": 1.6364,
+      "step": 4796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950024436533389e-07,
+      "loss": 1.7128,
+      "step": 4797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2947291449210993e-07,
+      "loss": 1.7426,
+      "step": 4798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944558220786061e-07,
+      "loss": 1.8208,
+      "step": 4799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941824751482183e-07,
+      "loss": 2.01,
+      "step": 4800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939091041522973e-07,
+      "loss": 1.8534,
+      "step": 4801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2936357091132057e-07,
+      "loss": 1.7612,
+      "step": 4802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933622900533093e-07,
+      "loss": 1.9604,
+      "step": 4803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930888469949742e-07,
+      "loss": 1.6337,
+      "step": 4804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928153799605695e-07,
+      "loss": 1.7349,
+      "step": 4805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292541888972467e-07,
+      "loss": 1.8356,
+      "step": 4806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922683740530385e-07,
+      "loss": 1.4734,
+      "step": 4807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919948352246596e-07,
+      "loss": 1.7051,
+      "step": 4808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917212725097062e-07,
+      "loss": 1.5899,
+      "step": 4809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291447685930558e-07,
+      "loss": 1.6902,
+      "step": 4810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291174075509595e-07,
+      "loss": 1.9017,
+      "step": 4811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909004412692003e-07,
+      "loss": 1.8538,
+      "step": 4812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.290626783231758e-07,
+      "loss": 1.8012,
+      "step": 4813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.290353101419655e-07,
+      "loss": 1.7262,
+      "step": 4814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.29007939585528e-07,
+      "loss": 1.5909,
+      "step": 4815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2898056665610232e-07,
+      "loss": 1.7144,
+      "step": 4816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2895319135592765e-07,
+      "loss": 1.9759,
+      "step": 4817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2892581368724347e-07,
+      "loss": 1.7134,
+      "step": 4818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.288984336522894e-07,
+      "loss": 1.8381,
+      "step": 4819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2887105125330525e-07,
+      "loss": 1.6157,
+      "step": 4820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2884366649253104e-07,
+      "loss": 1.5527,
+      "step": 4821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2881627937220696e-07,
+      "loss": 1.7097,
+      "step": 4822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.287888898945734e-07,
+      "loss": 1.7646,
+      "step": 4823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28761498061871e-07,
+      "loss": 1.7104,
+      "step": 4824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.287341038763405e-07,
+      "loss": 1.6627,
+      "step": 4825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870670734022286e-07,
+      "loss": 1.7533,
+      "step": 4826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286793084557593e-07,
+      "loss": 1.8212,
+      "step": 4827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865190722519113e-07,
+      "loss": 1.7563,
+      "step": 4828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862450365075995e-07,
+      "loss": 1.6103,
+      "step": 4829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859709773470746e-07,
+      "loss": 1.7987,
+      "step": 4830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285696894792756e-07,
+      "loss": 1.8947,
+      "step": 4831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285422788867065e-07,
+      "loss": 1.7584,
+      "step": 4832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285148659592425e-07,
+      "loss": 1.9385,
+      "step": 4833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2848745069912611e-07,
+      "loss": 1.6712,
+      "step": 4834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846003310860001e-07,
+      "loss": 1.7738,
+      "step": 4835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843261318990708e-07,
+      "loss": 1.6559,
+      "step": 4836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284051909452904e-07,
+      "loss": 1.6555,
+      "step": 4837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837776637699328e-07,
+      "loss": 1.8592,
+      "step": 4838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835033948725916e-07,
+      "loss": 1.8105,
+      "step": 4839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2832291027833166e-07,
+      "loss": 1.6675,
+      "step": 4840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829547875245466e-07,
+      "loss": 1.7417,
+      "step": 4841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282680449118722e-07,
+      "loss": 1.6173,
+      "step": 4842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824060875882844e-07,
+      "loss": 1.5794,
+      "step": 4843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821317029556785e-07,
+      "loss": 1.8648,
+      "step": 4844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818572952433497e-07,
+      "loss": 1.8685,
+      "step": 4845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2815828644737462e-07,
+      "loss": 1.8488,
+      "step": 4846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281308410669318e-07,
+      "loss": 1.583,
+      "step": 4847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810339338525164e-07,
+      "loss": 1.7808,
+      "step": 4848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280759434045795e-07,
+      "loss": 1.7977,
+      "step": 4849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804849112716092e-07,
+      "loss": 1.7688,
+      "step": 4850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802103655524164e-07,
+      "loss": 1.5705,
+      "step": 4851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279935796910675e-07,
+      "loss": 1.6805,
+      "step": 4852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796612053688472e-07,
+      "loss": 1.8157,
+      "step": 4853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793865909493953e-07,
+      "loss": 1.7496,
+      "step": 4854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791119536747842e-07,
+      "loss": 1.5168,
+      "step": 4855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27883729356748e-07,
+      "loss": 1.6713,
+      "step": 4856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278562610649952e-07,
+      "loss": 1.8062,
+      "step": 4857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27828790494467e-07,
+      "loss": 1.6777,
+      "step": 4858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780131764741063e-07,
+      "loss": 1.8233,
+      "step": 4859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277738425260736e-07,
+      "loss": 1.8328,
+      "step": 4860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774636513270337e-07,
+      "loss": 1.7595,
+      "step": 4861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771888546954778e-07,
+      "loss": 1.7642,
+      "step": 4862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2769140353885476e-07,
+      "loss": 1.7833,
+      "step": 4863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766391934287258e-07,
+      "loss": 1.7532,
+      "step": 4864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763643288384944e-07,
+      "loss": 1.7546,
+      "step": 4865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276089441640339e-07,
+      "loss": 1.5993,
+      "step": 4866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758145318567474e-07,
+      "loss": 1.6142,
+      "step": 4867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755395995102082e-07,
+      "loss": 1.8145,
+      "step": 4868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752646446232113e-07,
+      "loss": 1.7481,
+      "step": 4869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749896672182504e-07,
+      "loss": 1.9008,
+      "step": 4870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2747146673178195e-07,
+      "loss": 1.7236,
+      "step": 4871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2744396449444152e-07,
+      "loss": 1.7499,
+      "step": 4872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741646001205355e-07,
+      "loss": 1.7076,
+      "step": 4873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27388953286868e-07,
+      "loss": 1.5841,
+      "step": 4874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736144432113512e-07,
+      "loss": 1.8617,
+      "step": 4875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273339331171052e-07,
+      "loss": 1.7078,
+      "step": 4876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730641967702886e-07,
+      "loss": 1.8763,
+      "step": 4877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272789040031568e-07,
+      "loss": 1.5733,
+      "step": 4878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272513860977399e-07,
+      "loss": 1.6898,
+      "step": 4879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272238659630293e-07,
+      "loss": 1.5903,
+      "step": 4880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271963436012763e-07,
+      "loss": 1.7894,
+      "step": 4881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2716881901473234e-07,
+      "loss": 1.7576,
+      "step": 4882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27141292205649e-07,
+      "loss": 1.5693,
+      "step": 4883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711376317627817e-07,
+      "loss": 1.616,
+      "step": 4884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708623192887184e-07,
+      "loss": 1.9068,
+      "step": 4885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270586984656822e-07,
+      "loss": 2.0127,
+      "step": 4886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703116278896164e-07,
+      "loss": 1.7231,
+      "step": 4887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270036249009627e-07,
+      "loss": 1.7174,
+      "step": 4888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697608480393805e-07,
+      "loss": 1.7099,
+      "step": 4889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694854250014068e-07,
+      "loss": 1.8641,
+      "step": 4890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692099799182367e-07,
+      "loss": 1.6889,
+      "step": 4891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689345128124026e-07,
+      "loss": 1.7829,
+      "step": 4892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268659023706439e-07,
+      "loss": 1.7393,
+      "step": 4893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683835126228827e-07,
+      "loss": 1.7244,
+      "step": 4894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681079795842716e-07,
+      "loss": 1.5919,
+      "step": 4895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267832424613145e-07,
+      "loss": 1.8312,
+      "step": 4896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675568477320457e-07,
+      "loss": 1.5689,
+      "step": 4897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2672812489635165e-07,
+      "loss": 1.6536,
+      "step": 4898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267005628330103e-07,
+      "loss": 1.7028,
+      "step": 4899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667299858543516e-07,
+      "loss": 1.9624,
+      "step": 4900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664543215588126e-07,
+      "loss": 1.862,
+      "step": 4901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661786354660352e-07,
+      "loss": 1.8214,
+      "step": 4902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265902927598573e-07,
+      "loss": 1.5051,
+      "step": 4903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656271979789792e-07,
+      "loss": 1.6332,
+      "step": 4904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653514466298103e-07,
+      "loss": 1.9242,
+      "step": 4905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265075673573624e-07,
+      "loss": 1.8084,
+      "step": 4906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26479987883298e-07,
+      "loss": 1.8169,
+      "step": 4907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645240624304397e-07,
+      "loss": 1.5931,
+      "step": 4908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642482243885657e-07,
+      "loss": 1.7274,
+      "step": 4909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639723647299233e-07,
+      "loss": 1.8741,
+      "step": 4910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263696483477079e-07,
+      "loss": 1.727,
+      "step": 4911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634205806526014e-07,
+      "loss": 1.6134,
+      "step": 4912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631446562790602e-07,
+      "loss": 1.756,
+      "step": 4913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262868710379028e-07,
+      "loss": 1.7857,
+      "step": 4914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262592742975078e-07,
+      "loss": 1.9231,
+      "step": 4915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2623167540897858e-07,
+      "loss": 1.8167,
+      "step": 4916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620407437457286e-07,
+      "loss": 1.8435,
+      "step": 4917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617647119654856e-07,
+      "loss": 1.7285,
+      "step": 4918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614886587716367e-07,
+      "loss": 1.73,
+      "step": 4919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612125841867656e-07,
+      "loss": 1.9352,
+      "step": 4920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609364882334552e-07,
+      "loss": 1.7055,
+      "step": 4921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606603709342926e-07,
+      "loss": 1.7434,
+      "step": 4922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260384232311865e-07,
+      "loss": 1.6628,
+      "step": 4923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601080723887623e-07,
+      "loss": 1.9137,
+      "step": 4924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598318911875748e-07,
+      "loss": 1.6988,
+      "step": 4925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595556887308957e-07,
+      "loss": 1.6689,
+      "step": 4926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2592794650413207e-07,
+      "loss": 1.74,
+      "step": 4927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.259003220141445e-07,
+      "loss": 1.7763,
+      "step": 4928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2587269540538678e-07,
+      "loss": 1.6616,
+      "step": 4929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2584506668011873e-07,
+      "loss": 1.6291,
+      "step": 4930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2581743584060075e-07,
+      "loss": 1.9417,
+      "step": 4931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2578980288909296e-07,
+      "loss": 1.7375,
+      "step": 4932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25762167827856e-07,
+      "loss": 1.7072,
+      "step": 4933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2573453065915045e-07,
+      "loss": 1.9548,
+      "step": 4934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2570689138523728e-07,
+      "loss": 1.7814,
+      "step": 4935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2567925000837742e-07,
+      "loss": 1.7867,
+      "step": 4936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2565160653083213e-07,
+      "loss": 1.6151,
+      "step": 4937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2562396095486273e-07,
+      "loss": 1.8324,
+      "step": 4938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559631328273076e-07,
+      "loss": 1.8113,
+      "step": 4939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2556866351669796e-07,
+      "loss": 1.6242,
+      "step": 4940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554101165902621e-07,
+      "loss": 1.785,
+      "step": 4941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2551335771197759e-07,
+      "loss": 1.6706,
+      "step": 4942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2548570167781426e-07,
+      "loss": 1.6931,
+      "step": 4943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545804355879865e-07,
+      "loss": 1.6355,
+      "step": 4944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2543038335719334e-07,
+      "loss": 1.8906,
+      "step": 4945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2540272107526104e-07,
+      "loss": 1.7439,
+      "step": 4946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537505671526465e-07,
+      "loss": 1.8494,
+      "step": 4947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253473902794673e-07,
+      "loss": 1.8906,
+      "step": 4948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531972177013218e-07,
+      "loss": 1.7355,
+      "step": 4949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529205118952276e-07,
+      "loss": 1.8824,
+      "step": 4950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526437853990253e-07,
+      "loss": 1.8257,
+      "step": 4951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523670382353536e-07,
+      "loss": 1.4973,
+      "step": 4952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520902704268513e-07,
+      "loss": 1.6881,
+      "step": 4953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518134819961592e-07,
+      "loss": 1.849,
+      "step": 4954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515366729659197e-07,
+      "loss": 1.6437,
+      "step": 4955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512598433587776e-07,
+      "loss": 1.7408,
+      "step": 4956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250982993197379e-07,
+      "loss": 1.6853,
+      "step": 4957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507061225043706e-07,
+      "loss": 1.7578,
+      "step": 4958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250429231302403e-07,
+      "loss": 1.7576,
+      "step": 4959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501523196141263e-07,
+      "loss": 1.6301,
+      "step": 4960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498753874621938e-07,
+      "loss": 1.5483,
+      "step": 4961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2495984348692593e-07,
+      "loss": 1.8316,
+      "step": 4962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493214618579792e-07,
+      "loss": 1.7422,
+      "step": 4963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249044468451011e-07,
+      "loss": 1.8349,
+      "step": 4964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2487674546710144e-07,
+      "loss": 1.7282,
+      "step": 4965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484904205406503e-07,
+      "loss": 1.6429,
+      "step": 4966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482133660825812e-07,
+      "loss": 1.6665,
+      "step": 4967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247936291319472e-07,
+      "loss": 1.7556,
+      "step": 4968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476591962739883e-07,
+      "loss": 1.7527,
+      "step": 4969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247382080968798e-07,
+      "loss": 1.7278,
+      "step": 4970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471049454265703e-07,
+      "loss": 1.7297,
+      "step": 4971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246827789669976e-07,
+      "loss": 1.7293,
+      "step": 4972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465506137216887e-07,
+      "loss": 1.7952,
+      "step": 4973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246273417604382e-07,
+      "loss": 1.6367,
+      "step": 4974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459962013407315e-07,
+      "loss": 1.7669,
+      "step": 4975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457189649534154e-07,
+      "loss": 1.6631,
+      "step": 4976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245441708465113e-07,
+      "loss": 1.5891,
+      "step": 4977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245164431898505e-07,
+      "loss": 1.7285,
+      "step": 4978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244887135276274e-07,
+      "loss": 1.6683,
+      "step": 4979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244609818621104e-07,
+      "loss": 1.8534,
+      "step": 4980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443324819556811e-07,
+      "loss": 1.804,
+      "step": 4981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440551253026927e-07,
+      "loss": 1.7422,
+      "step": 4982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243777748684828e-07,
+      "loss": 1.7119,
+      "step": 4983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435003521247776e-07,
+      "loss": 1.8593,
+      "step": 4984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432229356452334e-07,
+      "loss": 1.8123,
+      "step": 4985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429454992688904e-07,
+      "loss": 1.6501,
+      "step": 4986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242668043018444e-07,
+      "loss": 1.785,
+      "step": 4987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423905669165904e-07,
+      "loss": 1.9725,
+      "step": 4988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421130709860292e-07,
+      "loss": 1.5025,
+      "step": 4989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418355552494612e-07,
+      "loss": 1.6594,
+      "step": 4990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241558019729588e-07,
+      "loss": 1.7269,
+      "step": 4991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241280464449114e-07,
+      "loss": 1.798,
+      "step": 4992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410028894307438e-07,
+      "loss": 1.8049,
+      "step": 4993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407252946971846e-07,
+      "loss": 1.7797,
+      "step": 4994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240447680271145e-07,
+      "loss": 1.7181,
+      "step": 4995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401700461753356e-07,
+      "loss": 1.679,
+      "step": 4996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398923924324675e-07,
+      "loss": 1.8115,
+      "step": 4997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396147190652544e-07,
+      "loss": 1.7021,
+      "step": 4998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393370260964115e-07,
+      "loss": 1.7546,
+      "step": 4999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390593135486552e-07,
+      "loss": 1.7473,
+      "step": 5000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2387815814447037e-07,
+      "loss": 1.7192,
+      "step": 5001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238503829807277e-07,
+      "loss": 1.6777,
+      "step": 5002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238226058659096e-07,
+      "loss": 1.7136,
+      "step": 5003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379482680228847e-07,
+      "loss": 1.6975,
+      "step": 5004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237670457921367e-07,
+      "loss": 1.7643,
+      "step": 5005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373926283772688e-07,
+      "loss": 1.7323,
+      "step": 5006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371147794133185e-07,
+      "loss": 1.7608,
+      "step": 5007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368369110522452e-07,
+      "loss": 1.7764,
+      "step": 5008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365590233167803e-07,
+      "loss": 1.8841,
+      "step": 5009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362811162296555e-07,
+      "loss": 1.7185,
+      "step": 5010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360031898136055e-07,
+      "loss": 1.6019,
+      "step": 5011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235725244091366e-07,
+      "loss": 1.8912,
+      "step": 5012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354472790856743e-07,
+      "loss": 1.6293,
+      "step": 5013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235169294819269e-07,
+      "loss": 1.7547,
+      "step": 5014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348912913148905e-07,
+      "loss": 1.773,
+      "step": 5015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234613268595281e-07,
+      "loss": 1.8872,
+      "step": 5016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343352266831845e-07,
+      "loss": 1.654,
+      "step": 5017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340571656013457e-07,
+      "loss": 1.8054,
+      "step": 5018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337790853725113e-07,
+      "loss": 1.7569,
+      "step": 5019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335009860194295e-07,
+      "loss": 1.8435,
+      "step": 5020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332228675648505e-07,
+      "loss": 1.7459,
+      "step": 5021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329447300315256e-07,
+      "loss": 1.7997,
+      "step": 5022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326665734422076e-07,
+      "loss": 1.6108,
+      "step": 5023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323883978196513e-07,
+      "loss": 1.6529,
+      "step": 5024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321102031866123e-07,
+      "loss": 1.7487,
+      "step": 5025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318319895658494e-07,
+      "loss": 1.8809,
+      "step": 5026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315537569801204e-07,
+      "loss": 1.8629,
+      "step": 5027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231275505452187e-07,
+      "loss": 1.6628,
+      "step": 5028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230997235004811e-07,
+      "loss": 1.8207,
+      "step": 5029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307189456607568e-07,
+      "loss": 1.6729,
+      "step": 5030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230440637442789e-07,
+      "loss": 1.9783,
+      "step": 5031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230162310373675e-07,
+      "loss": 1.7478,
+      "step": 5032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298839644761834e-07,
+      "loss": 1.6581,
+      "step": 5033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296055997730841e-07,
+      "loss": 1.7172,
+      "step": 5034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229327216287149e-07,
+      "loss": 1.7722,
+      "step": 5035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2290488140411504e-07,
+      "loss": 1.6594,
+      "step": 5036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287703930578634e-07,
+      "loss": 1.7561,
+      "step": 5037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284919533600644e-07,
+      "loss": 1.7478,
+      "step": 5038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228213494970531e-07,
+      "loss": 1.8429,
+      "step": 5039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.227935017912042e-07,
+      "loss": 1.7863,
+      "step": 5040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2276565222073786e-07,
+      "loss": 1.8744,
+      "step": 5041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2273780078793227e-07,
+      "loss": 1.7913,
+      "step": 5042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2270994749506586e-07,
+      "loss": 1.7897,
+      "step": 5043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2268209234441712e-07,
+      "loss": 1.6032,
+      "step": 5044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2265423533826474e-07,
+      "loss": 1.7961,
+      "step": 5045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.226263764788876e-07,
+      "loss": 1.7233,
+      "step": 5046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2259851576856462e-07,
+      "loss": 1.7056,
+      "step": 5047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2257065320957501e-07,
+      "loss": 1.8185,
+      "step": 5048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22542788804198e-07,
+      "loss": 1.8078,
+      "step": 5049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2251492255471305e-07,
+      "loss": 1.6842,
+      "step": 5050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2248705446339978e-07,
+      "loss": 1.5521,
+      "step": 5051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245918453253793e-07,
+      "loss": 1.5736,
+      "step": 5052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243131276440733e-07,
+      "loss": 1.7532,
+      "step": 5053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224034391612881e-07,
+      "loss": 1.6521,
+      "step": 5054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237556372546036e-07,
+      "loss": 1.7521,
+      "step": 5055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2234768645920454e-07,
+      "loss": 1.9772,
+      "step": 5056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223198073648011e-07,
+      "loss": 1.9298,
+      "step": 5057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222919264445306e-07,
+      "loss": 1.86,
+      "step": 5058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22264043700674e-07,
+      "loss": 1.7927,
+      "step": 5059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222361591355121e-07,
+      "loss": 1.8356,
+      "step": 5060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220827275132607e-07,
+      "loss": 1.72,
+      "step": 5061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218038455039708e-07,
+      "loss": 1.8295,
+      "step": 5062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221524945350066e-07,
+      "loss": 1.7517,
+      "step": 5063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212460270743607e-07,
+      "loss": 1.758,
+      "step": 5064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220967090699673e-07,
+      "loss": 1.705,
+      "step": 5065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206881362488198e-07,
+      "loss": 1.9108,
+      "step": 5066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204091637446222e-07,
+      "loss": 1.6531,
+      "step": 5067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201301732099005e-07,
+      "loss": 1.7847,
+      "step": 5068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198511646674783e-07,
+      "loss": 1.7252,
+      "step": 5069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219572138140179e-07,
+      "loss": 1.6082,
+      "step": 5070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192930936508286e-07,
+      "loss": 1.5395,
+      "step": 5071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190140312222548e-07,
+      "loss": 1.7828,
+      "step": 5072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187349508772854e-07,
+      "loss": 1.6441,
+      "step": 5073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184558526387516e-07,
+      "loss": 1.8653,
+      "step": 5074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181767365294836e-07,
+      "loss": 1.6959,
+      "step": 5075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178976025723155e-07,
+      "loss": 1.6634,
+      "step": 5076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176184507900813e-07,
+      "loss": 1.59,
+      "step": 5077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173392812056175e-07,
+      "loss": 1.6391,
+      "step": 5078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170600938417603e-07,
+      "loss": 1.8792,
+      "step": 5079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21678088872135e-07,
+      "loss": 1.5807,
+      "step": 5080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165016658672257e-07,
+      "loss": 1.7221,
+      "step": 5081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162224253022304e-07,
+      "loss": 1.7836,
+      "step": 5082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215943167049206e-07,
+      "loss": 1.8665,
+      "step": 5083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215663891130998e-07,
+      "loss": 1.7977,
+      "step": 5084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153845975704526e-07,
+      "loss": 1.7292,
+      "step": 5085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151052863904167e-07,
+      "loss": 1.6776,
+      "step": 5086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21482595761374e-07,
+      "loss": 1.7574,
+      "step": 5087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145466112632722e-07,
+      "loss": 1.7398,
+      "step": 5088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214267247361866e-07,
+      "loss": 1.597,
+      "step": 5089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139878659323737e-07,
+      "loss": 1.7607,
+      "step": 5090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213708466997651e-07,
+      "loss": 1.6049,
+      "step": 5091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134290505805532e-07,
+      "loss": 1.6713,
+      "step": 5092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213149616703939e-07,
+      "loss": 1.8541,
+      "step": 5093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128701653906664e-07,
+      "loss": 1.8372,
+      "step": 5094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125906966635963e-07,
+      "loss": 1.7125,
+      "step": 5095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21231121054559e-07,
+      "loss": 1.9309,
+      "step": 5096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212031707059512e-07,
+      "loss": 1.8324,
+      "step": 5097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117521862282263e-07,
+      "loss": 1.6889,
+      "step": 5098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114726480745988e-07,
+      "loss": 1.6865,
+      "step": 5099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2111930926214973e-07,
+      "loss": 1.802,
+      "step": 5100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210913519891791e-07,
+      "loss": 1.7712,
+      "step": 5101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106339299083502e-07,
+      "loss": 1.8046,
+      "step": 5102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2103543226940463e-07,
+      "loss": 1.823,
+      "step": 5103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100746982717527e-07,
+      "loss": 1.579,
+      "step": 5104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097950566643443e-07,
+      "loss": 1.9714,
+      "step": 5105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209515397894697e-07,
+      "loss": 1.7484,
+      "step": 5106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209235721985688e-07,
+      "loss": 1.7544,
+      "step": 5107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2089560289601968e-07,
+      "loss": 1.9131,
+      "step": 5108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086763188411023e-07,
+      "loss": 1.6682,
+      "step": 5109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083965916512875e-07,
+      "loss": 1.8302,
+      "step": 5110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2081168474136347e-07,
+      "loss": 1.6551,
+      "step": 5111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078370861510286e-07,
+      "loss": 1.9136,
+      "step": 5112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075573078863545e-07,
+      "loss": 1.7229,
+      "step": 5113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072775126425002e-07,
+      "loss": 1.748,
+      "step": 5114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069977004423542e-07,
+      "loss": 1.7571,
+      "step": 5115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067178713088062e-07,
+      "loss": 1.8339,
+      "step": 5116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064380252647479e-07,
+      "loss": 1.6247,
+      "step": 5117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2061581623330717e-07,
+      "loss": 1.8376,
+      "step": 5118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205878282536672e-07,
+      "loss": 1.6966,
+      "step": 5119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205598385898444e-07,
+      "loss": 1.5055,
+      "step": 5120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053184724412849e-07,
+      "loss": 1.833,
+      "step": 5121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050385421880926e-07,
+      "loss": 1.9387,
+      "step": 5122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047585951617673e-07,
+      "loss": 1.826,
+      "step": 5123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2044786313852096e-07,
+      "loss": 1.7187,
+      "step": 5124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2041986508813217e-07,
+      "loss": 1.7158,
+      "step": 5125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203918653673008e-07,
+      "loss": 1.6725,
+      "step": 5126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203638639783173e-07,
+      "loss": 1.8576,
+      "step": 5127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033586092347232e-07,
+      "loss": 1.8603,
+      "step": 5128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030785620505667e-07,
+      "loss": 1.5567,
+      "step": 5129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202798498253613e-07,
+      "loss": 1.7847,
+      "step": 5130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025184178667715e-07,
+      "loss": 1.766,
+      "step": 5131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022383209129557e-07,
+      "loss": 1.6317,
+      "step": 5132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019582074150776e-07,
+      "loss": 1.6863,
+      "step": 5133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016780773960525e-07,
+      "loss": 1.8565,
+      "step": 5134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201397930878796e-07,
+      "loss": 1.5629,
+      "step": 5135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2011177678862255e-07,
+      "loss": 1.5371,
+      "step": 5136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2008375884412598e-07,
+      "loss": 1.7617,
+      "step": 5137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005573925668188e-07,
+      "loss": 1.7604,
+      "step": 5138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002771802858242e-07,
+      "loss": 1.8553,
+      "step": 5139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999969516211982e-07,
+      "loss": 1.5958,
+      "step": 5140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997167065958652e-07,
+      "loss": 1.8349,
+      "step": 5141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994364452327505e-07,
+      "loss": 1.8388,
+      "step": 5142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991561675547807e-07,
+      "loss": 1.8469,
+      "step": 5143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198875873584884e-07,
+      "loss": 1.7295,
+      "step": 5144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19859556334599e-07,
+      "loss": 1.7793,
+      "step": 5145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1983152368610288e-07,
+      "loss": 1.9419,
+      "step": 5146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980348941529327e-07,
+      "loss": 1.61,
+      "step": 5147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977545352446353e-07,
+      "loss": 1.7897,
+      "step": 5148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974741601590714e-07,
+      "loss": 1.6832,
+      "step": 5149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971937689191765e-07,
+      "loss": 1.7097,
+      "step": 5150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969133615478882e-07,
+      "loss": 1.9236,
+      "step": 5151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966329380681453e-07,
+      "loss": 1.9377,
+      "step": 5152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1963524985028876e-07,
+      "loss": 1.8365,
+      "step": 5153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1960720428750569e-07,
+      "loss": 1.5623,
+      "step": 5154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.195791571207595e-07,
+      "loss": 1.65,
+      "step": 5155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1955110835234461e-07,
+      "loss": 1.7392,
+      "step": 5156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1952305798455557e-07,
+      "loss": 1.801,
+      "step": 5157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1949500601968705e-07,
+      "loss": 1.6033,
+      "step": 5158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1946695246003376e-07,
+      "loss": 1.6301,
+      "step": 5159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1943889730789066e-07,
+      "loss": 1.8805,
+      "step": 5160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1941084056555282e-07,
+      "loss": 1.7571,
+      "step": 5161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.193827822353154e-07,
+      "loss": 1.6744,
+      "step": 5162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1935472231947367e-07,
+      "loss": 1.7289,
+      "step": 5163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.193266608203231e-07,
+      "loss": 1.7691,
+      "step": 5164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1929859774015924e-07,
+      "loss": 1.6886,
+      "step": 5165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927053308127777e-07,
+      "loss": 1.6536,
+      "step": 5166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924246684597458e-07,
+      "loss": 1.6264,
+      "step": 5167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921439903654554e-07,
+      "loss": 1.6141,
+      "step": 5168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918632965528679e-07,
+      "loss": 1.6873,
+      "step": 5169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915825870449449e-07,
+      "loss": 1.8378,
+      "step": 5170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913018618646506e-07,
+      "loss": 1.9143,
+      "step": 5171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1910211210349485e-07,
+      "loss": 1.575,
+      "step": 5172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907403645788054e-07,
+      "loss": 1.8623,
+      "step": 5173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1904595925191882e-07,
+      "loss": 1.8349,
+      "step": 5174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901788048790657e-07,
+      "loss": 1.6898,
+      "step": 5175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898980016814074e-07,
+      "loss": 1.7793,
+      "step": 5176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896171829491841e-07,
+      "loss": 1.887,
+      "step": 5177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893363487053685e-07,
+      "loss": 1.5077,
+      "step": 5178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890554989729342e-07,
+      "loss": 1.7494,
+      "step": 5179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188774633774856e-07,
+      "loss": 1.5838,
+      "step": 5180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18849375313411e-07,
+      "loss": 2.0103,
+      "step": 5181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1882128570736732e-07,
+      "loss": 1.9128,
+      "step": 5182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1879319456165247e-07,
+      "loss": 1.7143,
+      "step": 5183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876510187856445e-07,
+      "loss": 1.6805,
+      "step": 5184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873700766040133e-07,
+      "loss": 1.7812,
+      "step": 5185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870891190946137e-07,
+      "loss": 1.7218,
+      "step": 5186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868081462804296e-07,
+      "loss": 1.8453,
+      "step": 5187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865271581844459e-07,
+      "loss": 1.7375,
+      "step": 5188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862461548296483e-07,
+      "loss": 1.7755,
+      "step": 5189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859651362390244e-07,
+      "loss": 1.9203,
+      "step": 5190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856841024355633e-07,
+      "loss": 1.7204,
+      "step": 5191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1854030534422544e-07,
+      "loss": 1.8431,
+      "step": 5192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851219892820892e-07,
+      "loss": 1.8265,
+      "step": 5193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1848409099780599e-07,
+      "loss": 1.7385,
+      "step": 5194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18455981555316e-07,
+      "loss": 1.6544,
+      "step": 5195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842787060303847e-07,
+      "loss": 1.7133,
+      "step": 5196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1839975814327301e-07,
+      "loss": 1.5462,
+      "step": 5197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837164417831931e-07,
+      "loss": 1.5336,
+      "step": 5198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834352871047726e-07,
+      "loss": 1.8967,
+      "step": 5199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831541174204685e-07,
+      "loss": 1.7673,
+      "step": 5200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1828729327532816e-07,
+      "loss": 1.5937,
+      "step": 5201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1825917331262143e-07,
+      "loss": 1.8291,
+      "step": 5202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823105185622697e-07,
+      "loss": 1.8302,
+      "step": 5203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182029289084453e-07,
+      "loss": 1.5842,
+      "step": 5204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817480447157699e-07,
+      "loss": 1.5883,
+      "step": 5205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1814667854792277e-07,
+      "loss": 1.8234,
+      "step": 5206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811855113978341e-07,
+      "loss": 1.7627,
+      "step": 5207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809042224945998e-07,
+      "loss": 1.8092,
+      "step": 5208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806229187925344e-07,
+      "loss": 1.8655,
+      "step": 5209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803416003146508e-07,
+      "loss": 1.8134,
+      "step": 5210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800602670839617e-07,
+      "loss": 1.7526,
+      "step": 5211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797789191234815e-07,
+      "loss": 1.7493,
+      "step": 5212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179497556456226e-07,
+      "loss": 1.7591,
+      "step": 5213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792161791052122e-07,
+      "loss": 1.8374,
+      "step": 5214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789347870934577e-07,
+      "loss": 1.8911,
+      "step": 5215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786533804439818e-07,
+      "loss": 1.7005,
+      "step": 5216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783719591798047e-07,
+      "loss": 1.7185,
+      "step": 5217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780905233239486e-07,
+      "loss": 1.7168,
+      "step": 5218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778090728994361e-07,
+      "loss": 1.7429,
+      "step": 5219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775276079292909e-07,
+      "loss": 1.719,
+      "step": 5220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772461284365385e-07,
+      "loss": 1.7847,
+      "step": 5221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769646344442048e-07,
+      "loss": 1.8837,
+      "step": 5222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1766831259753181e-07,
+      "loss": 1.7236,
+      "step": 5223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764016030529063e-07,
+      "loss": 1.6061,
+      "step": 5224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761200657000001e-07,
+      "loss": 1.691,
+      "step": 5225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758385139396301e-07,
+      "loss": 1.9847,
+      "step": 5226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755569477948289e-07,
+      "loss": 1.8596,
+      "step": 5227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752753672886293e-07,
+      "loss": 1.721,
+      "step": 5228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749937724440669e-07,
+      "loss": 1.7699,
+      "step": 5229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747121632841767e-07,
+      "loss": 1.8007,
+      "step": 5230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174430539831996e-07,
+      "loss": 1.671,
+      "step": 5231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741489021105633e-07,
+      "loss": 1.7496,
+      "step": 5232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173867250142917e-07,
+      "loss": 1.7697,
+      "step": 5233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735855839520987e-07,
+      "loss": 1.5583,
+      "step": 5234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733039035611492e-07,
+      "loss": 1.7866,
+      "step": 5235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730222089931116e-07,
+      "loss": 1.7987,
+      "step": 5236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727405002710298e-07,
+      "loss": 1.629,
+      "step": 5237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1724587774179491e-07,
+      "loss": 1.6527,
+      "step": 5238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1721770404569153e-07,
+      "loss": 1.8208,
+      "step": 5239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718952894109766e-07,
+      "loss": 1.6838,
+      "step": 5240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716135243031806e-07,
+      "loss": 1.6969,
+      "step": 5241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171331745156578e-07,
+      "loss": 1.7261,
+      "step": 5242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710499519942194e-07,
+      "loss": 1.8008,
+      "step": 5243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707681448391564e-07,
+      "loss": 1.7971,
+      "step": 5244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1704863237144424e-07,
+      "loss": 1.7523,
+      "step": 5245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702044886431317e-07,
+      "loss": 1.8724,
+      "step": 5246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699226396482804e-07,
+      "loss": 1.5759,
+      "step": 5247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696407767529441e-07,
+      "loss": 1.5851,
+      "step": 5248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1693588999801811e-07,
+      "loss": 1.6791,
+      "step": 5249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16907700935305e-07,
+      "loss": 1.8382,
+      "step": 5250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1687951048946109e-07,
+      "loss": 1.7506,
+      "step": 5251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685131866279252e-07,
+      "loss": 1.8407,
+      "step": 5252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682312545760549e-07,
+      "loss": 1.8412,
+      "step": 5253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679493087620632e-07,
+      "loss": 1.6668,
+      "step": 5254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167667349209015e-07,
+      "loss": 1.7658,
+      "step": 5255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673853759399756e-07,
+      "loss": 1.8783,
+      "step": 5256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671033889780124e-07,
+      "loss": 1.6687,
+      "step": 5257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1668213883461921e-07,
+      "loss": 1.9066,
+      "step": 5258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1665393740675849e-07,
+      "loss": 1.7469,
+      "step": 5259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662573461652602e-07,
+      "loss": 1.7356,
+      "step": 5260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659753046622898e-07,
+      "loss": 1.7592,
+      "step": 5261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1656932495817455e-07,
+      "loss": 1.8708,
+      "step": 5262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1654111809467008e-07,
+      "loss": 1.7183,
+      "step": 5263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165129098780231e-07,
+      "loss": 1.561,
+      "step": 5264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648470031054107e-07,
+      "loss": 1.7188,
+      "step": 5265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1645648939453172e-07,
+      "loss": 1.834,
+      "step": 5266
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1642827713230283e-07,
+      "loss": 1.7562,
+      "step": 5267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1640006352616234e-07,
+      "loss": 1.8602,
+      "step": 5268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.163718485784182e-07,
+      "loss": 1.6524,
+      "step": 5269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1634363229137855e-07,
+      "loss": 1.7318,
+      "step": 5270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.163154146673516e-07,
+      "loss": 1.7034,
+      "step": 5271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1628719570864572e-07,
+      "loss": 1.7835,
+      "step": 5272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1625897541756935e-07,
+      "loss": 1.6453,
+      "step": 5273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1623075379643101e-07,
+      "loss": 1.7663,
+      "step": 5274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.162025308475394e-07,
+      "loss": 1.8306,
+      "step": 5275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1617430657320327e-07,
+      "loss": 1.8233,
+      "step": 5276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1614608097573153e-07,
+      "loss": 1.8595,
+      "step": 5277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611785405743313e-07,
+      "loss": 1.5235,
+      "step": 5278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608962582061717e-07,
+      "loss": 1.874,
+      "step": 5279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606139626759288e-07,
+      "loss": 1.7307,
+      "step": 5280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603316540066958e-07,
+      "loss": 1.6894,
+      "step": 5281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1600493322215667e-07,
+      "loss": 1.6394,
+      "step": 5282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597669973436368e-07,
+      "loss": 1.6321,
+      "step": 5283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594846493960022e-07,
+      "loss": 1.7202,
+      "step": 5284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592022884017607e-07,
+      "loss": 1.5849,
+      "step": 5285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589199143840109e-07,
+      "loss": 1.8435,
+      "step": 5286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586375273658517e-07,
+      "loss": 1.7494,
+      "step": 5287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583551273703842e-07,
+      "loss": 1.8147,
+      "step": 5288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580727144207101e-07,
+      "loss": 1.743,
+      "step": 5289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157790288539932e-07,
+      "loss": 1.9025,
+      "step": 5290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575078497511534e-07,
+      "loss": 1.9814,
+      "step": 5291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572253980774795e-07,
+      "loss": 1.6447,
+      "step": 5292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569429335420162e-07,
+      "loss": 1.8269,
+      "step": 5293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566604561678705e-07,
+      "loss": 1.5848,
+      "step": 5294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566604561678705e-07,
+      "loss": 1.6975,
+      "step": 5295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15637796597815e-07,
+      "loss": 1.7854,
+      "step": 5296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560954629959642e-07,
+      "loss": 1.8047,
+      "step": 5297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155812947244423e-07,
+      "loss": 1.7881,
+      "step": 5298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555304187466372e-07,
+      "loss": 1.6569,
+      "step": 5299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15524787752572e-07,
+      "loss": 1.819,
+      "step": 5300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549653236047834e-07,
+      "loss": 1.7209,
+      "step": 5301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546827570069424e-07,
+      "loss": 1.7738,
+      "step": 5302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154400177755312e-07,
+      "loss": 1.7387,
+      "step": 5303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154117585873009e-07,
+      "loss": 1.8222,
+      "step": 5304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15383498138315e-07,
+      "loss": 1.6787,
+      "step": 5305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535523643088541e-07,
+      "loss": 1.5582,
+      "step": 5306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532697346732403e-07,
+      "loss": 1.8685,
+      "step": 5307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529870924994294e-07,
+      "loss": 1.7352,
+      "step": 5308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527044378105423e-07,
+      "loss": 1.8637,
+      "step": 5309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524217706297022e-07,
+      "loss": 1.7756,
+      "step": 5310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521390909800321e-07,
+      "loss": 1.6368,
+      "step": 5311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518563988846568e-07,
+      "loss": 1.9603,
+      "step": 5312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515736943667023e-07,
+      "loss": 1.9563,
+      "step": 5313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151290977449294e-07,
+      "loss": 1.8637,
+      "step": 5314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510082481555605e-07,
+      "loss": 1.696,
+      "step": 5315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1507255065086301e-07,
+      "loss": 1.7241,
+      "step": 5316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504427525316327e-07,
+      "loss": 1.8975,
+      "step": 5317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501599862476986e-07,
+      "loss": 1.8165,
+      "step": 5318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1498772076799589e-07,
+      "loss": 1.9882,
+      "step": 5319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1495944168515475e-07,
+      "loss": 1.646,
+      "step": 5320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493116137855971e-07,
+      "loss": 1.8565,
+      "step": 5321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490287985052428e-07,
+      "loss": 1.8329,
+      "step": 5322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1487459710336198e-07,
+      "loss": 1.8686,
+      "step": 5323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148463131393865e-07,
+      "loss": 1.8854,
+      "step": 5324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481802796091162e-07,
+      "loss": 1.7808,
+      "step": 5325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478974157025119e-07,
+      "loss": 1.9062,
+      "step": 5326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476145396971916e-07,
+      "loss": 1.5252,
+      "step": 5327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473316516162958e-07,
+      "loss": 1.6832,
+      "step": 5328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470487514829665e-07,
+      "loss": 1.7304,
+      "step": 5329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467658393203463e-07,
+      "loss": 1.7086,
+      "step": 5330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464829151515782e-07,
+      "loss": 1.8345,
+      "step": 5331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461999789998072e-07,
+      "loss": 1.5053,
+      "step": 5332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459170308881787e-07,
+      "loss": 1.687,
+      "step": 5333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456340708398394e-07,
+      "loss": 1.729,
+      "step": 5334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1453510988779364e-07,
+      "loss": 1.6152,
+      "step": 5335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1450681150256185e-07,
+      "loss": 1.8392,
+      "step": 5336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144785119306035e-07,
+      "loss": 1.7724,
+      "step": 5337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445021117423364e-07,
+      "loss": 1.8826,
+      "step": 5338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442190923576738e-07,
+      "loss": 1.8095,
+      "step": 5339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439360611751996e-07,
+      "loss": 1.7459,
+      "step": 5340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436530182180674e-07,
+      "loss": 1.776,
+      "step": 5341
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433699635094314e-07,
+      "loss": 1.7369,
+      "step": 5342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430868970724468e-07,
+      "loss": 1.7571,
+      "step": 5343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428038189302695e-07,
+      "loss": 1.9473,
+      "step": 5344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142520729106057e-07,
+      "loss": 1.5991,
+      "step": 5345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142237627622967e-07,
+      "loss": 1.9169,
+      "step": 5346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1419545145041592e-07,
+      "loss": 1.7728,
+      "step": 5347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141671389772793e-07,
+      "loss": 1.7696,
+      "step": 5348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413882534520296e-07,
+      "loss": 1.7817,
+      "step": 5349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411051055650309e-07,
+      "loss": 1.7194,
+      "step": 5350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408219461349598e-07,
+      "loss": 1.8189,
+      "step": 5351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14053877518498e-07,
+      "loss": 1.6463,
+      "step": 5352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140255592738256e-07,
+      "loss": 1.6477,
+      "step": 5353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399723988179541e-07,
+      "loss": 1.6253,
+      "step": 5354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396891934472403e-07,
+      "loss": 1.6733,
+      "step": 5355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139405976649283e-07,
+      "loss": 1.8278,
+      "step": 5356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391227484472494e-07,
+      "loss": 1.9343,
+      "step": 5357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13883950886431e-07,
+      "loss": 1.8023,
+      "step": 5358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385562579236347e-07,
+      "loss": 1.6776,
+      "step": 5359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382729956483949e-07,
+      "loss": 1.7814,
+      "step": 5360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379897220617629e-07,
+      "loss": 1.76,
+      "step": 5361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1377064371869115e-07,
+      "loss": 1.8104,
+      "step": 5362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374231410470149e-07,
+      "loss": 1.8117,
+      "step": 5363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371398336652484e-07,
+      "loss": 1.8185,
+      "step": 5364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368565150647875e-07,
+      "loss": 1.6489,
+      "step": 5365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136573185268809e-07,
+      "loss": 1.75,
+      "step": 5366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136289844300491e-07,
+      "loss": 1.8709,
+      "step": 5367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1360064921830119e-07,
+      "loss": 1.6872,
+      "step": 5368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1357231289395514e-07,
+      "loss": 1.8821,
+      "step": 5369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354397545932895e-07,
+      "loss": 1.5871,
+      "step": 5370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351563691674085e-07,
+      "loss": 1.7701,
+      "step": 5371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348729726850896e-07,
+      "loss": 1.8765,
+      "step": 5372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134589565169517e-07,
+      "loss": 1.7747,
+      "step": 5373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343061466438737e-07,
+      "loss": 1.7474,
+      "step": 5374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340227171313459e-07,
+      "loss": 1.611,
+      "step": 5375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337392766551186e-07,
+      "loss": 1.6464,
+      "step": 5376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133455825238379e-07,
+      "loss": 1.7806,
+      "step": 5377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1331723629043144e-07,
+      "loss": 1.7371,
+      "step": 5378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328888896761135e-07,
+      "loss": 1.8756,
+      "step": 5379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1326054055769665e-07,
+      "loss": 1.7426,
+      "step": 5380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1323219106300625e-07,
+      "loss": 1.7041,
+      "step": 5381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.132038404858594e-07,
+      "loss": 1.8718,
+      "step": 5382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1317548882857519e-07,
+      "loss": 1.6948,
+      "step": 5383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1314713609347304e-07,
+      "loss": 1.6269,
+      "step": 5384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1311878228287224e-07,
+      "loss": 1.8494,
+      "step": 5385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1309042739909233e-07,
+      "loss": 1.8791,
+      "step": 5386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1306207144445283e-07,
+      "loss": 1.6216,
+      "step": 5387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1303371442127348e-07,
+      "loss": 1.7491,
+      "step": 5388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.130053563318739e-07,
+      "loss": 1.7197,
+      "step": 5389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1297699717857401e-07,
+      "loss": 1.7758,
+      "step": 5390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1294863696369363e-07,
+      "loss": 1.6333,
+      "step": 5391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1292027568955285e-07,
+      "loss": 1.9473,
+      "step": 5392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289191335847177e-07,
+      "loss": 1.7303,
+      "step": 5393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1286354997277048e-07,
+      "loss": 1.6741,
+      "step": 5394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1283518553476929e-07,
+      "loss": 1.7363,
+      "step": 5395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280682004678852e-07,
+      "loss": 1.8184,
+      "step": 5396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277845351114867e-07,
+      "loss": 1.6435,
+      "step": 5397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275008593017018e-07,
+      "loss": 1.7802,
+      "step": 5398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127217173061737e-07,
+      "loss": 1.8456,
+      "step": 5399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269334764147987e-07,
+      "loss": 1.7834,
+      "step": 5400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266497693840955e-07,
+      "loss": 1.8072,
+      "step": 5401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263660519928351e-07,
+      "loss": 1.7705,
+      "step": 5402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260823242642274e-07,
+      "loss": 1.8185,
+      "step": 5403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257985862214825e-07,
+      "loss": 1.8904,
+      "step": 5404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255148378878119e-07,
+      "loss": 1.5907,
+      "step": 5405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125231079286427e-07,
+      "loss": 1.586,
+      "step": 5406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.124947310440541e-07,
+      "loss": 1.752,
+      "step": 5407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246635313733672e-07,
+      "loss": 1.6813,
+      "step": 5408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243797421081206e-07,
+      "loss": 1.823,
+      "step": 5409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1240959426680165e-07,
+      "loss": 1.5829,
+      "step": 5410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238121330762704e-07,
+      "loss": 1.6112,
+      "step": 5411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235283133560997e-07,
+      "loss": 1.845,
+      "step": 5412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232444835307224e-07,
+      "loss": 1.6945,
+      "step": 5413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229606436233572e-07,
+      "loss": 1.7872,
+      "step": 5414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226767936572227e-07,
+      "loss": 2.0158,
+      "step": 5415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223929336555404e-07,
+      "loss": 1.8516,
+      "step": 5416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221090636415302e-07,
+      "loss": 1.8765,
+      "step": 5417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121825183638415e-07,
+      "loss": 1.5932,
+      "step": 5418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121541293669417e-07,
+      "loss": 1.686,
+      "step": 5419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212573937577599e-07,
+      "loss": 1.6512,
+      "step": 5420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120973483926668e-07,
+      "loss": 1.7244,
+      "step": 5421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206895641993668e-07,
+      "loss": 1.6629,
+      "step": 5422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120405634599082e-07,
+      "loss": 1.5476,
+      "step": 5423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201216951490404e-07,
+      "loss": 1.6997,
+      "step": 5424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198377458724697e-07,
+      "loss": 1.7721,
+      "step": 5425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195537867925981e-07,
+      "loss": 1.7705,
+      "step": 5426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192698179326554e-07,
+      "loss": 1.882,
+      "step": 5427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189858393158708e-07,
+      "loss": 1.6354,
+      "step": 5428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187018509654756e-07,
+      "loss": 1.8227,
+      "step": 5429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184178529047013e-07,
+      "loss": 1.7727,
+      "step": 5430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181338451567804e-07,
+      "loss": 1.7294,
+      "step": 5431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117849827744946e-07,
+      "loss": 1.7119,
+      "step": 5432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175658006924318e-07,
+      "loss": 1.721,
+      "step": 5433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117281764022473e-07,
+      "loss": 1.7203,
+      "step": 5434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1169977177583051e-07,
+      "loss": 1.7219,
+      "step": 5435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167136619231645e-07,
+      "loss": 1.6008,
+      "step": 5436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1164295965402879e-07,
+      "loss": 1.8238,
+      "step": 5437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161455216329137e-07,
+      "loss": 1.7318,
+      "step": 5438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11586143722428e-07,
+      "loss": 1.7439,
+      "step": 5439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155773433376272e-07,
+      "loss": 1.9177,
+      "step": 5440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152932399961945e-07,
+      "loss": 1.8221,
+      "step": 5441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1150091272232237e-07,
+      "loss": 1.6158,
+      "step": 5442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147250050419562e-07,
+      "loss": 1.7053,
+      "step": 5443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144408734756345e-07,
+      "loss": 1.5229,
+      "step": 5444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114156732547502e-07,
+      "loss": 1.8024,
+      "step": 5445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138725822808028e-07,
+      "loss": 1.6747,
+      "step": 5446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1135884226987819e-07,
+      "loss": 1.6889,
+      "step": 5447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133042538246849e-07,
+      "loss": 1.6612,
+      "step": 5448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113020075681758e-07,
+      "loss": 1.6287,
+      "step": 5449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127358882932483e-07,
+      "loss": 1.6137,
+      "step": 5450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124516916824038e-07,
+      "loss": 1.7817,
+      "step": 5451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121674858724733e-07,
+      "loss": 1.6705,
+      "step": 5452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1118832708867058e-07,
+      "loss": 1.7577,
+      "step": 5453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1115990467483518e-07,
+      "loss": 1.8377,
+      "step": 5454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113148134806621e-07,
+      "loss": 1.7264,
+      "step": 5455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110305711068882e-07,
+      "loss": 1.7641,
+      "step": 5456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107463196502827e-07,
+      "loss": 1.7705,
+      "step": 5457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104620591340985e-07,
+      "loss": 1.6944,
+      "step": 5458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101777895815894e-07,
+      "loss": 1.7469,
+      "step": 5459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098935110160104e-07,
+      "loss": 1.7474,
+      "step": 5460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096092234606167e-07,
+      "loss": 1.6574,
+      "step": 5461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093249269386644e-07,
+      "loss": 1.7133,
+      "step": 5462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090406214734101e-07,
+      "loss": 1.8486,
+      "step": 5463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1087563070881115e-07,
+      "loss": 1.7329,
+      "step": 5464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1084719838060265e-07,
+      "loss": 1.8559,
+      "step": 5465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081876516504151e-07,
+      "loss": 1.9472,
+      "step": 5466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1079033106445361e-07,
+      "loss": 1.8126,
+      "step": 5467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076189608116502e-07,
+      "loss": 1.8402,
+      "step": 5468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073346021750185e-07,
+      "loss": 1.7308,
+      "step": 5469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070502347579033e-07,
+      "loss": 1.6332,
+      "step": 5470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1067658585835669e-07,
+      "loss": 1.9095,
+      "step": 5471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064814736752725e-07,
+      "loss": 1.8513,
+      "step": 5472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1061970800562843e-07,
+      "loss": 1.7268,
+      "step": 5473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059126777498672e-07,
+      "loss": 1.7116,
+      "step": 5474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056282667792867e-07,
+      "loss": 1.7927,
+      "step": 5475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1053438471678085e-07,
+      "loss": 1.8612,
+      "step": 5476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050594189386999e-07,
+      "loss": 1.8636,
+      "step": 5477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047749821152284e-07,
+      "loss": 1.6165,
+      "step": 5478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1044905367206625e-07,
+      "loss": 1.7852,
+      "step": 5479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042060827782706e-07,
+      "loss": 1.6663,
+      "step": 5480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103921620311323e-07,
+      "loss": 1.7539,
+      "step": 5481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1036371493430899e-07,
+      "loss": 2.0292,
+      "step": 5482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033526698968423e-07,
+      "loss": 1.6273,
+      "step": 5483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103068181995852e-07,
+      "loss": 1.4813,
+      "step": 5484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027836856633915e-07,
+      "loss": 1.9242,
+      "step": 5485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024991809227338e-07,
+      "loss": 1.6544,
+      "step": 5486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022146677971528e-07,
+      "loss": 1.5662,
+      "step": 5487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1019301463099237e-07,
+      "loss": 1.8471,
+      "step": 5488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1016456164843209e-07,
+      "loss": 1.7373,
+      "step": 5489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013610783436203e-07,
+      "loss": 1.9119,
+      "step": 5490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010765319110988e-07,
+      "loss": 1.7341,
+      "step": 5491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007919772100341e-07,
+      "loss": 1.7646,
+      "step": 5492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1005074142637032e-07,
+      "loss": 1.6387,
+      "step": 5493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.100222843095385e-07,
+      "loss": 1.6993,
+      "step": 5494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0999382637283591e-07,
+      "loss": 1.6467,
+      "step": 5495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0996536761859054e-07,
+      "loss": 1.4899,
+      "step": 5496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0993690804913042e-07,
+      "loss": 1.7096,
+      "step": 5497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0990844766678371e-07,
+      "loss": 1.8133,
+      "step": 5498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.098799864738786e-07,
+      "loss": 1.7904,
+      "step": 5499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0985152447274333e-07,
+      "loss": 1.7657,
+      "step": 5500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0982306166570627e-07,
+      "loss": 1.6281,
+      "step": 5501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0979459805509577e-07,
+      "loss": 1.8198,
+      "step": 5502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.097661336432403e-07,
+      "loss": 1.6392,
+      "step": 5503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.097376684324684e-07,
+      "loss": 1.793,
+      "step": 5504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0970920242510866e-07,
+      "loss": 1.6622,
+      "step": 5505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968073562348972e-07,
+      "loss": 1.6171,
+      "step": 5506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096522680299403e-07,
+      "loss": 1.6484,
+      "step": 5507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0962379964678921e-07,
+      "loss": 1.6712,
+      "step": 5508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959533047636528e-07,
+      "loss": 1.8462,
+      "step": 5509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956686052099742e-07,
+      "loss": 1.8189,
+      "step": 5510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953838978301461e-07,
+      "loss": 1.7724,
+      "step": 5511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950991826474592e-07,
+      "loss": 1.7286,
+      "step": 5512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0948144596852042e-07,
+      "loss": 1.4995,
+      "step": 5513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094529728966673e-07,
+      "loss": 1.7134,
+      "step": 5514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942449905151579e-07,
+      "loss": 1.663,
+      "step": 5515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093960244353952e-07,
+      "loss": 1.6972,
+      "step": 5516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0936754905063483e-07,
+      "loss": 1.7258,
+      "step": 5517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933907289956422e-07,
+      "loss": 1.7095,
+      "step": 5518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931059598451273e-07,
+      "loss": 1.6703,
+      "step": 5519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928211830780995e-07,
+      "loss": 1.5541,
+      "step": 5520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925363987178553e-07,
+      "loss": 1.654,
+      "step": 5521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0922516067876912e-07,
+      "loss": 1.8128,
+      "step": 5522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0919668073109043e-07,
+      "loss": 1.8758,
+      "step": 5523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916820003107927e-07,
+      "loss": 1.9264,
+      "step": 5524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091397185810655e-07,
+      "loss": 1.7029,
+      "step": 5525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0911123638337905e-07,
+      "loss": 1.757,
+      "step": 5526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908275344034992e-07,
+      "loss": 1.8665,
+      "step": 5527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905426975430805e-07,
+      "loss": 1.6678,
+      "step": 5528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902578532758367e-07,
+      "loss": 1.8115,
+      "step": 5529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0899730016250685e-07,
+      "loss": 1.7367,
+      "step": 5530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089688142614079e-07,
+      "loss": 1.8895,
+      "step": 5531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0894032762661698e-07,
+      "loss": 1.6384,
+      "step": 5532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0891184026046456e-07,
+      "loss": 1.7822,
+      "step": 5533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888335216528097e-07,
+      "loss": 1.8697,
+      "step": 5534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885486334339668e-07,
+      "loss": 1.6294,
+      "step": 5535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882637379714222e-07,
+      "loss": 1.6063,
+      "step": 5536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0879788352884818e-07,
+      "loss": 1.7141,
+      "step": 5537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0876939254084518e-07,
+      "loss": 1.8924,
+      "step": 5538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0874090083546395e-07,
+      "loss": 1.8323,
+      "step": 5539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871240841503524e-07,
+      "loss": 1.5642,
+      "step": 5540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086839152818898e-07,
+      "loss": 1.6259,
+      "step": 5541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865542143835862e-07,
+      "loss": 1.9338,
+      "step": 5542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862692688677252e-07,
+      "loss": 1.6798,
+      "step": 5543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085984316294626e-07,
+      "loss": 1.6163,
+      "step": 5544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0856993566875978e-07,
+      "loss": 1.7697,
+      "step": 5545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854143900699531e-07,
+      "loss": 1.8112,
+      "step": 5546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851294164650025e-07,
+      "loss": 1.8779,
+      "step": 5547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848444358960584e-07,
+      "loss": 1.9965,
+      "step": 5548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845594483864335e-07,
+      "loss": 1.9247,
+      "step": 5549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084274453959442e-07,
+      "loss": 1.7263,
+      "step": 5550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0839894526383966e-07,
+      "loss": 1.8058,
+      "step": 5551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837044444466124e-07,
+      "loss": 1.7297,
+      "step": 5552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834194294074042e-07,
+      "loss": 1.7736,
+      "step": 5553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831344075440878e-07,
+      "loss": 1.838,
+      "step": 5554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828493788799795e-07,
+      "loss": 1.7409,
+      "step": 5555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825643434383955e-07,
+      "loss": 1.703,
+      "step": 5556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822793012426536e-07,
+      "loss": 1.7932,
+      "step": 5557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0819942523160713e-07,
+      "loss": 1.7323,
+      "step": 5558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0817091966819669e-07,
+      "loss": 1.8819,
+      "step": 5559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814241343636594e-07,
+      "loss": 1.8219,
+      "step": 5560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811390653844688e-07,
+      "loss": 1.7299,
+      "step": 5561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0808539897677137e-07,
+      "loss": 1.7642,
+      "step": 5562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805689075367164e-07,
+      "loss": 1.8352,
+      "step": 5563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0802838187147969e-07,
+      "loss": 1.7498,
+      "step": 5564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0799987233252772e-07,
+      "loss": 1.7672,
+      "step": 5565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797136213914788e-07,
+      "loss": 1.6768,
+      "step": 5566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794285129367251e-07,
+      "loss": 1.7682,
+      "step": 5567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791433979843396e-07,
+      "loss": 1.7238,
+      "step": 5568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788582765576456e-07,
+      "loss": 1.7826,
+      "step": 5569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078573148679967e-07,
+      "loss": 1.868,
+      "step": 5570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782880143746294e-07,
+      "loss": 1.7723,
+      "step": 5571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780028736649579e-07,
+      "loss": 1.656,
+      "step": 5572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777177265742781e-07,
+      "loss": 1.6987,
+      "step": 5573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774325731259164e-07,
+      "loss": 1.7974,
+      "step": 5574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771474133432001e-07,
+      "loss": 1.7552,
+      "step": 5575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768622472494566e-07,
+      "loss": 1.7693,
+      "step": 5576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765770748680137e-07,
+      "loss": 1.8029,
+      "step": 5577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762918962221999e-07,
+      "loss": 1.7721,
+      "step": 5578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760067113353434e-07,
+      "loss": 1.753,
+      "step": 5579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075721520230775e-07,
+      "loss": 1.7402,
+      "step": 5580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754363229318242e-07,
+      "loss": 1.7616,
+      "step": 5581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751511194618212e-07,
+      "loss": 1.7519,
+      "step": 5582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748659098440974e-07,
+      "loss": 1.7445,
+      "step": 5583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074580694101984e-07,
+      "loss": 1.657,
+      "step": 5584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742954722588131e-07,
+      "loss": 1.5718,
+      "step": 5585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740102443379173e-07,
+      "loss": 1.9762,
+      "step": 5586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737250103626296e-07,
+      "loss": 1.7318,
+      "step": 5587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734397703562834e-07,
+      "loss": 1.6051,
+      "step": 5588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731545243422129e-07,
+      "loss": 1.6939,
+      "step": 5589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728692723437525e-07,
+      "loss": 1.7854,
+      "step": 5590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725840143842369e-07,
+      "loss": 1.5208,
+      "step": 5591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722987504870018e-07,
+      "loss": 1.8205,
+      "step": 5592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072013480675383e-07,
+      "loss": 1.8119,
+      "step": 5593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0717282049727176e-07,
+      "loss": 1.6963,
+      "step": 5594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714429234023417e-07,
+      "loss": 1.5726,
+      "step": 5595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711576359875928e-07,
+      "loss": 1.7886,
+      "step": 5596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708723427518094e-07,
+      "loss": 1.5632,
+      "step": 5597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705870437183294e-07,
+      "loss": 1.8212,
+      "step": 5598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703017389104914e-07,
+      "loss": 1.8812,
+      "step": 5599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700164283516348e-07,
+      "loss": 1.7134,
+      "step": 5600
+    },
+    {
+      "epoch": 0.49,
+      "step": 5600,
+      "total_flos": 4.621907792557507e+18,
+      "train_loss": 0.0,
+      "train_runtime": 1142.9323,
+      "train_samples_per_second": 0.001,
+      "train_steps_per_second": 0.001
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 4.621907792557507e+18,
+  "trial_name": null,
+  "trial_params": null
+}