diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5275 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.9928673323823105,
+  "global_step": 875,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98857142857143e-06,
+      "loss": 2.0202,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.977142857142858e-06,
+      "loss": 1.9754,
+      "step": 2
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.965714285714286e-06,
+      "loss": 1.9429,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.954285714285715e-06,
+      "loss": 2.0103,
+      "step": 4
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.942857142857145e-06,
+      "loss": 1.9785,
+      "step": 5
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.931428571428571e-06,
+      "loss": 2.012,
+      "step": 6
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 1.9989,
+      "step": 7
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.90857142857143e-06,
+      "loss": 1.9499,
+      "step": 8
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.897142857142858e-06,
+      "loss": 1.9444,
+      "step": 9
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.885714285714287e-06,
+      "loss": 1.9597,
+      "step": 10
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.874285714285715e-06,
+      "loss": 1.9654,
+      "step": 11
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.862857142857144e-06,
+      "loss": 1.9811,
+      "step": 12
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.851428571428572e-06,
+      "loss": 1.9988,
+      "step": 13
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84e-06,
+      "loss": 1.9983,
+      "step": 14
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828571428571429e-06,
+      "loss": 1.9565,
+      "step": 15
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81714285714286e-06,
+      "loss": 1.876,
+      "step": 16
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.805714285714286e-06,
+      "loss": 1.8638,
+      "step": 17
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.794285714285714e-06,
+      "loss": 1.9153,
+      "step": 18
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.782857142857145e-06,
+      "loss": 1.9132,
+      "step": 19
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.771428571428571e-06,
+      "loss": 1.9551,
+      "step": 20
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 1.8265,
+      "step": 21
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74857142857143e-06,
+      "loss": 1.9253,
+      "step": 22
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737142857142858e-06,
+      "loss": 1.8201,
+      "step": 23
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.725714285714287e-06,
+      "loss": 1.9096,
+      "step": 24
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 1.7755,
+      "step": 25
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.702857142857144e-06,
+      "loss": 1.8151,
+      "step": 26
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.691428571428572e-06,
+      "loss": 1.8359,
+      "step": 27
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68e-06,
+      "loss": 1.747,
+      "step": 28
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668571428571429e-06,
+      "loss": 1.8638,
+      "step": 29
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.657142857142859e-06,
+      "loss": 1.8103,
+      "step": 30
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.645714285714286e-06,
+      "loss": 1.805,
+      "step": 31
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.634285714285714e-06,
+      "loss": 1.7564,
+      "step": 32
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.622857142857144e-06,
+      "loss": 1.6883,
+      "step": 33
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.611428571428573e-06,
+      "loss": 1.8681,
+      "step": 34
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.0254,
+      "step": 35
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.58857142857143e-06,
+      "loss": 1.8567,
+      "step": 36
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.577142857142858e-06,
+      "loss": 1.9897,
+      "step": 37
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.565714285714287e-06,
+      "loss": 1.9687,
+      "step": 38
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.554285714285715e-06,
+      "loss": 2.0195,
+      "step": 39
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.542857142857143e-06,
+      "loss": 1.8665,
+      "step": 40
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.531428571428572e-06,
+      "loss": 1.7672,
+      "step": 41
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.52e-06,
+      "loss": 1.8378,
+      "step": 42
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.508571428571429e-06,
+      "loss": 1.8454,
+      "step": 43
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.497142857142859e-06,
+      "loss": 1.7435,
+      "step": 44
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.485714285714287e-06,
+      "loss": 1.7557,
+      "step": 45
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.474285714285714e-06,
+      "loss": 1.7289,
+      "step": 46
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.462857142857144e-06,
+      "loss": 1.6987,
+      "step": 47
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.451428571428573e-06,
+      "loss": 1.7178,
+      "step": 48
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 1.7105,
+      "step": 49
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 1.8543,
+      "step": 50
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.417142857142858e-06,
+      "loss": 1.7117,
+      "step": 51
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.405714285714286e-06,
+      "loss": 1.8157,
+      "step": 52
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.394285714285715e-06,
+      "loss": 1.6601,
+      "step": 53
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.382857142857143e-06,
+      "loss": 1.6842,
+      "step": 54
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.371428571428572e-06,
+      "loss": 1.7915,
+      "step": 55
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 1.7107,
+      "step": 56
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.348571428571429e-06,
+      "loss": 1.7052,
+      "step": 57
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.337142857142859e-06,
+      "loss": 1.6986,
+      "step": 58
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.325714285714287e-06,
+      "loss": 1.7981,
+      "step": 59
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.314285714285714e-06,
+      "loss": 1.6798,
+      "step": 60
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.302857142857144e-06,
+      "loss": 1.753,
+      "step": 61
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.291428571428572e-06,
+      "loss": 1.6584,
+      "step": 62
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 1.711,
+      "step": 63
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.26857142857143e-06,
+      "loss": 1.6625,
+      "step": 64
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.257142857142858e-06,
+      "loss": 1.6937,
+      "step": 65
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.245714285714286e-06,
+      "loss": 1.6583,
+      "step": 66
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.234285714285715e-06,
+      "loss": 1.6619,
+      "step": 67
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.222857142857143e-06,
+      "loss": 1.6572,
+      "step": 68
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.211428571428572e-06,
+      "loss": 1.6537,
+      "step": 69
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.6308,
+      "step": 70
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.188571428571428e-06,
+      "loss": 1.6518,
+      "step": 71
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.177142857142859e-06,
+      "loss": 1.7237,
+      "step": 72
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.165714285714287e-06,
+      "loss": 1.6625,
+      "step": 73
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.154285714285715e-06,
+      "loss": 1.6398,
+      "step": 74
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 1.6435,
+      "step": 75
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.131428571428572e-06,
+      "loss": 1.6497,
+      "step": 76
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.12e-06,
+      "loss": 1.6057,
+      "step": 77
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.10857142857143e-06,
+      "loss": 1.594,
+      "step": 78
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.097142857142858e-06,
+      "loss": 1.5912,
+      "step": 79
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.085714285714286e-06,
+      "loss": 1.6062,
+      "step": 80
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.074285714285716e-06,
+      "loss": 1.6505,
+      "step": 81
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.062857142857143e-06,
+      "loss": 1.6325,
+      "step": 82
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.051428571428571e-06,
+      "loss": 1.6868,
+      "step": 83
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 1.6513,
+      "step": 84
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.028571428571428e-06,
+      "loss": 1.5926,
+      "step": 85
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.017142857142858e-06,
+      "loss": 1.5749,
+      "step": 86
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.005714285714287e-06,
+      "loss": 1.5802,
+      "step": 87
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.994285714285715e-06,
+      "loss": 1.5887,
+      "step": 88
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.982857142857144e-06,
+      "loss": 1.5737,
+      "step": 89
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.971428571428572e-06,
+      "loss": 1.5834,
+      "step": 90
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.96e-06,
+      "loss": 1.6029,
+      "step": 91
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.948571428571429e-06,
+      "loss": 1.5517,
+      "step": 92
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.937142857142857e-06,
+      "loss": 1.5611,
+      "step": 93
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.925714285714286e-06,
+      "loss": 1.59,
+      "step": 94
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.914285714285716e-06,
+      "loss": 1.5953,
+      "step": 95
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902857142857143e-06,
+      "loss": 1.5782,
+      "step": 96
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891428571428571e-06,
+      "loss": 1.6462,
+      "step": 97
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 1.5581,
+      "step": 98
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.86857142857143e-06,
+      "loss": 1.6,
+      "step": 99
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.857142857142858e-06,
+      "loss": 1.5814,
+      "step": 100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.845714285714287e-06,
+      "loss": 1.5749,
+      "step": 101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.834285714285715e-06,
+      "loss": 1.5351,
+      "step": 102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.822857142857144e-06,
+      "loss": 1.56,
+      "step": 103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.811428571428572e-06,
+      "loss": 1.5687,
+      "step": 104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.8e-06,
+      "loss": 1.5763,
+      "step": 105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.788571428571429e-06,
+      "loss": 1.5875,
+      "step": 106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.777142857142857e-06,
+      "loss": 1.6139,
+      "step": 107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.765714285714286e-06,
+      "loss": 1.5734,
+      "step": 108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.754285714285716e-06,
+      "loss": 1.5117,
+      "step": 109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.742857142857144e-06,
+      "loss": 1.5611,
+      "step": 110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.731428571428571e-06,
+      "loss": 1.5481,
+      "step": 111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 1.5578,
+      "step": 112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.70857142857143e-06,
+      "loss": 1.6075,
+      "step": 113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.697142857142858e-06,
+      "loss": 1.6112,
+      "step": 114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.685714285714287e-06,
+      "loss": 1.4582,
+      "step": 115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.674285714285715e-06,
+      "loss": 1.5832,
+      "step": 116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.662857142857143e-06,
+      "loss": 1.7716,
+      "step": 117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.651428571428572e-06,
+      "loss": 1.9526,
+      "step": 118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.64e-06,
+      "loss": 1.7239,
+      "step": 119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.628571428571429e-06,
+      "loss": 1.6897,
+      "step": 120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.617142857142859e-06,
+      "loss": 1.5891,
+      "step": 121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.605714285714286e-06,
+      "loss": 1.7339,
+      "step": 122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.594285714285716e-06,
+      "loss": 1.7173,
+      "step": 123
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.582857142857144e-06,
+      "loss": 1.596,
+      "step": 124
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.548,
+      "step": 125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 1.5142,
+      "step": 126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.54857142857143e-06,
+      "loss": 1.5413,
+      "step": 127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.537142857142858e-06,
+      "loss": 1.5976,
+      "step": 128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.525714285714286e-06,
+      "loss": 1.5416,
+      "step": 129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.514285714285715e-06,
+      "loss": 1.5091,
+      "step": 130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.502857142857143e-06,
+      "loss": 1.56,
+      "step": 131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.491428571428572e-06,
+      "loss": 1.5188,
+      "step": 132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.48e-06,
+      "loss": 1.5378,
+      "step": 133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.468571428571429e-06,
+      "loss": 1.5595,
+      "step": 134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.457142857142859e-06,
+      "loss": 1.5591,
+      "step": 135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.445714285714285e-06,
+      "loss": 1.565,
+      "step": 136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.434285714285716e-06,
+      "loss": 1.5318,
+      "step": 137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.422857142857144e-06,
+      "loss": 1.6051,
+      "step": 138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.411428571428572e-06,
+      "loss": 1.6247,
+      "step": 139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.5086,
+      "step": 140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.38857142857143e-06,
+      "loss": 1.5235,
+      "step": 141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.377142857142858e-06,
+      "loss": 1.5342,
+      "step": 142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.365714285714286e-06,
+      "loss": 1.4861,
+      "step": 143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.354285714285715e-06,
+      "loss": 1.4781,
+      "step": 144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.342857142857143e-06,
+      "loss": 1.5155,
+      "step": 145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.331428571428573e-06,
+      "loss": 1.513,
+      "step": 146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.32e-06,
+      "loss": 1.4958,
+      "step": 147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.308571428571428e-06,
+      "loss": 1.4728,
+      "step": 148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.297142857142859e-06,
+      "loss": 1.504,
+      "step": 149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.285714285714287e-06,
+      "loss": 1.5638,
+      "step": 150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.274285714285715e-06,
+      "loss": 1.4473,
+      "step": 151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.262857142857144e-06,
+      "loss": 1.4839,
+      "step": 152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.251428571428572e-06,
+      "loss": 1.5174,
+      "step": 153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.24e-06,
+      "loss": 1.5251,
+      "step": 154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.22857142857143e-06,
+      "loss": 1.4409,
+      "step": 155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.217142857142858e-06,
+      "loss": 1.469,
+      "step": 156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.205714285714286e-06,
+      "loss": 1.4872,
+      "step": 157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.194285714285714e-06,
+      "loss": 1.4819,
+      "step": 158
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.182857142857143e-06,
+      "loss": 1.4822,
+      "step": 159
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.171428571428573e-06,
+      "loss": 1.4797,
+      "step": 160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.16e-06,
+      "loss": 1.4872,
+      "step": 161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.148571428571428e-06,
+      "loss": 1.4472,
+      "step": 162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.137142857142858e-06,
+      "loss": 1.4962,
+      "step": 163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.125714285714287e-06,
+      "loss": 1.5026,
+      "step": 164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.114285714285715e-06,
+      "loss": 1.5028,
+      "step": 165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.102857142857144e-06,
+      "loss": 1.5236,
+      "step": 166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.091428571428572e-06,
+      "loss": 1.4764,
+      "step": 167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.08e-06,
+      "loss": 1.5006,
+      "step": 168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.068571428571429e-06,
+      "loss": 1.4522,
+      "step": 169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.057142857142857e-06,
+      "loss": 1.4775,
+      "step": 170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.045714285714286e-06,
+      "loss": 1.4455,
+      "step": 171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.034285714285714e-06,
+      "loss": 1.4824,
+      "step": 172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.022857142857143e-06,
+      "loss": 1.4554,
+      "step": 173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.011428571428573e-06,
+      "loss": 1.4831,
+      "step": 174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4786,
+      "step": 175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.988571428571428e-06,
+      "loss": 1.9457,
+      "step": 176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.977142857142858e-06,
+      "loss": 1.8681,
+      "step": 177
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.965714285714287e-06,
+      "loss": 1.8721,
+      "step": 178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.954285714285715e-06,
+      "loss": 1.8807,
+      "step": 179
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.942857142857144e-06,
+      "loss": 1.8907,
+      "step": 180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.931428571428572e-06,
+      "loss": 1.868,
+      "step": 181
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.92e-06,
+      "loss": 1.9417,
+      "step": 182
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.908571428571429e-06,
+      "loss": 1.8749,
+      "step": 183
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.897142857142857e-06,
+      "loss": 1.8543,
+      "step": 184
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.885714285714286e-06,
+      "loss": 1.9108,
+      "step": 185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.874285714285716e-06,
+      "loss": 1.8662,
+      "step": 186
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.862857142857143e-06,
+      "loss": 1.8439,
+      "step": 187
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.851428571428573e-06,
+      "loss": 1.8629,
+      "step": 188
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 1.8611,
+      "step": 189
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.828571428571428e-06,
+      "loss": 1.9209,
+      "step": 190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.817142857142858e-06,
+      "loss": 1.8903,
+      "step": 191
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.805714285714286e-06,
+      "loss": 1.8397,
+      "step": 192
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.794285714285715e-06,
+      "loss": 1.8394,
+      "step": 193
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.782857142857143e-06,
+      "loss": 1.8822,
+      "step": 194
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.771428571428572e-06,
+      "loss": 1.8179,
+      "step": 195
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.76e-06,
+      "loss": 1.8568,
+      "step": 196
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.74857142857143e-06,
+      "loss": 1.8766,
+      "step": 197
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.737142857142857e-06,
+      "loss": 1.8984,
+      "step": 198
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.725714285714286e-06,
+      "loss": 1.9151,
+      "step": 199
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.714285714285716e-06,
+      "loss": 1.8517,
+      "step": 200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.702857142857142e-06,
+      "loss": 1.8674,
+      "step": 201
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.691428571428573e-06,
+      "loss": 1.8796,
+      "step": 202
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 1.9014,
+      "step": 203
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.66857142857143e-06,
+      "loss": 1.8437,
+      "step": 204
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.657142857142858e-06,
+      "loss": 1.8858,
+      "step": 205
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.645714285714286e-06,
+      "loss": 1.882,
+      "step": 206
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.634285714285715e-06,
+      "loss": 1.8613,
+      "step": 207
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.622857142857143e-06,
+      "loss": 1.907,
+      "step": 208
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.611428571428572e-06,
+      "loss": 1.9055,
+      "step": 209
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.8942,
+      "step": 210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.588571428571429e-06,
+      "loss": 1.8409,
+      "step": 211
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.577142857142857e-06,
+      "loss": 1.826,
+      "step": 212
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.565714285714286e-06,
+      "loss": 1.8701,
+      "step": 213
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.5542857142857155e-06,
+      "loss": 1.8298,
+      "step": 214
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.542857142857144e-06,
+      "loss": 1.9199,
+      "step": 215
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.5314285714285716e-06,
+      "loss": 1.8758,
+      "step": 216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 1.8709,
+      "step": 217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.508571428571429e-06,
+      "loss": 1.875,
+      "step": 218
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.497142857142857e-06,
+      "loss": 1.9255,
+      "step": 219
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.485714285714286e-06,
+      "loss": 1.8664,
+      "step": 220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.4742857142857154e-06,
+      "loss": 1.8337,
+      "step": 221
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.462857142857144e-06,
+      "loss": 1.9226,
+      "step": 222
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.4514285714285715e-06,
+      "loss": 1.8908,
+      "step": 223
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 1.8781,
+      "step": 224
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 1.9171,
+      "step": 225
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.417142857142857e-06,
+      "loss": 1.8577,
+      "step": 226
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.405714285714286e-06,
+      "loss": 1.8957,
+      "step": 227
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.394285714285715e-06,
+      "loss": 1.9199,
+      "step": 228
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.382857142857144e-06,
+      "loss": 1.949,
+      "step": 229
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.371428571428571e-06,
+      "loss": 1.9293,
+      "step": 230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 1.9471,
+      "step": 231
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.348571428571429e-06,
+      "loss": 1.8477,
+      "step": 232
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.337142857142858e-06,
+      "loss": 1.8893,
+      "step": 233
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.325714285714286e-06,
+      "loss": 1.8764,
+      "step": 234
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.314285714285715e-06,
+      "loss": 1.9136,
+      "step": 235
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.302857142857144e-06,
+      "loss": 1.9147,
+      "step": 236
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.291428571428571e-06,
+      "loss": 1.8703,
+      "step": 237
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 1.9123,
+      "step": 238
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.268571428571429e-06,
+      "loss": 1.9115,
+      "step": 239
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.257142857142858e-06,
+      "loss": 1.9358,
+      "step": 240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.245714285714286e-06,
+      "loss": 1.9283,
+      "step": 241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.234285714285715e-06,
+      "loss": 1.8585,
+      "step": 242
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.222857142857144e-06,
+      "loss": 1.8692,
+      "step": 243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.211428571428573e-06,
+      "loss": 1.8875,
+      "step": 244
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.9205,
+      "step": 245
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.188571428571429e-06,
+      "loss": 1.9152,
+      "step": 246
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.177142857142858e-06,
+      "loss": 1.8396,
+      "step": 247
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.165714285714286e-06,
+      "loss": 1.8779,
+      "step": 248
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.154285714285715e-06,
+      "loss": 1.8801,
+      "step": 249
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 1.8648,
+      "step": 250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.131428571428573e-06,
+      "loss": 1.8415,
+      "step": 251
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 1.9185,
+      "step": 252
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.108571428571429e-06,
+      "loss": 1.8344,
+      "step": 253
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.097142857142858e-06,
+      "loss": 1.8843,
+      "step": 254
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.085714285714286e-06,
+      "loss": 1.8984,
+      "step": 255
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.074285714285715e-06,
+      "loss": 1.8439,
+      "step": 256
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.0628571428571435e-06,
+      "loss": 1.8614,
+      "step": 257
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.051428571428573e-06,
+      "loss": 1.8259,
+      "step": 258
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.04e-06,
+      "loss": 1.8883,
+      "step": 259
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.028571428571429e-06,
+      "loss": 1.9634,
+      "step": 260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.017142857142858e-06,
+      "loss": 1.9071,
+      "step": 261
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.0057142857142865e-06,
+      "loss": 1.9218,
+      "step": 262
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.994285714285715e-06,
+      "loss": 1.8727,
+      "step": 263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.982857142857143e-06,
+      "loss": 1.8501,
+      "step": 264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.971428571428573e-06,
+      "loss": 1.8371,
+      "step": 265
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.96e-06,
+      "loss": 1.8941,
+      "step": 266
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.948571428571429e-06,
+      "loss": 1.8408,
+      "step": 267
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.937142857142858e-06,
+      "loss": 1.9015,
+      "step": 268
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.9257142857142864e-06,
+      "loss": 1.94,
+      "step": 269
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.914285714285715e-06,
+      "loss": 1.8656,
+      "step": 270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.902857142857143e-06,
+      "loss": 1.8429,
+      "step": 271
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.891428571428573e-06,
+      "loss": 1.9209,
+      "step": 272
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.88e-06,
+      "loss": 1.9166,
+      "step": 273
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.868571428571429e-06,
+      "loss": 1.9284,
+      "step": 274
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 1.8919,
+      "step": 275
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.845714285714286e-06,
+      "loss": 1.9217,
+      "step": 276
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.834285714285715e-06,
+      "loss": 1.8765,
+      "step": 277
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.822857142857143e-06,
+      "loss": 1.8971,
+      "step": 278
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.8114285714285725e-06,
+      "loss": 1.9175,
+      "step": 279
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.8901,
+      "step": 280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.7885714285714286e-06,
+      "loss": 1.912,
+      "step": 281
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.777142857142858e-06,
+      "loss": 1.8572,
+      "step": 282
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.765714285714286e-06,
+      "loss": 1.8419,
+      "step": 283
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.754285714285715e-06,
+      "loss": 1.8567,
+      "step": 284
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.742857142857143e-06,
+      "loss": 1.9234,
+      "step": 285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.7314285714285724e-06,
+      "loss": 1.8159,
+      "step": 286
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 1.8715,
+      "step": 287
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.7085714285714285e-06,
+      "loss": 1.8771,
+      "step": 288
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.697142857142858e-06,
+      "loss": 1.836,
+      "step": 289
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.685714285714286e-06,
+      "loss": 1.8798,
+      "step": 290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.6742857142857155e-06,
+      "loss": 1.9188,
+      "step": 291
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.662857142857143e-06,
+      "loss": 1.8665,
+      "step": 292
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.651428571428572e-06,
+      "loss": 1.8801,
+      "step": 293
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 1.8261,
+      "step": 294
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.628571428571428e-06,
+      "loss": 1.8659,
+      "step": 295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.617142857142858e-06,
+      "loss": 1.9501,
+      "step": 296
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.605714285714286e-06,
+      "loss": 1.8535,
+      "step": 297
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.594285714285715e-06,
+      "loss": 1.8963,
+      "step": 298
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.582857142857143e-06,
+      "loss": 1.8672,
+      "step": 299
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.571428571428572e-06,
+      "loss": 1.8863,
+      "step": 300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 1.8532,
+      "step": 301
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.548571428571428e-06,
+      "loss": 1.8971,
+      "step": 302
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.537142857142858e-06,
+      "loss": 1.8611,
+      "step": 303
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.525714285714286e-06,
+      "loss": 1.825,
+      "step": 304
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.514285714285715e-06,
+      "loss": 1.8707,
+      "step": 305
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.502857142857143e-06,
+      "loss": 1.9209,
+      "step": 306
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.491428571428572e-06,
+      "loss": 1.8413,
+      "step": 307
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 1.8566,
+      "step": 308
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.46857142857143e-06,
+      "loss": 1.8916,
+      "step": 309
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.4571428571428575e-06,
+      "loss": 1.9048,
+      "step": 310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.445714285714286e-06,
+      "loss": 1.8584,
+      "step": 311
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.434285714285715e-06,
+      "loss": 1.8349,
+      "step": 312
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.422857142857143e-06,
+      "loss": 1.89,
+      "step": 313
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.411428571428572e-06,
+      "loss": 1.8845,
+      "step": 314
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.7968,
+      "step": 315
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.38857142857143e-06,
+      "loss": 1.8606,
+      "step": 316
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.3771428571428574e-06,
+      "loss": 1.8538,
+      "step": 317
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.365714285714286e-06,
+      "loss": 1.8271,
+      "step": 318
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.354285714285715e-06,
+      "loss": 1.8151,
+      "step": 319
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.342857142857143e-06,
+      "loss": 1.9135,
+      "step": 320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.331428571428572e-06,
+      "loss": 1.8722,
+      "step": 321
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 1.9211,
+      "step": 322
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.30857142857143e-06,
+      "loss": 1.9172,
+      "step": 323
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.297142857142857e-06,
+      "loss": 1.9246,
+      "step": 324
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 1.8079,
+      "step": 325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.274285714285715e-06,
+      "loss": 1.9163,
+      "step": 326
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.2628571428571435e-06,
+      "loss": 1.924,
+      "step": 327
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.251428571428572e-06,
+      "loss": 1.8322,
+      "step": 328
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.24e-06,
+      "loss": 1.9303,
+      "step": 329
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.22857142857143e-06,
+      "loss": 1.9153,
+      "step": 330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.217142857142857e-06,
+      "loss": 1.9472,
+      "step": 331
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.205714285714286e-06,
+      "loss": 1.8898,
+      "step": 332
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.194285714285715e-06,
+      "loss": 1.8602,
+      "step": 333
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1828571428571434e-06,
+      "loss": 1.8837,
+      "step": 334
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.171428571428572e-06,
+      "loss": 1.869,
+      "step": 335
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.16e-06,
+      "loss": 1.8635,
+      "step": 336
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.14857142857143e-06,
+      "loss": 1.8479,
+      "step": 337
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.137142857142858e-06,
+      "loss": 1.842,
+      "step": 338
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.125714285714286e-06,
+      "loss": 1.8694,
+      "step": 339
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.114285714285715e-06,
+      "loss": 1.9413,
+      "step": 340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.102857142857143e-06,
+      "loss": 1.8717,
+      "step": 341
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.091428571428572e-06,
+      "loss": 1.9343,
+      "step": 342
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.08e-06,
+      "loss": 1.8665,
+      "step": 343
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.0685714285714295e-06,
+      "loss": 1.8946,
+      "step": 344
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.057142857142858e-06,
+      "loss": 1.8547,
+      "step": 345
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.0457142857142855e-06,
+      "loss": 1.8472,
+      "step": 346
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.034285714285715e-06,
+      "loss": 1.8862,
+      "step": 347
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.022857142857143e-06,
+      "loss": 1.894,
+      "step": 348
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.011428571428572e-06,
+      "loss": 1.8344,
+      "step": 349
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6e-06,
+      "loss": 1.8955,
+      "step": 350
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.9885714285714294e-06,
+      "loss": 1.4813,
+      "step": 351
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.977142857142858e-06,
+      "loss": 1.5577,
+      "step": 352
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.9657142857142855e-06,
+      "loss": 1.5384,
+      "step": 353
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.954285714285715e-06,
+      "loss": 1.5282,
+      "step": 354
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.942857142857143e-06,
+      "loss": 1.607,
+      "step": 355
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.9314285714285725e-06,
+      "loss": 1.5854,
+      "step": 356
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.92e-06,
+      "loss": 1.5224,
+      "step": 357
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.908571428571429e-06,
+      "loss": 1.5569,
+      "step": 358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.897142857142858e-06,
+      "loss": 1.517,
+      "step": 359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.885714285714285e-06,
+      "loss": 1.5195,
+      "step": 360
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.874285714285715e-06,
+      "loss": 1.506,
+      "step": 361
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.862857142857143e-06,
+      "loss": 1.5612,
+      "step": 362
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.851428571428572e-06,
+      "loss": 1.5164,
+      "step": 363
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.84e-06,
+      "loss": 1.5524,
+      "step": 364
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.828571428571429e-06,
+      "loss": 1.4835,
+      "step": 365
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.817142857142858e-06,
+      "loss": 1.4747,
+      "step": 366
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.805714285714285e-06,
+      "loss": 1.5536,
+      "step": 367
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.794285714285715e-06,
+      "loss": 1.5723,
+      "step": 368
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.782857142857143e-06,
+      "loss": 1.5643,
+      "step": 369
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.771428571428572e-06,
+      "loss": 1.5,
+      "step": 370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.76e-06,
+      "loss": 1.4934,
+      "step": 371
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.748571428571429e-06,
+      "loss": 1.537,
+      "step": 372
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.737142857142858e-06,
+      "loss": 1.5511,
+      "step": 373
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.725714285714287e-06,
+      "loss": 1.5645,
+      "step": 374
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.5472,
+      "step": 375
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.702857142857143e-06,
+      "loss": 1.5441,
+      "step": 376
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.691428571428572e-06,
+      "loss": 1.5462,
+      "step": 377
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.68e-06,
+      "loss": 1.5596,
+      "step": 378
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.668571428571429e-06,
+      "loss": 1.4952,
+      "step": 379
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.6571428571428576e-06,
+      "loss": 1.5553,
+      "step": 380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.645714285714287e-06,
+      "loss": 1.6068,
+      "step": 381
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6342857142857144e-06,
+      "loss": 1.5156,
+      "step": 382
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.622857142857143e-06,
+      "loss": 1.5715,
+      "step": 383
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.611428571428572e-06,
+      "loss": 1.5572,
+      "step": 384
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.4732,
+      "step": 385
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.588571428571429e-06,
+      "loss": 1.4676,
+      "step": 386
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.5771428571428575e-06,
+      "loss": 1.4884,
+      "step": 387
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.565714285714287e-06,
+      "loss": 1.5728,
+      "step": 388
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.554285714285714e-06,
+      "loss": 1.4687,
+      "step": 389
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.542857142857143e-06,
+      "loss": 1.5328,
+      "step": 390
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.531428571428572e-06,
+      "loss": 1.5801,
+      "step": 391
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 1.5507,
+      "step": 392
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.508571428571429e-06,
+      "loss": 1.4823,
+      "step": 393
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.497142857142857e-06,
+      "loss": 1.5014,
+      "step": 394
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.485714285714287e-06,
+      "loss": 1.5224,
+      "step": 395
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.474285714285714e-06,
+      "loss": 1.6566,
+      "step": 396
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.462857142857143e-06,
+      "loss": 1.5042,
+      "step": 397
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.451428571428572e-06,
+      "loss": 1.5212,
+      "step": 398
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 1.4854,
+      "step": 399
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 1.5316,
+      "step": 400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.417142857142857e-06,
+      "loss": 1.4371,
+      "step": 401
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.405714285714287e-06,
+      "loss": 1.5148,
+      "step": 402
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.394285714285715e-06,
+      "loss": 1.479,
+      "step": 403
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.382857142857143e-06,
+      "loss": 1.5122,
+      "step": 404
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.371428571428572e-06,
+      "loss": 1.5215,
+      "step": 405
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.36e-06,
+      "loss": 1.5144,
+      "step": 406
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.348571428571429e-06,
+      "loss": 1.7639,
+      "step": 407
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.337142857142857e-06,
+      "loss": 1.4448,
+      "step": 408
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.3257142857142865e-06,
+      "loss": 1.7201,
+      "step": 409
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.314285714285715e-06,
+      "loss": 1.4932,
+      "step": 410
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.3028571428571425e-06,
+      "loss": 1.4845,
+      "step": 411
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.291428571428572e-06,
+      "loss": 1.4798,
+      "step": 412
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.28e-06,
+      "loss": 1.4747,
+      "step": 413
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.268571428571429e-06,
+      "loss": 1.4672,
+      "step": 414
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.257142857142857e-06,
+      "loss": 1.4649,
+      "step": 415
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.2457142857142864e-06,
+      "loss": 1.4545,
+      "step": 416
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.234285714285715e-06,
+      "loss": 1.4801,
+      "step": 417
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.2228571428571425e-06,
+      "loss": 1.5474,
+      "step": 418
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.211428571428572e-06,
+      "loss": 1.4755,
+      "step": 419
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.2e-06,
+      "loss": 1.5061,
+      "step": 420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.1885714285714295e-06,
+      "loss": 1.4828,
+      "step": 421
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.177142857142857e-06,
+      "loss": 1.4669,
+      "step": 422
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.165714285714286e-06,
+      "loss": 1.4596,
+      "step": 423
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.154285714285715e-06,
+      "loss": 1.5299,
+      "step": 424
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 1.4452,
+      "step": 425
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.131428571428572e-06,
+      "loss": 1.5066,
+      "step": 426
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.12e-06,
+      "loss": 1.5446,
+      "step": 427
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.108571428571429e-06,
+      "loss": 1.4556,
+      "step": 428
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.097142857142857e-06,
+      "loss": 1.4567,
+      "step": 429
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.085714285714286e-06,
+      "loss": 1.5352,
+      "step": 430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.074285714285715e-06,
+      "loss": 1.5015,
+      "step": 431
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.062857142857144e-06,
+      "loss": 1.4837,
+      "step": 432
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.051428571428572e-06,
+      "loss": 1.5209,
+      "step": 433
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.04e-06,
+      "loss": 1.4885,
+      "step": 434
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.028571428571429e-06,
+      "loss": 1.6766,
+      "step": 435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.017142857142857e-06,
+      "loss": 1.6379,
+      "step": 436
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.005714285714286e-06,
+      "loss": 1.4389,
+      "step": 437
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.994285714285715e-06,
+      "loss": 1.4618,
+      "step": 438
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.982857142857143e-06,
+      "loss": 1.5679,
+      "step": 439
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.971428571428572e-06,
+      "loss": 1.5112,
+      "step": 440
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 1.5533,
+      "step": 441
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.948571428571429e-06,
+      "loss": 1.4795,
+      "step": 442
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.937142857142858e-06,
+      "loss": 1.4529,
+      "step": 443
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.925714285714286e-06,
+      "loss": 1.4912,
+      "step": 444
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.9142857142857145e-06,
+      "loss": 1.4789,
+      "step": 445
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.902857142857143e-06,
+      "loss": 1.4331,
+      "step": 446
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.891428571428572e-06,
+      "loss": 1.4892,
+      "step": 447
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 1.4924,
+      "step": 448
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.868571428571429e-06,
+      "loss": 1.459,
+      "step": 449
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 1.5136,
+      "step": 450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.845714285714286e-06,
+      "loss": 1.4582,
+      "step": 451
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8342857142857145e-06,
+      "loss": 1.4864,
+      "step": 452
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.822857142857143e-06,
+      "loss": 1.4838,
+      "step": 453
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.811428571428572e-06,
+      "loss": 1.4838,
+      "step": 454
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.5041,
+      "step": 455
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.788571428571429e-06,
+      "loss": 1.4823,
+      "step": 456
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.7771428571428575e-06,
+      "loss": 1.492,
+      "step": 457
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.765714285714286e-06,
+      "loss": 1.4655,
+      "step": 458
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.754285714285714e-06,
+      "loss": 1.4767,
+      "step": 459
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.742857142857144e-06,
+      "loss": 1.462,
+      "step": 460
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.731428571428572e-06,
+      "loss": 1.4796,
+      "step": 461
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 1.5112,
+      "step": 462
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.708571428571429e-06,
+      "loss": 1.4453,
+      "step": 463
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.6971428571428574e-06,
+      "loss": 1.4278,
+      "step": 464
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.685714285714286e-06,
+      "loss": 1.4257,
+      "step": 465
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.674285714285714e-06,
+      "loss": 1.4736,
+      "step": 466
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.662857142857144e-06,
+      "loss": 1.5285,
+      "step": 467
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.651428571428572e-06,
+      "loss": 1.507,
+      "step": 468
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 1.5101,
+      "step": 469
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.628571428571429e-06,
+      "loss": 1.4296,
+      "step": 470
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.617142857142857e-06,
+      "loss": 1.4725,
+      "step": 471
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.605714285714286e-06,
+      "loss": 1.4013,
+      "step": 472
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.594285714285714e-06,
+      "loss": 1.5,
+      "step": 473
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5828571428571435e-06,
+      "loss": 1.4106,
+      "step": 474
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 1.4365,
+      "step": 475
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.56e-06,
+      "loss": 1.4955,
+      "step": 476
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.548571428571429e-06,
+      "loss": 1.5075,
+      "step": 477
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.537142857142858e-06,
+      "loss": 1.4233,
+      "step": 478
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.525714285714286e-06,
+      "loss": 1.4557,
+      "step": 479
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.514285714285714e-06,
+      "loss": 1.4704,
+      "step": 480
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5028571428571434e-06,
+      "loss": 1.4758,
+      "step": 481
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.491428571428572e-06,
+      "loss": 1.4608,
+      "step": 482
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.48e-06,
+      "loss": 1.5033,
+      "step": 483
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.468571428571429e-06,
+      "loss": 1.4485,
+      "step": 484
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.457142857142858e-06,
+      "loss": 1.4785,
+      "step": 485
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.445714285714286e-06,
+      "loss": 1.4599,
+      "step": 486
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.434285714285715e-06,
+      "loss": 1.443,
+      "step": 487
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.422857142857143e-06,
+      "loss": 1.4236,
+      "step": 488
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.411428571428572e-06,
+      "loss": 1.4812,
+      "step": 489
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.4e-06,
+      "loss": 1.471,
+      "step": 490
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.388571428571429e-06,
+      "loss": 1.4846,
+      "step": 491
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.377142857142858e-06,
+      "loss": 1.7677,
+      "step": 492
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.3657142857142855e-06,
+      "loss": 1.4363,
+      "step": 493
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.354285714285715e-06,
+      "loss": 1.4643,
+      "step": 494
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.342857142857143e-06,
+      "loss": 1.4879,
+      "step": 495
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.331428571428572e-06,
+      "loss": 1.4658,
+      "step": 496
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.32e-06,
+      "loss": 1.4439,
+      "step": 497
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.3085714285714294e-06,
+      "loss": 1.4414,
+      "step": 498
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.297142857142858e-06,
+      "loss": 1.454,
+      "step": 499
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.473,
+      "step": 500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.274285714285715e-06,
+      "loss": 1.4548,
+      "step": 501
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.262857142857143e-06,
+      "loss": 1.448,
+      "step": 502
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.251428571428572e-06,
+      "loss": 1.4413,
+      "step": 503
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.24e-06,
+      "loss": 1.4646,
+      "step": 504
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.228571428571429e-06,
+      "loss": 1.4668,
+      "step": 505
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.217142857142858e-06,
+      "loss": 1.7769,
+      "step": 506
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.205714285714286e-06,
+      "loss": 1.7808,
+      "step": 507
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.194285714285715e-06,
+      "loss": 1.4353,
+      "step": 508
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.182857142857143e-06,
+      "loss": 1.494,
+      "step": 509
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.1714285714285715e-06,
+      "loss": 1.3975,
+      "step": 510
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.16e-06,
+      "loss": 1.4711,
+      "step": 511
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.148571428571429e-06,
+      "loss": 1.4339,
+      "step": 512
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.137142857142858e-06,
+      "loss": 1.438,
+      "step": 513
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.125714285714286e-06,
+      "loss": 1.4791,
+      "step": 514
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.114285714285715e-06,
+      "loss": 1.4116,
+      "step": 515
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.102857142857143e-06,
+      "loss": 1.4995,
+      "step": 516
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0914285714285715e-06,
+      "loss": 1.4826,
+      "step": 517
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.08e-06,
+      "loss": 1.4134,
+      "step": 518
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.068571428571429e-06,
+      "loss": 1.3754,
+      "step": 519
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.057142857142858e-06,
+      "loss": 1.4713,
+      "step": 520
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.045714285714286e-06,
+      "loss": 1.4444,
+      "step": 521
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0342857142857145e-06,
+      "loss": 1.4322,
+      "step": 522
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.022857142857143e-06,
+      "loss": 1.4162,
+      "step": 523
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.011428571428571e-06,
+      "loss": 1.3937,
+      "step": 524
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.4016,
+      "step": 525
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.988571428571429e-06,
+      "loss": 1.7411,
+      "step": 526
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.9771428571428575e-06,
+      "loss": 1.9399,
+      "step": 527
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.965714285714286e-06,
+      "loss": 1.963,
+      "step": 528
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.954285714285714e-06,
+      "loss": 1.8972,
+      "step": 529
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.942857142857143e-06,
+      "loss": 1.8315,
+      "step": 530
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.931428571428571e-06,
+      "loss": 1.883,
+      "step": 531
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 1.9201,
+      "step": 532
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.908571428571429e-06,
+      "loss": 1.8479,
+      "step": 533
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.8971428571428575e-06,
+      "loss": 1.8917,
+      "step": 534
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.885714285714286e-06,
+      "loss": 1.8961,
+      "step": 535
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.874285714285715e-06,
+      "loss": 1.847,
+      "step": 536
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.862857142857143e-06,
+      "loss": 1.8268,
+      "step": 537
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.851428571428571e-06,
+      "loss": 1.9209,
+      "step": 538
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 1.9007,
+      "step": 539
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.828571428571429e-06,
+      "loss": 1.9178,
+      "step": 540
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.817142857142857e-06,
+      "loss": 1.885,
+      "step": 541
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.805714285714286e-06,
+      "loss": 1.8167,
+      "step": 542
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.7942857142857147e-06,
+      "loss": 1.8998,
+      "step": 543
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.782857142857143e-06,
+      "loss": 1.8859,
+      "step": 544
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.771428571428572e-06,
+      "loss": 1.8759,
+      "step": 545
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 1.8849,
+      "step": 546
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.7485714285714284e-06,
+      "loss": 1.9288,
+      "step": 547
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.7371428571428577e-06,
+      "loss": 1.8733,
+      "step": 548
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.7257142857142857e-06,
+      "loss": 1.8456,
+      "step": 549
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 1.9075,
+      "step": 550
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.702857142857143e-06,
+      "loss": 1.8708,
+      "step": 551
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.691428571428572e-06,
+      "loss": 1.8631,
+      "step": 552
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 1.9257,
+      "step": 553
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.668571428571429e-06,
+      "loss": 1.8945,
+      "step": 554
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.6571428571428576e-06,
+      "loss": 1.9192,
+      "step": 555
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.6457142857142857e-06,
+      "loss": 1.93,
+      "step": 556
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.6342857142857145e-06,
+      "loss": 1.9129,
+      "step": 557
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.622857142857143e-06,
+      "loss": 1.8748,
+      "step": 558
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.611428571428572e-06,
+      "loss": 1.8528,
+      "step": 559
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 1.892,
+      "step": 560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.588571428571429e-06,
+      "loss": 1.8771,
+      "step": 561
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.5771428571428576e-06,
+      "loss": 1.8586,
+      "step": 562
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.5657142857142864e-06,
+      "loss": 1.8654,
+      "step": 563
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.5542857142857144e-06,
+      "loss": 1.8428,
+      "step": 564
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.542857142857143e-06,
+      "loss": 1.8909,
+      "step": 565
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.5314285714285717e-06,
+      "loss": 1.8652,
+      "step": 566
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.52e-06,
+      "loss": 1.8372,
+      "step": 567
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.508571428571429e-06,
+      "loss": 1.8633,
+      "step": 568
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.4971428571428575e-06,
+      "loss": 1.8877,
+      "step": 569
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.4857142857142863e-06,
+      "loss": 1.9261,
+      "step": 570
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.4742857142857144e-06,
+      "loss": 1.9024,
+      "step": 571
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.4628571428571432e-06,
+      "loss": 1.8371,
+      "step": 572
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.4514285714285717e-06,
+      "loss": 1.8502,
+      "step": 573
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.44e-06,
+      "loss": 1.9052,
+      "step": 574
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 1.9039,
+      "step": 575
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.4171428571428574e-06,
+      "loss": 1.837,
+      "step": 576
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.4057142857142863e-06,
+      "loss": 1.8864,
+      "step": 577
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3942857142857143e-06,
+      "loss": 1.9116,
+      "step": 578
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.382857142857143e-06,
+      "loss": 1.8871,
+      "step": 579
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3714285714285716e-06,
+      "loss": 1.7812,
+      "step": 580
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 1.844,
+      "step": 581
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.348571428571429e-06,
+      "loss": 1.9072,
+      "step": 582
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3371428571428577e-06,
+      "loss": 1.8789,
+      "step": 583
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.325714285714286e-06,
+      "loss": 1.8951,
+      "step": 584
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.314285714285714e-06,
+      "loss": 1.8319,
+      "step": 585
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.302857142857143e-06,
+      "loss": 1.8207,
+      "step": 586
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2914285714285715e-06,
+      "loss": 1.8575,
+      "step": 587
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 1.9078,
+      "step": 588
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.268571428571429e-06,
+      "loss": 1.8926,
+      "step": 589
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2571428571428577e-06,
+      "loss": 1.8988,
+      "step": 590
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.245714285714286e-06,
+      "loss": 1.8542,
+      "step": 591
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.234285714285715e-06,
+      "loss": 1.8345,
+      "step": 592
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.222857142857143e-06,
+      "loss": 1.8951,
+      "step": 593
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2114285714285714e-06,
+      "loss": 1.9398,
+      "step": 594
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.9409,
+      "step": 595
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.1885714285714287e-06,
+      "loss": 1.8817,
+      "step": 596
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.1771428571428576e-06,
+      "loss": 1.9209,
+      "step": 597
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.165714285714286e-06,
+      "loss": 1.8879,
+      "step": 598
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.154285714285715e-06,
+      "loss": 1.8764,
+      "step": 599
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.142857142857143e-06,
+      "loss": 1.8648,
+      "step": 600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.1314285714285718e-06,
+      "loss": 1.9254,
+      "step": 601
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.12e-06,
+      "loss": 1.8743,
+      "step": 602
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1085714285714286e-06,
+      "loss": 1.865,
+      "step": 603
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0971428571428575e-06,
+      "loss": 1.8978,
+      "step": 604
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.085714285714286e-06,
+      "loss": 1.8673,
+      "step": 605
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.074285714285715e-06,
+      "loss": 1.8956,
+      "step": 606
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.062857142857143e-06,
+      "loss": 1.8259,
+      "step": 607
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0514285714285717e-06,
+      "loss": 1.8632,
+      "step": 608
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.04e-06,
+      "loss": 1.8239,
+      "step": 609
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.028571428571429e-06,
+      "loss": 1.8832,
+      "step": 610
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0171428571428574e-06,
+      "loss": 1.9016,
+      "step": 611
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.005714285714286e-06,
+      "loss": 1.8118,
+      "step": 612
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.9942857142857147e-06,
+      "loss": 1.9122,
+      "step": 613
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.9828571428571427e-06,
+      "loss": 1.9052,
+      "step": 614
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.9714285714285716e-06,
+      "loss": 1.8897,
+      "step": 615
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.96e-06,
+      "loss": 1.9204,
+      "step": 616
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.948571428571429e-06,
+      "loss": 1.9011,
+      "step": 617
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.9371428571428573e-06,
+      "loss": 1.8924,
+      "step": 618
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.925714285714286e-06,
+      "loss": 1.8645,
+      "step": 619
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9142857142857146e-06,
+      "loss": 1.9246,
+      "step": 620
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9028571428571427e-06,
+      "loss": 1.9291,
+      "step": 621
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.8914285714285715e-06,
+      "loss": 1.9409,
+      "step": 622
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.88e-06,
+      "loss": 1.8722,
+      "step": 623
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.868571428571429e-06,
+      "loss": 1.8968,
+      "step": 624
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.8497,
+      "step": 625
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.845714285714286e-06,
+      "loss": 1.854,
+      "step": 626
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.8342857142857146e-06,
+      "loss": 1.9165,
+      "step": 627
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.8228571428571434e-06,
+      "loss": 1.8735,
+      "step": 628
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.8114285714285714e-06,
+      "loss": 1.8905,
+      "step": 629
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.8628,
+      "step": 630
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.7885714285714287e-06,
+      "loss": 1.86,
+      "step": 631
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.777142857142857e-06,
+      "loss": 1.8703,
+      "step": 632
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.765714285714286e-06,
+      "loss": 1.8255,
+      "step": 633
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.7542857142857145e-06,
+      "loss": 1.8789,
+      "step": 634
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.7428571428571433e-06,
+      "loss": 1.8921,
+      "step": 635
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.7314285714285714e-06,
+      "loss": 1.9307,
+      "step": 636
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 1.8286,
+      "step": 637
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.7085714285714287e-06,
+      "loss": 1.9071,
+      "step": 638
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.6971428571428575e-06,
+      "loss": 1.9268,
+      "step": 639
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.685714285714286e-06,
+      "loss": 1.8395,
+      "step": 640
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.6742857142857144e-06,
+      "loss": 1.857,
+      "step": 641
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.6628571428571433e-06,
+      "loss": 1.9159,
+      "step": 642
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6514285714285713e-06,
+      "loss": 1.8864,
+      "step": 643
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.64e-06,
+      "loss": 1.9121,
+      "step": 644
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.6285714285714286e-06,
+      "loss": 1.8867,
+      "step": 645
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.6171428571428574e-06,
+      "loss": 1.8625,
+      "step": 646
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.605714285714286e-06,
+      "loss": 1.9116,
+      "step": 647
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.5942857142857147e-06,
+      "loss": 1.8708,
+      "step": 648
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.582857142857143e-06,
+      "loss": 1.8573,
+      "step": 649
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 1.8341,
+      "step": 650
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.56e-06,
+      "loss": 1.955,
+      "step": 651
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5485714285714285e-06,
+      "loss": 1.9266,
+      "step": 652
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.5371428571428574e-06,
+      "loss": 1.8667,
+      "step": 653
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.525714285714286e-06,
+      "loss": 1.8991,
+      "step": 654
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.5142857142857147e-06,
+      "loss": 1.8957,
+      "step": 655
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.502857142857143e-06,
+      "loss": 1.9198,
+      "step": 656
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.4914285714285715e-06,
+      "loss": 1.8965,
+      "step": 657
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 1.86,
+      "step": 658
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.468571428571429e-06,
+      "loss": 1.8685,
+      "step": 659
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.4571428571428573e-06,
+      "loss": 1.8905,
+      "step": 660
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.445714285714286e-06,
+      "loss": 1.8697,
+      "step": 661
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.4342857142857146e-06,
+      "loss": 1.8578,
+      "step": 662
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.422857142857143e-06,
+      "loss": 1.8943,
+      "step": 663
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.4114285714285715e-06,
+      "loss": 1.8709,
+      "step": 664
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.8787,
+      "step": 665
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.3885714285714288e-06,
+      "loss": 1.8286,
+      "step": 666
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.377142857142857e-06,
+      "loss": 1.8774,
+      "step": 667
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.365714285714286e-06,
+      "loss": 1.8848,
+      "step": 668
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.3542857142857145e-06,
+      "loss": 1.9164,
+      "step": 669
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.342857142857143e-06,
+      "loss": 1.8708,
+      "step": 670
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.331428571428572e-06,
+      "loss": 1.8526,
+      "step": 671
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 1.9009,
+      "step": 672
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.3085714285714287e-06,
+      "loss": 1.8904,
+      "step": 673
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.297142857142857e-06,
+      "loss": 1.8988,
+      "step": 674
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 1.8882,
+      "step": 675
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.2742857142857144e-06,
+      "loss": 1.8642,
+      "step": 676
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.262857142857143e-06,
+      "loss": 1.835,
+      "step": 677
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.2514285714285717e-06,
+      "loss": 1.8289,
+      "step": 678
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.24e-06,
+      "loss": 1.9342,
+      "step": 679
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.228571428571429e-06,
+      "loss": 1.8783,
+      "step": 680
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.2171428571428575e-06,
+      "loss": 1.8279,
+      "step": 681
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.205714285714286e-06,
+      "loss": 1.8467,
+      "step": 682
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.1942857142857143e-06,
+      "loss": 1.8277,
+      "step": 683
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.1828571428571428e-06,
+      "loss": 1.8546,
+      "step": 684
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.1714285714285716e-06,
+      "loss": 1.8878,
+      "step": 685
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.16e-06,
+      "loss": 1.8328,
+      "step": 686
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.148571428571429e-06,
+      "loss": 1.9074,
+      "step": 687
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.1371428571428574e-06,
+      "loss": 1.8791,
+      "step": 688
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.125714285714286e-06,
+      "loss": 1.8258,
+      "step": 689
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.1142857142857147e-06,
+      "loss": 1.9425,
+      "step": 690
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.102857142857143e-06,
+      "loss": 1.8249,
+      "step": 691
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.0914285714285716e-06,
+      "loss": 1.8544,
+      "step": 692
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.08e-06,
+      "loss": 1.8397,
+      "step": 693
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.068571428571429e-06,
+      "loss": 1.9088,
+      "step": 694
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0571428571428573e-06,
+      "loss": 1.8075,
+      "step": 695
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0457142857142857e-06,
+      "loss": 1.858,
+      "step": 696
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0342857142857146e-06,
+      "loss": 1.9248,
+      "step": 697
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.022857142857143e-06,
+      "loss": 1.8725,
+      "step": 698
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0114285714285715e-06,
+      "loss": 1.8586,
+      "step": 699
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.8891,
+      "step": 700
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.9885714285714288e-06,
+      "loss": 1.6464,
+      "step": 701
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.977142857142857e-06,
+      "loss": 1.5097,
+      "step": 702
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9657142857142856e-06,
+      "loss": 1.5016,
+      "step": 703
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9542857142857145e-06,
+      "loss": 1.7568,
+      "step": 704
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.942857142857143e-06,
+      "loss": 1.4996,
+      "step": 705
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9314285714285714e-06,
+      "loss": 1.5447,
+      "step": 706
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.491,
+      "step": 707
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9085714285714287e-06,
+      "loss": 1.4641,
+      "step": 708
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8971428571428573e-06,
+      "loss": 1.5629,
+      "step": 709
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.885714285714286e-06,
+      "loss": 1.5122,
+      "step": 710
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8742857142857142e-06,
+      "loss": 1.5143,
+      "step": 711
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8628571428571429e-06,
+      "loss": 1.4936,
+      "step": 712
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8514285714285715e-06,
+      "loss": 1.5177,
+      "step": 713
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 1.5246,
+      "step": 714
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.8285714285714288e-06,
+      "loss": 1.4653,
+      "step": 715
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.8171428571428573e-06,
+      "loss": 1.5703,
+      "step": 716
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.805714285714286e-06,
+      "loss": 1.4921,
+      "step": 717
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7942857142857146e-06,
+      "loss": 1.4894,
+      "step": 718
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7828571428571432e-06,
+      "loss": 1.4682,
+      "step": 719
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7714285714285714e-06,
+      "loss": 1.5342,
+      "step": 720
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.76e-06,
+      "loss": 1.4958,
+      "step": 721
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7485714285714287e-06,
+      "loss": 1.4758,
+      "step": 722
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7371428571428572e-06,
+      "loss": 1.5241,
+      "step": 723
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7257142857142858e-06,
+      "loss": 1.4942,
+      "step": 724
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 1.5299,
+      "step": 725
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7028571428571431e-06,
+      "loss": 1.4982,
+      "step": 726
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.6914285714285716e-06,
+      "loss": 1.5077,
+      "step": 727
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 1.5504,
+      "step": 728
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6685714285714289e-06,
+      "loss": 1.4379,
+      "step": 729
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.657142857142857e-06,
+      "loss": 1.4829,
+      "step": 730
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6457142857142857e-06,
+      "loss": 1.5232,
+      "step": 731
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6342857142857144e-06,
+      "loss": 1.5023,
+      "step": 732
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.622857142857143e-06,
+      "loss": 1.4581,
+      "step": 733
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6114285714285715e-06,
+      "loss": 1.5294,
+      "step": 734
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.4823,
+      "step": 735
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.5885714285714288e-06,
+      "loss": 1.5193,
+      "step": 736
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.5771428571428574e-06,
+      "loss": 1.5021,
+      "step": 737
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.5657142857142859e-06,
+      "loss": 1.5075,
+      "step": 738
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.5542857142857143e-06,
+      "loss": 1.4578,
+      "step": 739
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.542857142857143e-06,
+      "loss": 1.646,
+      "step": 740
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.5314285714285714e-06,
+      "loss": 1.445,
+      "step": 741
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.52e-06,
+      "loss": 1.4514,
+      "step": 742
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.5085714285714287e-06,
+      "loss": 1.4697,
+      "step": 743
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.4971428571428574e-06,
+      "loss": 1.4633,
+      "step": 744
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.4857142857142858e-06,
+      "loss": 1.4788,
+      "step": 745
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.4742857142857144e-06,
+      "loss": 1.4946,
+      "step": 746
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.462857142857143e-06,
+      "loss": 1.4891,
+      "step": 747
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.4514285714285713e-06,
+      "loss": 1.4945,
+      "step": 748
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.44e-06,
+      "loss": 1.4317,
+      "step": 749
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.5761,
+      "step": 750
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4171428571428573e-06,
+      "loss": 1.5298,
+      "step": 751
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4057142857142857e-06,
+      "loss": 1.4761,
+      "step": 752
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3942857142857144e-06,
+      "loss": 1.4827,
+      "step": 753
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.382857142857143e-06,
+      "loss": 1.479,
+      "step": 754
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3714285714285717e-06,
+      "loss": 1.5391,
+      "step": 755
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 1.5073,
+      "step": 756
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3485714285714288e-06,
+      "loss": 1.5017,
+      "step": 757
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3371428571428572e-06,
+      "loss": 1.4506,
+      "step": 758
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3257142857142856e-06,
+      "loss": 1.5273,
+      "step": 759
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3142857142857143e-06,
+      "loss": 1.515,
+      "step": 760
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.302857142857143e-06,
+      "loss": 1.5218,
+      "step": 761
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2914285714285716e-06,
+      "loss": 1.4744,
+      "step": 762
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.28e-06,
+      "loss": 1.4799,
+      "step": 763
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2685714285714287e-06,
+      "loss": 1.5036,
+      "step": 764
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2571428571428573e-06,
+      "loss": 1.4557,
+      "step": 765
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2457142857142858e-06,
+      "loss": 1.5085,
+      "step": 766
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2342857142857144e-06,
+      "loss": 1.5359,
+      "step": 767
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.222857142857143e-06,
+      "loss": 1.4674,
+      "step": 768
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2114285714285715e-06,
+      "loss": 1.593,
+      "step": 769
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.4956,
+      "step": 770
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1885714285714286e-06,
+      "loss": 1.5141,
+      "step": 771
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1771428571428572e-06,
+      "loss": 1.5328,
+      "step": 772
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.165714285714286e-06,
+      "loss": 1.5455,
+      "step": 773
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1542857142857143e-06,
+      "loss": 1.4508,
+      "step": 774
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 1.5287,
+      "step": 775
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1314285714285714e-06,
+      "loss": 1.508,
+      "step": 776
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.12e-06,
+      "loss": 1.4928,
+      "step": 777
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1085714285714287e-06,
+      "loss": 1.4682,
+      "step": 778
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0971428571428572e-06,
+      "loss": 1.4761,
+      "step": 779
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0857142857142858e-06,
+      "loss": 1.499,
+      "step": 780
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0742857142857145e-06,
+      "loss": 1.531,
+      "step": 781
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.062857142857143e-06,
+      "loss": 1.4764,
+      "step": 782
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0514285714285716e-06,
+      "loss": 1.5061,
+      "step": 783
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.04e-06,
+      "loss": 1.5677,
+      "step": 784
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0285714285714286e-06,
+      "loss": 1.4716,
+      "step": 785
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0171428571428573e-06,
+      "loss": 1.4379,
+      "step": 786
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0057142857142857e-06,
+      "loss": 1.5114,
+      "step": 787
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 9.942857142857144e-07,
+      "loss": 1.5163,
+      "step": 788
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 9.828571428571428e-07,
+      "loss": 1.4867,
+      "step": 789
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.714285714285715e-07,
+      "loss": 1.4943,
+      "step": 790
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.5142,
+      "step": 791
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.485714285714287e-07,
+      "loss": 1.4688,
+      "step": 792
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.371428571428571e-07,
+      "loss": 1.5287,
+      "step": 793
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.257142857142858e-07,
+      "loss": 1.534,
+      "step": 794
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.142857142857144e-07,
+      "loss": 1.5041,
+      "step": 795
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.02857142857143e-07,
+      "loss": 1.5213,
+      "step": 796
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.914285714285716e-07,
+      "loss": 1.5111,
+      "step": 797
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.8e-07,
+      "loss": 1.5181,
+      "step": 798
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.685714285714286e-07,
+      "loss": 1.4652,
+      "step": 799
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 1.5007,
+      "step": 800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.457142857142858e-07,
+      "loss": 1.5182,
+      "step": 801
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.342857142857144e-07,
+      "loss": 1.5377,
+      "step": 802
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.228571428571429e-07,
+      "loss": 1.5312,
+      "step": 803
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.114285714285715e-07,
+      "loss": 1.56,
+      "step": 804
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.5159,
+      "step": 805
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.885714285714287e-07,
+      "loss": 1.5334,
+      "step": 806
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.771428571428572e-07,
+      "loss": 1.4815,
+      "step": 807
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.657142857142857e-07,
+      "loss": 1.4832,
+      "step": 808
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.542857142857144e-07,
+      "loss": 1.526,
+      "step": 809
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.428571428571429e-07,
+      "loss": 1.5323,
+      "step": 810
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.314285714285715e-07,
+      "loss": 1.5741,
+      "step": 811
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.2e-07,
+      "loss": 1.5122,
+      "step": 812
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.085714285714286e-07,
+      "loss": 1.524,
+      "step": 813
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.971428571428572e-07,
+      "loss": 1.5032,
+      "step": 814
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.857142857142858e-07,
+      "loss": 1.5023,
+      "step": 815
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.742857142857144e-07,
+      "loss": 1.5635,
+      "step": 816
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.628571428571428e-07,
+      "loss": 1.5251,
+      "step": 817
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.514285714285715e-07,
+      "loss": 1.5297,
+      "step": 818
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.4e-07,
+      "loss": 1.5207,
+      "step": 819
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.285714285714287e-07,
+      "loss": 1.5084,
+      "step": 820
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.171428571428572e-07,
+      "loss": 1.5176,
+      "step": 821
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.057142857142858e-07,
+      "loss": 1.5485,
+      "step": 822
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.942857142857143e-07,
+      "loss": 1.6003,
+      "step": 823
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.82857142857143e-07,
+      "loss": 1.5627,
+      "step": 824
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 1.547,
+      "step": 825
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.6e-07,
+      "loss": 1.5134,
+      "step": 826
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.485714285714286e-07,
+      "loss": 1.6044,
+      "step": 827
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.371428571428572e-07,
+      "loss": 1.532,
+      "step": 828
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.257142857142858e-07,
+      "loss": 1.4849,
+      "step": 829
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.142857142857143e-07,
+      "loss": 1.5417,
+      "step": 830
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.028571428571429e-07,
+      "loss": 1.6036,
+      "step": 831
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.914285714285714e-07,
+      "loss": 1.5157,
+      "step": 832
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.5236,
+      "step": 833
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.6857142857142855e-07,
+      "loss": 1.5127,
+      "step": 834
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.571428571428572e-07,
+      "loss": 1.5084,
+      "step": 835
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.457142857142858e-07,
+      "loss": 1.5011,
+      "step": 836
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.342857142857143e-07,
+      "loss": 1.4924,
+      "step": 837
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.228571428571429e-07,
+      "loss": 1.5125,
+      "step": 838
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.1142857142857144e-07,
+      "loss": 1.5937,
+      "step": 839
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.504,
+      "step": 840
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.885714285714286e-07,
+      "loss": 1.5465,
+      "step": 841
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.771428571428572e-07,
+      "loss": 1.5221,
+      "step": 842
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.657142857142858e-07,
+      "loss": 1.5342,
+      "step": 843
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.542857142857143e-07,
+      "loss": 1.5579,
+      "step": 844
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.428571428571429e-07,
+      "loss": 1.5855,
+      "step": 845
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.314285714285714e-07,
+      "loss": 1.5642,
+      "step": 846
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.2e-07,
+      "loss": 1.5323,
+      "step": 847
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.085714285714286e-07,
+      "loss": 1.6015,
+      "step": 848
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.9714285714285715e-07,
+      "loss": 1.4954,
+      "step": 849
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 1.499,
+      "step": 850
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.742857142857143e-07,
+      "loss": 1.5401,
+      "step": 851
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.628571428571429e-07,
+      "loss": 1.5234,
+      "step": 852
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5142857142857143e-07,
+      "loss": 1.494,
+      "step": 853
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 1.5839,
+      "step": 854
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.285714285714286e-07,
+      "loss": 1.4961,
+      "step": 855
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.1714285714285715e-07,
+      "loss": 1.553,
+      "step": 856
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.0571428571428572e-07,
+      "loss": 1.5256,
+      "step": 857
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.942857142857143e-07,
+      "loss": 1.5505,
+      "step": 858
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.828571428571429e-07,
+      "loss": 1.5722,
+      "step": 859
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7142857142857146e-07,
+      "loss": 1.5449,
+      "step": 860
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6e-07,
+      "loss": 1.5505,
+      "step": 861
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4857142857142857e-07,
+      "loss": 1.5537,
+      "step": 862
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3714285714285715e-07,
+      "loss": 1.5144,
+      "step": 863
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2571428571428572e-07,
+      "loss": 1.594,
+      "step": 864
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.142857142857143e-07,
+      "loss": 1.6201,
+      "step": 865
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0285714285714286e-07,
+      "loss": 1.5398,
+      "step": 866
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 9.142857142857144e-08,
+      "loss": 1.5405,
+      "step": 867
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8e-08,
+      "loss": 1.5522,
+      "step": 868
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.857142857142857e-08,
+      "loss": 1.5109,
+      "step": 869
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.714285714285715e-08,
+      "loss": 1.5823,
+      "step": 870
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.571428571428572e-08,
+      "loss": 1.5888,
+      "step": 871
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.4285714285714286e-08,
+      "loss": 1.5141,
+      "step": 872
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.285714285714286e-08,
+      "loss": 1.5193,
+      "step": 873
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.142857142857143e-08,
+      "loss": 1.5431,
+      "step": 874
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0,
+      "loss": 1.5294,
+      "step": 875
+    },
+    {
+      "epoch": 4.99,
+      "step": 875,
+      "total_flos": 5.021505014464512e+17,
+      "train_loss": 1.6896061137063163,
+      "train_runtime": 6113.9003,
+      "train_samples_per_second": 36.667,
+      "train_steps_per_second": 0.143
+    }
+  ],
+  "max_steps": 875,
+  "num_train_epochs": 5,
+  "total_flos": 5.021505014464512e+17,
+  "trial_name": null,
+  "trial_params": null
+}