diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6619 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 11.12867530825166,
+  "eval_steps": 500,
+  "global_step": 1100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996811224489796e-05,
+      "loss": 3.2244,
+      "step": 1
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993622448979592e-05,
+      "loss": 2.9431,
+      "step": 2
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990433673469388e-05,
+      "loss": 2.7933,
+      "step": 3
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987244897959184e-05,
+      "loss": 2.7027,
+      "step": 4
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984056122448981e-05,
+      "loss": 2.5159,
+      "step": 5
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980867346938776e-05,
+      "loss": 2.5234,
+      "step": 6
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.977678571428571e-05,
+      "loss": 2.4048,
+      "step": 7
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974489795918368e-05,
+      "loss": 2.297,
+      "step": 8
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971301020408164e-05,
+      "loss": 2.2237,
+      "step": 9
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96811224489796e-05,
+      "loss": 2.1927,
+      "step": 10
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.964923469387755e-05,
+      "loss": 2.1405,
+      "step": 11
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.961734693877551e-05,
+      "loss": 2.2251,
+      "step": 12
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.958545918367348e-05,
+      "loss": 2.1525,
+      "step": 13
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.955357142857144e-05,
+      "loss": 2.1014,
+      "step": 14
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.952168367346938e-05,
+      "loss": 2.061,
+      "step": 15
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.948979591836736e-05,
+      "loss": 2.05,
+      "step": 16
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.945790816326531e-05,
+      "loss": 2.101,
+      "step": 17
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.942602040816327e-05,
+      "loss": 1.8366,
+      "step": 18
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.939413265306123e-05,
+      "loss": 1.8752,
+      "step": 19
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.936224489795919e-05,
+      "loss": 2.0378,
+      "step": 20
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.933035714285714e-05,
+      "loss": 2.0911,
+      "step": 21
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.929846938775512e-05,
+      "loss": 1.9428,
+      "step": 22
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.926658163265307e-05,
+      "loss": 1.9914,
+      "step": 23
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.923469387755102e-05,
+      "loss": 1.949,
+      "step": 24
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.920280612244899e-05,
+      "loss": 2.0358,
+      "step": 25
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.917091836734695e-05,
+      "loss": 2.1013,
+      "step": 26
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.91390306122449e-05,
+      "loss": 1.9653,
+      "step": 27
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.910714285714286e-05,
+      "loss": 1.8327,
+      "step": 28
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.907525510204082e-05,
+      "loss": 1.6626,
+      "step": 29
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.904336734693878e-05,
+      "loss": 1.9772,
+      "step": 30
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.901147959183675e-05,
+      "loss": 1.9107,
+      "step": 31
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.897959183673469e-05,
+      "loss": 1.8967,
+      "step": 32
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.894770408163265e-05,
+      "loss": 2.0814,
+      "step": 33
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.891581632653062e-05,
+      "loss": 1.8678,
+      "step": 34
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.888392857142858e-05,
+      "loss": 1.8817,
+      "step": 35
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.885204081632652e-05,
+      "loss": 1.9398,
+      "step": 36
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.88201530612245e-05,
+      "loss": 1.8375,
+      "step": 37
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.878826530612245e-05,
+      "loss": 2.0266,
+      "step": 38
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.875637755102042e-05,
+      "loss": 1.937,
+      "step": 39
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.872448979591837e-05,
+      "loss": 1.9531,
+      "step": 40
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.869260204081633e-05,
+      "loss": 1.9794,
+      "step": 41
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.86607142857143e-05,
+      "loss": 1.8376,
+      "step": 42
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.862882653061225e-05,
+      "loss": 1.8657,
+      "step": 43
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.859693877551021e-05,
+      "loss": 1.7837,
+      "step": 44
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.856505102040817e-05,
+      "loss": 1.8824,
+      "step": 45
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.853316326530613e-05,
+      "loss": 1.8274,
+      "step": 46
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.850127551020409e-05,
+      "loss": 1.9559,
+      "step": 47
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.846938775510204e-05,
+      "loss": 1.8214,
+      "step": 48
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.84375e-05,
+      "loss": 1.9955,
+      "step": 49
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.840561224489796e-05,
+      "loss": 1.8716,
+      "step": 50
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837372448979593e-05,
+      "loss": 1.8892,
+      "step": 51
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.834183673469389e-05,
+      "loss": 1.9172,
+      "step": 52
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.830994897959183e-05,
+      "loss": 1.8013,
+      "step": 53
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.82780612244898e-05,
+      "loss": 1.9853,
+      "step": 54
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.824617346938776e-05,
+      "loss": 1.8744,
+      "step": 55
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.821428571428572e-05,
+      "loss": 1.8443,
+      "step": 56
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.818239795918368e-05,
+      "loss": 1.9072,
+      "step": 57
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.815051020408163e-05,
+      "loss": 1.8902,
+      "step": 58
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.811862244897959e-05,
+      "loss": 1.9012,
+      "step": 59
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.808673469387756e-05,
+      "loss": 1.8728,
+      "step": 60
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.805484693877551e-05,
+      "loss": 1.7778,
+      "step": 61
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.802295918367348e-05,
+      "loss": 1.9753,
+      "step": 62
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.799107142857144e-05,
+      "loss": 1.9134,
+      "step": 63
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.79591836734694e-05,
+      "loss": 1.7386,
+      "step": 64
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.792729591836735e-05,
+      "loss": 1.7398,
+      "step": 65
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.789540816326531e-05,
+      "loss": 1.8563,
+      "step": 66
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.786352040816327e-05,
+      "loss": 1.9333,
+      "step": 67
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.783163265306124e-05,
+      "loss": 1.8351,
+      "step": 68
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.77997448979592e-05,
+      "loss": 1.8199,
+      "step": 69
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.776785714285714e-05,
+      "loss": 1.9057,
+      "step": 70
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.773596938775511e-05,
+      "loss": 1.9722,
+      "step": 71
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.770408163265307e-05,
+      "loss": 1.8762,
+      "step": 72
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.767219387755103e-05,
+      "loss": 1.8591,
+      "step": 73
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.764030612244898e-05,
+      "loss": 1.7435,
+      "step": 74
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.760841836734694e-05,
+      "loss": 1.8729,
+      "step": 75
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.75765306122449e-05,
+      "loss": 1.8089,
+      "step": 76
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.754464285714287e-05,
+      "loss": 1.9033,
+      "step": 77
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.751275510204082e-05,
+      "loss": 1.8569,
+      "step": 78
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.748086734693877e-05,
+      "loss": 1.8682,
+      "step": 79
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.744897959183674e-05,
+      "loss": 1.6905,
+      "step": 80
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.74170918367347e-05,
+      "loss": 1.8561,
+      "step": 81
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.738520408163265e-05,
+      "loss": 1.8295,
+      "step": 82
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.735331632653062e-05,
+      "loss": 1.848,
+      "step": 83
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.732142857142858e-05,
+      "loss": 1.722,
+      "step": 84
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.728954081632653e-05,
+      "loss": 1.6927,
+      "step": 85
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.725765306122449e-05,
+      "loss": 1.7642,
+      "step": 86
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.722576530612245e-05,
+      "loss": 1.8603,
+      "step": 87
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.719387755102042e-05,
+      "loss": 1.8246,
+      "step": 88
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.716198979591838e-05,
+      "loss": 1.9409,
+      "step": 89
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.713010204081633e-05,
+      "loss": 1.7765,
+      "step": 90
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.709821428571429e-05,
+      "loss": 1.7607,
+      "step": 91
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.706632653061225e-05,
+      "loss": 1.863,
+      "step": 92
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.703443877551021e-05,
+      "loss": 1.7051,
+      "step": 93
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.700255102040817e-05,
+      "loss": 1.8005,
+      "step": 94
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.697066326530612e-05,
+      "loss": 1.7598,
+      "step": 95
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.693877551020408e-05,
+      "loss": 1.8445,
+      "step": 96
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.690688775510205e-05,
+      "loss": 1.9542,
+      "step": 97
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.687500000000001e-05,
+      "loss": 1.7595,
+      "step": 98
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.684311224489795e-05,
+      "loss": 1.8095,
+      "step": 99
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.681122448979593e-05,
+      "loss": 1.7445,
+      "step": 100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.677933673469388e-05,
+      "loss": 1.9506,
+      "step": 101
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.674744897959184e-05,
+      "loss": 1.7708,
+      "step": 102
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.67155612244898e-05,
+      "loss": 1.898,
+      "step": 103
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.668367346938776e-05,
+      "loss": 1.7326,
+      "step": 104
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.665178571428571e-05,
+      "loss": 1.8423,
+      "step": 105
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.661989795918369e-05,
+      "loss": 1.7541,
+      "step": 106
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.658801020408163e-05,
+      "loss": 1.9325,
+      "step": 107
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.655612244897959e-05,
+      "loss": 1.8437,
+      "step": 108
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.652423469387756e-05,
+      "loss": 1.8415,
+      "step": 109
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.649234693877552e-05,
+      "loss": 1.8457,
+      "step": 110
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.646045918367347e-05,
+      "loss": 1.6994,
+      "step": 111
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.642857142857143e-05,
+      "loss": 1.8482,
+      "step": 112
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.639668367346939e-05,
+      "loss": 1.8053,
+      "step": 113
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.636479591836736e-05,
+      "loss": 1.8064,
+      "step": 114
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.633290816326532e-05,
+      "loss": 1.7199,
+      "step": 115
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.630102040816326e-05,
+      "loss": 1.7226,
+      "step": 116
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.626913265306123e-05,
+      "loss": 1.8804,
+      "step": 117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.623724489795919e-05,
+      "loss": 1.7488,
+      "step": 118
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.620535714285715e-05,
+      "loss": 1.8766,
+      "step": 119
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.617346938775511e-05,
+      "loss": 1.8261,
+      "step": 120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.614158163265306e-05,
+      "loss": 1.7813,
+      "step": 121
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.610969387755102e-05,
+      "loss": 1.6967,
+      "step": 122
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.6077806122449e-05,
+      "loss": 1.6953,
+      "step": 123
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.604591836734694e-05,
+      "loss": 1.6754,
+      "step": 124
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.60140306122449e-05,
+      "loss": 1.7052,
+      "step": 125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.598214285714287e-05,
+      "loss": 1.8196,
+      "step": 126
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.595025510204082e-05,
+      "loss": 1.7346,
+      "step": 127
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.591836734693878e-05,
+      "loss": 1.8132,
+      "step": 128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.588647959183674e-05,
+      "loss": 1.7865,
+      "step": 129
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.58545918367347e-05,
+      "loss": 1.745,
+      "step": 130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.582270408163266e-05,
+      "loss": 1.8189,
+      "step": 131
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.579081632653061e-05,
+      "loss": 1.7351,
+      "step": 132
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.575892857142857e-05,
+      "loss": 1.8111,
+      "step": 133
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.572704081632653e-05,
+      "loss": 1.815,
+      "step": 134
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.56951530612245e-05,
+      "loss": 1.7898,
+      "step": 135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.566326530612246e-05,
+      "loss": 1.8174,
+      "step": 136
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.563137755102042e-05,
+      "loss": 1.7156,
+      "step": 137
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.559948979591837e-05,
+      "loss": 1.8901,
+      "step": 138
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.556760204081633e-05,
+      "loss": 1.7957,
+      "step": 139
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.553571428571429e-05,
+      "loss": 1.8229,
+      "step": 140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.550382653061225e-05,
+      "loss": 1.7423,
+      "step": 141
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.54719387755102e-05,
+      "loss": 1.8035,
+      "step": 142
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.544005102040818e-05,
+      "loss": 1.8071,
+      "step": 143
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.540816326530613e-05,
+      "loss": 1.7434,
+      "step": 144
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.537627551020408e-05,
+      "loss": 1.7771,
+      "step": 145
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.534438775510205e-05,
+      "loss": 1.7994,
+      "step": 146
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.53125e-05,
+      "loss": 1.8915,
+      "step": 147
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.528061224489796e-05,
+      "loss": 1.7729,
+      "step": 148
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.524872448979592e-05,
+      "loss": 1.8204,
+      "step": 149
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.521683673469388e-05,
+      "loss": 1.8064,
+      "step": 150
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.518494897959184e-05,
+      "loss": 1.8689,
+      "step": 151
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.515306122448981e-05,
+      "loss": 1.7717,
+      "step": 152
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.512117346938775e-05,
+      "loss": 1.819,
+      "step": 153
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.508928571428571e-05,
+      "loss": 1.754,
+      "step": 154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.505739795918368e-05,
+      "loss": 1.9207,
+      "step": 155
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.502551020408164e-05,
+      "loss": 1.8492,
+      "step": 156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49936224489796e-05,
+      "loss": 1.8649,
+      "step": 157
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.496173469387755e-05,
+      "loss": 1.8203,
+      "step": 158
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.492984693877551e-05,
+      "loss": 1.8118,
+      "step": 159
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.489795918367348e-05,
+      "loss": 1.7663,
+      "step": 160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.486607142857144e-05,
+      "loss": 1.8782,
+      "step": 161
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.483418367346939e-05,
+      "loss": 1.8452,
+      "step": 162
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.480229591836736e-05,
+      "loss": 1.7889,
+      "step": 163
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.477040816326531e-05,
+      "loss": 1.7076,
+      "step": 164
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.473852040816327e-05,
+      "loss": 1.845,
+      "step": 165
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.470663265306123e-05,
+      "loss": 1.7403,
+      "step": 166
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.467474489795919e-05,
+      "loss": 1.8155,
+      "step": 167
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.464285714285715e-05,
+      "loss": 1.8515,
+      "step": 168
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.461096938775512e-05,
+      "loss": 1.7782,
+      "step": 169
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.457908163265306e-05,
+      "loss": 1.7943,
+      "step": 170
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.454719387755102e-05,
+      "loss": 1.7188,
+      "step": 171
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.451530612244899e-05,
+      "loss": 1.7785,
+      "step": 172
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.448341836734695e-05,
+      "loss": 1.8056,
+      "step": 173
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.44515306122449e-05,
+      "loss": 1.7764,
+      "step": 174
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.441964285714286e-05,
+      "loss": 1.8921,
+      "step": 175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.438775510204082e-05,
+      "loss": 1.762,
+      "step": 176
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.435586734693878e-05,
+      "loss": 1.7324,
+      "step": 177
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.432397959183674e-05,
+      "loss": 1.7451,
+      "step": 178
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.42920918367347e-05,
+      "loss": 1.7564,
+      "step": 179
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.426020408163265e-05,
+      "loss": 1.7764,
+      "step": 180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.422831632653062e-05,
+      "loss": 1.7353,
+      "step": 181
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.419642857142858e-05,
+      "loss": 1.8021,
+      "step": 182
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.416454081632652e-05,
+      "loss": 1.7088,
+      "step": 183
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.41326530612245e-05,
+      "loss": 1.8425,
+      "step": 184
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.410076530612245e-05,
+      "loss": 1.7693,
+      "step": 185
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.406887755102041e-05,
+      "loss": 1.6315,
+      "step": 186
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.403698979591837e-05,
+      "loss": 1.7466,
+      "step": 187
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.400510204081633e-05,
+      "loss": 1.8778,
+      "step": 188
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.39732142857143e-05,
+      "loss": 1.8097,
+      "step": 189
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.394132653061226e-05,
+      "loss": 1.8514,
+      "step": 190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.39094387755102e-05,
+      "loss": 1.7642,
+      "step": 191
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.387755102040817e-05,
+      "loss": 1.7704,
+      "step": 192
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.384566326530613e-05,
+      "loss": 1.8643,
+      "step": 193
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.381377551020409e-05,
+      "loss": 1.7102,
+      "step": 194
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.378188775510204e-05,
+      "loss": 1.8454,
+      "step": 195
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.375e-05,
+      "loss": 1.7541,
+      "step": 196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.371811224489796e-05,
+      "loss": 1.6068,
+      "step": 197
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.368622448979593e-05,
+      "loss": 1.8284,
+      "step": 198
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.365433673469388e-05,
+      "loss": 1.7188,
+      "step": 199
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.362244897959183e-05,
+      "loss": 1.7339,
+      "step": 200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.35905612244898e-05,
+      "loss": 1.851,
+      "step": 201
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.355867346938776e-05,
+      "loss": 1.7182,
+      "step": 202
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.352678571428572e-05,
+      "loss": 1.6969,
+      "step": 203
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.349489795918368e-05,
+      "loss": 1.7887,
+      "step": 204
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.346301020408163e-05,
+      "loss": 1.7422,
+      "step": 205
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.343112244897959e-05,
+      "loss": 1.6576,
+      "step": 206
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.339923469387756e-05,
+      "loss": 1.8358,
+      "step": 207
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.336734693877551e-05,
+      "loss": 1.8129,
+      "step": 208
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.333545918367348e-05,
+      "loss": 1.7009,
+      "step": 209
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.330357142857144e-05,
+      "loss": 1.6702,
+      "step": 210
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.32716836734694e-05,
+      "loss": 1.7322,
+      "step": 211
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.323979591836735e-05,
+      "loss": 1.7945,
+      "step": 212
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.320790816326531e-05,
+      "loss": 1.7738,
+      "step": 213
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.317602040816327e-05,
+      "loss": 1.7397,
+      "step": 214
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.314413265306124e-05,
+      "loss": 1.8265,
+      "step": 215
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.311224489795918e-05,
+      "loss": 1.8222,
+      "step": 216
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.308035714285714e-05,
+      "loss": 1.7259,
+      "step": 217
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.304846938775511e-05,
+      "loss": 1.7603,
+      "step": 218
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.301658163265307e-05,
+      "loss": 1.7456,
+      "step": 219
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.298469387755103e-05,
+      "loss": 1.7605,
+      "step": 220
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.295280612244899e-05,
+      "loss": 1.7271,
+      "step": 221
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.292091836734694e-05,
+      "loss": 1.778,
+      "step": 222
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.28890306122449e-05,
+      "loss": 1.762,
+      "step": 223
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.285714285714286e-05,
+      "loss": 1.8079,
+      "step": 224
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.282525510204082e-05,
+      "loss": 1.7605,
+      "step": 225
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.279336734693877e-05,
+      "loss": 1.8419,
+      "step": 226
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.276147959183675e-05,
+      "loss": 1.7739,
+      "step": 227
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.27295918367347e-05,
+      "loss": 1.7588,
+      "step": 228
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.269770408163265e-05,
+      "loss": 1.6675,
+      "step": 229
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.266581632653062e-05,
+      "loss": 1.7239,
+      "step": 230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.263392857142858e-05,
+      "loss": 1.7963,
+      "step": 231
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.260204081632653e-05,
+      "loss": 1.6988,
+      "step": 232
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.257015306122449e-05,
+      "loss": 1.8134,
+      "step": 233
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.253826530612245e-05,
+      "loss": 1.7699,
+      "step": 234
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.250637755102042e-05,
+      "loss": 1.8085,
+      "step": 235
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.247448979591838e-05,
+      "loss": 1.6914,
+      "step": 236
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.244260204081632e-05,
+      "loss": 1.5951,
+      "step": 237
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.24107142857143e-05,
+      "loss": 1.7358,
+      "step": 238
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.237882653061225e-05,
+      "loss": 1.7977,
+      "step": 239
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.234693877551021e-05,
+      "loss": 1.7313,
+      "step": 240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.231505102040817e-05,
+      "loss": 1.76,
+      "step": 241
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.228316326530612e-05,
+      "loss": 1.7217,
+      "step": 242
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.225127551020408e-05,
+      "loss": 1.7061,
+      "step": 243
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.221938775510205e-05,
+      "loss": 1.6406,
+      "step": 244
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.21875e-05,
+      "loss": 1.7745,
+      "step": 245
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.215561224489796e-05,
+      "loss": 1.6836,
+      "step": 246
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.212372448979593e-05,
+      "loss": 1.9216,
+      "step": 247
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.209183673469388e-05,
+      "loss": 1.6483,
+      "step": 248
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.205994897959184e-05,
+      "loss": 1.8161,
+      "step": 249
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.20280612244898e-05,
+      "loss": 1.7031,
+      "step": 250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.199617346938776e-05,
+      "loss": 1.8476,
+      "step": 251
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.196428571428572e-05,
+      "loss": 1.7902,
+      "step": 252
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.193239795918369e-05,
+      "loss": 1.6055,
+      "step": 253
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.190051020408163e-05,
+      "loss": 1.7595,
+      "step": 254
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.186862244897959e-05,
+      "loss": 1.7044,
+      "step": 255
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.183673469387756e-05,
+      "loss": 1.7589,
+      "step": 256
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.180484693877552e-05,
+      "loss": 1.7683,
+      "step": 257
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.177295918367348e-05,
+      "loss": 1.711,
+      "step": 258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.174107142857143e-05,
+      "loss": 1.7786,
+      "step": 259
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.170918367346939e-05,
+      "loss": 1.7657,
+      "step": 260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.167729591836736e-05,
+      "loss": 1.7338,
+      "step": 261
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.16454081632653e-05,
+      "loss": 1.7197,
+      "step": 262
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.161352040816326e-05,
+      "loss": 1.7484,
+      "step": 263
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.158163265306124e-05,
+      "loss": 1.7471,
+      "step": 264
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.154974489795919e-05,
+      "loss": 1.8983,
+      "step": 265
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.151785714285715e-05,
+      "loss": 1.6745,
+      "step": 266
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.148596938775511e-05,
+      "loss": 1.711,
+      "step": 267
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.145408163265307e-05,
+      "loss": 1.7758,
+      "step": 268
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.142219387755102e-05,
+      "loss": 1.7229,
+      "step": 269
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.139030612244898e-05,
+      "loss": 1.7072,
+      "step": 270
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.135841836734694e-05,
+      "loss": 1.7081,
+      "step": 271
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.13265306122449e-05,
+      "loss": 1.7859,
+      "step": 272
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.129464285714287e-05,
+      "loss": 1.7052,
+      "step": 273
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.126275510204083e-05,
+      "loss": 1.7401,
+      "step": 274
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.123086734693877e-05,
+      "loss": 1.7259,
+      "step": 275
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.119897959183674e-05,
+      "loss": 1.7349,
+      "step": 276
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.11670918367347e-05,
+      "loss": 1.8151,
+      "step": 277
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.113520408163266e-05,
+      "loss": 1.7264,
+      "step": 278
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.110331632653061e-05,
+      "loss": 1.8593,
+      "step": 279
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.107142857142857e-05,
+      "loss": 1.765,
+      "step": 280
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.103954081632653e-05,
+      "loss": 1.7441,
+      "step": 281
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.10076530612245e-05,
+      "loss": 1.7231,
+      "step": 282
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.097576530612245e-05,
+      "loss": 1.6782,
+      "step": 283
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.094387755102042e-05,
+      "loss": 1.6981,
+      "step": 284
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.091198979591837e-05,
+      "loss": 1.6902,
+      "step": 285
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.088010204081633e-05,
+      "loss": 1.7088,
+      "step": 286
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.084821428571429e-05,
+      "loss": 1.8195,
+      "step": 287
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.081632653061225e-05,
+      "loss": 1.7754,
+      "step": 288
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.07844387755102e-05,
+      "loss": 1.6354,
+      "step": 289
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.075255102040818e-05,
+      "loss": 1.8996,
+      "step": 290
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.072066326530612e-05,
+      "loss": 1.8391,
+      "step": 291
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.068877551020408e-05,
+      "loss": 1.6725,
+      "step": 292
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.065688775510205e-05,
+      "loss": 1.7103,
+      "step": 293
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.062500000000001e-05,
+      "loss": 1.7725,
+      "step": 294
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.059311224489797e-05,
+      "loss": 1.6804,
+      "step": 295
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.056122448979592e-05,
+      "loss": 1.7693,
+      "step": 296
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.052933673469388e-05,
+      "loss": 1.7712,
+      "step": 297
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.049744897959184e-05,
+      "loss": 1.7508,
+      "step": 298
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.046556122448981e-05,
+      "loss": 1.6661,
+      "step": 299
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.043367346938775e-05,
+      "loss": 1.7509,
+      "step": 300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 9.040178571428571e-05,
+      "loss": 1.5842,
+      "step": 301
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.036989795918368e-05,
+      "loss": 1.7407,
+      "step": 302
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.033801020408164e-05,
+      "loss": 1.733,
+      "step": 303
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.030612244897958e-05,
+      "loss": 1.6372,
+      "step": 304
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.027423469387756e-05,
+      "loss": 1.6975,
+      "step": 305
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.024234693877551e-05,
+      "loss": 1.7982,
+      "step": 306
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.021045918367348e-05,
+      "loss": 1.7448,
+      "step": 307
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.017857142857143e-05,
+      "loss": 1.6857,
+      "step": 308
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.014668367346939e-05,
+      "loss": 1.7282,
+      "step": 309
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.011479591836736e-05,
+      "loss": 1.7198,
+      "step": 310
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.008290816326532e-05,
+      "loss": 1.7697,
+      "step": 311
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.005102040816327e-05,
+      "loss": 1.6301,
+      "step": 312
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.001913265306123e-05,
+      "loss": 1.6895,
+      "step": 313
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.998724489795919e-05,
+      "loss": 1.6554,
+      "step": 314
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.995535714285715e-05,
+      "loss": 1.7577,
+      "step": 315
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.99234693877551e-05,
+      "loss": 1.7406,
+      "step": 316
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.989158163265306e-05,
+      "loss": 1.6299,
+      "step": 317
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.985969387755102e-05,
+      "loss": 1.6744,
+      "step": 318
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.982780612244899e-05,
+      "loss": 1.7713,
+      "step": 319
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.979591836734695e-05,
+      "loss": 1.66,
+      "step": 320
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.976403061224489e-05,
+      "loss": 1.6957,
+      "step": 321
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.973214285714286e-05,
+      "loss": 1.6484,
+      "step": 322
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.970025510204082e-05,
+      "loss": 1.5804,
+      "step": 323
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.966836734693878e-05,
+      "loss": 1.762,
+      "step": 324
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.963647959183674e-05,
+      "loss": 1.8231,
+      "step": 325
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.96045918367347e-05,
+      "loss": 1.7701,
+      "step": 326
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.957270408163265e-05,
+      "loss": 1.7539,
+      "step": 327
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.954081632653062e-05,
+      "loss": 1.7987,
+      "step": 328
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.950892857142857e-05,
+      "loss": 1.6533,
+      "step": 329
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.947704081632653e-05,
+      "loss": 1.8039,
+      "step": 330
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.94451530612245e-05,
+      "loss": 1.7719,
+      "step": 331
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.941326530612245e-05,
+      "loss": 1.7582,
+      "step": 332
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.938137755102041e-05,
+      "loss": 1.6774,
+      "step": 333
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.934948979591837e-05,
+      "loss": 1.6298,
+      "step": 334
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.931760204081633e-05,
+      "loss": 1.7936,
+      "step": 335
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.92857142857143e-05,
+      "loss": 1.8324,
+      "step": 336
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.925382653061224e-05,
+      "loss": 1.7726,
+      "step": 337
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.92219387755102e-05,
+      "loss": 1.7102,
+      "step": 338
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.919005102040817e-05,
+      "loss": 1.6567,
+      "step": 339
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.915816326530613e-05,
+      "loss": 1.6084,
+      "step": 340
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.912627551020409e-05,
+      "loss": 1.5728,
+      "step": 341
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.909438775510205e-05,
+      "loss": 1.6946,
+      "step": 342
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.90625e-05,
+      "loss": 1.7973,
+      "step": 343
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.903061224489796e-05,
+      "loss": 1.796,
+      "step": 344
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.899872448979593e-05,
+      "loss": 1.691,
+      "step": 345
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.896683673469388e-05,
+      "loss": 1.7176,
+      "step": 346
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.893494897959183e-05,
+      "loss": 1.634,
+      "step": 347
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.89030612244898e-05,
+      "loss": 1.766,
+      "step": 348
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.887117346938776e-05,
+      "loss": 1.8859,
+      "step": 349
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.883928571428571e-05,
+      "loss": 1.7391,
+      "step": 350
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.880739795918368e-05,
+      "loss": 1.7228,
+      "step": 351
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.877551020408164e-05,
+      "loss": 1.7388,
+      "step": 352
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.87436224489796e-05,
+      "loss": 1.6822,
+      "step": 353
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.871173469387755e-05,
+      "loss": 1.7068,
+      "step": 354
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.867984693877551e-05,
+      "loss": 1.8398,
+      "step": 355
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.864795918367348e-05,
+      "loss": 1.7962,
+      "step": 356
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.861607142857144e-05,
+      "loss": 1.6781,
+      "step": 357
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.85841836734694e-05,
+      "loss": 1.6459,
+      "step": 358
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.855229591836735e-05,
+      "loss": 1.7337,
+      "step": 359
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.852040816326531e-05,
+      "loss": 1.678,
+      "step": 360
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.848852040816327e-05,
+      "loss": 1.7358,
+      "step": 361
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.845663265306123e-05,
+      "loss": 1.7226,
+      "step": 362
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.842474489795918e-05,
+      "loss": 1.7129,
+      "step": 363
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.839285714285714e-05,
+      "loss": 1.6493,
+      "step": 364
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.836096938775511e-05,
+      "loss": 1.7302,
+      "step": 365
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.832908163265307e-05,
+      "loss": 1.6787,
+      "step": 366
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.829719387755102e-05,
+      "loss": 1.6729,
+      "step": 367
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.826530612244899e-05,
+      "loss": 1.7308,
+      "step": 368
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.823341836734694e-05,
+      "loss": 1.6517,
+      "step": 369
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.82015306122449e-05,
+      "loss": 1.7198,
+      "step": 370
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.816964285714286e-05,
+      "loss": 1.6133,
+      "step": 371
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.813775510204082e-05,
+      "loss": 1.6732,
+      "step": 372
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.810586734693878e-05,
+      "loss": 1.729,
+      "step": 373
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.807397959183675e-05,
+      "loss": 1.699,
+      "step": 374
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.804209183673469e-05,
+      "loss": 1.7103,
+      "step": 375
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.801020408163265e-05,
+      "loss": 1.7119,
+      "step": 376
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.797831632653062e-05,
+      "loss": 1.646,
+      "step": 377
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.794642857142858e-05,
+      "loss": 1.5956,
+      "step": 378
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.791454081632654e-05,
+      "loss": 1.7735,
+      "step": 379
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.788265306122449e-05,
+      "loss": 1.7089,
+      "step": 380
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.785076530612245e-05,
+      "loss": 1.7068,
+      "step": 381
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.781887755102042e-05,
+      "loss": 1.6947,
+      "step": 382
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.778698979591838e-05,
+      "loss": 1.7792,
+      "step": 383
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.775510204081632e-05,
+      "loss": 1.6787,
+      "step": 384
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.77232142857143e-05,
+      "loss": 1.825,
+      "step": 385
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.769132653061225e-05,
+      "loss": 1.7203,
+      "step": 386
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.765943877551021e-05,
+      "loss": 1.6524,
+      "step": 387
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.762755102040817e-05,
+      "loss": 1.6665,
+      "step": 388
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.759566326530613e-05,
+      "loss": 1.7011,
+      "step": 389
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.756377551020408e-05,
+      "loss": 1.7353,
+      "step": 390
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.753188775510205e-05,
+      "loss": 1.7717,
+      "step": 391
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.75e-05,
+      "loss": 1.6323,
+      "step": 392
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.746811224489796e-05,
+      "loss": 1.6933,
+      "step": 393
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.743622448979593e-05,
+      "loss": 1.6924,
+      "step": 394
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.740433673469389e-05,
+      "loss": 1.797,
+      "step": 395
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.737244897959183e-05,
+      "loss": 1.6189,
+      "step": 396
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.73405612244898e-05,
+      "loss": 1.6157,
+      "step": 397
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.730867346938776e-05,
+      "loss": 1.5831,
+      "step": 398
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.727678571428572e-05,
+      "loss": 1.6539,
+      "step": 399
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.724489795918367e-05,
+      "loss": 1.6136,
+      "step": 400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.721301020408163e-05,
+      "loss": 1.6864,
+      "step": 401
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.718112244897959e-05,
+      "loss": 1.5279,
+      "step": 402
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.714923469387756e-05,
+      "loss": 1.6846,
+      "step": 403
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.711734693877552e-05,
+      "loss": 1.7603,
+      "step": 404
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.708545918367348e-05,
+      "loss": 1.7768,
+      "step": 405
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.705357142857143e-05,
+      "loss": 1.751,
+      "step": 406
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.702168367346939e-05,
+      "loss": 1.6345,
+      "step": 407
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.698979591836735e-05,
+      "loss": 1.6333,
+      "step": 408
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.695790816326531e-05,
+      "loss": 1.6771,
+      "step": 409
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.692602040816327e-05,
+      "loss": 1.6764,
+      "step": 410
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.689413265306124e-05,
+      "loss": 1.7229,
+      "step": 411
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.68622448979592e-05,
+      "loss": 1.6311,
+      "step": 412
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.683035714285714e-05,
+      "loss": 1.7222,
+      "step": 413
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.679846938775511e-05,
+      "loss": 1.6414,
+      "step": 414
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.676658163265307e-05,
+      "loss": 1.7688,
+      "step": 415
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.673469387755102e-05,
+      "loss": 1.7213,
+      "step": 416
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.670280612244898e-05,
+      "loss": 1.6997,
+      "step": 417
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.667091836734694e-05,
+      "loss": 1.7518,
+      "step": 418
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.66390306122449e-05,
+      "loss": 1.7478,
+      "step": 419
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.660714285714287e-05,
+      "loss": 1.7175,
+      "step": 420
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.657525510204081e-05,
+      "loss": 1.6769,
+      "step": 421
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.654336734693877e-05,
+      "loss": 1.7689,
+      "step": 422
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.651147959183674e-05,
+      "loss": 1.6298,
+      "step": 423
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.64795918367347e-05,
+      "loss": 1.6876,
+      "step": 424
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.644770408163266e-05,
+      "loss": 1.7329,
+      "step": 425
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.641581632653062e-05,
+      "loss": 1.7087,
+      "step": 426
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.638392857142857e-05,
+      "loss": 1.6905,
+      "step": 427
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.635204081632653e-05,
+      "loss": 1.7005,
+      "step": 428
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.63201530612245e-05,
+      "loss": 1.6099,
+      "step": 429
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.628826530612245e-05,
+      "loss": 1.7045,
+      "step": 430
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.625637755102042e-05,
+      "loss": 1.5877,
+      "step": 431
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.622448979591838e-05,
+      "loss": 1.6743,
+      "step": 432
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.619260204081633e-05,
+      "loss": 1.7205,
+      "step": 433
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.616071428571429e-05,
+      "loss": 1.566,
+      "step": 434
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.612882653061225e-05,
+      "loss": 1.5954,
+      "step": 435
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.60969387755102e-05,
+      "loss": 1.6791,
+      "step": 436
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.606505102040818e-05,
+      "loss": 1.5968,
+      "step": 437
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.603316326530612e-05,
+      "loss": 1.7167,
+      "step": 438
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.600127551020408e-05,
+      "loss": 1.6694,
+      "step": 439
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.596938775510205e-05,
+      "loss": 1.7976,
+      "step": 440
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.593750000000001e-05,
+      "loss": 1.7609,
+      "step": 441
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.590561224489795e-05,
+      "loss": 1.8266,
+      "step": 442
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.587372448979592e-05,
+      "loss": 1.7249,
+      "step": 443
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.584183673469388e-05,
+      "loss": 1.7351,
+      "step": 444
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.580994897959184e-05,
+      "loss": 1.7943,
+      "step": 445
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.57780612244898e-05,
+      "loss": 1.7614,
+      "step": 446
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.574617346938775e-05,
+      "loss": 1.6992,
+      "step": 447
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 1.6828,
+      "step": 448
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.568239795918368e-05,
+      "loss": 1.7782,
+      "step": 449
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.565051020408164e-05,
+      "loss": 1.6556,
+      "step": 450
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.561862244897959e-05,
+      "loss": 1.6431,
+      "step": 451
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.558673469387756e-05,
+      "loss": 1.6783,
+      "step": 452
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.555484693877551e-05,
+      "loss": 1.6774,
+      "step": 453
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.552295918367347e-05,
+      "loss": 1.6588,
+      "step": 454
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.549107142857143e-05,
+      "loss": 1.6892,
+      "step": 455
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.545918367346939e-05,
+      "loss": 1.6022,
+      "step": 456
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.542729591836736e-05,
+      "loss": 1.7083,
+      "step": 457
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.539540816326532e-05,
+      "loss": 1.746,
+      "step": 458
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.536352040816326e-05,
+      "loss": 1.7238,
+      "step": 459
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.533163265306123e-05,
+      "loss": 1.6144,
+      "step": 460
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.529974489795919e-05,
+      "loss": 1.5568,
+      "step": 461
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.526785714285715e-05,
+      "loss": 1.6344,
+      "step": 462
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.52359693877551e-05,
+      "loss": 1.6804,
+      "step": 463
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.520408163265306e-05,
+      "loss": 1.5241,
+      "step": 464
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.517219387755102e-05,
+      "loss": 1.7548,
+      "step": 465
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.514030612244899e-05,
+      "loss": 1.6902,
+      "step": 466
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.510841836734694e-05,
+      "loss": 1.761,
+      "step": 467
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.50765306122449e-05,
+      "loss": 1.6532,
+      "step": 468
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.504464285714287e-05,
+      "loss": 1.6625,
+      "step": 469
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.501275510204082e-05,
+      "loss": 1.6393,
+      "step": 470
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.498086734693878e-05,
+      "loss": 1.6935,
+      "step": 471
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.494897959183674e-05,
+      "loss": 1.6177,
+      "step": 472
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.49170918367347e-05,
+      "loss": 1.6962,
+      "step": 473
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.488520408163265e-05,
+      "loss": 1.6814,
+      "step": 474
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.485331632653063e-05,
+      "loss": 1.631,
+      "step": 475
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.482142857142857e-05,
+      "loss": 1.6923,
+      "step": 476
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.478954081632653e-05,
+      "loss": 1.6235,
+      "step": 477
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.47576530612245e-05,
+      "loss": 1.6599,
+      "step": 478
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.472576530612246e-05,
+      "loss": 1.6514,
+      "step": 479
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.469387755102041e-05,
+      "loss": 1.6919,
+      "step": 480
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.466198979591837e-05,
+      "loss": 1.647,
+      "step": 481
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.463010204081633e-05,
+      "loss": 1.6861,
+      "step": 482
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.45982142857143e-05,
+      "loss": 1.7036,
+      "step": 483
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.456632653061224e-05,
+      "loss": 1.7445,
+      "step": 484
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.45344387755102e-05,
+      "loss": 1.6108,
+      "step": 485
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8.450255102040817e-05,
+      "loss": 1.5834,
+      "step": 486
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.447066326530613e-05,
+      "loss": 1.6655,
+      "step": 487
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.443877551020409e-05,
+      "loss": 1.6679,
+      "step": 488
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.440688775510205e-05,
+      "loss": 1.6832,
+      "step": 489
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.4375e-05,
+      "loss": 1.7003,
+      "step": 490
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.434311224489796e-05,
+      "loss": 1.7136,
+      "step": 491
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.431122448979592e-05,
+      "loss": 1.6938,
+      "step": 492
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.427933673469388e-05,
+      "loss": 1.6589,
+      "step": 493
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.424744897959184e-05,
+      "loss": 1.7169,
+      "step": 494
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 8.42155612244898e-05,
+      "loss": 1.6738,
+      "step": 495
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 8.418367346938776e-05,
+      "loss": 1.5954,
+      "step": 496
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 8.415178571428571e-05,
+      "loss": 1.6149,
+      "step": 497
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 8.411989795918368e-05,
+      "loss": 1.7302,
+      "step": 498
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 8.408801020408164e-05,
+      "loss": 1.6168,
+      "step": 499
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 8.40561224489796e-05,
+      "loss": 1.572,
+      "step": 500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 8.402423469387755e-05,
+      "loss": 1.7744,
+      "step": 501
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 8.399234693877551e-05,
+      "loss": 1.711,
+      "step": 502
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 8.396045918367348e-05,
+      "loss": 1.6042,
+      "step": 503
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 8.392857142857144e-05,
+      "loss": 1.7177,
+      "step": 504
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 8.389668367346938e-05,
+      "loss": 1.7372,
+      "step": 505
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 8.386479591836736e-05,
+      "loss": 1.6844,
+      "step": 506
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 8.383290816326531e-05,
+      "loss": 1.5831,
+      "step": 507
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 8.380102040816327e-05,
+      "loss": 1.6598,
+      "step": 508
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 8.376913265306123e-05,
+      "loss": 1.5603,
+      "step": 509
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 8.373724489795919e-05,
+      "loss": 1.6364,
+      "step": 510
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 8.370535714285714e-05,
+      "loss": 1.591,
+      "step": 511
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 8.367346938775511e-05,
+      "loss": 1.8037,
+      "step": 512
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 8.364158163265306e-05,
+      "loss": 1.5908,
+      "step": 513
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 8.360969387755102e-05,
+      "loss": 1.6017,
+      "step": 514
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.357780612244899e-05,
+      "loss": 1.718,
+      "step": 515
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.354591836734695e-05,
+      "loss": 1.669,
+      "step": 516
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.35140306122449e-05,
+      "loss": 1.5423,
+      "step": 517
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.348214285714286e-05,
+      "loss": 1.6151,
+      "step": 518
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.345025510204082e-05,
+      "loss": 1.623,
+      "step": 519
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 8.341836734693878e-05,
+      "loss": 1.5912,
+      "step": 520
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 8.338647959183675e-05,
+      "loss": 1.6331,
+      "step": 521
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 8.335459183673469e-05,
+      "loss": 1.6982,
+      "step": 522
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 8.332270408163265e-05,
+      "loss": 1.5511,
+      "step": 523
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 8.329081632653062e-05,
+      "loss": 1.5958,
+      "step": 524
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 8.325892857142858e-05,
+      "loss": 1.5648,
+      "step": 525
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 8.322704081632652e-05,
+      "loss": 1.4833,
+      "step": 526
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 8.31951530612245e-05,
+      "loss": 1.5844,
+      "step": 527
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 8.316326530612245e-05,
+      "loss": 1.6717,
+      "step": 528
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 8.313137755102042e-05,
+      "loss": 1.6574,
+      "step": 529
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 8.309948979591837e-05,
+      "loss": 1.622,
+      "step": 530
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 8.306760204081633e-05,
+      "loss": 1.5806,
+      "step": 531
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 8.30357142857143e-05,
+      "loss": 1.6023,
+      "step": 532
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 8.300382653061225e-05,
+      "loss": 1.6616,
+      "step": 533
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 8.297193877551021e-05,
+      "loss": 1.6409,
+      "step": 534
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 8.294005102040817e-05,
+      "loss": 1.6206,
+      "step": 535
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 8.290816326530613e-05,
+      "loss": 1.5892,
+      "step": 536
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 8.287627551020408e-05,
+      "loss": 1.6957,
+      "step": 537
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 8.284438775510204e-05,
+      "loss": 1.713,
+      "step": 538
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 8.28125e-05,
+      "loss": 1.6176,
+      "step": 539
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 8.278061224489796e-05,
+      "loss": 1.7442,
+      "step": 540
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.274872448979593e-05,
+      "loss": 1.6822,
+      "step": 541
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.271683673469389e-05,
+      "loss": 1.7747,
+      "step": 542
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.268494897959183e-05,
+      "loss": 1.5352,
+      "step": 543
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.26530612244898e-05,
+      "loss": 1.6603,
+      "step": 544
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.262117346938776e-05,
+      "loss": 1.6696,
+      "step": 545
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.258928571428572e-05,
+      "loss": 1.6286,
+      "step": 546
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.255739795918368e-05,
+      "loss": 1.6475,
+      "step": 547
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.252551020408163e-05,
+      "loss": 1.7275,
+      "step": 548
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.249362244897959e-05,
+      "loss": 1.681,
+      "step": 549
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.246173469387756e-05,
+      "loss": 1.5995,
+      "step": 550
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.24298469387755e-05,
+      "loss": 1.6233,
+      "step": 551
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.239795918367348e-05,
+      "loss": 1.6292,
+      "step": 552
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.236607142857144e-05,
+      "loss": 1.6524,
+      "step": 553
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.23341836734694e-05,
+      "loss": 1.735,
+      "step": 554
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.230229591836735e-05,
+      "loss": 1.5992,
+      "step": 555
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.227040816326531e-05,
+      "loss": 1.7104,
+      "step": 556
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.223852040816327e-05,
+      "loss": 1.6503,
+      "step": 557
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.220663265306124e-05,
+      "loss": 1.6799,
+      "step": 558
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 8.217474489795918e-05,
+      "loss": 1.7083,
+      "step": 559
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 8.214285714285714e-05,
+      "loss": 1.539,
+      "step": 560
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 8.211096938775511e-05,
+      "loss": 1.6676,
+      "step": 561
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 8.207908163265307e-05,
+      "loss": 1.5834,
+      "step": 562
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 8.204719387755103e-05,
+      "loss": 1.7084,
+      "step": 563
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 8.201530612244898e-05,
+      "loss": 1.746,
+      "step": 564
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 8.198341836734694e-05,
+      "loss": 1.6726,
+      "step": 565
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 8.19515306122449e-05,
+      "loss": 1.7247,
+      "step": 566
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 8.191964285714287e-05,
+      "loss": 1.7044,
+      "step": 567
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.188775510204081e-05,
+      "loss": 1.7063,
+      "step": 568
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.185586734693877e-05,
+      "loss": 1.6217,
+      "step": 569
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 8.182397959183674e-05,
+      "loss": 1.6446,
+      "step": 570
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 8.17920918367347e-05,
+      "loss": 1.5033,
+      "step": 571
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 8.176020408163265e-05,
+      "loss": 1.6669,
+      "step": 572
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 8.172831632653062e-05,
+      "loss": 1.6104,
+      "step": 573
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 8.169642857142857e-05,
+      "loss": 1.7092,
+      "step": 574
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 8.166454081632653e-05,
+      "loss": 1.6563,
+      "step": 575
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.163265306122449e-05,
+      "loss": 1.6982,
+      "step": 576
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.160076530612245e-05,
+      "loss": 1.6593,
+      "step": 577
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 8.156887755102042e-05,
+      "loss": 1.6762,
+      "step": 578
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 8.153698979591838e-05,
+      "loss": 1.6478,
+      "step": 579
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 8.150510204081633e-05,
+      "loss": 1.6337,
+      "step": 580
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 8.147321428571429e-05,
+      "loss": 1.6629,
+      "step": 581
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 8.144132653061225e-05,
+      "loss": 1.6911,
+      "step": 582
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 8.140943877551021e-05,
+      "loss": 1.6673,
+      "step": 583
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 8.137755102040817e-05,
+      "loss": 1.669,
+      "step": 584
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.134566326530612e-05,
+      "loss": 1.7474,
+      "step": 585
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 8.131377551020408e-05,
+      "loss": 1.7361,
+      "step": 586
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 8.128188775510205e-05,
+      "loss": 1.5395,
+      "step": 587
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 1.667,
+      "step": 588
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 8.121811224489795e-05,
+      "loss": 1.6239,
+      "step": 589
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 8.118622448979593e-05,
+      "loss": 1.6123,
+      "step": 590
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 8.115433673469388e-05,
+      "loss": 1.6097,
+      "step": 591
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 8.112244897959184e-05,
+      "loss": 1.5834,
+      "step": 592
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 8.10905612244898e-05,
+      "loss": 1.616,
+      "step": 593
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 8.105867346938776e-05,
+      "loss": 1.7001,
+      "step": 594
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 8.102678571428571e-05,
+      "loss": 1.5911,
+      "step": 595
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 8.099489795918369e-05,
+      "loss": 1.6972,
+      "step": 596
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 8.096301020408163e-05,
+      "loss": 1.5402,
+      "step": 597
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 8.093112244897959e-05,
+      "loss": 1.4818,
+      "step": 598
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 8.089923469387756e-05,
+      "loss": 1.6306,
+      "step": 599
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 8.086734693877552e-05,
+      "loss": 1.663,
+      "step": 600
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 8.083545918367347e-05,
+      "loss": 1.7275,
+      "step": 601
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 8.080357142857143e-05,
+      "loss": 1.7,
+      "step": 602
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 8.077168367346939e-05,
+      "loss": 1.6751,
+      "step": 603
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 8.073979591836736e-05,
+      "loss": 1.6002,
+      "step": 604
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 8.07079081632653e-05,
+      "loss": 1.5782,
+      "step": 605
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 8.067602040816326e-05,
+      "loss": 1.5804,
+      "step": 606
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 8.064413265306123e-05,
+      "loss": 1.607,
+      "step": 607
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 8.061224489795919e-05,
+      "loss": 1.5291,
+      "step": 608
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 8.058035714285715e-05,
+      "loss": 1.6181,
+      "step": 609
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 8.05484693877551e-05,
+      "loss": 1.5749,
+      "step": 610
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 8.051658163265306e-05,
+      "loss": 1.6551,
+      "step": 611
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 8.048469387755102e-05,
+      "loss": 1.6582,
+      "step": 612
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 8.0452806122449e-05,
+      "loss": 1.5973,
+      "step": 613
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 8.042091836734694e-05,
+      "loss": 1.6114,
+      "step": 614
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 8.03890306122449e-05,
+      "loss": 1.6425,
+      "step": 615
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 8.035714285714287e-05,
+      "loss": 1.6389,
+      "step": 616
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 8.032525510204082e-05,
+      "loss": 1.6543,
+      "step": 617
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 8.029336734693877e-05,
+      "loss": 1.7075,
+      "step": 618
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 8.026147959183674e-05,
+      "loss": 1.7116,
+      "step": 619
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 8.02295918367347e-05,
+      "loss": 1.6724,
+      "step": 620
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 8.019770408163266e-05,
+      "loss": 1.6999,
+      "step": 621
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 8.016581632653061e-05,
+      "loss": 1.6217,
+      "step": 622
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 8.013392857142857e-05,
+      "loss": 1.4954,
+      "step": 623
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 8.010204081632653e-05,
+      "loss": 1.5963,
+      "step": 624
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 8.00701530612245e-05,
+      "loss": 1.5706,
+      "step": 625
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 8.003826530612246e-05,
+      "loss": 1.5399,
+      "step": 626
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 8.000637755102042e-05,
+      "loss": 1.5439,
+      "step": 627
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.997448979591837e-05,
+      "loss": 1.6247,
+      "step": 628
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.994260204081633e-05,
+      "loss": 1.5136,
+      "step": 629
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.991071428571429e-05,
+      "loss": 1.6517,
+      "step": 630
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.987882653061225e-05,
+      "loss": 1.4643,
+      "step": 631
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.98469387755102e-05,
+      "loss": 1.6165,
+      "step": 632
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.981505102040817e-05,
+      "loss": 1.6965,
+      "step": 633
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.978316326530613e-05,
+      "loss": 1.6557,
+      "step": 634
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.975127551020408e-05,
+      "loss": 1.589,
+      "step": 635
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.971938775510205e-05,
+      "loss": 1.5587,
+      "step": 636
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.96875e-05,
+      "loss": 1.6603,
+      "step": 637
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.965561224489796e-05,
+      "loss": 1.628,
+      "step": 638
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.962372448979592e-05,
+      "loss": 1.5991,
+      "step": 639
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.959183673469388e-05,
+      "loss": 1.5695,
+      "step": 640
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.955994897959184e-05,
+      "loss": 1.5779,
+      "step": 641
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.952806122448981e-05,
+      "loss": 1.6066,
+      "step": 642
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.949617346938775e-05,
+      "loss": 1.5591,
+      "step": 643
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.946428571428571e-05,
+      "loss": 1.7178,
+      "step": 644
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.943239795918368e-05,
+      "loss": 1.5171,
+      "step": 645
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.940051020408164e-05,
+      "loss": 1.5264,
+      "step": 646
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.93686224489796e-05,
+      "loss": 1.6441,
+      "step": 647
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.933673469387755e-05,
+      "loss": 1.6895,
+      "step": 648
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.930484693877551e-05,
+      "loss": 1.5807,
+      "step": 649
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.927295918367348e-05,
+      "loss": 1.6415,
+      "step": 650
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.924107142857143e-05,
+      "loss": 1.5917,
+      "step": 651
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.920918367346939e-05,
+      "loss": 1.683,
+      "step": 652
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.917729591836736e-05,
+      "loss": 1.7768,
+      "step": 653
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.914540816326531e-05,
+      "loss": 1.6131,
+      "step": 654
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.911352040816327e-05,
+      "loss": 1.6095,
+      "step": 655
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.908163265306123e-05,
+      "loss": 1.5111,
+      "step": 656
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.904974489795919e-05,
+      "loss": 1.5455,
+      "step": 657
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.901785714285714e-05,
+      "loss": 1.6746,
+      "step": 658
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.898596938775512e-05,
+      "loss": 1.635,
+      "step": 659
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.895408163265306e-05,
+      "loss": 1.4963,
+      "step": 660
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.892219387755102e-05,
+      "loss": 1.6298,
+      "step": 661
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.889030612244899e-05,
+      "loss": 1.6339,
+      "step": 662
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.885841836734695e-05,
+      "loss": 1.6626,
+      "step": 663
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.882653061224489e-05,
+      "loss": 1.6076,
+      "step": 664
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.879464285714286e-05,
+      "loss": 1.6432,
+      "step": 665
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.876275510204082e-05,
+      "loss": 1.713,
+      "step": 666
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.873086734693878e-05,
+      "loss": 1.6984,
+      "step": 667
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.869897959183674e-05,
+      "loss": 1.6848,
+      "step": 668
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.86670918367347e-05,
+      "loss": 1.6054,
+      "step": 669
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.863520408163265e-05,
+      "loss": 1.6365,
+      "step": 670
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.860331632653062e-05,
+      "loss": 1.5795,
+      "step": 671
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 7.857142857142858e-05,
+      "loss": 1.5735,
+      "step": 672
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 7.853954081632652e-05,
+      "loss": 1.5976,
+      "step": 673
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 7.85076530612245e-05,
+      "loss": 1.6806,
+      "step": 674
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 7.847576530612245e-05,
+      "loss": 1.5685,
+      "step": 675
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 7.844387755102041e-05,
+      "loss": 1.5815,
+      "step": 676
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 7.841198979591837e-05,
+      "loss": 1.5752,
+      "step": 677
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 7.838010204081633e-05,
+      "loss": 1.73,
+      "step": 678
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 7.83482142857143e-05,
+      "loss": 1.6906,
+      "step": 679
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 7.831632653061226e-05,
+      "loss": 1.4941,
+      "step": 680
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 7.82844387755102e-05,
+      "loss": 1.6237,
+      "step": 681
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 7.825255102040817e-05,
+      "loss": 1.6532,
+      "step": 682
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 7.822066326530613e-05,
+      "loss": 1.5407,
+      "step": 683
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 7.818877551020409e-05,
+      "loss": 1.6433,
+      "step": 684
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 7.815688775510204e-05,
+      "loss": 1.5911,
+      "step": 685
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 7.8125e-05,
+      "loss": 1.5864,
+      "step": 686
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 7.809311224489796e-05,
+      "loss": 1.5878,
+      "step": 687
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 7.806122448979593e-05,
+      "loss": 1.575,
+      "step": 688
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 7.802933673469387e-05,
+      "loss": 1.6028,
+      "step": 689
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 7.799744897959183e-05,
+      "loss": 1.6365,
+      "step": 690
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 7.79655612244898e-05,
+      "loss": 1.5638,
+      "step": 691
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 7.793367346938776e-05,
+      "loss": 1.5945,
+      "step": 692
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 7.790178571428572e-05,
+      "loss": 1.5479,
+      "step": 693
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 7.786989795918368e-05,
+      "loss": 1.5982,
+      "step": 694
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 7.783801020408163e-05,
+      "loss": 1.5528,
+      "step": 695
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 7.780612244897959e-05,
+      "loss": 1.5581,
+      "step": 696
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 7.777423469387755e-05,
+      "loss": 1.5725,
+      "step": 697
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 7.774234693877551e-05,
+      "loss": 1.6888,
+      "step": 698
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 7.771045918367348e-05,
+      "loss": 1.5152,
+      "step": 699
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 7.767857142857144e-05,
+      "loss": 1.5641,
+      "step": 700
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 7.76466836734694e-05,
+      "loss": 1.644,
+      "step": 701
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 7.761479591836735e-05,
+      "loss": 1.6291,
+      "step": 702
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 7.758290816326531e-05,
+      "loss": 1.6123,
+      "step": 703
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 7.755102040816327e-05,
+      "loss": 1.5258,
+      "step": 704
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 7.751913265306124e-05,
+      "loss": 1.5356,
+      "step": 705
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 7.748724489795918e-05,
+      "loss": 1.6682,
+      "step": 706
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 7.745535714285714e-05,
+      "loss": 1.5796,
+      "step": 707
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 7.742346938775511e-05,
+      "loss": 1.6159,
+      "step": 708
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 7.739158163265307e-05,
+      "loss": 1.5525,
+      "step": 709
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 7.735969387755101e-05,
+      "loss": 1.7342,
+      "step": 710
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 7.732780612244899e-05,
+      "loss": 1.5667,
+      "step": 711
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.729591836734694e-05,
+      "loss": 1.5964,
+      "step": 712
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.72640306122449e-05,
+      "loss": 1.5842,
+      "step": 713
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.723214285714286e-05,
+      "loss": 1.541,
+      "step": 714
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.720025510204082e-05,
+      "loss": 1.6021,
+      "step": 715
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.716836734693877e-05,
+      "loss": 1.591,
+      "step": 716
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.713647959183675e-05,
+      "loss": 1.6173,
+      "step": 717
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.71045918367347e-05,
+      "loss": 1.5007,
+      "step": 718
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.707270408163265e-05,
+      "loss": 1.5286,
+      "step": 719
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.704081632653062e-05,
+      "loss": 1.5203,
+      "step": 720
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.700892857142858e-05,
+      "loss": 1.5964,
+      "step": 721
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.697704081632653e-05,
+      "loss": 1.5882,
+      "step": 722
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 7.694515306122449e-05,
+      "loss": 1.5834,
+      "step": 723
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 7.691326530612245e-05,
+      "loss": 1.729,
+      "step": 724
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 7.688137755102042e-05,
+      "loss": 1.5524,
+      "step": 725
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 7.684948979591838e-05,
+      "loss": 1.5673,
+      "step": 726
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.681760204081632e-05,
+      "loss": 1.5539,
+      "step": 727
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.67857142857143e-05,
+      "loss": 1.6296,
+      "step": 728
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.675382653061225e-05,
+      "loss": 1.5992,
+      "step": 729
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.672193877551021e-05,
+      "loss": 1.5906,
+      "step": 730
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.669005102040817e-05,
+      "loss": 1.491,
+      "step": 731
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.665816326530612e-05,
+      "loss": 1.5692,
+      "step": 732
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.662627551020408e-05,
+      "loss": 1.5475,
+      "step": 733
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.659438775510205e-05,
+      "loss": 1.6955,
+      "step": 734
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.65625e-05,
+      "loss": 1.6321,
+      "step": 735
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.653061224489796e-05,
+      "loss": 1.6069,
+      "step": 736
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.649872448979593e-05,
+      "loss": 1.5381,
+      "step": 737
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.646683673469388e-05,
+      "loss": 1.5282,
+      "step": 738
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.643494897959184e-05,
+      "loss": 1.5751,
+      "step": 739
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.64030612244898e-05,
+      "loss": 1.7236,
+      "step": 740
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.637117346938776e-05,
+      "loss": 1.4974,
+      "step": 741
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.633928571428572e-05,
+      "loss": 1.5998,
+      "step": 742
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.630739795918369e-05,
+      "loss": 1.5756,
+      "step": 743
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.627551020408163e-05,
+      "loss": 1.5504,
+      "step": 744
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.624362244897959e-05,
+      "loss": 1.596,
+      "step": 745
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.621173469387756e-05,
+      "loss": 1.5373,
+      "step": 746
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.617984693877552e-05,
+      "loss": 1.343,
+      "step": 747
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.614795918367347e-05,
+      "loss": 1.6185,
+      "step": 748
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.611607142857143e-05,
+      "loss": 1.5853,
+      "step": 749
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.608418367346939e-05,
+      "loss": 1.5827,
+      "step": 750
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.605229591836736e-05,
+      "loss": 1.5355,
+      "step": 751
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.60204081632653e-05,
+      "loss": 1.6442,
+      "step": 752
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.598852040816326e-05,
+      "loss": 1.5472,
+      "step": 753
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.595663265306123e-05,
+      "loss": 1.7043,
+      "step": 754
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.592474489795919e-05,
+      "loss": 1.618,
+      "step": 755
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.589285714285714e-05,
+      "loss": 1.5461,
+      "step": 756
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.586096938775511e-05,
+      "loss": 1.6406,
+      "step": 757
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.582908163265307e-05,
+      "loss": 1.6037,
+      "step": 758
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.579719387755102e-05,
+      "loss": 1.5769,
+      "step": 759
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.576530612244898e-05,
+      "loss": 1.6035,
+      "step": 760
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.573341836734694e-05,
+      "loss": 1.6155,
+      "step": 761
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.57015306122449e-05,
+      "loss": 1.5729,
+      "step": 762
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.566964285714287e-05,
+      "loss": 1.5519,
+      "step": 763
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.563775510204083e-05,
+      "loss": 1.6702,
+      "step": 764
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.560586734693877e-05,
+      "loss": 1.6641,
+      "step": 765
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.557397959183674e-05,
+      "loss": 1.6938,
+      "step": 766
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.55420918367347e-05,
+      "loss": 1.6093,
+      "step": 767
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.551020408163266e-05,
+      "loss": 1.5468,
+      "step": 768
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.547831632653061e-05,
+      "loss": 1.6407,
+      "step": 769
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.544642857142857e-05,
+      "loss": 1.6491,
+      "step": 770
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.541454081632653e-05,
+      "loss": 1.4902,
+      "step": 771
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.53826530612245e-05,
+      "loss": 1.562,
+      "step": 772
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.535076530612244e-05,
+      "loss": 1.5765,
+      "step": 773
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.531887755102042e-05,
+      "loss": 1.541,
+      "step": 774
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.528698979591837e-05,
+      "loss": 1.6346,
+      "step": 775
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.525510204081633e-05,
+      "loss": 1.482,
+      "step": 776
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 7.522321428571429e-05,
+      "loss": 1.6146,
+      "step": 777
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 7.519132653061225e-05,
+      "loss": 1.569,
+      "step": 778
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 7.51594387755102e-05,
+      "loss": 1.5714,
+      "step": 779
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 7.512755102040818e-05,
+      "loss": 1.5187,
+      "step": 780
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 7.509566326530612e-05,
+      "loss": 1.5353,
+      "step": 781
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 7.506377551020408e-05,
+      "loss": 1.471,
+      "step": 782
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.503188775510205e-05,
+      "loss": 1.5398,
+      "step": 783
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.5894,
+      "step": 784
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 7.496811224489796e-05,
+      "loss": 1.6253,
+      "step": 785
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 7.493622448979592e-05,
+      "loss": 1.5551,
+      "step": 786
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 7.490433673469388e-05,
+      "loss": 1.625,
+      "step": 787
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 7.487244897959184e-05,
+      "loss": 1.6302,
+      "step": 788
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 7.484056122448981e-05,
+      "loss": 1.6363,
+      "step": 789
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 7.480867346938775e-05,
+      "loss": 1.5215,
+      "step": 790
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 7.477678571428571e-05,
+      "loss": 1.4867,
+      "step": 791
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 7.474489795918368e-05,
+      "loss": 1.6359,
+      "step": 792
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 7.471301020408164e-05,
+      "loss": 1.5422,
+      "step": 793
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 7.468112244897958e-05,
+      "loss": 1.5507,
+      "step": 794
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 7.464923469387756e-05,
+      "loss": 1.5375,
+      "step": 795
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 7.461734693877551e-05,
+      "loss": 1.5598,
+      "step": 796
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 7.458545918367348e-05,
+      "loss": 1.5592,
+      "step": 797
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 7.455357142857143e-05,
+      "loss": 1.5317,
+      "step": 798
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 7.452168367346939e-05,
+      "loss": 1.5153,
+      "step": 799
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 7.448979591836736e-05,
+      "loss": 1.6207,
+      "step": 800
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 7.445790816326532e-05,
+      "loss": 1.5162,
+      "step": 801
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 7.442602040816326e-05,
+      "loss": 1.606,
+      "step": 802
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 7.439413265306123e-05,
+      "loss": 1.5156,
+      "step": 803
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 7.436224489795919e-05,
+      "loss": 1.5116,
+      "step": 804
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 7.433035714285715e-05,
+      "loss": 1.6724,
+      "step": 805
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 7.42984693877551e-05,
+      "loss": 1.4939,
+      "step": 806
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 7.426658163265306e-05,
+      "loss": 1.5348,
+      "step": 807
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 7.423469387755102e-05,
+      "loss": 1.5413,
+      "step": 808
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 7.420280612244899e-05,
+      "loss": 1.6815,
+      "step": 809
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 7.417091836734695e-05,
+      "loss": 1.4869,
+      "step": 810
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 7.413903061224489e-05,
+      "loss": 1.5318,
+      "step": 811
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 7.410714285714286e-05,
+      "loss": 1.5961,
+      "step": 812
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 7.407525510204082e-05,
+      "loss": 1.4803,
+      "step": 813
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 7.404336734693878e-05,
+      "loss": 1.5162,
+      "step": 814
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 7.401147959183674e-05,
+      "loss": 1.3663,
+      "step": 815
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 7.39795918367347e-05,
+      "loss": 1.5663,
+      "step": 816
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 7.394770408163265e-05,
+      "loss": 1.448,
+      "step": 817
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 7.391581632653062e-05,
+      "loss": 1.5293,
+      "step": 818
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 7.388392857142857e-05,
+      "loss": 1.5237,
+      "step": 819
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 7.385204081632653e-05,
+      "loss": 1.5021,
+      "step": 820
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 7.38201530612245e-05,
+      "loss": 1.6319,
+      "step": 821
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 7.378826530612245e-05,
+      "loss": 1.5267,
+      "step": 822
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 7.375637755102041e-05,
+      "loss": 1.5979,
+      "step": 823
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 7.372448979591837e-05,
+      "loss": 1.5628,
+      "step": 824
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 7.369260204081633e-05,
+      "loss": 1.6452,
+      "step": 825
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 7.36607142857143e-05,
+      "loss": 1.5668,
+      "step": 826
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 7.362882653061224e-05,
+      "loss": 1.5252,
+      "step": 827
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 7.35969387755102e-05,
+      "loss": 1.4068,
+      "step": 828
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 7.356505102040817e-05,
+      "loss": 1.5646,
+      "step": 829
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 7.353316326530613e-05,
+      "loss": 1.572,
+      "step": 830
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.350127551020409e-05,
+      "loss": 1.6101,
+      "step": 831
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.346938775510205e-05,
+      "loss": 1.4802,
+      "step": 832
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.34375e-05,
+      "loss": 1.5475,
+      "step": 833
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.340561224489796e-05,
+      "loss": 1.6726,
+      "step": 834
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.337372448979593e-05,
+      "loss": 1.5066,
+      "step": 835
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.334183673469388e-05,
+      "loss": 1.4366,
+      "step": 836
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.330994897959183e-05,
+      "loss": 1.4615,
+      "step": 837
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.32780612244898e-05,
+      "loss": 1.5023,
+      "step": 838
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.324617346938776e-05,
+      "loss": 1.5047,
+      "step": 839
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.321428571428571e-05,
+      "loss": 1.5154,
+      "step": 840
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.318239795918368e-05,
+      "loss": 1.5953,
+      "step": 841
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.315051020408164e-05,
+      "loss": 1.5582,
+      "step": 842
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.31186224489796e-05,
+      "loss": 1.4305,
+      "step": 843
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.308673469387755e-05,
+      "loss": 1.5189,
+      "step": 844
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.305484693877551e-05,
+      "loss": 1.5471,
+      "step": 845
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.302295918367348e-05,
+      "loss": 1.5957,
+      "step": 846
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.299107142857144e-05,
+      "loss": 1.6169,
+      "step": 847
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.29591836734694e-05,
+      "loss": 1.527,
+      "step": 848
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.292729591836735e-05,
+      "loss": 1.5946,
+      "step": 849
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.289540816326531e-05,
+      "loss": 1.5168,
+      "step": 850
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 7.286352040816327e-05,
+      "loss": 1.611,
+      "step": 851
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 7.283163265306123e-05,
+      "loss": 1.579,
+      "step": 852
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 7.279974489795918e-05,
+      "loss": 1.5112,
+      "step": 853
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 7.276785714285714e-05,
+      "loss": 1.6126,
+      "step": 854
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 7.273596938775511e-05,
+      "loss": 1.6139,
+      "step": 855
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 7.270408163265307e-05,
+      "loss": 1.5538,
+      "step": 856
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.267219387755102e-05,
+      "loss": 1.4936,
+      "step": 857
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.264030612244899e-05,
+      "loss": 1.5517,
+      "step": 858
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.260841836734694e-05,
+      "loss": 1.5221,
+      "step": 859
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.25765306122449e-05,
+      "loss": 1.552,
+      "step": 860
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.254464285714286e-05,
+      "loss": 1.5297,
+      "step": 861
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.251275510204082e-05,
+      "loss": 1.4874,
+      "step": 862
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.248086734693878e-05,
+      "loss": 1.6291,
+      "step": 863
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.244897959183675e-05,
+      "loss": 1.5594,
+      "step": 864
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.241709183673469e-05,
+      "loss": 1.5455,
+      "step": 865
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.238520408163265e-05,
+      "loss": 1.4842,
+      "step": 866
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.235331632653062e-05,
+      "loss": 1.676,
+      "step": 867
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.232142857142858e-05,
+      "loss": 1.4955,
+      "step": 868
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.228954081632653e-05,
+      "loss": 1.5832,
+      "step": 869
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.225765306122449e-05,
+      "loss": 1.5278,
+      "step": 870
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.222576530612245e-05,
+      "loss": 1.6077,
+      "step": 871
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.219387755102042e-05,
+      "loss": 1.5612,
+      "step": 872
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.216198979591837e-05,
+      "loss": 1.549,
+      "step": 873
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.213010204081632e-05,
+      "loss": 1.6173,
+      "step": 874
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.20982142857143e-05,
+      "loss": 1.6918,
+      "step": 875
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.206632653061225e-05,
+      "loss": 1.5562,
+      "step": 876
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.203443877551021e-05,
+      "loss": 1.5568,
+      "step": 877
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.200255102040817e-05,
+      "loss": 1.6067,
+      "step": 878
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.197066326530613e-05,
+      "loss": 1.493,
+      "step": 879
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.193877551020408e-05,
+      "loss": 1.597,
+      "step": 880
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.190688775510205e-05,
+      "loss": 1.5958,
+      "step": 881
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.1875e-05,
+      "loss": 1.5026,
+      "step": 882
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.184311224489796e-05,
+      "loss": 1.4857,
+      "step": 883
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.181122448979593e-05,
+      "loss": 1.5722,
+      "step": 884
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.177933673469389e-05,
+      "loss": 1.5176,
+      "step": 885
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.174744897959183e-05,
+      "loss": 1.4883,
+      "step": 886
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.17155612244898e-05,
+      "loss": 1.5146,
+      "step": 887
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.168367346938776e-05,
+      "loss": 1.5528,
+      "step": 888
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.165178571428572e-05,
+      "loss": 1.5486,
+      "step": 889
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.161989795918367e-05,
+      "loss": 1.6233,
+      "step": 890
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 7.158801020408163e-05,
+      "loss": 1.4514,
+      "step": 891
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 7.155612244897959e-05,
+      "loss": 1.4619,
+      "step": 892
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 7.152423469387756e-05,
+      "loss": 1.5101,
+      "step": 893
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 7.149234693877552e-05,
+      "loss": 1.5235,
+      "step": 894
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 7.146045918367348e-05,
+      "loss": 1.6027,
+      "step": 895
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 1.5341,
+      "step": 896
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 7.139668367346939e-05,
+      "loss": 1.4323,
+      "step": 897
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 7.136479591836735e-05,
+      "loss": 1.5753,
+      "step": 898
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 7.133290816326531e-05,
+      "loss": 1.5127,
+      "step": 899
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 7.130102040816326e-05,
+      "loss": 1.5789,
+      "step": 900
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 7.126913265306124e-05,
+      "loss": 1.473,
+      "step": 901
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 7.12372448979592e-05,
+      "loss": 1.4387,
+      "step": 902
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 7.120535714285714e-05,
+      "loss": 1.4895,
+      "step": 903
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 7.117346938775511e-05,
+      "loss": 1.4464,
+      "step": 904
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 7.114158163265307e-05,
+      "loss": 1.5466,
+      "step": 905
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 7.110969387755102e-05,
+      "loss": 1.5301,
+      "step": 906
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 7.107780612244898e-05,
+      "loss": 1.5272,
+      "step": 907
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 7.104591836734694e-05,
+      "loss": 1.58,
+      "step": 908
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 7.10140306122449e-05,
+      "loss": 1.4687,
+      "step": 909
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 7.098214285714287e-05,
+      "loss": 1.4929,
+      "step": 910
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 7.095025510204081e-05,
+      "loss": 1.4503,
+      "step": 911
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 7.091836734693877e-05,
+      "loss": 1.4659,
+      "step": 912
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 7.088647959183674e-05,
+      "loss": 1.5948,
+      "step": 913
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 7.08545918367347e-05,
+      "loss": 1.4717,
+      "step": 914
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 7.082270408163266e-05,
+      "loss": 1.5466,
+      "step": 915
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 7.079081632653062e-05,
+      "loss": 1.4722,
+      "step": 916
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 7.075892857142857e-05,
+      "loss": 1.4694,
+      "step": 917
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 7.072704081632653e-05,
+      "loss": 1.3661,
+      "step": 918
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 7.069515306122449e-05,
+      "loss": 1.5341,
+      "step": 919
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 7.066326530612245e-05,
+      "loss": 1.4553,
+      "step": 920
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 7.063137755102042e-05,
+      "loss": 1.4315,
+      "step": 921
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 7.059948979591838e-05,
+      "loss": 1.5519,
+      "step": 922
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 7.056760204081633e-05,
+      "loss": 1.4633,
+      "step": 923
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 7.053571428571429e-05,
+      "loss": 1.5942,
+      "step": 924
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 7.050382653061225e-05,
+      "loss": 1.3865,
+      "step": 925
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 7.04719387755102e-05,
+      "loss": 1.5227,
+      "step": 926
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 7.044005102040818e-05,
+      "loss": 1.5258,
+      "step": 927
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 7.040816326530612e-05,
+      "loss": 1.576,
+      "step": 928
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 7.037627551020408e-05,
+      "loss": 1.5317,
+      "step": 929
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 7.034438775510205e-05,
+      "loss": 1.537,
+      "step": 930
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 7.031250000000001e-05,
+      "loss": 1.3364,
+      "step": 931
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 7.028061224489795e-05,
+      "loss": 1.499,
+      "step": 932
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 7.024872448979592e-05,
+      "loss": 1.4468,
+      "step": 933
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 7.021683673469388e-05,
+      "loss": 1.5918,
+      "step": 934
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 7.018494897959184e-05,
+      "loss": 1.5661,
+      "step": 935
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 7.01530612244898e-05,
+      "loss": 1.5732,
+      "step": 936
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 7.012117346938775e-05,
+      "loss": 1.5791,
+      "step": 937
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 7.008928571428571e-05,
+      "loss": 1.5933,
+      "step": 938
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 7.005739795918368e-05,
+      "loss": 1.5665,
+      "step": 939
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 7.002551020408164e-05,
+      "loss": 1.4839,
+      "step": 940
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 6.999362244897959e-05,
+      "loss": 1.4768,
+      "step": 941
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 6.996173469387756e-05,
+      "loss": 1.6085,
+      "step": 942
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 6.992984693877551e-05,
+      "loss": 1.4922,
+      "step": 943
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 6.989795918367347e-05,
+      "loss": 1.5707,
+      "step": 944
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 6.986607142857143e-05,
+      "loss": 1.5311,
+      "step": 945
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 6.983418367346939e-05,
+      "loss": 1.5081,
+      "step": 946
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 6.980229591836736e-05,
+      "loss": 1.512,
+      "step": 947
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 6.977040816326532e-05,
+      "loss": 1.5491,
+      "step": 948
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 6.973852040816326e-05,
+      "loss": 1.5359,
+      "step": 949
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 6.970663265306123e-05,
+      "loss": 1.5589,
+      "step": 950
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 6.967474489795919e-05,
+      "loss": 1.4852,
+      "step": 951
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 6.964285714285715e-05,
+      "loss": 1.5861,
+      "step": 952
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 6.96109693877551e-05,
+      "loss": 1.3901,
+      "step": 953
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 6.957908163265306e-05,
+      "loss": 1.5414,
+      "step": 954
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 6.954719387755102e-05,
+      "loss": 1.5375,
+      "step": 955
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 6.951530612244899e-05,
+      "loss": 1.5343,
+      "step": 956
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 6.948341836734694e-05,
+      "loss": 1.424,
+      "step": 957
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 6.94515306122449e-05,
+      "loss": 1.4389,
+      "step": 958
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 6.941964285714286e-05,
+      "loss": 1.4984,
+      "step": 959
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 6.938775510204082e-05,
+      "loss": 1.5261,
+      "step": 960
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 6.935586734693878e-05,
+      "loss": 1.5071,
+      "step": 961
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 6.932397959183674e-05,
+      "loss": 1.5013,
+      "step": 962
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 6.92920918367347e-05,
+      "loss": 1.6516,
+      "step": 963
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 6.926020408163265e-05,
+      "loss": 1.5105,
+      "step": 964
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 6.922831632653061e-05,
+      "loss": 1.48,
+      "step": 965
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 6.919642857142857e-05,
+      "loss": 1.5517,
+      "step": 966
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.916454081632653e-05,
+      "loss": 1.364,
+      "step": 967
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.91326530612245e-05,
+      "loss": 1.5156,
+      "step": 968
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.910076530612246e-05,
+      "loss": 1.5559,
+      "step": 969
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.906887755102041e-05,
+      "loss": 1.4075,
+      "step": 970
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.903698979591837e-05,
+      "loss": 1.5884,
+      "step": 971
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 6.900510204081633e-05,
+      "loss": 1.554,
+      "step": 972
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 6.89732142857143e-05,
+      "loss": 1.6034,
+      "step": 973
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 6.894132653061224e-05,
+      "loss": 1.4644,
+      "step": 974
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.89094387755102e-05,
+      "loss": 1.5907,
+      "step": 975
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.887755102040817e-05,
+      "loss": 1.4513,
+      "step": 976
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.884566326530613e-05,
+      "loss": 1.6045,
+      "step": 977
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.881377551020408e-05,
+      "loss": 1.5816,
+      "step": 978
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.878188775510205e-05,
+      "loss": 1.5426,
+      "step": 979
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.875e-05,
+      "loss": 1.4962,
+      "step": 980
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.871811224489796e-05,
+      "loss": 1.5634,
+      "step": 981
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 6.868622448979592e-05,
+      "loss": 1.4918,
+      "step": 982
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.865433673469388e-05,
+      "loss": 1.5547,
+      "step": 983
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 6.862244897959184e-05,
+      "loss": 1.6126,
+      "step": 984
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 6.85905612244898e-05,
+      "loss": 1.4723,
+      "step": 985
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.855867346938776e-05,
+      "loss": 1.5965,
+      "step": 986
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 6.852678571428571e-05,
+      "loss": 1.5439,
+      "step": 987
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 6.849489795918368e-05,
+      "loss": 1.5178,
+      "step": 988
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 6.846301020408164e-05,
+      "loss": 1.4874,
+      "step": 989
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 6.84311224489796e-05,
+      "loss": 1.5047,
+      "step": 990
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 6.839923469387755e-05,
+      "loss": 1.4825,
+      "step": 991
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 6.836734693877551e-05,
+      "loss": 1.4454,
+      "step": 992
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 6.833545918367348e-05,
+      "loss": 1.5514,
+      "step": 993
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 6.830357142857144e-05,
+      "loss": 1.5191,
+      "step": 994
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 6.827168367346938e-05,
+      "loss": 1.5448,
+      "step": 995
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 6.823979591836735e-05,
+      "loss": 1.5009,
+      "step": 996
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 6.820790816326531e-05,
+      "loss": 1.5012,
+      "step": 997
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 6.817602040816327e-05,
+      "loss": 1.4249,
+      "step": 998
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 6.814413265306123e-05,
+      "loss": 1.5253,
+      "step": 999
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 6.811224489795919e-05,
+      "loss": 1.5544,
+      "step": 1000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 6.808035714285714e-05,
+      "loss": 1.4549,
+      "step": 1001
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 6.804846938775511e-05,
+      "loss": 1.4832,
+      "step": 1002
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 6.801658163265306e-05,
+      "loss": 1.5244,
+      "step": 1003
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 6.798469387755102e-05,
+      "loss": 1.4185,
+      "step": 1004
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 6.795280612244899e-05,
+      "loss": 1.4247,
+      "step": 1005
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 6.792091836734695e-05,
+      "loss": 1.5257,
+      "step": 1006
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 6.78890306122449e-05,
+      "loss": 1.4686,
+      "step": 1007
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 6.785714285714286e-05,
+      "loss": 1.4543,
+      "step": 1008
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 6.782525510204082e-05,
+      "loss": 1.4694,
+      "step": 1009
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 6.779336734693878e-05,
+      "loss": 1.4686,
+      "step": 1010
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 6.776147959183673e-05,
+      "loss": 1.4029,
+      "step": 1011
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 6.772959183673469e-05,
+      "loss": 1.4833,
+      "step": 1012
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 6.769770408163265e-05,
+      "loss": 1.3575,
+      "step": 1013
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 6.766581632653062e-05,
+      "loss": 1.5344,
+      "step": 1014
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 6.763392857142858e-05,
+      "loss": 1.3947,
+      "step": 1015
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 6.760204081632652e-05,
+      "loss": 1.5571,
+      "step": 1016
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 6.75701530612245e-05,
+      "loss": 1.3519,
+      "step": 1017
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 6.753826530612245e-05,
+      "loss": 1.4303,
+      "step": 1018
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 6.750637755102042e-05,
+      "loss": 1.5916,
+      "step": 1019
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 6.747448979591837e-05,
+      "loss": 1.4956,
+      "step": 1020
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 6.744260204081632e-05,
+      "loss": 1.3845,
+      "step": 1021
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 6.74107142857143e-05,
+      "loss": 1.5454,
+      "step": 1022
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 6.737882653061225e-05,
+      "loss": 1.4286,
+      "step": 1023
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 6.73469387755102e-05,
+      "loss": 1.4626,
+      "step": 1024
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 6.731505102040817e-05,
+      "loss": 1.5129,
+      "step": 1025
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 6.728316326530613e-05,
+      "loss": 1.4496,
+      "step": 1026
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 6.725127551020408e-05,
+      "loss": 1.5002,
+      "step": 1027
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 6.721938775510204e-05,
+      "loss": 1.4472,
+      "step": 1028
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 6.71875e-05,
+      "loss": 1.5916,
+      "step": 1029
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 6.715561224489796e-05,
+      "loss": 1.4294,
+      "step": 1030
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 6.712372448979593e-05,
+      "loss": 1.5925,
+      "step": 1031
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 6.709183673469389e-05,
+      "loss": 1.3663,
+      "step": 1032
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 6.705994897959183e-05,
+      "loss": 1.4723,
+      "step": 1033
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 6.70280612244898e-05,
+      "loss": 1.5188,
+      "step": 1034
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 6.699617346938776e-05,
+      "loss": 1.5211,
+      "step": 1035
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 6.696428571428572e-05,
+      "loss": 1.493,
+      "step": 1036
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 6.693239795918368e-05,
+      "loss": 1.5321,
+      "step": 1037
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 6.690051020408163e-05,
+      "loss": 1.5492,
+      "step": 1038
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 6.686862244897959e-05,
+      "loss": 1.494,
+      "step": 1039
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 6.683673469387756e-05,
+      "loss": 1.534,
+      "step": 1040
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 6.68048469387755e-05,
+      "loss": 1.4434,
+      "step": 1041
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 6.677295918367348e-05,
+      "loss": 1.4481,
+      "step": 1042
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 6.674107142857144e-05,
+      "loss": 1.4684,
+      "step": 1043
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 6.670918367346939e-05,
+      "loss": 1.5014,
+      "step": 1044
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 6.667729591836735e-05,
+      "loss": 1.5881,
+      "step": 1045
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 6.664540816326531e-05,
+      "loss": 1.5053,
+      "step": 1046
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 6.661352040816327e-05,
+      "loss": 1.4924,
+      "step": 1047
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 6.658163265306124e-05,
+      "loss": 1.4137,
+      "step": 1048
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 6.654974489795918e-05,
+      "loss": 1.4442,
+      "step": 1049
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 6.651785714285714e-05,
+      "loss": 1.4903,
+      "step": 1050
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 6.648596938775511e-05,
+      "loss": 1.5714,
+      "step": 1051
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 6.645408163265307e-05,
+      "loss": 1.4596,
+      "step": 1052
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 6.642219387755103e-05,
+      "loss": 1.482,
+      "step": 1053
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 6.639030612244898e-05,
+      "loss": 1.5267,
+      "step": 1054
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 6.635841836734694e-05,
+      "loss": 1.4587,
+      "step": 1055
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 6.63265306122449e-05,
+      "loss": 1.5019,
+      "step": 1056
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 6.629464285714286e-05,
+      "loss": 1.4034,
+      "step": 1057
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 6.626275510204081e-05,
+      "loss": 1.4521,
+      "step": 1058
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 6.623086734693877e-05,
+      "loss": 1.5502,
+      "step": 1059
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 6.619897959183674e-05,
+      "loss": 1.4779,
+      "step": 1060
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 6.61670918367347e-05,
+      "loss": 1.4686,
+      "step": 1061
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 6.613520408163265e-05,
+      "loss": 1.4792,
+      "step": 1062
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 6.610331632653062e-05,
+      "loss": 1.4339,
+      "step": 1063
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 6.607142857142857e-05,
+      "loss": 1.5343,
+      "step": 1064
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 6.603954081632653e-05,
+      "loss": 1.4648,
+      "step": 1065
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 6.600765306122449e-05,
+      "loss": 1.5308,
+      "step": 1066
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 6.597576530612245e-05,
+      "loss": 1.5239,
+      "step": 1067
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 6.594387755102042e-05,
+      "loss": 1.4896,
+      "step": 1068
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 6.591198979591838e-05,
+      "loss": 1.5079,
+      "step": 1069
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 6.588010204081632e-05,
+      "loss": 1.4364,
+      "step": 1070
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 6.584821428571429e-05,
+      "loss": 1.5552,
+      "step": 1071
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 6.581632653061225e-05,
+      "loss": 1.4662,
+      "step": 1072
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 6.578443877551021e-05,
+      "loss": 1.5869,
+      "step": 1073
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 6.575255102040817e-05,
+      "loss": 1.3986,
+      "step": 1074
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 6.572066326530612e-05,
+      "loss": 1.486,
+      "step": 1075
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 6.568877551020408e-05,
+      "loss": 1.4775,
+      "step": 1076
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 6.565688775510205e-05,
+      "loss": 1.5488,
+      "step": 1077
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 6.562500000000001e-05,
+      "loss": 1.4099,
+      "step": 1078
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 6.559311224489795e-05,
+      "loss": 1.4595,
+      "step": 1079
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 6.556122448979592e-05,
+      "loss": 1.4397,
+      "step": 1080
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 6.552933673469388e-05,
+      "loss": 1.5166,
+      "step": 1081
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 6.549744897959184e-05,
+      "loss": 1.4361,
+      "step": 1082
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 6.54655612244898e-05,
+      "loss": 1.4449,
+      "step": 1083
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 6.543367346938776e-05,
+      "loss": 1.4578,
+      "step": 1084
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 6.540178571428571e-05,
+      "loss": 1.4669,
+      "step": 1085
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 6.536989795918368e-05,
+      "loss": 1.5328,
+      "step": 1086
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 6.533801020408163e-05,
+      "loss": 1.4579,
+      "step": 1087
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 6.530612244897959e-05,
+      "loss": 1.5361,
+      "step": 1088
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 6.527423469387756e-05,
+      "loss": 1.5363,
+      "step": 1089
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 6.524234693877552e-05,
+      "loss": 1.4183,
+      "step": 1090
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 6.521045918367347e-05,
+      "loss": 1.3972,
+      "step": 1091
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 6.517857142857143e-05,
+      "loss": 1.4249,
+      "step": 1092
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 6.514668367346939e-05,
+      "loss": 1.3283,
+      "step": 1093
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 6.511479591836736e-05,
+      "loss": 1.4536,
+      "step": 1094
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 6.50829081632653e-05,
+      "loss": 1.4411,
+      "step": 1095
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 6.505102040816326e-05,
+      "loss": 1.4507,
+      "step": 1096
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 6.501913265306123e-05,
+      "loss": 1.4749,
+      "step": 1097
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 6.498724489795919e-05,
+      "loss": 1.5126,
+      "step": 1098
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 6.495535714285715e-05,
+      "loss": 1.6189,
+      "step": 1099
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 6.49234693877551e-05,
+      "loss": 1.3727,
+      "step": 1100
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3136,
+  "num_train_epochs": 32,
+  "save_steps": 100,
+  "total_flos": 8.142360683945779e+17,
+  "trial_name": null,
+  "trial_params": null
+}