diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16674 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2774,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.952380952380953e-08,
+      "loss": 1.1294,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1904761904761906e-07,
+      "loss": 1.1387,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7857142857142858e-07,
+      "loss": 1.1772,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3809523809523811e-07,
+      "loss": 1.1333,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9761904761904765e-07,
+      "loss": 1.1484,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 1.1582,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1666666666666667e-07,
+      "loss": 1.0933,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7619047619047623e-07,
+      "loss": 1.1504,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.357142857142857e-07,
+      "loss": 1.1104,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.952380952380953e-07,
+      "loss": 1.1162,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.547619047619048e-07,
+      "loss": 1.1558,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 1.165,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.738095238095239e-07,
+      "loss": 1.1729,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.1426,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.928571428571429e-07,
+      "loss": 1.0425,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 1.1709,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.011904761904762e-06,
+      "loss": 1.1748,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 1.2041,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.130952380952381e-06,
+      "loss": 1.1641,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1904761904761906e-06,
+      "loss": 1.0566,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.25e-06,
+      "loss": 1.1689,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3095238095238096e-06,
+      "loss": 1.1099,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3690476190476193e-06,
+      "loss": 1.104,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.126,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4880952380952381e-06,
+      "loss": 1.0576,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5476190476190479e-06,
+      "loss": 1.0488,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6071428571428574e-06,
+      "loss": 1.1729,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.1636,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7261904761904764e-06,
+      "loss": 1.0781,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 1.0566,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8452380952380954e-06,
+      "loss": 1.082,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 1.042,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9642857142857144e-06,
+      "loss": 1.0249,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.023809523809524e-06,
+      "loss": 1.0791,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 1.0693,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 1.0415,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2023809523809525e-06,
+      "loss": 1.0752,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.261904761904762e-06,
+      "loss": 1.0649,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.321428571428572e-06,
+      "loss": 1.0972,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.380952380952381e-06,
+      "loss": 1.0835,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4404761904761905e-06,
+      "loss": 1.0435,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-06,
+      "loss": 1.105,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5595238095238095e-06,
+      "loss": 1.1113,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6190476190476192e-06,
+      "loss": 1.002,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6785714285714285e-06,
+      "loss": 1.0391,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7380952380952387e-06,
+      "loss": 1.0635,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.797619047619048e-06,
+      "loss": 1.0591,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.083,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.916666666666667e-06,
+      "loss": 1.0903,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9761904761904763e-06,
+      "loss": 1.0776,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0357142857142856e-06,
+      "loss": 0.9985,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0952380952380957e-06,
+      "loss": 1.002,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.154761904761905e-06,
+      "loss": 1.0967,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 1.0996,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.273809523809524e-06,
+      "loss": 1.0845,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.0469,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3928571428571435e-06,
+      "loss": 1.0381,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4523809523809528e-06,
+      "loss": 1.0674,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.511904761904762e-06,
+      "loss": 1.0181,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 1.0278,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.630952380952381e-06,
+      "loss": 0.9951,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.690476190476191e-06,
+      "loss": 1.0386,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 1.0269,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 1.0552,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.869047619047619e-06,
+      "loss": 1.063,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 1.0244,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9880952380952386e-06,
+      "loss": 1.105,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.047619047619048e-06,
+      "loss": 1.043,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.107142857142857e-06,
+      "loss": 1.0854,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.9614,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.226190476190477e-06,
+      "loss": 1.0498,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.0669,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.345238095238096e-06,
+      "loss": 1.042,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.404761904761905e-06,
+      "loss": 1.083,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.464285714285715e-06,
+      "loss": 1.0146,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.523809523809524e-06,
+      "loss": 1.0522,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.583333333333333e-06,
+      "loss": 1.0,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 1.0752,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.702380952380953e-06,
+      "loss": 1.0649,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 1.0405,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.821428571428572e-06,
+      "loss": 1.0566,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.880952380952381e-06,
+      "loss": 1.0557,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.940476190476191e-06,
+      "loss": 0.9883,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 0.9775,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999998295075511e-06,
+      "loss": 0.9976,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999931803043675e-06,
+      "loss": 1.0503,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999984655693547e-06,
+      "loss": 1.0117,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999972721254676e-06,
+      "loss": 1.0293,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999957377004031e-06,
+      "loss": 1.0674,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999386229625436e-06,
+      "loss": 1.0532,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999916459155791e-06,
+      "loss": 0.9653,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999890885614004e-06,
+      "loss": 1.0649,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999861902372063e-06,
+      "loss": 1.0786,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9998295094694995e-06,
+      "loss": 1.0527,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999793706950496e-06,
+      "loss": 0.979,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999754494863884e-06,
+      "loss": 1.0317,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999711873263148e-06,
+      "loss": 1.0347,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9996658422064195e-06,
+      "loss": 1.0322,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9996164017564835e-06,
+      "loss": 1.0366,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999563551980773e-06,
+      "loss": 1.0151,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999507292951371e-06,
+      "loss": 0.9834,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9994476247450145e-06,
+      "loss": 1.0244,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999384547443084e-06,
+      "loss": 0.9766,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999318061131614e-06,
+      "loss": 1.0312,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999248165901289e-06,
+      "loss": 1.0693,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999174861847441e-06,
+      "loss": 1.0796,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999098149070052e-06,
+      "loss": 1.0425,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999018027673754e-06,
+      "loss": 1.0151,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998934497767829e-06,
+      "loss": 1.0215,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998847559466204e-06,
+      "loss": 1.0576,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99875721288746e-06,
+      "loss": 1.0854,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9986634581548235e-06,
+      "loss": 1.042,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998566295396169e-06,
+      "loss": 1.0664,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998465724744023e-06,
+      "loss": 1.0088,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998361746335556e-06,
+      "loss": 1.0615,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998254360312589e-06,
+      "loss": 1.042,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998143566821589e-06,
+      "loss": 1.0137,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998029366013674e-06,
+      "loss": 1.0347,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997911758044605e-06,
+      "loss": 1.0532,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997790743074793e-06,
+      "loss": 1.0752,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997666321269294e-06,
+      "loss": 0.9868,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997538492797813e-06,
+      "loss": 1.0439,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9974072578347e-06,
+      "loss": 0.9834,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997272616558952e-06,
+      "loss": 1.106,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99713456915421e-06,
+      "loss": 1.0312,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996993115808765e-06,
+      "loss": 0.9814,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996848256715547e-06,
+      "loss": 1.0161,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996699992072139e-06,
+      "loss": 1.0122,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996548322080763e-06,
+      "loss": 1.0459,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996393246948288e-06,
+      "loss": 1.0337,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996234766886227e-06,
+      "loss": 1.0166,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996072882110737e-06,
+      "loss": 1.0342,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995907592842619e-06,
+      "loss": 1.0039,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995738899307319e-06,
+      "loss": 1.0557,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995566801734923e-06,
+      "loss": 1.0161,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9953913003601625e-06,
+      "loss": 1.0039,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995212395422412e-06,
+      "loss": 1.0215,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995030087165684e-06,
+      "loss": 1.0527,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994844375838639e-06,
+      "loss": 0.9849,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994655261694575e-06,
+      "loss": 1.0498,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994462744991431e-06,
+      "loss": 1.0322,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994266825991788e-06,
+      "loss": 1.0396,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9940675049628715e-06,
+      "loss": 1.0088,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993864782176539e-06,
+      "loss": 1.0078,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993658657909294e-06,
+      "loss": 0.9629,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993449132442278e-06,
+      "loss": 0.9785,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9932362060612694e-06,
+      "loss": 1.0596,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993019879056689e-06,
+      "loss": 1.0303,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992800151723592e-06,
+      "loss": 1.0684,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9925770243616745e-06,
+      "loss": 0.9956,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992350497275268e-06,
+      "loss": 1.0322,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992120570773342e-06,
+      "loss": 1.0742,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991887245169502e-06,
+      "loss": 1.0566,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99165052078199e-06,
+      "loss": 1.0605,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991410397933685e-06,
+      "loss": 1.0112,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991166876952098e-06,
+      "loss": 1.0498,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990919958169379e-06,
+      "loss": 1.0659,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99066964192231e-06,
+      "loss": 0.9556,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990415928552306e-06,
+      "loss": 1.0527,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990158818405417e-06,
+      "loss": 1.1377,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9898983118323265e-06,
+      "loss": 1.0562,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989634409188349e-06,
+      "loss": 1.0586,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989367110833432e-06,
+      "loss": 1.0205,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9890964171321535e-06,
+      "loss": 1.042,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988822328453725e-06,
+      "loss": 1.0288,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988544845171986e-06,
+      "loss": 1.0459,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9882639676654075e-06,
+      "loss": 1.0356,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987979696317088e-06,
+      "loss": 1.0381,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987692031514758e-06,
+      "loss": 1.0044,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9874009736507745e-06,
+      "loss": 1.0757,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987106523122122e-06,
+      "loss": 1.0215,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986808680330415e-06,
+      "loss": 1.0396,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9865074456818906e-06,
+      "loss": 1.0537,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9862028195874165e-06,
+      "loss": 1.0645,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985894802462485e-06,
+      "loss": 0.999,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985583394727211e-06,
+      "loss": 1.0571,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985268596806336e-06,
+      "loss": 0.9673,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9849504091292264e-06,
+      "loss": 1.0605,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98462883212987e-06,
+      "loss": 1.0771,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984303866246879e-06,
+      "loss": 1.1079,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983975511923488e-06,
+      "loss": 0.9976,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.98364376960755e-06,
+      "loss": 0.9878,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983308639751544e-06,
+      "loss": 1.0107,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982970122812566e-06,
+      "loss": 1.0322,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9826282192523315e-06,
+      "loss": 0.9829,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982282929537179e-06,
+      "loss": 1.0029,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.98193425413806e-06,
+      "loss": 1.0386,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9815821935305475e-06,
+      "loss": 1.0215,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981226748194833e-06,
+      "loss": 1.0249,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980867918615719e-06,
+      "loss": 1.0068,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980505705282629e-06,
+      "loss": 1.0366,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980140108689602e-06,
+      "loss": 1.0464,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979771129335286e-06,
+      "loss": 0.9741,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979398767722949e-06,
+      "loss": 1.041,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97902302436047e-06,
+      "loss": 0.9717,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9786438997603385e-06,
+      "loss": 1.0186,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978261394439658e-06,
+      "loss": 0.9966,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9778755089201445e-06,
+      "loss": 1.0659,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97748624372812e-06,
+      "loss": 1.0732,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97709359939452e-06,
+      "loss": 1.0776,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976697576454889e-06,
+      "loss": 1.166,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9762981754493755e-06,
+      "loss": 1.0479,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97589539692274e-06,
+      "loss": 1.0161,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975489241424347e-06,
+      "loss": 1.0459,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975079709508171e-06,
+      "loss": 1.0908,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9746668017327845e-06,
+      "loss": 1.0581,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974250518661371e-06,
+      "loss": 0.9844,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.973830860861717e-06,
+      "loss": 0.999,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973407828906208e-06,
+      "loss": 1.0444,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9729814233718345e-06,
+      "loss": 1.084,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972551644840188e-06,
+      "loss": 0.9941,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972118493897461e-06,
+      "loss": 1.0229,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9716819711344446e-06,
+      "loss": 1.0664,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.97124207714653e-06,
+      "loss": 1.0444,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9707988125337056e-06,
+      "loss": 0.9995,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970352177900558e-06,
+      "loss": 1.0269,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9699021738562705e-06,
+      "loss": 1.0503,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9694488010146195e-06,
+      "loss": 1.043,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968992059993979e-06,
+      "loss": 1.0435,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9685319514173165e-06,
+      "loss": 0.9829,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968068475912192e-06,
+      "loss": 1.0205,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967601634110758e-06,
+      "loss": 1.0195,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9671314266497595e-06,
+      "loss": 1.02,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96665785417053e-06,
+      "loss": 1.0938,
+      "step": 224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966180917318994e-06,
+      "loss": 0.9912,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965700616745665e-06,
+      "loss": 1.0371,
+      "step": 226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965216953105644e-06,
+      "loss": 1.0537,
+      "step": 227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964729927058618e-06,
+      "loss": 1.0176,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964239539268861e-06,
+      "loss": 1.0005,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963745790405234e-06,
+      "loss": 0.9961,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963248681141179e-06,
+      "loss": 1.02,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962748212154724e-06,
+      "loss": 1.0684,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9622443841284786e-06,
+      "loss": 1.0229,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961737197749633e-06,
+      "loss": 1.0518,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961226653709959e-06,
+      "loss": 1.0249,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960712752705808e-06,
+      "loss": 1.0688,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.96019549543811e-06,
+      "loss": 1.0439,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959674882612372e-06,
+      "loss": 1.0122,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95915091493868e-06,
+      "loss": 1.0776,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958623593131691e-06,
+      "loss": 1.0278,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958092917910646e-06,
+      "loss": 0.9966,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9575588899993464e-06,
+      "loss": 1.0532,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9570215101261796e-06,
+      "loss": 1.0562,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956480779024098e-06,
+      "loss": 1.0918,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955936697430625e-06,
+      "loss": 1.0913,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955389266087856e-06,
+      "loss": 1.0386,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954838485742453e-06,
+      "loss": 1.0195,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95428435714565e-06,
+      "loss": 1.0044,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953726881053242e-06,
+      "loss": 0.9722,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9531660582255934e-06,
+      "loss": 1.0186,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952601889427634e-06,
+      "loss": 1.0244,
+      "step": 251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9520343754288545e-06,
+      "loss": 1.0239,
+      "step": 252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.951463517003311e-06,
+      "loss": 1.0142,
+      "step": 253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.950889314929618e-06,
+      "loss": 1.0381,
+      "step": 254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9503117699909545e-06,
+      "loss": 1.0464,
+      "step": 255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.949730882975055e-06,
+      "loss": 1.0361,
+      "step": 256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.949146654674216e-06,
+      "loss": 1.1025,
+      "step": 257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948559085885288e-06,
+      "loss": 1.0479,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947968177409681e-06,
+      "loss": 1.0273,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9473739300533575e-06,
+      "loss": 1.02,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946776344626834e-06,
+      "loss": 1.022,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9461754219451844e-06,
+      "loss": 0.936,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945571162828027e-06,
+      "loss": 1.0244,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9449635680995375e-06,
+      "loss": 1.0488,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.944352638588436e-06,
+      "loss": 1.0425,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.943738375127996e-06,
+      "loss": 0.9795,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.943120778556034e-06,
+      "loss": 1.0034,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942499849714915e-06,
+      "loss": 1.0171,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.941875589451548e-06,
+      "loss": 1.002,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9412479986173854e-06,
+      "loss": 1.0405,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940617078068426e-06,
+      "loss": 1.0332,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9399828286652056e-06,
+      "loss": 1.0444,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939345251272802e-06,
+      "loss": 1.0015,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938704346760832e-06,
+      "loss": 0.9541,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938060116003452e-06,
+      "loss": 1.0674,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937412559879352e-06,
+      "loss": 1.0449,
+      "step": 276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936761679271761e-06,
+      "loss": 1.0322,
+      "step": 277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9361074750684404e-06,
+      "loss": 1.0488,
+      "step": 278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935449948161684e-06,
+      "loss": 1.0098,
+      "step": 279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.93478909944832e-06,
+      "loss": 1.0303,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.934124929829706e-06,
+      "loss": 1.0347,
+      "step": 281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9334574402117295e-06,
+      "loss": 1.0425,
+      "step": 282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932786631504805e-06,
+      "loss": 0.9795,
+      "step": 283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932112504623876e-06,
+      "loss": 1.0762,
+      "step": 284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.931435060488411e-06,
+      "loss": 1.0469,
+      "step": 285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9307543000224024e-06,
+      "loss": 0.9966,
+      "step": 286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930070224154366e-06,
+      "loss": 1.0791,
+      "step": 287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.92938283381734e-06,
+      "loss": 1.0449,
+      "step": 288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928692129948884e-06,
+      "loss": 1.1064,
+      "step": 289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.927998113491076e-06,
+      "loss": 0.9844,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.927300785390513e-06,
+      "loss": 1.0459,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.926600146598307e-06,
+      "loss": 1.0132,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925896198070088e-06,
+      "loss": 1.0474,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925188940766e-06,
+      "loss": 1.0391,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9244783756506975e-06,
+      "loss": 1.0317,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9237645036933505e-06,
+      "loss": 1.0244,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.923047325867635e-06,
+      "loss": 0.9746,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.922326843151739e-06,
+      "loss": 1.0181,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.921603056528358e-06,
+      "loss": 1.0024,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.920875966984693e-06,
+      "loss": 1.0146,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.92014557551245e-06,
+      "loss": 1.0181,
+      "step": 301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91941188310784e-06,
+      "loss": 1.0601,
+      "step": 302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.918674890771573e-06,
+      "loss": 1.0024,
+      "step": 303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9179345995088625e-06,
+      "loss": 1.0527,
+      "step": 304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.917191010329423e-06,
+      "loss": 1.0332,
+      "step": 305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.916444124247463e-06,
+      "loss": 1.0356,
+      "step": 306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.915693942281691e-06,
+      "loss": 1.064,
+      "step": 307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91494046545531e-06,
+      "loss": 1.0547,
+      "step": 308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914183694796017e-06,
+      "loss": 1.002,
+      "step": 309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.913423631336e-06,
+      "loss": 1.0356,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912660276111941e-06,
+      "loss": 1.0054,
+      "step": 311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911893630165011e-06,
+      "loss": 1.0439,
+      "step": 312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911123694540868e-06,
+      "loss": 1.0205,
+      "step": 313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910350470289656e-06,
+      "loss": 1.0732,
+      "step": 314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.90957395846601e-06,
+      "loss": 1.019,
+      "step": 315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9087941601290416e-06,
+      "loss": 1.0181,
+      "step": 316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.90801107634235e-06,
+      "loss": 1.0522,
+      "step": 317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.907224708174014e-06,
+      "loss": 1.0625,
+      "step": 318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9064350566965925e-06,
+      "loss": 0.9829,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.905642122987123e-06,
+      "loss": 1.042,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.904845908127119e-06,
+      "loss": 1.019,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.904046413202568e-06,
+      "loss": 1.0566,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.903243639303934e-06,
+      "loss": 1.0234,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.902437587526152e-06,
+      "loss": 1.0107,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901628258968628e-06,
+      "loss": 1.0151,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900815654735237e-06,
+      "loss": 1.0229,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8999997759343225e-06,
+      "loss": 0.9951,
+      "step": 327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.899180623678693e-06,
+      "loss": 1.0283,
+      "step": 328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.898358199085624e-06,
+      "loss": 1.0195,
+      "step": 329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.897532503276852e-06,
+      "loss": 1.0605,
+      "step": 330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896703537378577e-06,
+      "loss": 1.0337,
+      "step": 331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.895871302521457e-06,
+      "loss": 1.0493,
+      "step": 332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89503579984061e-06,
+      "loss": 1.0229,
+      "step": 333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894197030475614e-06,
+      "loss": 0.9893,
+      "step": 334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893354995570497e-06,
+      "loss": 1.0786,
+      "step": 335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892509696273745e-06,
+      "loss": 1.0552,
+      "step": 336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891661133738295e-06,
+      "loss": 1.0103,
+      "step": 337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8908093091215344e-06,
+      "loss": 1.0234,
+      "step": 338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889954223585301e-06,
+      "loss": 1.021,
+      "step": 339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.88909587829588e-06,
+      "loss": 1.1162,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8882342744240015e-06,
+      "loss": 1.0635,
+      "step": 341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8873694131448425e-06,
+      "loss": 0.978,
+      "step": 342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886501295638021e-06,
+      "loss": 1.0059,
+      "step": 343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.885629923087597e-06,
+      "loss": 1.0684,
+      "step": 344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.88475529668207e-06,
+      "loss": 0.9941,
+      "step": 345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.883877417614376e-06,
+      "loss": 1.0059,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882996287081892e-06,
+      "loss": 1.0825,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882111906286425e-06,
+      "loss": 1.0459,
+      "step": 348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8812242764342165e-06,
+      "loss": 1.002,
+      "step": 349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880333398735941e-06,
+      "loss": 0.998,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.879439274406702e-06,
+      "loss": 1.062,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.87854190466603e-06,
+      "loss": 0.9668,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8776412907378845e-06,
+      "loss": 1.0552,
+      "step": 353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876737433850647e-06,
+      "loss": 1.0029,
+      "step": 354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875830335237125e-06,
+      "loss": 1.0596,
+      "step": 355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874919996134546e-06,
+      "loss": 1.001,
+      "step": 356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874006417784557e-06,
+      "loss": 1.0957,
+      "step": 357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873089601433223e-06,
+      "loss": 1.0806,
+      "step": 358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.872169548331028e-06,
+      "loss": 1.0415,
+      "step": 359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.871246259732867e-06,
+      "loss": 1.0732,
+      "step": 360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.870319736898052e-06,
+      "loss": 1.0498,
+      "step": 361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.869389981090302e-06,
+      "loss": 1.0068,
+      "step": 362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.868456993577749e-06,
+      "loss": 1.0596,
+      "step": 363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.867520775632931e-06,
+      "loss": 1.0474,
+      "step": 364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.866581328532793e-06,
+      "loss": 1.0449,
+      "step": 365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.865638653558684e-06,
+      "loss": 1.0161,
+      "step": 366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.864692751996356e-06,
+      "loss": 1.0024,
+      "step": 367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.863743625135962e-06,
+      "loss": 1.0278,
+      "step": 368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.862791274272053e-06,
+      "loss": 1.0132,
+      "step": 369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.861835700703578e-06,
+      "loss": 1.0083,
+      "step": 370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.860876905733881e-06,
+      "loss": 1.02,
+      "step": 371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.859914890670701e-06,
+      "loss": 0.9839,
+      "step": 372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.85894965682617e-06,
+      "loss": 1.043,
+      "step": 373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.857981205516807e-06,
+      "loss": 1.0449,
+      "step": 374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8570095380635215e-06,
+      "loss": 1.0703,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.856034655791608e-06,
+      "loss": 0.9351,
+      "step": 376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.85505656003075e-06,
+      "loss": 1.0044,
+      "step": 377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.854075252115007e-06,
+      "loss": 1.0781,
+      "step": 378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.853090733382827e-06,
+      "loss": 1.0605,
+      "step": 379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.852103005177033e-06,
+      "loss": 1.0381,
+      "step": 380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.851112068844827e-06,
+      "loss": 1.0171,
+      "step": 381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.850117925737784e-06,
+      "loss": 1.0132,
+      "step": 382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8491205772118585e-06,
+      "loss": 1.0059,
+      "step": 383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.848120024627372e-06,
+      "loss": 1.0347,
+      "step": 384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.847116269349018e-06,
+      "loss": 1.0317,
+      "step": 385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.846109312745857e-06,
+      "loss": 0.9907,
+      "step": 386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.845099156191319e-06,
+      "loss": 1.0088,
+      "step": 387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.844085801063195e-06,
+      "loss": 1.0156,
+      "step": 388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.843069248743641e-06,
+      "loss": 1.0845,
+      "step": 389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.842049500619173e-06,
+      "loss": 1.0801,
+      "step": 390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8410265580806645e-06,
+      "loss": 1.0278,
+      "step": 391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.840000422523348e-06,
+      "loss": 1.0737,
+      "step": 392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.838971095346811e-06,
+      "loss": 1.0405,
+      "step": 393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8379385779549944e-06,
+      "loss": 1.021,
+      "step": 394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.836902871756187e-06,
+      "loss": 1.0415,
+      "step": 395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.835863978163032e-06,
+      "loss": 1.0488,
+      "step": 396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.834821898592516e-06,
+      "loss": 1.0195,
+      "step": 397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.833776634465973e-06,
+      "loss": 1.0327,
+      "step": 398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.83272818720908e-06,
+      "loss": 1.0293,
+      "step": 399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8316765582518565e-06,
+      "loss": 1.0093,
+      "step": 400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.830621749028659e-06,
+      "loss": 0.9956,
+      "step": 401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.829563760978186e-06,
+      "loss": 1.0186,
+      "step": 402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.828502595543467e-06,
+      "loss": 1.0161,
+      "step": 403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8274382541718695e-06,
+      "loss": 1.0298,
+      "step": 404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.82637073831509e-06,
+      "loss": 0.9888,
+      "step": 405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.825300049429155e-06,
+      "loss": 1.0068,
+      "step": 406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.82422618897442e-06,
+      "loss": 0.9551,
+      "step": 407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8231491584155665e-06,
+      "loss": 1.0566,
+      "step": 408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.822068959221599e-06,
+      "loss": 1.0229,
+      "step": 409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8209855928658425e-06,
+      "loss": 1.0591,
+      "step": 410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.819899060825943e-06,
+      "loss": 1.0337,
+      "step": 411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8188093645838674e-06,
+      "loss": 1.0273,
+      "step": 412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.817716505625894e-06,
+      "loss": 1.0576,
+      "step": 413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.816620485442616e-06,
+      "loss": 0.9683,
+      "step": 414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.815521305528939e-06,
+      "loss": 1.041,
+      "step": 415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.814418967384078e-06,
+      "loss": 1.0493,
+      "step": 416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.813313472511555e-06,
+      "loss": 1.0293,
+      "step": 417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.812204822419199e-06,
+      "loss": 0.9956,
+      "step": 418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.811093018619143e-06,
+      "loss": 1.0215,
+      "step": 419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.809978062627818e-06,
+      "loss": 1.0117,
+      "step": 420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.808859955965957e-06,
+      "loss": 1.0,
+      "step": 421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.807738700158592e-06,
+      "loss": 1.0322,
+      "step": 422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.806614296735045e-06,
+      "loss": 1.0566,
+      "step": 423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.805486747228936e-06,
+      "loss": 1.0752,
+      "step": 424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.804356053178175e-06,
+      "loss": 1.0317,
+      "step": 425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8032222161249595e-06,
+      "loss": 1.0098,
+      "step": 426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.802085237615776e-06,
+      "loss": 0.9946,
+      "step": 427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.800945119201392e-06,
+      "loss": 1.001,
+      "step": 428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.799801862436863e-06,
+      "loss": 0.9922,
+      "step": 429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.798655468881519e-06,
+      "loss": 1.0137,
+      "step": 430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.797505940098975e-06,
+      "loss": 1.0474,
+      "step": 431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.796353277657117e-06,
+      "loss": 1.1255,
+      "step": 432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.795197483128107e-06,
+      "loss": 1.0562,
+      "step": 433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.794038558088378e-06,
+      "loss": 1.0557,
+      "step": 434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.792876504118636e-06,
+      "loss": 1.0801,
+      "step": 435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.791711322803852e-06,
+      "loss": 1.041,
+      "step": 436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.79054301573326e-06,
+      "loss": 1.0317,
+      "step": 437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.789371584500364e-06,
+      "loss": 1.062,
+      "step": 438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.788197030702924e-06,
+      "loss": 1.0542,
+      "step": 439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.787019355942959e-06,
+      "loss": 1.0288,
+      "step": 440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.785838561826749e-06,
+      "loss": 1.0215,
+      "step": 441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7846546499648224e-06,
+      "loss": 1.0317,
+      "step": 442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.783467621971966e-06,
+      "loss": 0.9839,
+      "step": 443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.782277479467216e-06,
+      "loss": 1.0156,
+      "step": 444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.78108422407385e-06,
+      "loss": 1.0405,
+      "step": 445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7798878574194e-06,
+      "loss": 1.0566,
+      "step": 446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.778688381135636e-06,
+      "loss": 1.0425,
+      "step": 447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.777485796858572e-06,
+      "loss": 1.0542,
+      "step": 448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.77628010622846e-06,
+      "loss": 1.0361,
+      "step": 449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.775071310889791e-06,
+      "loss": 0.9922,
+      "step": 450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.773859412491285e-06,
+      "loss": 1.0073,
+      "step": 451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7726444126859015e-06,
+      "loss": 1.0298,
+      "step": 452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.771426313130826e-06,
+      "loss": 1.0654,
+      "step": 453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.770205115487471e-06,
+      "loss": 1.0249,
+      "step": 454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.76898082142148e-06,
+      "loss": 0.9624,
+      "step": 455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.767753432602713e-06,
+      "loss": 1.001,
+      "step": 456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7665229507052545e-06,
+      "loss": 1.0176,
+      "step": 457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.765289377407409e-06,
+      "loss": 1.0342,
+      "step": 458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.764052714391695e-06,
+      "loss": 1.0298,
+      "step": 459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.762812963344845e-06,
+      "loss": 1.0415,
+      "step": 460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7615701259578065e-06,
+      "loss": 1.0098,
+      "step": 461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.760324203925735e-06,
+      "loss": 1.0547,
+      "step": 462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.75907519894799e-06,
+      "loss": 1.0513,
+      "step": 463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.757823112728141e-06,
+      "loss": 1.0586,
+      "step": 464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.756567946973958e-06,
+      "loss": 1.0977,
+      "step": 465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.75530970339741e-06,
+      "loss": 0.9995,
+      "step": 466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7540483837146675e-06,
+      "loss": 1.0361,
+      "step": 467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.752783989646092e-06,
+      "loss": 1.0293,
+      "step": 468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.751516522916242e-06,
+      "loss": 1.0107,
+      "step": 469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.750245985253864e-06,
+      "loss": 1.0586,
+      "step": 470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.748972378391897e-06,
+      "loss": 1.0527,
+      "step": 471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.747695704067462e-06,
+      "loss": 1.0381,
+      "step": 472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.746415964021866e-06,
+      "loss": 1.0132,
+      "step": 473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.745133160000598e-06,
+      "loss": 1.0107,
+      "step": 474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.743847293753323e-06,
+      "loss": 1.0391,
+      "step": 475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7425583670338885e-06,
+      "loss": 0.9688,
+      "step": 476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.741266381600309e-06,
+      "loss": 1.0645,
+      "step": 477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.739971339214776e-06,
+      "loss": 1.0747,
+      "step": 478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.73867324164365e-06,
+      "loss": 1.0225,
+      "step": 479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.737372090657458e-06,
+      "loss": 1.0137,
+      "step": 480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.736067888030888e-06,
+      "loss": 1.0283,
+      "step": 481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.734760635542797e-06,
+      "loss": 1.0547,
+      "step": 482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733450334976197e-06,
+      "loss": 1.0171,
+      "step": 483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.732136988118259e-06,
+      "loss": 1.0459,
+      "step": 484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.73082059676031e-06,
+      "loss": 0.9922,
+      "step": 485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7295011626978255e-06,
+      "loss": 1.021,
+      "step": 486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.728178687730436e-06,
+      "loss": 1.02,
+      "step": 487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.726853173661917e-06,
+      "loss": 1.0688,
+      "step": 488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.725524622300191e-06,
+      "loss": 1.0483,
+      "step": 489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.724193035457319e-06,
+      "loss": 1.063,
+      "step": 490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.722858414949506e-06,
+      "loss": 1.0371,
+      "step": 491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.721520762597095e-06,
+      "loss": 1.0298,
+      "step": 492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.720180080224563e-06,
+      "loss": 1.0337,
+      "step": 493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718836369660517e-06,
+      "loss": 1.0405,
+      "step": 494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7174896327377e-06,
+      "loss": 1.0986,
+      "step": 495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7161398712929785e-06,
+      "loss": 1.0879,
+      "step": 496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714787087167346e-06,
+      "loss": 1.0732,
+      "step": 497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.713431282205919e-06,
+      "loss": 1.0303,
+      "step": 498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712072458257932e-06,
+      "loss": 1.0791,
+      "step": 499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710710617176739e-06,
+      "loss": 1.0293,
+      "step": 500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70934576081981e-06,
+      "loss": 1.0601,
+      "step": 501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7079778910487264e-06,
+      "loss": 0.9668,
+      "step": 502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70660700972918e-06,
+      "loss": 1.0669,
+      "step": 503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705233118730969e-06,
+      "loss": 1.1055,
+      "step": 504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703856219927999e-06,
+      "loss": 1.0156,
+      "step": 505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702476315198275e-06,
+      "loss": 1.0615,
+      "step": 506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701093406423907e-06,
+      "loss": 1.0332,
+      "step": 507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699707495491096e-06,
+      "loss": 1.1064,
+      "step": 508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698318584290141e-06,
+      "loss": 1.0078,
+      "step": 509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696926674715435e-06,
+      "loss": 1.0317,
+      "step": 510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695531768665456e-06,
+      "loss": 1.0947,
+      "step": 511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694133868042775e-06,
+      "loss": 1.0269,
+      "step": 512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.692732974754041e-06,
+      "loss": 0.9688,
+      "step": 513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691329090709989e-06,
+      "loss": 1.0244,
+      "step": 514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689922217825431e-06,
+      "loss": 1.0576,
+      "step": 515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6885123580192575e-06,
+      "loss": 1.0073,
+      "step": 516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687099513214433e-06,
+      "loss": 1.1006,
+      "step": 517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685683685337991e-06,
+      "loss": 1.0708,
+      "step": 518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684264876321035e-06,
+      "loss": 1.061,
+      "step": 519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682843088098736e-06,
+      "loss": 0.9976,
+      "step": 520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681418322610327e-06,
+      "loss": 1.0098,
+      "step": 521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679990581799102e-06,
+      "loss": 1.0615,
+      "step": 522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678559867612412e-06,
+      "loss": 1.0679,
+      "step": 523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.677126182001667e-06,
+      "loss": 1.0898,
+      "step": 524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.675689526922324e-06,
+      "loss": 1.0562,
+      "step": 525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6742499043338985e-06,
+      "loss": 1.0024,
+      "step": 526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.672807316199946e-06,
+      "loss": 0.9907,
+      "step": 527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.671361764488069e-06,
+      "loss": 1.0269,
+      "step": 528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.669913251169914e-06,
+      "loss": 1.085,
+      "step": 529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.668461778221165e-06,
+      "loss": 1.0078,
+      "step": 530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6670073476215435e-06,
+      "loss": 0.9917,
+      "step": 531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.665549961354806e-06,
+      "loss": 1.0391,
+      "step": 532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.664089621408738e-06,
+      "loss": 1.0791,
+      "step": 533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.662626329775155e-06,
+      "loss": 1.0063,
+      "step": 534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6611600884498994e-06,
+      "loss": 1.0649,
+      "step": 535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.659690899432835e-06,
+      "loss": 1.0093,
+      "step": 536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.658218764727847e-06,
+      "loss": 1.0449,
+      "step": 537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.656743686342838e-06,
+      "loss": 1.0474,
+      "step": 538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.655265666289727e-06,
+      "loss": 1.061,
+      "step": 539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.653784706584443e-06,
+      "loss": 1.0562,
+      "step": 540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6523008092469255e-06,
+      "loss": 1.0645,
+      "step": 541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6508139763011205e-06,
+      "loss": 1.0762,
+      "step": 542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.649324209774979e-06,
+      "loss": 1.0303,
+      "step": 543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.647831511700453e-06,
+      "loss": 1.0508,
+      "step": 544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.646335884113492e-06,
+      "loss": 1.0303,
+      "step": 545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.644837329054042e-06,
+      "loss": 1.0537,
+      "step": 546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6433358485660405e-06,
+      "loss": 1.0654,
+      "step": 547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.641831444697417e-06,
+      "loss": 1.0566,
+      "step": 548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.640324119500087e-06,
+      "loss": 1.0347,
+      "step": 549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.638813875029952e-06,
+      "loss": 1.0337,
+      "step": 550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.637300713346894e-06,
+      "loss": 0.9717,
+      "step": 551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.635784636514773e-06,
+      "loss": 1.0137,
+      "step": 552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.634265646601427e-06,
+      "loss": 1.0098,
+      "step": 553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.632743745678667e-06,
+      "loss": 1.0166,
+      "step": 554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.631218935822273e-06,
+      "loss": 1.0649,
+      "step": 555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.629691219111993e-06,
+      "loss": 1.0801,
+      "step": 556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.628160597631543e-06,
+      "loss": 1.0073,
+      "step": 557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.626627073468596e-06,
+      "loss": 1.0127,
+      "step": 558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6250906487147865e-06,
+      "loss": 1.061,
+      "step": 559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.623551325465705e-06,
+      "loss": 1.0293,
+      "step": 560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.622009105820896e-06,
+      "loss": 0.9629,
+      "step": 561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.620463991883853e-06,
+      "loss": 1.0488,
+      "step": 562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6189159857620194e-06,
+      "loss": 1.0161,
+      "step": 563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.617365089566782e-06,
+      "loss": 0.9761,
+      "step": 564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.615811305413468e-06,
+      "loss": 1.0459,
+      "step": 565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.614254635421347e-06,
+      "loss": 1.0386,
+      "step": 566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.61269508171362e-06,
+      "loss": 1.0571,
+      "step": 567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.611132646417428e-06,
+      "loss": 0.9976,
+      "step": 568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.609567331663836e-06,
+      "loss": 0.9829,
+      "step": 569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.607999139587838e-06,
+      "loss": 1.0737,
+      "step": 570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.606428072328355e-06,
+      "loss": 1.0352,
+      "step": 571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.604854132028227e-06,
+      "loss": 1.0537,
+      "step": 572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.603277320834213e-06,
+      "loss": 1.0244,
+      "step": 573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6016976408969895e-06,
+      "loss": 1.0391,
+      "step": 574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.600115094371144e-06,
+      "loss": 0.9937,
+      "step": 575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5985296834151735e-06,
+      "loss": 1.0244,
+      "step": 576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5969414101914846e-06,
+      "loss": 1.0186,
+      "step": 577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.595350276866384e-06,
+      "loss": 1.0601,
+      "step": 578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.593756285610083e-06,
+      "loss": 1.0034,
+      "step": 579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.592159438596688e-06,
+      "loss": 1.0093,
+      "step": 580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.590559738004203e-06,
+      "loss": 1.0259,
+      "step": 581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.588957186014523e-06,
+      "loss": 1.0225,
+      "step": 582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.587351784813431e-06,
+      "loss": 1.0181,
+      "step": 583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.585743536590599e-06,
+      "loss": 1.0469,
+      "step": 584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5841324435395785e-06,
+      "loss": 1.0615,
+      "step": 585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.582518507857804e-06,
+      "loss": 1.0869,
+      "step": 586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.580901731746587e-06,
+      "loss": 0.9849,
+      "step": 587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.579282117411111e-06,
+      "loss": 1.0464,
+      "step": 588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.577659667060432e-06,
+      "loss": 1.0444,
+      "step": 589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.576034382907476e-06,
+      "loss": 1.0205,
+      "step": 590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.574406267169031e-06,
+      "loss": 1.0474,
+      "step": 591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.57277532206575e-06,
+      "loss": 1.0615,
+      "step": 592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.571141549822142e-06,
+      "loss": 1.0107,
+      "step": 593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.569504952666574e-06,
+      "loss": 1.0454,
+      "step": 594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.567865532831266e-06,
+      "loss": 1.0732,
+      "step": 595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.566223292552287e-06,
+      "loss": 1.0122,
+      "step": 596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.564578234069556e-06,
+      "loss": 1.0415,
+      "step": 597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5629303596268295e-06,
+      "loss": 1.0215,
+      "step": 598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.561279671471711e-06,
+      "loss": 1.0142,
+      "step": 599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.55962617185564e-06,
+      "loss": 1.043,
+      "step": 600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.557969863033889e-06,
+      "loss": 1.0596,
+      "step": 601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.556310747265562e-06,
+      "loss": 1.0308,
+      "step": 602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.554648826813595e-06,
+      "loss": 1.0361,
+      "step": 603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5529841039447466e-06,
+      "loss": 0.9897,
+      "step": 604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.551316580929597e-06,
+      "loss": 1.0264,
+      "step": 605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5496462600425474e-06,
+      "loss": 1.0728,
+      "step": 606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.547973143561816e-06,
+      "loss": 1.0391,
+      "step": 607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.54629723376943e-06,
+      "loss": 1.04,
+      "step": 608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.544618532951232e-06,
+      "loss": 1.0068,
+      "step": 609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.542937043396865e-06,
+      "loss": 1.0098,
+      "step": 610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.541252767399783e-06,
+      "loss": 1.0566,
+      "step": 611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5395657072572345e-06,
+      "loss": 1.0215,
+      "step": 612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.537875865270267e-06,
+      "loss": 1.0396,
+      "step": 613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.536183243743726e-06,
+      "loss": 1.084,
+      "step": 614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.534487844986241e-06,
+      "loss": 1.0376,
+      "step": 615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.532789671310236e-06,
+      "loss": 0.9961,
+      "step": 616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.531088725031917e-06,
+      "loss": 1.0186,
+      "step": 617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.529385008471272e-06,
+      "loss": 1.0024,
+      "step": 618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.527678523952067e-06,
+      "loss": 1.0552,
+      "step": 619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.525969273801845e-06,
+      "loss": 1.0347,
+      "step": 620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.524257260351917e-06,
+      "loss": 1.0371,
+      "step": 621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.522542485937369e-06,
+      "loss": 1.0161,
+      "step": 622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.520824952897048e-06,
+      "loss": 1.0449,
+      "step": 623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.519104663573567e-06,
+      "loss": 1.085,
+      "step": 624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.517381620313295e-06,
+      "loss": 1.0317,
+      "step": 625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.515655825466359e-06,
+      "loss": 1.0098,
+      "step": 626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5139272813866395e-06,
+      "loss": 1.0015,
+      "step": 627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.512195990431767e-06,
+      "loss": 1.0483,
+      "step": 628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.510461954963116e-06,
+      "loss": 1.0479,
+      "step": 629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.508725177345809e-06,
+      "loss": 1.0059,
+      "step": 630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5069856599487014e-06,
+      "loss": 1.0029,
+      "step": 631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.505243405144394e-06,
+      "loss": 1.0659,
+      "step": 632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5034984153092145e-06,
+      "loss": 1.0967,
+      "step": 633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.501750692823225e-06,
+      "loss": 0.999,
+      "step": 634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.500000240070212e-06,
+      "loss": 1.0054,
+      "step": 635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.498247059437689e-06,
+      "loss": 1.0859,
+      "step": 636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.496491153316887e-06,
+      "loss": 1.0498,
+      "step": 637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.494732524102757e-06,
+      "loss": 1.0977,
+      "step": 638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.492971174193963e-06,
+      "loss": 1.022,
+      "step": 639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4912071059928794e-06,
+      "loss": 1.0728,
+      "step": 640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.489440321905588e-06,
+      "loss": 1.0312,
+      "step": 641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.487670824341877e-06,
+      "loss": 1.04,
+      "step": 642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.485898615715233e-06,
+      "loss": 1.0278,
+      "step": 643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4841236984428426e-06,
+      "loss": 1.083,
+      "step": 644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.482346074945585e-06,
+      "loss": 1.0161,
+      "step": 645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4805657476480305e-06,
+      "loss": 1.0615,
+      "step": 646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4787827189784395e-06,
+      "loss": 1.0811,
+      "step": 647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.476996991368755e-06,
+      "loss": 1.0283,
+      "step": 648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4752085672546005e-06,
+      "loss": 1.0137,
+      "step": 649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.47341744907528e-06,
+      "loss": 1.0586,
+      "step": 650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.47162363927377e-06,
+      "loss": 1.0518,
+      "step": 651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.469827140296719e-06,
+      "loss": 1.0103,
+      "step": 652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.468027954594442e-06,
+      "loss": 1.0269,
+      "step": 653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.466226084620919e-06,
+      "loss": 0.998,
+      "step": 654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.464421532833794e-06,
+      "loss": 1.02,
+      "step": 655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.462614301694367e-06,
+      "loss": 1.0112,
+      "step": 656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.460804393667589e-06,
+      "loss": 1.0928,
+      "step": 657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.458991811222067e-06,
+      "loss": 1.0337,
+      "step": 658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.457176556830054e-06,
+      "loss": 1.04,
+      "step": 659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4553586329674484e-06,
+      "loss": 1.063,
+      "step": 660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4535380421137865e-06,
+      "loss": 1.0332,
+      "step": 661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.451714786752245e-06,
+      "loss": 1.0156,
+      "step": 662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.449888869369634e-06,
+      "loss": 1.0312,
+      "step": 663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.448060292456395e-06,
+      "loss": 1.083,
+      "step": 664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.446229058506596e-06,
+      "loss": 1.0181,
+      "step": 665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.44439517001793e-06,
+      "loss": 1.0503,
+      "step": 666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.44255862949171e-06,
+      "loss": 0.9688,
+      "step": 667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.440719439432866e-06,
+      "loss": 1.0264,
+      "step": 668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.438877602349941e-06,
+      "loss": 1.0029,
+      "step": 669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.437033120755092e-06,
+      "loss": 1.0894,
+      "step": 670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.435185997164079e-06,
+      "loss": 1.0142,
+      "step": 671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.433336234096267e-06,
+      "loss": 1.0151,
+      "step": 672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.431483834074621e-06,
+      "loss": 1.0249,
+      "step": 673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.429628799625704e-06,
+      "loss": 0.9697,
+      "step": 674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4277711332796695e-06,
+      "loss": 1.0215,
+      "step": 675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.425910837570263e-06,
+      "loss": 0.9863,
+      "step": 676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4240479150348145e-06,
+      "loss": 1.0327,
+      "step": 677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4221823682142385e-06,
+      "loss": 1.04,
+      "step": 678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.420314199653028e-06,
+      "loss": 1.0503,
+      "step": 679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4184434118992525e-06,
+      "loss": 1.0229,
+      "step": 680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4165700075045525e-06,
+      "loss": 0.9966,
+      "step": 681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.41469398902414e-06,
+      "loss": 1.0381,
+      "step": 682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.412815359016789e-06,
+      "loss": 1.0498,
+      "step": 683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.410934120044838e-06,
+      "loss": 1.0483,
+      "step": 684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4090502746741845e-06,
+      "loss": 1.0933,
+      "step": 685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4071638254742795e-06,
+      "loss": 0.9897,
+      "step": 686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4052747750181245e-06,
+      "loss": 1.0376,
+      "step": 687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.40338312588227e-06,
+      "loss": 1.0547,
+      "step": 688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.401488880646813e-06,
+      "loss": 1.0352,
+      "step": 689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.399592041895389e-06,
+      "loss": 1.0547,
+      "step": 690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.397692612215169e-06,
+      "loss": 1.0288,
+      "step": 691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.395790594196864e-06,
+      "loss": 1.0493,
+      "step": 692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.39388599043471e-06,
+      "loss": 1.0122,
+      "step": 693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.391978803526471e-06,
+      "loss": 1.0723,
+      "step": 694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.390069036073436e-06,
+      "loss": 1.0811,
+      "step": 695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3881566906804105e-06,
+      "loss": 1.0112,
+      "step": 696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.386241769955721e-06,
+      "loss": 1.0562,
+      "step": 697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3843242765112006e-06,
+      "loss": 1.084,
+      "step": 698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.382404212962196e-06,
+      "loss": 0.9971,
+      "step": 699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3804815819275585e-06,
+      "loss": 0.9771,
+      "step": 700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.378556386029638e-06,
+      "loss": 1.0542,
+      "step": 701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.37662862789429e-06,
+      "loss": 1.0195,
+      "step": 702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.374698310150856e-06,
+      "loss": 1.0898,
+      "step": 703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.372765435432176e-06,
+      "loss": 1.0264,
+      "step": 704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.370830006374571e-06,
+      "loss": 1.0366,
+      "step": 705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.368892025617852e-06,
+      "loss": 1.0386,
+      "step": 706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.366951495805306e-06,
+      "loss": 1.0352,
+      "step": 707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3650084195837e-06,
+      "loss": 1.0571,
+      "step": 708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.363062799603271e-06,
+      "loss": 1.0244,
+      "step": 709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.361114638517728e-06,
+      "loss": 1.0942,
+      "step": 710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.359163938984245e-06,
+      "loss": 1.0469,
+      "step": 711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.357210703663458e-06,
+      "loss": 1.0942,
+      "step": 712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.355254935219462e-06,
+      "loss": 1.0527,
+      "step": 713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.353296636319808e-06,
+      "loss": 1.0215,
+      "step": 714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3513358096354966e-06,
+      "loss": 1.0835,
+      "step": 715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3493724578409756e-06,
+      "loss": 1.0234,
+      "step": 716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.347406583614141e-06,
+      "loss": 1.0161,
+      "step": 717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3454381896363245e-06,
+      "loss": 1.04,
+      "step": 718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.343467278592297e-06,
+      "loss": 1.0562,
+      "step": 719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.341493853170263e-06,
+      "loss": 1.0371,
+      "step": 720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3395179160618545e-06,
+      "loss": 1.0205,
+      "step": 721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.337539469962131e-06,
+      "loss": 1.0317,
+      "step": 722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.335558517569573e-06,
+      "loss": 1.022,
+      "step": 723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.333575061586079e-06,
+      "loss": 1.0137,
+      "step": 724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.331589104716965e-06,
+      "loss": 1.1064,
+      "step": 725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.329600649670955e-06,
+      "loss": 1.0703,
+      "step": 726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.327609699160183e-06,
+      "loss": 1.083,
+      "step": 727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.325616255900183e-06,
+      "loss": 1.042,
+      "step": 728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.323620322609894e-06,
+      "loss": 0.9995,
+      "step": 729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.321621902011645e-06,
+      "loss": 1.0562,
+      "step": 730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.319620996831164e-06,
+      "loss": 0.9985,
+      "step": 731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3176176097975635e-06,
+      "loss": 0.9839,
+      "step": 732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.315611743643342e-06,
+      "loss": 1.0054,
+      "step": 733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.31360340110438e-06,
+      "loss": 1.0654,
+      "step": 734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.311592584919936e-06,
+      "loss": 1.0303,
+      "step": 735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.309579297832642e-06,
+      "loss": 1.0601,
+      "step": 736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.307563542588498e-06,
+      "loss": 1.0293,
+      "step": 737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.305545321936875e-06,
+      "loss": 1.0166,
+      "step": 738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.303524638630503e-06,
+      "loss": 0.9912,
+      "step": 739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.301501495425472e-06,
+      "loss": 1.0156,
+      "step": 740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.299475895081226e-06,
+      "loss": 1.0273,
+      "step": 741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.297447840360562e-06,
+      "loss": 1.0278,
+      "step": 742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.295417334029626e-06,
+      "loss": 0.9575,
+      "step": 743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.293384378857903e-06,
+      "loss": 0.9702,
+      "step": 744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.291348977618224e-06,
+      "loss": 1.0752,
+      "step": 745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.289311133086751e-06,
+      "loss": 1.1016,
+      "step": 746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.287270848042982e-06,
+      "loss": 1.0044,
+      "step": 747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.285228125269742e-06,
+      "loss": 0.9775,
+      "step": 748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.283182967553183e-06,
+      "loss": 1.0493,
+      "step": 749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.281135377682775e-06,
+      "loss": 1.02,
+      "step": 750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.279085358451307e-06,
+      "loss": 1.084,
+      "step": 751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.277032912654881e-06,
+      "loss": 1.041,
+      "step": 752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.27497804309291e-06,
+      "loss": 1.0649,
+      "step": 753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.272920752568112e-06,
+      "loss": 1.043,
+      "step": 754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.270861043886506e-06,
+      "loss": 1.0254,
+      "step": 755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.268798919857412e-06,
+      "loss": 1.0215,
+      "step": 756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.266734383293441e-06,
+      "loss": 1.0444,
+      "step": 757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.264667437010497e-06,
+      "loss": 1.0283,
+      "step": 758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.262598083827769e-06,
+      "loss": 1.0664,
+      "step": 759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.26052632656773e-06,
+      "loss": 1.0234,
+      "step": 760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.258452168056132e-06,
+      "loss": 1.0239,
+      "step": 761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.256375611122003e-06,
+      "loss": 1.0439,
+      "step": 762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.25429665859764e-06,
+      "loss": 1.0244,
+      "step": 763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.252215313318608e-06,
+      "loss": 1.0332,
+      "step": 764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.250131578123737e-06,
+      "loss": 1.0469,
+      "step": 765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.248045455855116e-06,
+      "loss": 1.0312,
+      "step": 766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.24595694935809e-06,
+      "loss": 1.0728,
+      "step": 767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.243866061481256e-06,
+      "loss": 1.0049,
+      "step": 768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.241772795076458e-06,
+      "loss": 1.0488,
+      "step": 769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.239677152998784e-06,
+      "loss": 1.0186,
+      "step": 770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2375791381065654e-06,
+      "loss": 0.9648,
+      "step": 771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.235478753261366e-06,
+      "loss": 1.0234,
+      "step": 772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.233376001327984e-06,
+      "loss": 0.9854,
+      "step": 773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.231270885174448e-06,
+      "loss": 1.0298,
+      "step": 774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.229163407672007e-06,
+      "loss": 1.042,
+      "step": 775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2270535716951345e-06,
+      "loss": 1.0093,
+      "step": 776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.224941380121518e-06,
+      "loss": 1.0083,
+      "step": 777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2228268358320605e-06,
+      "loss": 1.0776,
+      "step": 778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.220709941710871e-06,
+      "loss": 1.0464,
+      "step": 779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.218590700645267e-06,
+      "loss": 1.0151,
+      "step": 780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.216469115525763e-06,
+      "loss": 1.0562,
+      "step": 781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.214345189246077e-06,
+      "loss": 1.0498,
+      "step": 782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.212218924703111e-06,
+      "loss": 1.0576,
+      "step": 783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.210090324796965e-06,
+      "loss": 1.0464,
+      "step": 784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.207959392430921e-06,
+      "loss": 1.04,
+      "step": 785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.205826130511439e-06,
+      "loss": 1.0273,
+      "step": 786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2036905419481615e-06,
+      "loss": 1.064,
+      "step": 787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.201552629653902e-06,
+      "loss": 1.0757,
+      "step": 788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1994123965446435e-06,
+      "loss": 0.9976,
+      "step": 789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.197269845539535e-06,
+      "loss": 0.9971,
+      "step": 790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1951249795608865e-06,
+      "loss": 1.0527,
+      "step": 791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.192977801534165e-06,
+      "loss": 1.0518,
+      "step": 792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1908283143879925e-06,
+      "loss": 1.063,
+      "step": 793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.188676521054139e-06,
+      "loss": 0.9517,
+      "step": 794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.186522424467522e-06,
+      "loss": 1.0679,
+      "step": 795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1843660275661964e-06,
+      "loss": 1.0723,
+      "step": 796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1822073332913605e-06,
+      "loss": 1.0864,
+      "step": 797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1800463445873405e-06,
+      "loss": 1.0532,
+      "step": 798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.177883064401596e-06,
+      "loss": 0.998,
+      "step": 799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.175717495684709e-06,
+      "loss": 0.9844,
+      "step": 800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1735496413903855e-06,
+      "loss": 1.0322,
+      "step": 801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.171379504475447e-06,
+      "loss": 1.0605,
+      "step": 802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.169207087899829e-06,
+      "loss": 1.041,
+      "step": 803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.167032394626579e-06,
+      "loss": 1.0391,
+      "step": 804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.164855427621844e-06,
+      "loss": 1.0249,
+      "step": 805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.162676189854877e-06,
+      "loss": 1.0601,
+      "step": 806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.160494684298027e-06,
+      "loss": 1.085,
+      "step": 807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.158310913926735e-06,
+      "loss": 1.0444,
+      "step": 808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.156124881719533e-06,
+      "loss": 0.9932,
+      "step": 809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1539365906580354e-06,
+      "loss": 1.0425,
+      "step": 810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.15174604372694e-06,
+      "loss": 1.0381,
+      "step": 811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.14955324391402e-06,
+      "loss": 1.02,
+      "step": 812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.147358194210122e-06,
+      "loss": 1.019,
+      "step": 813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.14516089760916e-06,
+      "loss": 0.9517,
+      "step": 814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1429613571081164e-06,
+      "loss": 1.0571,
+      "step": 815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.140759575707031e-06,
+      "loss": 1.0913,
+      "step": 816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.138555556408998e-06,
+      "loss": 1.0903,
+      "step": 817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.13634930222017e-06,
+      "loss": 1.0503,
+      "step": 818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.134140816149742e-06,
+      "loss": 1.0332,
+      "step": 819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1319301012099575e-06,
+      "loss": 1.0962,
+      "step": 820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1297171604160965e-06,
+      "loss": 1.0664,
+      "step": 821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.127501996786477e-06,
+      "loss": 1.1011,
+      "step": 822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.125284613342449e-06,
+      "loss": 0.9912,
+      "step": 823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1230650131083884e-06,
+      "loss": 1.0684,
+      "step": 824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.120843199111697e-06,
+      "loss": 1.0308,
+      "step": 825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.118619174382794e-06,
+      "loss": 1.0405,
+      "step": 826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.116392941955117e-06,
+      "loss": 1.0684,
+      "step": 827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.114164504865108e-06,
+      "loss": 0.9971,
+      "step": 828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.111933866152225e-06,
+      "loss": 1.0957,
+      "step": 829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1097010288589225e-06,
+      "loss": 1.1035,
+      "step": 830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.107465996030657e-06,
+      "loss": 1.0396,
+      "step": 831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.105228770715876e-06,
+      "loss": 1.061,
+      "step": 832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.102989355966021e-06,
+      "loss": 1.0474,
+      "step": 833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.100747754835518e-06,
+      "loss": 1.0752,
+      "step": 834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.098503970381777e-06,
+      "loss": 1.0083,
+      "step": 835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0962580056651835e-06,
+      "loss": 0.9746,
+      "step": 836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0940098637490964e-06,
+      "loss": 0.9878,
+      "step": 837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.091759547699848e-06,
+      "loss": 1.063,
+      "step": 838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.089507060586731e-06,
+      "loss": 1.0576,
+      "step": 839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.087252405482002e-06,
+      "loss": 1.0542,
+      "step": 840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.084995585460877e-06,
+      "loss": 1.0767,
+      "step": 841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.082736603601519e-06,
+      "loss": 1.0278,
+      "step": 842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0804754629850445e-06,
+      "loss": 1.0537,
+      "step": 843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.078212166695513e-06,
+      "loss": 1.0283,
+      "step": 844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.075946717819923e-06,
+      "loss": 1.0361,
+      "step": 845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.07367911944821e-06,
+      "loss": 1.0493,
+      "step": 846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.071409374673241e-06,
+      "loss": 1.0215,
+      "step": 847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.069137486590812e-06,
+      "loss": 0.9629,
+      "step": 848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.06686345829964e-06,
+      "loss": 1.0264,
+      "step": 849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0645872929013626e-06,
+      "loss": 1.0,
+      "step": 850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.062308993500531e-06,
+      "loss": 1.0107,
+      "step": 851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.06002856320461e-06,
+      "loss": 1.0898,
+      "step": 852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.057746005123966e-06,
+      "loss": 1.0215,
+      "step": 853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.055461322371873e-06,
+      "loss": 1.0684,
+      "step": 854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.053174518064499e-06,
+      "loss": 0.9888,
+      "step": 855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.050885595320906e-06,
+      "loss": 1.0635,
+      "step": 856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.048594557263049e-06,
+      "loss": 1.0342,
+      "step": 857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.046301407015763e-06,
+      "loss": 0.9648,
+      "step": 858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.044006147706768e-06,
+      "loss": 1.063,
+      "step": 859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.041708782466657e-06,
+      "loss": 1.085,
+      "step": 860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.039409314428899e-06,
+      "loss": 1.0498,
+      "step": 861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0371077467298305e-06,
+      "loss": 1.0103,
+      "step": 862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0348040825086485e-06,
+      "loss": 1.0767,
+      "step": 863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.032498324907413e-06,
+      "loss": 0.9951,
+      "step": 864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.030190477071039e-06,
+      "loss": 1.0034,
+      "step": 865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.02788054214729e-06,
+      "loss": 1.0073,
+      "step": 866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.025568523286778e-06,
+      "loss": 0.9995,
+      "step": 867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.023254423642957e-06,
+      "loss": 1.0288,
+      "step": 868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.020938246372119e-06,
+      "loss": 1.02,
+      "step": 869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.018619994633391e-06,
+      "loss": 1.0273,
+      "step": 870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.016299671588728e-06,
+      "loss": 0.9927,
+      "step": 871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.01397728040291e-06,
+      "loss": 1.0015,
+      "step": 872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.011652824243538e-06,
+      "loss": 1.0078,
+      "step": 873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0093263062810305e-06,
+      "loss": 1.0513,
+      "step": 874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.006997729688616e-06,
+      "loss": 1.0601,
+      "step": 875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.004667097642334e-06,
+      "loss": 1.0576,
+      "step": 876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.002334413321026e-06,
+      "loss": 1.0464,
+      "step": 877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999999679906331e-06,
+      "loss": 1.0293,
+      "step": 878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.997662900582685e-06,
+      "loss": 1.0732,
+      "step": 879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9953240785373145e-06,
+      "loss": 1.0327,
+      "step": 880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.992983216960231e-06,
+      "loss": 1.002,
+      "step": 881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.990640319044228e-06,
+      "loss": 0.9893,
+      "step": 882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9882953879848764e-06,
+      "loss": 0.9893,
+      "step": 883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9859484269805215e-06,
+      "loss": 1.061,
+      "step": 884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9835994392322755e-06,
+      "loss": 1.0249,
+      "step": 885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.981248427944017e-06,
+      "loss": 0.9839,
+      "step": 886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9788953963223815e-06,
+      "loss": 1.0376,
+      "step": 887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.976540347576763e-06,
+      "loss": 1.0083,
+      "step": 888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.974183284919306e-06,
+      "loss": 1.0679,
+      "step": 889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.971824211564902e-06,
+      "loss": 1.0908,
+      "step": 890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.969463130731183e-06,
+      "loss": 1.0347,
+      "step": 891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.967100045638522e-06,
+      "loss": 1.0439,
+      "step": 892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.964734959510024e-06,
+      "loss": 0.9761,
+      "step": 893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.962367875571522e-06,
+      "loss": 1.0459,
+      "step": 894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.959998797051578e-06,
+      "loss": 1.0366,
+      "step": 895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.957627727181472e-06,
+      "loss": 1.0112,
+      "step": 896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.955254669195198e-06,
+      "loss": 0.9858,
+      "step": 897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.952879626329464e-06,
+      "loss": 1.02,
+      "step": 898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.950502601823686e-06,
+      "loss": 1.0513,
+      "step": 899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.948123598919982e-06,
+      "loss": 1.0151,
+      "step": 900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9457426208631674e-06,
+      "loss": 1.0107,
+      "step": 901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.943359670900753e-06,
+      "loss": 1.0273,
+      "step": 902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.940974752282939e-06,
+      "loss": 1.0801,
+      "step": 903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9385878682626085e-06,
+      "loss": 0.9902,
+      "step": 904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.93619902209533e-06,
+      "loss": 0.9917,
+      "step": 905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.933808217039343e-06,
+      "loss": 0.9932,
+      "step": 906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.931415456355562e-06,
+      "loss": 1.0381,
+      "step": 907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.929020743307567e-06,
+      "loss": 1.084,
+      "step": 908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.926624081161604e-06,
+      "loss": 1.0977,
+      "step": 909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9242254731865734e-06,
+      "loss": 0.9834,
+      "step": 910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.921824922654033e-06,
+      "loss": 1.04,
+      "step": 911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.919422432838188e-06,
+      "loss": 1.0034,
+      "step": 912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.91701800701589e-06,
+      "loss": 1.0654,
+      "step": 913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.914611648466629e-06,
+      "loss": 1.0283,
+      "step": 914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.912203360472535e-06,
+      "loss": 1.0371,
+      "step": 915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.909793146318366e-06,
+      "loss": 1.0063,
+      "step": 916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.907381009291509e-06,
+      "loss": 1.0693,
+      "step": 917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.904966952681972e-06,
+      "loss": 1.0557,
+      "step": 918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.902550979782384e-06,
+      "loss": 1.0449,
+      "step": 919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.900133093887984e-06,
+      "loss": 1.0718,
+      "step": 920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.897713298296625e-06,
+      "loss": 1.0659,
+      "step": 921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.89529159630876e-06,
+      "loss": 1.0146,
+      "step": 922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.892867991227445e-06,
+      "loss": 1.0591,
+      "step": 923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.890442486358332e-06,
+      "loss": 1.0298,
+      "step": 924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88801508500966e-06,
+      "loss": 1.125,
+      "step": 925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88558579049226e-06,
+      "loss": 1.0981,
+      "step": 926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.883154606119544e-06,
+      "loss": 1.0635,
+      "step": 927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8807215352074975e-06,
+      "loss": 1.04,
+      "step": 928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.878286581074685e-06,
+      "loss": 1.1113,
+      "step": 929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.875849747042236e-06,
+      "loss": 1.0601,
+      "step": 930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.873411036433845e-06,
+      "loss": 1.0752,
+      "step": 931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.870970452575765e-06,
+      "loss": 1.0366,
+      "step": 932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.868527998796807e-06,
+      "loss": 1.0059,
+      "step": 933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.866083678428328e-06,
+      "loss": 1.0298,
+      "step": 934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.863637494804235e-06,
+      "loss": 1.0698,
+      "step": 935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.861189451260974e-06,
+      "loss": 1.0244,
+      "step": 936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.858739551137528e-06,
+      "loss": 1.0205,
+      "step": 937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.856287797775414e-06,
+      "loss": 0.9458,
+      "step": 938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.853834194518675e-06,
+      "loss": 1.0527,
+      "step": 939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.851378744713879e-06,
+      "loss": 1.0571,
+      "step": 940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.848921451710107e-06,
+      "loss": 1.0156,
+      "step": 941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.846462318858962e-06,
+      "loss": 1.0229,
+      "step": 942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.844001349514553e-06,
+      "loss": 1.0952,
+      "step": 943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8415385470334906e-06,
+      "loss": 1.0186,
+      "step": 944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.83907391477489e-06,
+      "loss": 1.0308,
+      "step": 945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.836607456100362e-06,
+      "loss": 1.0142,
+      "step": 946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.834139174374005e-06,
+      "loss": 1.0366,
+      "step": 947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.831669072962408e-06,
+      "loss": 1.0117,
+      "step": 948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.82919715523464e-06,
+      "loss": 0.9795,
+      "step": 949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.826723424562246e-06,
+      "loss": 1.0293,
+      "step": 950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.824247884319245e-06,
+      "loss": 1.0405,
+      "step": 951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.821770537882126e-06,
+      "loss": 0.9946,
+      "step": 952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.81929138862984e-06,
+      "loss": 0.9727,
+      "step": 953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.816810439943795e-06,
+      "loss": 1.019,
+      "step": 954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.814327695207858e-06,
+      "loss": 1.0215,
+      "step": 955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.81184315780834e-06,
+      "loss": 1.0723,
+      "step": 956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8093568311340007e-06,
+      "loss": 1.0044,
+      "step": 957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8068687185760406e-06,
+      "loss": 1.0259,
+      "step": 958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.804378823528093e-06,
+      "loss": 1.0493,
+      "step": 959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8018871493862265e-06,
+      "loss": 1.0259,
+      "step": 960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.799393699548932e-06,
+      "loss": 1.0288,
+      "step": 961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7968984774171268e-06,
+      "loss": 1.0537,
+      "step": 962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7944014863941415e-06,
+      "loss": 0.9907,
+      "step": 963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7919027298857213e-06,
+      "loss": 1.0645,
+      "step": 964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7894022113000184e-06,
+      "loss": 1.0532,
+      "step": 965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.786899934047591e-06,
+      "loss": 1.002,
+      "step": 966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.784395901541393e-06,
+      "loss": 1.0059,
+      "step": 967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7818901171967727e-06,
+      "loss": 1.0615,
+      "step": 968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7793825844314702e-06,
+      "loss": 0.9937,
+      "step": 969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7768733066656075e-06,
+      "loss": 1.0366,
+      "step": 970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7743622873216886e-06,
+      "loss": 1.0361,
+      "step": 971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7718495298245917e-06,
+      "loss": 1.0063,
+      "step": 972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.769335037601566e-06,
+      "loss": 1.0449,
+      "step": 973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.766818814082228e-06,
+      "loss": 1.0537,
+      "step": 974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7643008626985532e-06,
+      "loss": 1.021,
+      "step": 975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7617811868848774e-06,
+      "loss": 1.0664,
+      "step": 976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7592597900778836e-06,
+      "loss": 1.0269,
+      "step": 977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.756736675716606e-06,
+      "loss": 0.9629,
+      "step": 978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7542118472424207e-06,
+      "loss": 1.0464,
+      "step": 979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7516853080990403e-06,
+      "loss": 1.0469,
+      "step": 980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.749157061732511e-06,
+      "loss": 1.002,
+      "step": 981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.74662711159121e-06,
+      "loss": 1.0425,
+      "step": 982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.744095461125835e-06,
+      "loss": 1.0137,
+      "step": 983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7415621137894055e-06,
+      "loss": 1.0259,
+      "step": 984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.739027073037253e-06,
+      "loss": 1.0522,
+      "step": 985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7364903423270204e-06,
+      "loss": 1.0464,
+      "step": 986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.733951925118655e-06,
+      "loss": 1.0625,
+      "step": 987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7314118248744045e-06,
+      "loss": 1.0444,
+      "step": 988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7288700450588134e-06,
+      "loss": 1.0332,
+      "step": 989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.726326589138714e-06,
+      "loss": 1.106,
+      "step": 990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.723781460583228e-06,
+      "loss": 1.0371,
+      "step": 991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7212346628637557e-06,
+      "loss": 1.0503,
+      "step": 992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7186861994539763e-06,
+      "loss": 1.042,
+      "step": 993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.716136073829839e-06,
+      "loss": 1.0264,
+      "step": 994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.713584289469563e-06,
+      "loss": 1.0088,
+      "step": 995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7110308498536264e-06,
+      "loss": 1.0127,
+      "step": 996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7084757584647662e-06,
+      "loss": 1.0225,
+      "step": 997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.705919018787974e-06,
+      "loss": 1.1157,
+      "step": 998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7033606343104877e-06,
+      "loss": 1.0571,
+      "step": 999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.700800608521789e-06,
+      "loss": 1.0698,
+      "step": 1000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6982389449135986e-06,
+      "loss": 1.0522,
+      "step": 1001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.695675646979871e-06,
+      "loss": 1.0659,
+      "step": 1002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6931107182167904e-06,
+      "loss": 1.0371,
+      "step": 1003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.690544162122763e-06,
+      "loss": 1.0786,
+      "step": 1004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6879759821984175e-06,
+      "loss": 1.0322,
+      "step": 1005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.685406181946596e-06,
+      "loss": 1.0098,
+      "step": 1006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.682834764872351e-06,
+      "loss": 1.04,
+      "step": 1007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6802617344829393e-06,
+      "loss": 1.0801,
+      "step": 1008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6776870942878196e-06,
+      "loss": 0.9507,
+      "step": 1009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.675110847798645e-06,
+      "loss": 0.9746,
+      "step": 1010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6725329985292614e-06,
+      "loss": 1.0527,
+      "step": 1011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.669953549995698e-06,
+      "loss": 0.9746,
+      "step": 1012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6673725057161676e-06,
+      "loss": 1.0317,
+      "step": 1013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6647898692110578e-06,
+      "loss": 1.0029,
+      "step": 1014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6622056440029303e-06,
+      "loss": 1.0073,
+      "step": 1015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6596198336165107e-06,
+      "loss": 1.0317,
+      "step": 1016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.657032441578689e-06,
+      "loss": 1.0498,
+      "step": 1017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6544434714185117e-06,
+      "loss": 0.9985,
+      "step": 1018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6518529266671764e-06,
+      "loss": 1.0889,
+      "step": 1019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.649260810858031e-06,
+      "loss": 1.0239,
+      "step": 1020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6466671275265653e-06,
+      "loss": 1.0254,
+      "step": 1021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.644071880210405e-06,
+      "loss": 1.0371,
+      "step": 1022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.641475072449312e-06,
+      "loss": 1.0396,
+      "step": 1023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6388767077851745e-06,
+      "loss": 1.0586,
+      "step": 1024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6362767897620054e-06,
+      "loss": 1.0269,
+      "step": 1025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.633675321925936e-06,
+      "loss": 1.0015,
+      "step": 1026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6310723078252103e-06,
+      "loss": 1.103,
+      "step": 1027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6284677510101827e-06,
+      "loss": 1.0278,
+      "step": 1028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6258616550333128e-06,
+      "loss": 1.04,
+      "step": 1029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.623254023449156e-06,
+      "loss": 1.042,
+      "step": 1030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.620644859814365e-06,
+      "loss": 1.061,
+      "step": 1031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6180341676876818e-06,
+      "loss": 1.0278,
+      "step": 1032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.615421950629932e-06,
+      "loss": 1.0215,
+      "step": 1033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6128082122040224e-06,
+      "loss": 1.0264,
+      "step": 1034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.610192955974935e-06,
+      "loss": 0.9805,
+      "step": 1035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.60757618550972e-06,
+      "loss": 1.0176,
+      "step": 1036
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6049579043774946e-06,
+      "loss": 1.0298,
+      "step": 1037
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.602338116149437e-06,
+      "loss": 1.0312,
+      "step": 1038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.599716824398779e-06,
+      "loss": 1.0151,
+      "step": 1039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5970940327008043e-06,
+      "loss": 1.0146,
+      "step": 1040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.594469744632843e-06,
+      "loss": 1.0918,
+      "step": 1041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5918439637742648e-06,
+      "loss": 1.0337,
+      "step": 1042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5892166937064765e-06,
+      "loss": 1.04,
+      "step": 1043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5865879380129157e-06,
+      "loss": 1.0474,
+      "step": 1044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.583957700279047e-06,
+      "loss": 1.0444,
+      "step": 1045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5813259840923543e-06,
+      "loss": 1.0015,
+      "step": 1046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5786927930423408e-06,
+      "loss": 1.0195,
+      "step": 1047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.57605813072052e-06,
+      "loss": 1.0327,
+      "step": 1048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5734220007204114e-06,
+      "loss": 1.0444,
+      "step": 1049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5707844066375373e-06,
+      "loss": 1.0054,
+      "step": 1050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5681453520694164e-06,
+      "loss": 1.0435,
+      "step": 1051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.565504840615561e-06,
+      "loss": 1.02,
+      "step": 1052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5628628758774685e-06,
+      "loss": 1.1133,
+      "step": 1053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5602194614586184e-06,
+      "loss": 1.0444,
+      "step": 1054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5575746009644696e-06,
+      "loss": 0.9902,
+      "step": 1055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.554928298002451e-06,
+      "loss": 0.9507,
+      "step": 1056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5522805561819605e-06,
+      "loss": 1.0088,
+      "step": 1057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5496313791143578e-06,
+      "loss": 1.0659,
+      "step": 1058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.54698077041296e-06,
+      "loss": 1.0249,
+      "step": 1059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.544328733693038e-06,
+      "loss": 0.9556,
+      "step": 1060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.54167527257181e-06,
+      "loss": 1.0688,
+      "step": 1061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5390203906684356e-06,
+      "loss": 0.979,
+      "step": 1062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5363640916040137e-06,
+      "loss": 1.0298,
+      "step": 1063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.533706379001577e-06,
+      "loss": 1.0249,
+      "step": 1064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.531047256486082e-06,
+      "loss": 1.0166,
+      "step": 1065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5283867276844147e-06,
+      "loss": 1.0186,
+      "step": 1066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5257247962253727e-06,
+      "loss": 1.0537,
+      "step": 1067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.523061465739671e-06,
+      "loss": 1.0591,
+      "step": 1068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.520396739859932e-06,
+      "loss": 0.9956,
+      "step": 1069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5177306222206797e-06,
+      "loss": 1.0176,
+      "step": 1070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5150631164583393e-06,
+      "loss": 0.9819,
+      "step": 1071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5123942262112255e-06,
+      "loss": 1.0278,
+      "step": 1072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.509723955119544e-06,
+      "loss": 1.0381,
+      "step": 1073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5070523068253835e-06,
+      "loss": 1.0088,
+      "step": 1074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5043792849727116e-06,
+      "loss": 1.042,
+      "step": 1075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5017048932073674e-06,
+      "loss": 1.0908,
+      "step": 1076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.49902913517706e-06,
+      "loss": 1.0137,
+      "step": 1077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.496352014531361e-06,
+      "loss": 1.083,
+      "step": 1078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.493673534921703e-06,
+      "loss": 1.0103,
+      "step": 1079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4909937000013706e-06,
+      "loss": 1.04,
+      "step": 1080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.488312513425495e-06,
+      "loss": 0.9487,
+      "step": 1081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.485629978851053e-06,
+      "loss": 1.0898,
+      "step": 1082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4829460999368597e-06,
+      "loss": 1.0166,
+      "step": 1083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.480260880343565e-06,
+      "loss": 1.0083,
+      "step": 1084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.477574323733645e-06,
+      "loss": 1.0151,
+      "step": 1085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.474886433771401e-06,
+      "loss": 1.0444,
+      "step": 1086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.472197214122953e-06,
+      "loss": 1.043,
+      "step": 1087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.469506668456234e-06,
+      "loss": 1.0225,
+      "step": 1088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.466814800440985e-06,
+      "loss": 0.9492,
+      "step": 1089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.464121613748752e-06,
+      "loss": 1.0264,
+      "step": 1090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4614271120528787e-06,
+      "loss": 1.041,
+      "step": 1091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.458731299028503e-06,
+      "loss": 1.022,
+      "step": 1092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.456034178352551e-06,
+      "loss": 1.0337,
+      "step": 1093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4533357537037315e-06,
+      "loss": 1.0239,
+      "step": 1094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4506360287625337e-06,
+      "loss": 1.0645,
+      "step": 1095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4479350072112183e-06,
+      "loss": 1.0762,
+      "step": 1096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.445232692733817e-06,
+      "loss": 1.0298,
+      "step": 1097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.442529089016123e-06,
+      "loss": 1.0229,
+      "step": 1098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.439824199745688e-06,
+      "loss": 0.9966,
+      "step": 1099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4371180286118172e-06,
+      "loss": 1.0884,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.434410579305565e-06,
+      "loss": 1.021,
+      "step": 1101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4317018555197303e-06,
+      "loss": 1.0591,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4289918609488453e-06,
+      "loss": 1.0356,
+      "step": 1103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.426280599289182e-06,
+      "loss": 1.0093,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4235680742387355e-06,
+      "loss": 1.064,
+      "step": 1105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4208542894972272e-06,
+      "loss": 0.9858,
+      "step": 1106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4181392487660964e-06,
+      "loss": 1.0391,
+      "step": 1107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4154229557484924e-06,
+      "loss": 1.0908,
+      "step": 1108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.412705414149276e-06,
+      "loss": 1.0366,
+      "step": 1109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4099866276750106e-06,
+      "loss": 1.0205,
+      "step": 1110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.407266600033955e-06,
+      "loss": 1.0029,
+      "step": 1111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4045453349360643e-06,
+      "loss": 1.0264,
+      "step": 1112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.401822836092977e-06,
+      "loss": 1.0835,
+      "step": 1113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.39909910721802e-06,
+      "loss": 0.9907,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.396374152026194e-06,
+      "loss": 0.9399,
+      "step": 1115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3936479742341734e-06,
+      "loss": 1.0205,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.390920577560299e-06,
+      "loss": 1.0361,
+      "step": 1117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.388191965724576e-06,
+      "loss": 1.0571,
+      "step": 1118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3854621424486663e-06,
+      "loss": 1.02,
+      "step": 1119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3827311114558834e-06,
+      "loss": 1.0649,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3799988764711883e-06,
+      "loss": 0.9761,
+      "step": 1121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3772654412211854e-06,
+      "loss": 1.0967,
+      "step": 1122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3745308094341144e-06,
+      "loss": 1.0664,
+      "step": 1123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3717949848398485e-06,
+      "loss": 1.0107,
+      "step": 1124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.369057971169888e-06,
+      "loss": 0.9727,
+      "step": 1125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3663197721573516e-06,
+      "loss": 1.0039,
+      "step": 1126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3635803915369795e-06,
+      "loss": 1.0557,
+      "step": 1127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3608398330451206e-06,
+      "loss": 1.0635,
+      "step": 1128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3580981004197323e-06,
+      "loss": 1.0112,
+      "step": 1129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.35535519740037e-06,
+      "loss": 1.0054,
+      "step": 1130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3526111277281897e-06,
+      "loss": 1.0371,
+      "step": 1131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3498658951459357e-06,
+      "loss": 1.0132,
+      "step": 1132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3471195033979405e-06,
+      "loss": 1.0361,
+      "step": 1133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3443719562301147e-06,
+      "loss": 1.0649,
+      "step": 1134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.341623257389949e-06,
+      "loss": 1.0249,
+      "step": 1135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3388734106264997e-06,
+      "loss": 1.002,
+      "step": 1136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.336122419690394e-06,
+      "loss": 1.0347,
+      "step": 1137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.333370288333817e-06,
+      "loss": 1.0264,
+      "step": 1138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3306170203105086e-06,
+      "loss": 1.0127,
+      "step": 1139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3278626193757607e-06,
+      "loss": 1.0225,
+      "step": 1140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3251070892864097e-06,
+      "loss": 1.063,
+      "step": 1141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.322350433800832e-06,
+      "loss": 0.9995,
+      "step": 1142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3195926566789405e-06,
+      "loss": 1.0444,
+      "step": 1143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.316833761682175e-06,
+      "loss": 1.0435,
+      "step": 1144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3140737525735017e-06,
+      "loss": 1.0327,
+      "step": 1145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.311312633117407e-06,
+      "loss": 1.0522,
+      "step": 1146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3085504070798915e-06,
+      "loss": 1.0029,
+      "step": 1147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.305787078228463e-06,
+      "loss": 1.0493,
+      "step": 1148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.303022650332136e-06,
+      "loss": 1.0,
+      "step": 1149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3002571271614233e-06,
+      "loss": 1.0249,
+      "step": 1150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2974905124883315e-06,
+      "loss": 1.0166,
+      "step": 1151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2947228100863558e-06,
+      "loss": 1.1074,
+      "step": 1152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2919540237304746e-06,
+      "loss": 1.0884,
+      "step": 1153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2891841571971463e-06,
+      "loss": 0.9922,
+      "step": 1154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2864132142643e-06,
+      "loss": 0.9536,
+      "step": 1155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.283641198711337e-06,
+      "loss": 1.0347,
+      "step": 1156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2808681143191162e-06,
+      "loss": 1.0542,
+      "step": 1157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.278093964869959e-06,
+      "loss": 1.0859,
+      "step": 1158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.275318754147636e-06,
+      "loss": 1.0015,
+      "step": 1159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.272542485937369e-06,
+      "loss": 1.0317,
+      "step": 1160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2697651640258195e-06,
+      "loss": 1.0337,
+      "step": 1161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.266986792201086e-06,
+      "loss": 0.9932,
+      "step": 1162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2642073742527e-06,
+      "loss": 1.0225,
+      "step": 1163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.26142691397162e-06,
+      "loss": 1.0181,
+      "step": 1164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.258645415150226e-06,
+      "loss": 1.0171,
+      "step": 1165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2558628815823144e-06,
+      "loss": 1.0332,
+      "step": 1166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2530793170630926e-06,
+      "loss": 1.0835,
+      "step": 1167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2502947253891742e-06,
+      "loss": 1.0527,
+      "step": 1168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.247509110358575e-06,
+      "loss": 0.9717,
+      "step": 1169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.244722475770705e-06,
+      "loss": 1.0449,
+      "step": 1170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2419348254263653e-06,
+      "loss": 0.978,
+      "step": 1171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.239146163127743e-06,
+      "loss": 0.9688,
+      "step": 1172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.236356492678404e-06,
+      "loss": 1.0488,
+      "step": 1173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2335658178832926e-06,
+      "loss": 1.0083,
+      "step": 1174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.230774142548718e-06,
+      "loss": 1.0234,
+      "step": 1175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2279814704823575e-06,
+      "loss": 0.98,
+      "step": 1176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2251878054932482e-06,
+      "loss": 1.0581,
+      "step": 1177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.222393151391779e-06,
+      "loss": 1.0107,
+      "step": 1178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2195975119896907e-06,
+      "loss": 1.0454,
+      "step": 1179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.216800891100065e-06,
+      "loss": 1.0195,
+      "step": 1180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.214003292537325e-06,
+      "loss": 1.0405,
+      "step": 1181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.211204720117225e-06,
+      "loss": 1.0732,
+      "step": 1182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2084051776568504e-06,
+      "loss": 1.0366,
+      "step": 1183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.205604668974607e-06,
+      "loss": 1.0669,
+      "step": 1184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2028031978902186e-06,
+      "loss": 1.0674,
+      "step": 1185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2000007682247243e-06,
+      "loss": 1.0845,
+      "step": 1186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1971973838004673e-06,
+      "loss": 1.0503,
+      "step": 1187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1943930484410963e-06,
+      "loss": 1.0073,
+      "step": 1188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.191587765971553e-06,
+      "loss": 1.0361,
+      "step": 1189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1887815402180756e-06,
+      "loss": 1.0171,
+      "step": 1190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1859743750081853e-06,
+      "loss": 1.0322,
+      "step": 1191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1831662741706853e-06,
+      "loss": 1.0059,
+      "step": 1192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1803572415356576e-06,
+      "loss": 1.0601,
+      "step": 1193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.177547280934451e-06,
+      "loss": 1.0273,
+      "step": 1194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1747363961996823e-06,
+      "loss": 0.9937,
+      "step": 1195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.171924591165229e-06,
+      "loss": 1.0278,
+      "step": 1196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1691118696662245e-06,
+      "loss": 1.0942,
+      "step": 1197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.166298235539048e-06,
+      "loss": 1.0977,
+      "step": 1198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1634836926213287e-06,
+      "loss": 1.0049,
+      "step": 1199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1606682447519333e-06,
+      "loss": 1.0566,
+      "step": 1200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.157851895770961e-06,
+      "loss": 1.0459,
+      "step": 1201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1550346495197433e-06,
+      "loss": 1.0112,
+      "step": 1202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1522165098408332e-06,
+      "loss": 1.0371,
+      "step": 1203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.149397480578002e-06,
+      "loss": 1.1455,
+      "step": 1204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1465775655762377e-06,
+      "loss": 1.0488,
+      "step": 1205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1437567686817317e-06,
+      "loss": 1.0068,
+      "step": 1206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.140935093741882e-06,
+      "loss": 1.04,
+      "step": 1207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.138112544605282e-06,
+      "loss": 1.0581,
+      "step": 1208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1352891251217183e-06,
+      "loss": 1.0122,
+      "step": 1209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.132464839142165e-06,
+      "loss": 0.9976,
+      "step": 1210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.129639690518777e-06,
+      "loss": 0.9946,
+      "step": 1211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.126813683104887e-06,
+      "loss": 1.0156,
+      "step": 1212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1239868207549974e-06,
+      "loss": 1.0679,
+      "step": 1213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.121159107324778e-06,
+      "loss": 1.1006,
+      "step": 1214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1183305466710605e-06,
+      "loss": 1.0308,
+      "step": 1215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.115501142651829e-06,
+      "loss": 1.0269,
+      "step": 1216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1126708991262205e-06,
+      "loss": 1.0254,
+      "step": 1217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.109839819954516e-06,
+      "loss": 1.0059,
+      "step": 1218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1070079089981364e-06,
+      "loss": 1.019,
+      "step": 1219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1041751701196377e-06,
+      "loss": 1.0669,
+      "step": 1220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1013416071827034e-06,
+      "loss": 0.9658,
+      "step": 1221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0985072240521434e-06,
+      "loss": 1.0435,
+      "step": 1222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0956720245938845e-06,
+      "loss": 0.9946,
+      "step": 1223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.092836012674968e-06,
+      "loss": 1.0352,
+      "step": 1224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.089999192163542e-06,
+      "loss": 1.1304,
+      "step": 1225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0871615669288584e-06,
+      "loss": 0.9897,
+      "step": 1226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0843231408412675e-06,
+      "loss": 1.0098,
+      "step": 1227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0814839177722108e-06,
+      "loss": 1.0762,
+      "step": 1228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.078643901594216e-06,
+      "loss": 0.9956,
+      "step": 1229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0758030961808954e-06,
+      "loss": 1.0835,
+      "step": 1230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0729615054069338e-06,
+      "loss": 1.0317,
+      "step": 1231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0701191331480905e-06,
+      "loss": 1.0166,
+      "step": 1232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0672759832811904e-06,
+      "loss": 1.0029,
+      "step": 1233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.064432059684117e-06,
+      "loss": 1.0884,
+      "step": 1234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.06158736623581e-06,
+      "loss": 1.0776,
+      "step": 1235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0587419068162605e-06,
+      "loss": 1.0776,
+      "step": 1236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0558956853065024e-06,
+      "loss": 1.0825,
+      "step": 1237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.053048705588611e-06,
+      "loss": 1.0132,
+      "step": 1238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.050200971545693e-06,
+      "loss": 1.0645,
+      "step": 1239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.047352487061887e-06,
+      "loss": 0.9956,
+      "step": 1240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.044503256022353e-06,
+      "loss": 1.1069,
+      "step": 1241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.041653282313271e-06,
+      "loss": 1.0874,
+      "step": 1242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0388025698218315e-06,
+      "loss": 1.0337,
+      "step": 1243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0359511224362353e-06,
+      "loss": 0.981,
+      "step": 1244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0330989440456837e-06,
+      "loss": 1.0,
+      "step": 1245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0302460385403763e-06,
+      "loss": 0.9878,
+      "step": 1246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0273924098115045e-06,
+      "loss": 1.0156,
+      "step": 1247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.024538061751243e-06,
+      "loss": 1.0747,
+      "step": 1248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.021682998252753e-06,
+      "loss": 0.9688,
+      "step": 1249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0188272232101666e-06,
+      "loss": 1.1143,
+      "step": 1250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.01597074051859e-06,
+      "loss": 1.0503,
+      "step": 1251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0131135540740915e-06,
+      "loss": 0.9985,
+      "step": 1252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0102556677737024e-06,
+      "loss": 0.9858,
+      "step": 1253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0073970855154057e-06,
+      "loss": 1.0024,
+      "step": 1254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.004537811198135e-06,
+      "loss": 0.9883,
+      "step": 1255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0016778487217683e-06,
+      "loss": 1.0713,
+      "step": 1256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9988172019871216e-06,
+      "loss": 0.9922,
+      "step": 1257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.995955874895944e-06,
+      "loss": 1.0132,
+      "step": 1258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9930938713509127e-06,
+      "loss": 1.0488,
+      "step": 1259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9902311952556286e-06,
+      "loss": 1.0405,
+      "step": 1260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9873678505146077e-06,
+      "loss": 0.979,
+      "step": 1261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9845038410332793e-06,
+      "loss": 1.061,
+      "step": 1262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9816391707179802e-06,
+      "loss": 1.0059,
+      "step": 1263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9787738434759472e-06,
+      "loss": 1.0088,
+      "step": 1264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9759078632153145e-06,
+      "loss": 1.0557,
+      "step": 1265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9730412338451044e-06,
+      "loss": 1.0107,
+      "step": 1266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9701739592752265e-06,
+      "loss": 1.0137,
+      "step": 1267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9673060434164712e-06,
+      "loss": 1.0244,
+      "step": 1268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9644374901805025e-06,
+      "loss": 1.0547,
+      "step": 1269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9615683034798514e-06,
+      "loss": 0.9863,
+      "step": 1270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9586984872279178e-06,
+      "loss": 0.9966,
+      "step": 1271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.955828045338957e-06,
+      "loss": 1.0317,
+      "step": 1272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.952956981728078e-06,
+      "loss": 1.0034,
+      "step": 1273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9500853003112384e-06,
+      "loss": 1.0479,
+      "step": 1274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9472130050052385e-06,
+      "loss": 1.0254,
+      "step": 1275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.944340099727715e-06,
+      "loss": 1.0239,
+      "step": 1276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9414665883971365e-06,
+      "loss": 1.0601,
+      "step": 1277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.938592474932801e-06,
+      "loss": 1.0093,
+      "step": 1278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9357177632548234e-06,
+      "loss": 1.0312,
+      "step": 1279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9328424572841375e-06,
+      "loss": 1.042,
+      "step": 1280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.929966560942487e-06,
+      "loss": 1.0854,
+      "step": 1281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9270900781524216e-06,
+      "loss": 1.0332,
+      "step": 1282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9242130128372896e-06,
+      "loss": 1.0566,
+      "step": 1283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9213353689212337e-06,
+      "loss": 1.0054,
+      "step": 1284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9184571503291865e-06,
+      "loss": 1.0156,
+      "step": 1285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.915578360986865e-06,
+      "loss": 1.1016,
+      "step": 1286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9126990048207633e-06,
+      "loss": 0.9805,
+      "step": 1287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9098190857581493e-06,
+      "loss": 1.0142,
+      "step": 1288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.906938607727059e-06,
+      "loss": 1.0479,
+      "step": 1289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.90405757465629e-06,
+      "loss": 1.0278,
+      "step": 1290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.901175990475398e-06,
+      "loss": 1.0518,
+      "step": 1291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8982938591146892e-06,
+      "loss": 1.0542,
+      "step": 1292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.895411184505217e-06,
+      "loss": 1.0151,
+      "step": 1293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.892527970578775e-06,
+      "loss": 0.9927,
+      "step": 1294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8896442212678933e-06,
+      "loss": 1.0298,
+      "step": 1295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8867599405058315e-06,
+      "loss": 0.9819,
+      "step": 1296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8838751322265746e-06,
+      "loss": 1.0771,
+      "step": 1297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.880989800364826e-06,
+      "loss": 1.0503,
+      "step": 1298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8781039488560055e-06,
+      "loss": 1.0527,
+      "step": 1299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8752175816362384e-06,
+      "loss": 1.0547,
+      "step": 1300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8723307026423565e-06,
+      "loss": 1.0244,
+      "step": 1301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.869443315811889e-06,
+      "loss": 1.0107,
+      "step": 1302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.866555425083055e-06,
+      "loss": 1.0742,
+      "step": 1303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8636670343947646e-06,
+      "loss": 1.0879,
+      "step": 1304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.860778147686608e-06,
+      "loss": 1.0054,
+      "step": 1305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.857888768898852e-06,
+      "loss": 1.0205,
+      "step": 1306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8549989019724344e-06,
+      "loss": 1.062,
+      "step": 1307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.852108550848959e-06,
+      "loss": 1.0127,
+      "step": 1308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.849217719470691e-06,
+      "loss": 1.1143,
+      "step": 1309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.84632641178055e-06,
+      "loss": 1.04,
+      "step": 1310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8434346317221033e-06,
+      "loss": 1.0347,
+      "step": 1311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.840542383239565e-06,
+      "loss": 0.9863,
+      "step": 1312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8376496702777884e-06,
+      "loss": 1.0469,
+      "step": 1313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8347564967822583e-06,
+      "loss": 1.0361,
+      "step": 1314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.831862866699089e-06,
+      "loss": 1.0122,
+      "step": 1315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8289687839750157e-06,
+      "loss": 1.0396,
+      "step": 1316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8260742525573944e-06,
+      "loss": 1.02,
+      "step": 1317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8231792763941894e-06,
+      "loss": 0.9785,
+      "step": 1318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8202838594339756e-06,
+      "loss": 1.0073,
+      "step": 1319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.817388005625924e-06,
+      "loss": 1.0801,
+      "step": 1320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8144917189198055e-06,
+      "loss": 0.9878,
+      "step": 1321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.811595003265981e-06,
+      "loss": 1.0576,
+      "step": 1322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.808697862615395e-06,
+      "loss": 1.0518,
+      "step": 1323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.805800300919572e-06,
+      "loss": 1.0015,
+      "step": 1324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8029023221306117e-06,
+      "loss": 0.9849,
+      "step": 1325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8000039302011817e-06,
+      "loss": 1.0156,
+      "step": 1326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7971051290845137e-06,
+      "loss": 1.0195,
+      "step": 1327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7942059227343974e-06,
+      "loss": 1.0244,
+      "step": 1328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7913063151051744e-06,
+      "loss": 1.021,
+      "step": 1329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7884063101517354e-06,
+      "loss": 1.0317,
+      "step": 1330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7855059118295114e-06,
+      "loss": 1.0361,
+      "step": 1331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7826051240944706e-06,
+      "loss": 1.0029,
+      "step": 1332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.779703950903112e-06,
+      "loss": 1.083,
+      "step": 1333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7768023962124613e-06,
+      "loss": 1.0205,
+      "step": 1334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7739004639800628e-06,
+      "loss": 1.0757,
+      "step": 1335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7709981581639772e-06,
+      "loss": 0.9976,
+      "step": 1336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.768095482722775e-06,
+      "loss": 1.0337,
+      "step": 1337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7651924416155298e-06,
+      "loss": 1.0156,
+      "step": 1338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7622890388018133e-06,
+      "loss": 1.0645,
+      "step": 1339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7593852782416923e-06,
+      "loss": 1.061,
+      "step": 1340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.756481163895722e-06,
+      "loss": 1.0259,
+      "step": 1341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.753576699724936e-06,
+      "loss": 1.0225,
+      "step": 1342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.750671889690851e-06,
+      "loss": 1.0308,
+      "step": 1343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7477667377554506e-06,
+      "loss": 1.0288,
+      "step": 1344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7448612478811878e-06,
+      "loss": 1.0298,
+      "step": 1345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7419554240309737e-06,
+      "loss": 1.0244,
+      "step": 1346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.739049270168177e-06,
+      "loss": 0.9937,
+      "step": 1347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7361427902566175e-06,
+      "loss": 0.9971,
+      "step": 1348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7332359882605563e-06,
+      "loss": 1.0059,
+      "step": 1349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7303288681446966e-06,
+      "loss": 1.0342,
+      "step": 1350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.727421433874175e-06,
+      "loss": 1.0073,
+      "step": 1351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7245136894145556e-06,
+      "loss": 1.0786,
+      "step": 1352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7216056387318257e-06,
+      "loss": 1.0142,
+      "step": 1353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7186972857923922e-06,
+      "loss": 1.0845,
+      "step": 1354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.715788634563072e-06,
+      "loss": 1.0879,
+      "step": 1355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.712879689011089e-06,
+      "loss": 1.0654,
+      "step": 1356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.70997045310407e-06,
+      "loss": 1.0576,
+      "step": 1357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.707060930810037e-06,
+      "loss": 1.0181,
+      "step": 1358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.704151126097403e-06,
+      "loss": 1.0396,
+      "step": 1359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7012410429349656e-06,
+      "loss": 0.9819,
+      "step": 1360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.698330685291902e-06,
+      "loss": 1.0513,
+      "step": 1361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.695420057137764e-06,
+      "loss": 1.0181,
+      "step": 1362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.692509162442473e-06,
+      "loss": 1.0244,
+      "step": 1363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6895980051763145e-06,
+      "loss": 1.0537,
+      "step": 1364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6866865893099298e-06,
+      "loss": 1.0776,
+      "step": 1365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.683774918814314e-06,
+      "loss": 0.9785,
+      "step": 1366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6808629976608114e-06,
+      "loss": 1.0386,
+      "step": 1367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6779508298211055e-06,
+      "loss": 1.0732,
+      "step": 1368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6750384192672172e-06,
+      "loss": 1.0068,
+      "step": 1369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6721257699714985e-06,
+      "loss": 0.9795,
+      "step": 1370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6692128859066283e-06,
+      "loss": 1.0039,
+      "step": 1371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.666299771045603e-06,
+      "loss": 1.0806,
+      "step": 1372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.663386429361736e-06,
+      "loss": 0.9761,
+      "step": 1373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6604728648286494e-06,
+      "loss": 1.0244,
+      "step": 1374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.657559081420269e-06,
+      "loss": 1.0557,
+      "step": 1375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6546450831108187e-06,
+      "loss": 0.9741,
+      "step": 1376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6517308738748178e-06,
+      "loss": 1.0015,
+      "step": 1377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6488164576870706e-06,
+      "loss": 0.9722,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6459018385226643e-06,
+      "loss": 0.9961,
+      "step": 1379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.642987020356964e-06,
+      "loss": 1.0605,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640072007165606e-06,
+      "loss": 1.0288,
+      "step": 1381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.637156802924492e-06,
+      "loss": 0.9858,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6342414116097838e-06,
+      "loss": 1.019,
+      "step": 1383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6313258371978996e-06,
+      "loss": 1.0024,
+      "step": 1384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.628410083665506e-06,
+      "loss": 1.0571,
+      "step": 1385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6254941549895156e-06,
+      "loss": 1.0059,
+      "step": 1386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.62257805514708e-06,
+      "loss": 1.0396,
+      "step": 1387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.61966178811558e-06,
+      "loss": 1.0405,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6167453578726303e-06,
+      "loss": 1.0308,
+      "step": 1389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.613828768396065e-06,
+      "loss": 1.0508,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.610912023663936e-06,
+      "loss": 1.0996,
+      "step": 1391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6079951276545067e-06,
+      "loss": 1.0029,
+      "step": 1392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.605078084346247e-06,
+      "loss": 0.9922,
+      "step": 1393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.602160897717828e-06,
+      "loss": 0.9883,
+      "step": 1394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.599243571748116e-06,
+      "loss": 1.0234,
+      "step": 1395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.596326110416167e-06,
+      "loss": 1.0361,
+      "step": 1396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.593408517701222e-06,
+      "loss": 0.9829,
+      "step": 1397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5904907975827015e-06,
+      "loss": 1.0044,
+      "step": 1398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5875729540401993e-06,
+      "loss": 1.04,
+      "step": 1399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.584654991053479e-06,
+      "loss": 1.0386,
+      "step": 1400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.581736912602464e-06,
+      "loss": 1.0483,
+      "step": 1401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.578818722667238e-06,
+      "loss": 1.0571,
+      "step": 1402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.575900425228035e-06,
+      "loss": 1.083,
+      "step": 1403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5729820242652376e-06,
+      "loss": 1.0425,
+      "step": 1404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.570063523759368e-06,
+      "loss": 0.9502,
+      "step": 1405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5671449276910836e-06,
+      "loss": 1.0723,
+      "step": 1406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5642262400411745e-06,
+      "loss": 1.0703,
+      "step": 1407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.561307464790554e-06,
+      "loss": 1.0239,
+      "step": 1408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.558388605920255e-06,
+      "loss": 0.9731,
+      "step": 1409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5554696674114243e-06,
+      "loss": 0.979,
+      "step": 1410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.552550653245318e-06,
+      "loss": 0.9814,
+      "step": 1411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5496315674032952e-06,
+      "loss": 1.0835,
+      "step": 1412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5467124138668126e-06,
+      "loss": 1.0508,
+      "step": 1413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.54379319661742e-06,
+      "loss": 1.0312,
+      "step": 1414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.540873919636752e-06,
+      "loss": 1.0566,
+      "step": 1415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.537954586906527e-06,
+      "loss": 1.0889,
+      "step": 1416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5350352024085383e-06,
+      "loss": 1.0518,
+      "step": 1417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5321157701246503e-06,
+      "loss": 1.0479,
+      "step": 1418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5291962940367915e-06,
+      "loss": 1.0264,
+      "step": 1419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.526276778126951e-06,
+      "loss": 1.0513,
+      "step": 1420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5233572263771727e-06,
+      "loss": 1.0171,
+      "step": 1421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.520437642769549e-06,
+      "loss": 1.0981,
+      "step": 1422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5175180312862145e-06,
+      "loss": 1.0522,
+      "step": 1423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.514598395909344e-06,
+      "loss": 1.0879,
+      "step": 1424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.511678740621143e-06,
+      "loss": 1.0459,
+      "step": 1425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5087590694038455e-06,
+      "loss": 1.0586,
+      "step": 1426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5058393862397067e-06,
+      "loss": 1.0493,
+      "step": 1427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5029196951109975e-06,
+      "loss": 1.0288,
+      "step": 1428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5e-06,
+      "loss": 0.9868,
+      "step": 1429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4970803048890033e-06,
+      "loss": 1.0098,
+      "step": 1430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4941606137602937e-06,
+      "loss": 1.0249,
+      "step": 1431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.491240930596155e-06,
+      "loss": 1.0,
+      "step": 1432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.488321259378857e-06,
+      "loss": 1.0854,
+      "step": 1433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4854016040906574e-06,
+      "loss": 1.0322,
+      "step": 1434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.482481968713787e-06,
+      "loss": 1.0215,
+      "step": 1435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4795623572304523e-06,
+      "loss": 1.0977,
+      "step": 1436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4766427736228277e-06,
+      "loss": 1.0264,
+      "step": 1437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4737232218730495e-06,
+      "loss": 1.0391,
+      "step": 1438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4708037059632094e-06,
+      "loss": 1.1123,
+      "step": 1439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.467884229875351e-06,
+      "loss": 1.0186,
+      "step": 1440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.464964797591462e-06,
+      "loss": 0.9585,
+      "step": 1441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4620454130934732e-06,
+      "loss": 1.1006,
+      "step": 1442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4591260803632484e-06,
+      "loss": 1.0444,
+      "step": 1443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4562068033825807e-06,
+      "loss": 1.0112,
+      "step": 1444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.453287586133188e-06,
+      "loss": 1.0176,
+      "step": 1445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.450368432596705e-06,
+      "loss": 1.0479,
+      "step": 1446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4474493467546828e-06,
+      "loss": 1.105,
+      "step": 1447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4445303325885765e-06,
+      "loss": 1.0073,
+      "step": 1448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4416113940797457e-06,
+      "loss": 1.0127,
+      "step": 1449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4386925352094464e-06,
+      "loss": 1.0103,
+      "step": 1450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4357737599588255e-06,
+      "loss": 1.0479,
+      "step": 1451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4328550723089173e-06,
+      "loss": 1.0073,
+      "step": 1452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.429936476240633e-06,
+      "loss": 1.0278,
+      "step": 1453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4270179757347633e-06,
+      "loss": 1.0059,
+      "step": 1454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4240995747719657e-06,
+      "loss": 1.0088,
+      "step": 1455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.421181277332763e-06,
+      "loss": 1.0039,
+      "step": 1456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.418263087397537e-06,
+      "loss": 1.0498,
+      "step": 1457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.415345008946522e-06,
+      "loss": 1.0488,
+      "step": 1458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4124270459598007e-06,
+      "loss": 1.0903,
+      "step": 1459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4095092024172994e-06,
+      "loss": 1.0298,
+      "step": 1460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.406591482298779e-06,
+      "loss": 1.0024,
+      "step": 1461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.403673889583835e-06,
+      "loss": 1.0576,
+      "step": 1462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4007564282518854e-06,
+      "loss": 1.0254,
+      "step": 1463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.397839102282173e-06,
+      "loss": 1.0337,
+      "step": 1464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.394921915653754e-06,
+      "loss": 1.04,
+      "step": 1465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3920048723454938e-06,
+      "loss": 0.9746,
+      "step": 1466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3890879763360643e-06,
+      "loss": 0.999,
+      "step": 1467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.386171231603935e-06,
+      "loss": 1.0801,
+      "step": 1468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3832546421273693e-06,
+      "loss": 0.9653,
+      "step": 1469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.380338211884421e-06,
+      "loss": 1.0659,
+      "step": 1470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.377421944852922e-06,
+      "loss": 1.0488,
+      "step": 1471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.374505845010485e-06,
+      "loss": 1.0283,
+      "step": 1472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3715899163344947e-06,
+      "loss": 1.0571,
+      "step": 1473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3686741628021016e-06,
+      "loss": 1.0449,
+      "step": 1474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.365758588390217e-06,
+      "loss": 1.0811,
+      "step": 1475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3628431970755087e-06,
+      "loss": 0.9927,
+      "step": 1476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.359927992834394e-06,
+      "loss": 1.0483,
+      "step": 1477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.357012979643036e-06,
+      "loss": 0.9756,
+      "step": 1478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3540981614773366e-06,
+      "loss": 1.022,
+      "step": 1479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3511835423129307e-06,
+      "loss": 1.0703,
+      "step": 1480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3482691261251835e-06,
+      "loss": 0.9927,
+      "step": 1481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3453549168891817e-06,
+      "loss": 1.0234,
+      "step": 1482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.342440918579732e-06,
+      "loss": 1.0459,
+      "step": 1483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3395271351713515e-06,
+      "loss": 0.9639,
+      "step": 1484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3366135706382644e-06,
+      "loss": 1.0854,
+      "step": 1485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.333700228954398e-06,
+      "loss": 1.083,
+      "step": 1486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3307871140933725e-06,
+      "loss": 1.0493,
+      "step": 1487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.327874230028502e-06,
+      "loss": 1.0029,
+      "step": 1488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3249615807327836e-06,
+      "loss": 1.1045,
+      "step": 1489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3220491701788953e-06,
+      "loss": 1.085,
+      "step": 1490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3191370023391894e-06,
+      "loss": 1.0215,
+      "step": 1491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3162250811856863e-06,
+      "loss": 1.0093,
+      "step": 1492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.313313410690071e-06,
+      "loss": 1.0586,
+      "step": 1493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3104019948236864e-06,
+      "loss": 1.1016,
+      "step": 1494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3074908375575273e-06,
+      "loss": 1.0366,
+      "step": 1495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3045799428622366e-06,
+      "loss": 1.0449,
+      "step": 1496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3016693147081e-06,
+      "loss": 1.0093,
+      "step": 1497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.298758957065036e-06,
+      "loss": 0.9878,
+      "step": 1498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.295848873902598e-06,
+      "loss": 0.9653,
+      "step": 1499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2929390691899635e-06,
+      "loss": 1.0483,
+      "step": 1500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2900295468959304e-06,
+      "loss": 1.0049,
+      "step": 1501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2871203109889117e-06,
+      "loss": 0.9829,
+      "step": 1502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.284211365436929e-06,
+      "loss": 1.0122,
+      "step": 1503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.281302714207608e-06,
+      "loss": 1.019,
+      "step": 1504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2783943612681743e-06,
+      "loss": 1.0117,
+      "step": 1505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2754863105854456e-06,
+      "loss": 0.9893,
+      "step": 1506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.272578566125826e-06,
+      "loss": 1.0283,
+      "step": 1507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.269671131855304e-06,
+      "loss": 1.0322,
+      "step": 1508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.266764011739444e-06,
+      "loss": 1.0518,
+      "step": 1509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.263857209743383e-06,
+      "loss": 1.0142,
+      "step": 1510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2609507298318235e-06,
+      "loss": 1.0156,
+      "step": 1511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.258044575969027e-06,
+      "loss": 1.0151,
+      "step": 1512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2551387521188135e-06,
+      "loss": 1.0464,
+      "step": 1513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25223326224455e-06,
+      "loss": 1.0225,
+      "step": 1514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.24932811030915e-06,
+      "loss": 1.0718,
+      "step": 1515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.246423300275065e-06,
+      "loss": 1.0127,
+      "step": 1516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2435188361042794e-06,
+      "loss": 1.0371,
+      "step": 1517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.240614721758308e-06,
+      "loss": 0.9985,
+      "step": 1518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2377109611981875e-06,
+      "loss": 1.0176,
+      "step": 1519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.234807558384471e-06,
+      "loss": 1.0322,
+      "step": 1520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2319045172772254e-06,
+      "loss": 1.0449,
+      "step": 1521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2290018418360228e-06,
+      "loss": 0.9907,
+      "step": 1522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2260995360199376e-06,
+      "loss": 0.9746,
+      "step": 1523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2231976037875404e-06,
+      "loss": 1.0635,
+      "step": 1524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.220296049096889e-06,
+      "loss": 1.0654,
+      "step": 1525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2173948759055306e-06,
+      "loss": 1.0605,
+      "step": 1526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.21449408817049e-06,
+      "loss": 1.021,
+      "step": 1527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2115936898482654e-06,
+      "loss": 0.9937,
+      "step": 1528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.208693684894826e-06,
+      "loss": 1.0864,
+      "step": 1529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2057940772656034e-06,
+      "loss": 1.0703,
+      "step": 1530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2028948709154867e-06,
+      "loss": 1.0186,
+      "step": 1531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.199996069798819e-06,
+      "loss": 1.061,
+      "step": 1532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.197097677869389e-06,
+      "loss": 1.0508,
+      "step": 1533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1941996990804287e-06,
+      "loss": 1.0029,
+      "step": 1534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1913021373846056e-06,
+      "loss": 1.0044,
+      "step": 1535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1884049967340193e-06,
+      "loss": 0.9658,
+      "step": 1536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.185508281080195e-06,
+      "loss": 1.0859,
+      "step": 1537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.182611994374077e-06,
+      "loss": 1.0078,
+      "step": 1538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1797161405660257e-06,
+      "loss": 1.0024,
+      "step": 1539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1768207236058106e-06,
+      "loss": 1.0522,
+      "step": 1540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.173925747442606e-06,
+      "loss": 0.9976,
+      "step": 1541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1710312160249856e-06,
+      "loss": 1.085,
+      "step": 1542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1681371333009127e-06,
+      "loss": 0.9932,
+      "step": 1543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1652435032177425e-06,
+      "loss": 1.0488,
+      "step": 1544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1623503297222124e-06,
+      "loss": 1.0444,
+      "step": 1545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1594576167604355e-06,
+      "loss": 1.0547,
+      "step": 1546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1565653682778975e-06,
+      "loss": 1.0273,
+      "step": 1547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.153673588219451e-06,
+      "loss": 1.0239,
+      "step": 1548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.150782280529309e-06,
+      "loss": 1.0684,
+      "step": 1549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1478914491510412e-06,
+      "loss": 1.0542,
+      "step": 1550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.145001098027567e-06,
+      "loss": 0.9854,
+      "step": 1551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1421112311011493e-06,
+      "loss": 0.9917,
+      "step": 1552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1392218523133927e-06,
+      "loss": 1.0078,
+      "step": 1553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.136332965605236e-06,
+      "loss": 1.0273,
+      "step": 1554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1334445749169457e-06,
+      "loss": 1.0586,
+      "step": 1555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.130556684188112e-06,
+      "loss": 1.0327,
+      "step": 1556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.127669297357644e-06,
+      "loss": 1.0215,
+      "step": 1557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.124782418363762e-06,
+      "loss": 1.019,
+      "step": 1558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1218960511439953e-06,
+      "loss": 1.0552,
+      "step": 1559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1190101996351745e-06,
+      "loss": 0.9746,
+      "step": 1560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1161248677734263e-06,
+      "loss": 1.0146,
+      "step": 1561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1132400594941697e-06,
+      "loss": 0.9775,
+      "step": 1562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1103557787321076e-06,
+      "loss": 1.0454,
+      "step": 1563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.107472029421226e-06,
+      "loss": 1.04,
+      "step": 1564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.104588815494784e-06,
+      "loss": 1.0537,
+      "step": 1565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.101706140885311e-06,
+      "loss": 1.041,
+      "step": 1566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0988240095246025e-06,
+      "loss": 1.0352,
+      "step": 1567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.09594242534371e-06,
+      "loss": 1.0142,
+      "step": 1568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0930613922729424e-06,
+      "loss": 0.9736,
+      "step": 1569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.090180914241852e-06,
+      "loss": 1.0083,
+      "step": 1570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.087300995179238e-06,
+      "loss": 1.04,
+      "step": 1571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.084421639013136e-06,
+      "loss": 1.0029,
+      "step": 1572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0815428496708143e-06,
+      "loss": 1.0522,
+      "step": 1573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.078664631078767e-06,
+      "loss": 1.0327,
+      "step": 1574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0757869871627112e-06,
+      "loss": 0.9668,
+      "step": 1575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0729099218475784e-06,
+      "loss": 1.0615,
+      "step": 1576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0700334390575126e-06,
+      "loss": 1.0366,
+      "step": 1577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0671575427158638e-06,
+      "loss": 0.9565,
+      "step": 1578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0642822367451783e-06,
+      "loss": 1.0215,
+      "step": 1579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0614075250672006e-06,
+      "loss": 1.0239,
+      "step": 1580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.058533411602864e-06,
+      "loss": 1.0464,
+      "step": 1581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.055659900272286e-06,
+      "loss": 1.0146,
+      "step": 1582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.052786994994763e-06,
+      "loss": 1.0396,
+      "step": 1583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.049914699688762e-06,
+      "loss": 0.9941,
+      "step": 1584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.047043018271922e-06,
+      "loss": 1.0786,
+      "step": 1585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.044171954661043e-06,
+      "loss": 0.9897,
+      "step": 1586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0413015127720826e-06,
+      "loss": 1.0459,
+      "step": 1587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0384316965201494e-06,
+      "loss": 1.0244,
+      "step": 1588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.035562509819499e-06,
+      "loss": 1.0049,
+      "step": 1589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0326939565835296e-06,
+      "loss": 1.0215,
+      "step": 1590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.029826040724774e-06,
+      "loss": 1.0376,
+      "step": 1591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0269587661548964e-06,
+      "loss": 1.0303,
+      "step": 1592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0240921367846863e-06,
+      "loss": 1.0366,
+      "step": 1593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0212261565240528e-06,
+      "loss": 1.0078,
+      "step": 1594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0183608292820197e-06,
+      "loss": 1.0356,
+      "step": 1595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.015496158966722e-06,
+      "loss": 1.019,
+      "step": 1596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0126321494853936e-06,
+      "loss": 1.082,
+      "step": 1597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0097688047443727e-06,
+      "loss": 1.085,
+      "step": 1598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0069061286490877e-06,
+      "loss": 1.0591,
+      "step": 1599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.004044125104057e-06,
+      "loss": 1.0796,
+      "step": 1600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0011827980128788e-06,
+      "loss": 1.0679,
+      "step": 1601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998322151278232e-06,
+      "loss": 1.0166,
+      "step": 1602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9954621888018656e-06,
+      "loss": 1.0493,
+      "step": 1603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9926029144845956e-06,
+      "loss": 1.0366,
+      "step": 1604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9897443322262985e-06,
+      "loss": 1.0444,
+      "step": 1605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.986886445925909e-06,
+      "loss": 1.0464,
+      "step": 1606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.984029259481411e-06,
+      "loss": 1.0278,
+      "step": 1607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.981172776789834e-06,
+      "loss": 1.0576,
+      "step": 1608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.978317001747248e-06,
+      "loss": 1.0083,
+      "step": 1609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9754619382487572e-06,
+      "loss": 0.9878,
+      "step": 1610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9726075901884964e-06,
+      "loss": 1.0107,
+      "step": 1611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9697539614596237e-06,
+      "loss": 1.0063,
+      "step": 1612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9669010559543163e-06,
+      "loss": 1.0801,
+      "step": 1613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9640488775637647e-06,
+      "loss": 1.0244,
+      "step": 1614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611974301781693e-06,
+      "loss": 1.0552,
+      "step": 1615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.95834671768673e-06,
+      "loss": 0.9683,
+      "step": 1616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9554967439776474e-06,
+      "loss": 1.0337,
+      "step": 1617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.952647512938114e-06,
+      "loss": 1.0786,
+      "step": 1618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9497990284543076e-06,
+      "loss": 1.0259,
+      "step": 1619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.94695129441139e-06,
+      "loss": 0.9878,
+      "step": 1620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.944104314693498e-06,
+      "loss": 1.0674,
+      "step": 1621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.94125809318374e-06,
+      "loss": 1.0474,
+      "step": 1622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.93841263376419e-06,
+      "loss": 1.0312,
+      "step": 1623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9355679403158843e-06,
+      "loss": 0.9824,
+      "step": 1624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.932724016718811e-06,
+      "loss": 1.0137,
+      "step": 1625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.92988086685191e-06,
+      "loss": 1.0581,
+      "step": 1626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9270384945930667e-06,
+      "loss": 1.0234,
+      "step": 1627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9241969038191055e-06,
+      "loss": 1.0142,
+      "step": 1628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9213560984057844e-06,
+      "loss": 1.0376,
+      "step": 1629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9185160822277896e-06,
+      "loss": 1.0186,
+      "step": 1630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.915676859158733e-06,
+      "loss": 0.9902,
+      "step": 1631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9128384330711416e-06,
+      "loss": 1.043,
+      "step": 1632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9100008078364586e-06,
+      "loss": 1.02,
+      "step": 1633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9071639873250333e-06,
+      "loss": 1.0117,
+      "step": 1634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9043279754061164e-06,
+      "loss": 1.0488,
+      "step": 1635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9014927759478575e-06,
+      "loss": 1.0405,
+      "step": 1636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8986583928172972e-06,
+      "loss": 1.001,
+      "step": 1637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8958248298803634e-06,
+      "loss": 1.0308,
+      "step": 1638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.892992091001864e-06,
+      "loss": 0.98,
+      "step": 1639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8901601800454845e-06,
+      "loss": 1.0264,
+      "step": 1640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8873291008737795e-06,
+      "loss": 1.0312,
+      "step": 1641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8844988573481722e-06,
+      "loss": 1.062,
+      "step": 1642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8816694533289405e-06,
+      "loss": 0.9956,
+      "step": 1643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8788408926752225e-06,
+      "loss": 1.0181,
+      "step": 1644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8760131792450034e-06,
+      "loss": 1.0698,
+      "step": 1645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8731863168951142e-06,
+      "loss": 1.0098,
+      "step": 1646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8703603094812236e-06,
+      "loss": 0.9658,
+      "step": 1647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8675351608578358e-06,
+      "loss": 1.0034,
+      "step": 1648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.864710874878282e-06,
+      "loss": 1.0171,
+      "step": 1649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8618874553947189e-06,
+      "loss": 1.0508,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8590649062581192e-06,
+      "loss": 1.0151,
+      "step": 1651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8562432313182692e-06,
+      "loss": 0.9712,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534224344237634e-06,
+      "loss": 1.0508,
+      "step": 1653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506025194219984e-06,
+      "loss": 1.0146,
+      "step": 1654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8477834901591678e-06,
+      "loss": 1.0083,
+      "step": 1655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8449653504802573e-06,
+      "loss": 0.9971,
+      "step": 1656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8421481042290393e-06,
+      "loss": 1.0532,
+      "step": 1657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8393317552480672e-06,
+      "loss": 1.0298,
+      "step": 1658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8365163073786712e-06,
+      "loss": 1.0259,
+      "step": 1659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8337017644609532e-06,
+      "loss": 0.9697,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8308881303337772e-06,
+      "loss": 1.0239,
+      "step": 1661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8280754088347714e-06,
+      "loss": 1.0366,
+      "step": 1662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8252636038003181e-06,
+      "loss": 1.0098,
+      "step": 1663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.82245271906555e-06,
+      "loss": 1.0171,
+      "step": 1664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8196427584643433e-06,
+      "loss": 1.0806,
+      "step": 1665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8168337258293145e-06,
+      "loss": 1.0269,
+      "step": 1666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8140256249918153e-06,
+      "loss": 1.0005,
+      "step": 1667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8112184597819246e-06,
+      "loss": 0.9678,
+      "step": 1668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.808412234028448e-06,
+      "loss": 1.0684,
+      "step": 1669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8056069515589054e-06,
+      "loss": 0.9517,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8028026161995335e-06,
+      "loss": 1.0537,
+      "step": 1671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7999992317752768e-06,
+      "loss": 1.0259,
+      "step": 1672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7971968021097818e-06,
+      "loss": 0.9946,
+      "step": 1673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7943953310253939e-06,
+      "loss": 1.0249,
+      "step": 1674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7915948223431506e-06,
+      "loss": 0.9741,
+      "step": 1675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.788795279882775e-06,
+      "loss": 1.0264,
+      "step": 1676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7859967074626756e-06,
+      "loss": 1.0107,
+      "step": 1677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7831991088999357e-06,
+      "loss": 1.0093,
+      "step": 1678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7804024880103101e-06,
+      "loss": 1.0088,
+      "step": 1679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7776068486082215e-06,
+      "loss": 1.0386,
+      "step": 1680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7748121945067526e-06,
+      "loss": 1.0269,
+      "step": 1681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.772018529517643e-06,
+      "loss": 1.0269,
+      "step": 1682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7692258574512827e-06,
+      "loss": 1.0674,
+      "step": 1683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.766434182116708e-06,
+      "loss": 1.0,
+      "step": 1684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7636435073215956e-06,
+      "loss": 1.0254,
+      "step": 1685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7608538368722572e-06,
+      "loss": 1.0176,
+      "step": 1686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7580651745736357e-06,
+      "loss": 1.0669,
+      "step": 1687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.755277524229296e-06,
+      "loss": 0.9888,
+      "step": 1688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.752490889641426e-06,
+      "loss": 1.0381,
+      "step": 1689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7497052746108262e-06,
+      "loss": 0.9927,
+      "step": 1690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7469206829369085e-06,
+      "loss": 1.0376,
+      "step": 1691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7441371184176865e-06,
+      "loss": 0.9658,
+      "step": 1692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7413545848497745e-06,
+      "loss": 1.0371,
+      "step": 1693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7385730860283806e-06,
+      "loss": 1.0171,
+      "step": 1694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7357926257473007e-06,
+      "loss": 1.0537,
+      "step": 1695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7330132077989159e-06,
+      "loss": 1.0767,
+      "step": 1696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7302348359741821e-06,
+      "loss": 1.0366,
+      "step": 1697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7274575140626318e-06,
+      "loss": 1.0205,
+      "step": 1698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7246812458523642e-06,
+      "loss": 1.0396,
+      "step": 1699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7219060351300417e-06,
+      "loss": 1.0181,
+      "step": 1700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7191318856808848e-06,
+      "loss": 1.0186,
+      "step": 1701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.716358801288664e-06,
+      "loss": 1.0146,
+      "step": 1702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7135867857356998e-06,
+      "loss": 1.0576,
+      "step": 1703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7108158428028537e-06,
+      "loss": 1.0127,
+      "step": 1704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7080459762695262e-06,
+      "loss": 1.0298,
+      "step": 1705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7052771899136453e-06,
+      "loss": 0.9614,
+      "step": 1706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7025094875116693e-06,
+      "loss": 1.0728,
+      "step": 1707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6997428728385773e-06,
+      "loss": 1.0068,
+      "step": 1708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6969773496678648e-06,
+      "loss": 1.061,
+      "step": 1709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6942129217715382e-06,
+      "loss": 1.0859,
+      "step": 1710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6914495929201098e-06,
+      "loss": 1.0635,
+      "step": 1711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6886873668825932e-06,
+      "loss": 1.0327,
+      "step": 1712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6859262474264985e-06,
+      "loss": 1.0698,
+      "step": 1713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6831662383178262e-06,
+      "loss": 1.0859,
+      "step": 1714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6804073433210605e-06,
+      "loss": 1.0762,
+      "step": 1715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6776495661991682e-06,
+      "loss": 1.0142,
+      "step": 1716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.674892910713591e-06,
+      "loss": 1.0317,
+      "step": 1717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.67213738062424e-06,
+      "loss": 1.0332,
+      "step": 1718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6693829796894923e-06,
+      "loss": 1.0479,
+      "step": 1719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.666629711666184e-06,
+      "loss": 1.0342,
+      "step": 1720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.663877580309607e-06,
+      "loss": 0.9551,
+      "step": 1721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6611265893735007e-06,
+      "loss": 1.0151,
+      "step": 1722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6583767426100528e-06,
+      "loss": 1.084,
+      "step": 1723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6556280437698857e-06,
+      "loss": 0.998,
+      "step": 1724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6528804966020603e-06,
+      "loss": 1.0239,
+      "step": 1725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6501341048540647e-06,
+      "loss": 1.0386,
+      "step": 1726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.647388872271811e-06,
+      "loss": 1.0283,
+      "step": 1727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6446448025996303e-06,
+      "loss": 1.0229,
+      "step": 1728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6419018995802685e-06,
+      "loss": 0.9937,
+      "step": 1729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6391601669548796e-06,
+      "loss": 1.0156,
+      "step": 1730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6364196084630207e-06,
+      "loss": 1.0571,
+      "step": 1731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6336802278426494e-06,
+      "loss": 1.0527,
+      "step": 1732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6309420288301136e-06,
+      "loss": 1.0278,
+      "step": 1733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.628205015160152e-06,
+      "loss": 0.9946,
+      "step": 1734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.625469190565886e-06,
+      "loss": 1.0776,
+      "step": 1735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6227345587788152e-06,
+      "loss": 1.0142,
+      "step": 1736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.620001123528812e-06,
+      "loss": 0.9854,
+      "step": 1737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6172688885441174e-06,
+      "loss": 0.9805,
+      "step": 1738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6145378575513343e-06,
+      "loss": 1.0596,
+      "step": 1739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.611808034275424e-06,
+      "loss": 1.062,
+      "step": 1740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.609079422439702e-06,
+      "loss": 1.063,
+      "step": 1741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6063520257658278e-06,
+      "loss": 1.0278,
+      "step": 1742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6036258479738065e-06,
+      "loss": 1.0264,
+      "step": 1743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6009008927819802e-06,
+      "loss": 1.0161,
+      "step": 1744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.598177163907023e-06,
+      "loss": 1.0464,
+      "step": 1745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5954546650639368e-06,
+      "loss": 0.9849,
+      "step": 1746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5927333999660457e-06,
+      "loss": 0.9863,
+      "step": 1747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.59001337232499e-06,
+      "loss": 1.0112,
+      "step": 1748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5872945858507239e-06,
+      "loss": 0.9531,
+      "step": 1749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5845770442515082e-06,
+      "loss": 1.0688,
+      "step": 1750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5818607512339048e-06,
+      "loss": 1.0796,
+      "step": 1751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.579145710502773e-06,
+      "loss": 1.0728,
+      "step": 1752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5764319257612649e-06,
+      "loss": 1.0781,
+      "step": 1753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.573719400710819e-06,
+      "loss": 1.0649,
+      "step": 1754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.571008139051155e-06,
+      "loss": 1.0737,
+      "step": 1755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5682981444802708e-06,
+      "loss": 1.0762,
+      "step": 1756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.565589420694435e-06,
+      "loss": 1.0669,
+      "step": 1757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5628819713881832e-06,
+      "loss": 1.0376,
+      "step": 1758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5601758002543138e-06,
+      "loss": 1.0342,
+      "step": 1759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5574709109838782e-06,
+      "loss": 0.9917,
+      "step": 1760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5547673072661837e-06,
+      "loss": 1.0225,
+      "step": 1761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.552064992788782e-06,
+      "loss": 0.9888,
+      "step": 1762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5493639712374672e-06,
+      "loss": 0.959,
+      "step": 1763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5466642462962695e-06,
+      "loss": 0.9824,
+      "step": 1764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.54396582164745e-06,
+      "loss": 1.042,
+      "step": 1765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5412687009714974e-06,
+      "loss": 0.9873,
+      "step": 1766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5385728879471217e-06,
+      "loss": 1.0161,
+      "step": 1767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.535878386251249e-06,
+      "loss": 0.9897,
+      "step": 1768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5331851995590159e-06,
+      "loss": 1.0239,
+      "step": 1769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.530493331543767e-06,
+      "loss": 1.0444,
+      "step": 1770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5278027858770472e-06,
+      "loss": 0.9868,
+      "step": 1771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5251135662285993e-06,
+      "loss": 1.0654,
+      "step": 1772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5224256762663556e-06,
+      "loss": 1.0332,
+      "step": 1773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5197391196564357e-06,
+      "loss": 0.9829,
+      "step": 1774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5170539000631407e-06,
+      "loss": 1.019,
+      "step": 1775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5143700211489476e-06,
+      "loss": 1.0684,
+      "step": 1776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5116874865745069e-06,
+      "loss": 1.0308,
+      "step": 1777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5090062999986304e-06,
+      "loss": 1.0015,
+      "step": 1778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5063264650782972e-06,
+      "loss": 1.0645,
+      "step": 1779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5036479854686392e-06,
+      "loss": 1.0039,
+      "step": 1780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5009708648229409e-06,
+      "loss": 1.0532,
+      "step": 1781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4982951067926335e-06,
+      "loss": 1.0522,
+      "step": 1782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.495620715027289e-06,
+      "loss": 0.999,
+      "step": 1783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4929476931746167e-06,
+      "loss": 1.0737,
+      "step": 1784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4902760448804559e-06,
+      "loss": 1.0142,
+      "step": 1785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4876057737887755e-06,
+      "loss": 1.0449,
+      "step": 1786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.484936883541662e-06,
+      "loss": 1.0527,
+      "step": 1787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4822693777793207e-06,
+      "loss": 0.998,
+      "step": 1788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4796032601400687e-06,
+      "loss": 1.0063,
+      "step": 1789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4769385342603292e-06,
+      "loss": 0.9961,
+      "step": 1790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4742752037746277e-06,
+      "loss": 0.957,
+      "step": 1791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4716132723155864e-06,
+      "loss": 1.0806,
+      "step": 1792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4689527435139184e-06,
+      "loss": 1.0098,
+      "step": 1793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4662936209984242e-06,
+      "loss": 1.0205,
+      "step": 1794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4636359083959867e-06,
+      "loss": 1.0142,
+      "step": 1795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.460979609331565e-06,
+      "loss": 0.9731,
+      "step": 1796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.458324727428191e-06,
+      "loss": 1.0439,
+      "step": 1797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4556712663069622e-06,
+      "loss": 0.9717,
+      "step": 1798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4530192295870405e-06,
+      "loss": 0.9766,
+      "step": 1799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4503686208856426e-06,
+      "loss": 1.0596,
+      "step": 1800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4477194438180403e-06,
+      "loss": 1.0288,
+      "step": 1801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.445071701997549e-06,
+      "loss": 1.0347,
+      "step": 1802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4424253990355308e-06,
+      "loss": 1.0371,
+      "step": 1803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.439780538541382e-06,
+      "loss": 1.021,
+      "step": 1804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4371371241225326e-06,
+      "loss": 1.0654,
+      "step": 1805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4344951593844391e-06,
+      "loss": 1.0342,
+      "step": 1806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.431854647930584e-06,
+      "loss": 0.9966,
+      "step": 1807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4292155933624642e-06,
+      "loss": 1.0718,
+      "step": 1808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4265779992795894e-06,
+      "loss": 1.0454,
+      "step": 1809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4239418692794813e-06,
+      "loss": 1.1113,
+      "step": 1810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4213072069576594e-06,
+      "loss": 0.9775,
+      "step": 1811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4186740159076461e-06,
+      "loss": 0.9922,
+      "step": 1812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4160422997209543e-06,
+      "loss": 1.0044,
+      "step": 1813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4134120619870855e-06,
+      "loss": 1.0215,
+      "step": 1814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4107833062935244e-06,
+      "loss": 0.9731,
+      "step": 1815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4081560362257365e-06,
+      "loss": 1.0439,
+      "step": 1816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.405530255367158e-06,
+      "loss": 0.9956,
+      "step": 1817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.402905967299197e-06,
+      "loss": 1.0713,
+      "step": 1818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4002831756012215e-06,
+      "loss": 0.9941,
+      "step": 1819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3976618838505637e-06,
+      "loss": 0.9834,
+      "step": 1820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3950420956225052e-06,
+      "loss": 0.9604,
+      "step": 1821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3924238144902813e-06,
+      "loss": 1.0059,
+      "step": 1822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3898070440250656e-06,
+      "loss": 1.0239,
+      "step": 1823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.387191787795978e-06,
+      "loss": 1.0474,
+      "step": 1824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3845780493700684e-06,
+      "loss": 0.9966,
+      "step": 1825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3819658323123193e-06,
+      "loss": 1.0156,
+      "step": 1826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3793551401856353e-06,
+      "loss": 0.9736,
+      "step": 1827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3767459765508448e-06,
+      "loss": 1.0249,
+      "step": 1828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3741383449666885e-06,
+      "loss": 0.9917,
+      "step": 1829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3715322489898169e-06,
+      "loss": 0.98,
+      "step": 1830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3689276921747901e-06,
+      "loss": 1.0195,
+      "step": 1831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3663246780740653e-06,
+      "loss": 1.0381,
+      "step": 1832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.363723210237996e-06,
+      "loss": 1.0317,
+      "step": 1833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.361123292214826e-06,
+      "loss": 1.0469,
+      "step": 1834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.358524927550689e-06,
+      "loss": 1.0088,
+      "step": 1835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3559281197895955e-06,
+      "loss": 1.0225,
+      "step": 1836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3533328724734358e-06,
+      "loss": 0.9814,
+      "step": 1837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3507391891419689e-06,
+      "loss": 0.9575,
+      "step": 1838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3481470733328238e-06,
+      "loss": 1.0278,
+      "step": 1839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3455565285814898e-06,
+      "loss": 1.041,
+      "step": 1840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3429675584213122e-06,
+      "loss": 1.0508,
+      "step": 1841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3403801663834897e-06,
+      "loss": 1.0044,
+      "step": 1842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3377943559970707e-06,
+      "loss": 0.979,
+      "step": 1843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3352101307889422e-06,
+      "loss": 1.0791,
+      "step": 1844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3326274942838333e-06,
+      "loss": 1.0098,
+      "step": 1845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.330046450004302e-06,
+      "loss": 1.0166,
+      "step": 1846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3274670014707392e-06,
+      "loss": 0.9639,
+      "step": 1847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3248891522013546e-06,
+      "loss": 1.002,
+      "step": 1848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3223129057121816e-06,
+      "loss": 1.0405,
+      "step": 1849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3197382655170616e-06,
+      "loss": 1.0806,
+      "step": 1850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3171652351276505e-06,
+      "loss": 0.9961,
+      "step": 1851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3145938180534045e-06,
+      "loss": 1.0498,
+      "step": 1852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3120240178015834e-06,
+      "loss": 1.0029,
+      "step": 1853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3094558378772383e-06,
+      "loss": 1.0596,
+      "step": 1854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3068892817832108e-06,
+      "loss": 1.0264,
+      "step": 1855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.30432435302013e-06,
+      "loss": 1.0283,
+      "step": 1856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3017610550864019e-06,
+      "loss": 1.0142,
+      "step": 1857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.299199391478212e-06,
+      "loss": 1.0156,
+      "step": 1858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2966393656895136e-06,
+      "loss": 1.0464,
+      "step": 1859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2940809812120276e-06,
+      "loss": 1.0332,
+      "step": 1860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2915242415352346e-06,
+      "loss": 0.9932,
+      "step": 1861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2889691501463753e-06,
+      "loss": 1.0288,
+      "step": 1862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2864157105304376e-06,
+      "loss": 1.0151,
+      "step": 1863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2838639261701614e-06,
+      "loss": 1.0317,
+      "step": 1864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2813138005460241e-06,
+      "loss": 0.9688,
+      "step": 1865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.278765337136245e-06,
+      "loss": 1.0659,
+      "step": 1866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.276218539416773e-06,
+      "loss": 1.0635,
+      "step": 1867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.273673410861287e-06,
+      "loss": 0.9834,
+      "step": 1868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.271129954941187e-06,
+      "loss": 1.0117,
+      "step": 1869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2685881751255957e-06,
+      "loss": 1.0557,
+      "step": 1870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2660480748813453e-06,
+      "loss": 0.9854,
+      "step": 1871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2635096576729804e-06,
+      "loss": 1.0508,
+      "step": 1872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.260972926962747e-06,
+      "loss": 1.0557,
+      "step": 1873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.258437886210595e-06,
+      "loss": 1.0615,
+      "step": 1874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2559045388741654e-06,
+      "loss": 1.0059,
+      "step": 1875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2533728884087909e-06,
+      "loss": 1.064,
+      "step": 1876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.250842938267489e-06,
+      "loss": 1.0615,
+      "step": 1877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2483146919009608e-06,
+      "loss": 0.9751,
+      "step": 1878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2457881527575808e-06,
+      "loss": 1.0312,
+      "step": 1879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2432633242833943e-06,
+      "loss": 1.0439,
+      "step": 1880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2407402099221174e-06,
+      "loss": 1.0547,
+      "step": 1881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2382188131151234e-06,
+      "loss": 1.0283,
+      "step": 1882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.235699137301447e-06,
+      "loss": 1.0449,
+      "step": 1883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2331811859177722e-06,
+      "loss": 1.126,
+      "step": 1884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2306649623984355e-06,
+      "loss": 1.0024,
+      "step": 1885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2281504701754094e-06,
+      "loss": 1.0249,
+      "step": 1886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2256377126783128e-06,
+      "loss": 1.0093,
+      "step": 1887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.223126693334393e-06,
+      "loss": 1.0488,
+      "step": 1888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2206174155685308e-06,
+      "loss": 0.9512,
+      "step": 1889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2181098828032273e-06,
+      "loss": 1.0112,
+      "step": 1890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2156040984586079e-06,
+      "loss": 1.0454,
+      "step": 1891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.213100065952409e-06,
+      "loss": 1.0679,
+      "step": 1892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2105977886999814e-06,
+      "loss": 1.082,
+      "step": 1893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2080972701142795e-06,
+      "loss": 1.0444,
+      "step": 1894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2055985136058595e-06,
+      "loss": 1.0151,
+      "step": 1895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2031015225828734e-06,
+      "loss": 1.0059,
+      "step": 1896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.200606300451068e-06,
+      "loss": 0.9873,
+      "step": 1897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1981128506137737e-06,
+      "loss": 1.0366,
+      "step": 1898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1956211764719072e-06,
+      "loss": 0.9766,
+      "step": 1899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1931312814239607e-06,
+      "loss": 0.9917,
+      "step": 1900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1906431688659995e-06,
+      "loss": 1.0361,
+      "step": 1901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.188156842191661e-06,
+      "loss": 1.0063,
+      "step": 1902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1856723047921434e-06,
+      "loss": 1.0659,
+      "step": 1903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1831895600562046e-06,
+      "loss": 1.0317,
+      "step": 1904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1807086113701608e-06,
+      "loss": 1.0171,
+      "step": 1905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.178229462117875e-06,
+      "loss": 1.0205,
+      "step": 1906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1757521156807556e-06,
+      "loss": 1.0215,
+      "step": 1907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1732765754377558e-06,
+      "loss": 1.0469,
+      "step": 1908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1708028447653614e-06,
+      "loss": 1.064,
+      "step": 1909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1683309270375928e-06,
+      "loss": 1.0127,
+      "step": 1910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.165860825625995e-06,
+      "loss": 1.0303,
+      "step": 1911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.16339254389964e-06,
+      "loss": 1.0195,
+      "step": 1912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1609260852251105e-06,
+      "loss": 1.0151,
+      "step": 1913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.158461452966511e-06,
+      "loss": 1.0171,
+      "step": 1914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1559986504854481e-06,
+      "loss": 1.0654,
+      "step": 1915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1535376811410384e-06,
+      "loss": 1.0532,
+      "step": 1916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1510785482898928e-06,
+      "loss": 1.0601,
+      "step": 1917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1486212552861225e-06,
+      "loss": 1.0205,
+      "step": 1918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1461658054813244e-06,
+      "loss": 1.0308,
+      "step": 1919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1437122022245859e-06,
+      "loss": 1.085,
+      "step": 1920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1412604488624721e-06,
+      "loss": 1.04,
+      "step": 1921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1388105487390273e-06,
+      "loss": 0.9932,
+      "step": 1922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1363625051957655e-06,
+      "loss": 0.999,
+      "step": 1923
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1339163215716728e-06,
+      "loss": 1.0474,
+      "step": 1924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1314720012031935e-06,
+      "loss": 1.0659,
+      "step": 1925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.129029547424235e-06,
+      "loss": 0.9482,
+      "step": 1926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1265889635661558e-06,
+      "loss": 1.0518,
+      "step": 1927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1241502529577642e-06,
+      "loss": 1.0073,
+      "step": 1928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1217134189253155e-06,
+      "loss": 1.0137,
+      "step": 1929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1192784647925031e-06,
+      "loss": 1.0093,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.116845393880458e-06,
+      "loss": 0.998,
+      "step": 1931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1144142095077406e-06,
+      "loss": 1.0015,
+      "step": 1932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1119849149903414e-06,
+      "loss": 0.9766,
+      "step": 1933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1095575136416695e-06,
+      "loss": 1.0454,
+      "step": 1934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1071320087725557e-06,
+      "loss": 1.0303,
+      "step": 1935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.10470840369124e-06,
+      "loss": 1.0181,
+      "step": 1936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1022867017033757e-06,
+      "loss": 1.0171,
+      "step": 1937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0998669061120157e-06,
+      "loss": 1.0244,
+      "step": 1938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.097449020217617e-06,
+      "loss": 1.0327,
+      "step": 1939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0950330473180287e-06,
+      "loss": 0.9478,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0926189907084922e-06,
+      "loss": 1.0537,
+      "step": 1941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.090206853681634e-06,
+      "loss": 1.0225,
+      "step": 1942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0877966395274654e-06,
+      "loss": 0.999,
+      "step": 1943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.08538835153337e-06,
+      "loss": 1.0176,
+      "step": 1944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0829819929841104e-06,
+      "loss": 1.0073,
+      "step": 1945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0805775671618124e-06,
+      "loss": 1.0464,
+      "step": 1946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.078175077345967e-06,
+      "loss": 1.0552,
+      "step": 1947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.075774526813427e-06,
+      "loss": 1.021,
+      "step": 1948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.073375918838397e-06,
+      "loss": 1.0107,
+      "step": 1949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0709792566924333e-06,
+      "loss": 1.04,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0685845436444391e-06,
+      "loss": 1.0396,
+      "step": 1951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0661917829606585e-06,
+      "loss": 0.9595,
+      "step": 1952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0638009779046707e-06,
+      "loss": 1.0322,
+      "step": 1953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.061412131737392e-06,
+      "loss": 1.0132,
+      "step": 1954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0590252477170614e-06,
+      "loss": 1.0742,
+      "step": 1955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0566403290992471e-06,
+      "loss": 1.0288,
+      "step": 1956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0542573791368323e-06,
+      "loss": 1.1035,
+      "step": 1957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0518764010800193e-06,
+      "loss": 1.0181,
+      "step": 1958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0494973981763145e-06,
+      "loss": 1.0112,
+      "step": 1959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0471203736705371e-06,
+      "loss": 0.9917,
+      "step": 1960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.044745330804803e-06,
+      "loss": 1.0576,
+      "step": 1961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0423722728185292e-06,
+      "loss": 1.0181,
+      "step": 1962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0400012029484216e-06,
+      "loss": 0.998,
+      "step": 1963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0376321244284778e-06,
+      "loss": 1.0181,
+      "step": 1964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0352650404899765e-06,
+      "loss": 0.9922,
+      "step": 1965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0328999543614782e-06,
+      "loss": 1.0283,
+      "step": 1966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0305368692688175e-06,
+      "loss": 1.061,
+      "step": 1967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.028175788435099e-06,
+      "loss": 1.0615,
+      "step": 1968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0258167150806938e-06,
+      "loss": 1.0083,
+      "step": 1969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0234596524232374e-06,
+      "loss": 1.0039,
+      "step": 1970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0211046036776187e-06,
+      "loss": 1.0176,
+      "step": 1971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.018751572055984e-06,
+      "loss": 1.0649,
+      "step": 1972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0164005607677253e-06,
+      "loss": 1.0034,
+      "step": 1973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.014051573019479e-06,
+      "loss": 0.9907,
+      "step": 1974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0117046120151242e-06,
+      "loss": 1.0874,
+      "step": 1975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0093596809557732e-06,
+      "loss": 1.0483,
+      "step": 1976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0070167830397702e-06,
+      "loss": 1.02,
+      "step": 1977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.004675921462686e-06,
+      "loss": 1.0254,
+      "step": 1978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0023370994173155e-06,
+      "loss": 0.9995,
+      "step": 1979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.000000320093669e-06,
+      "loss": 1.0137,
+      "step": 1980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.976655866789745e-07,
+      "loss": 0.9404,
+      "step": 1981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.953329023576655e-07,
+      "loss": 0.9761,
+      "step": 1982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.93002270311384e-07,
+      "loss": 1.0913,
+      "step": 1983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.9067369371897e-07,
+      "loss": 0.9878,
+      "step": 1984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.883471757564634e-07,
+      "loss": 1.019,
+      "step": 1985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.860227195970906e-07,
+      "loss": 1.0146,
+      "step": 1986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.837003284112727e-07,
+      "loss": 0.9927,
+      "step": 1987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.813800053666086e-07,
+      "loss": 1.0522,
+      "step": 1988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.790617536278809e-07,
+      "loss": 1.0073,
+      "step": 1989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.767455763570433e-07,
+      "loss": 1.0352,
+      "step": 1990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.74431476713223e-07,
+      "loss": 1.0088,
+      "step": 1991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.721194578527112e-07,
+      "loss": 0.9961,
+      "step": 1992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.698095229289614e-07,
+      "loss": 0.9985,
+      "step": 1993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.67501675092587e-07,
+      "loss": 0.9868,
+      "step": 1994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.65195917491352e-07,
+      "loss": 1.0405,
+      "step": 1995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.6289225327017e-07,
+      "loss": 1.0435,
+      "step": 1996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.605906855711011e-07,
+      "loss": 0.9395,
+      "step": 1997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.582912175333438e-07,
+      "loss": 1.0776,
+      "step": 1998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.55993852293233e-07,
+      "loss": 1.0024,
+      "step": 1999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.53698592984238e-07,
+      "loss": 1.04,
+      "step": 2000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.514054427369515e-07,
+      "loss": 1.0435,
+      "step": 2001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.491144046790939e-07,
+      "loss": 1.0259,
+      "step": 2002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.468254819355019e-07,
+      "loss": 1.0303,
+      "step": 2003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.445386776281282e-07,
+      "loss": 1.0024,
+      "step": 2004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.422539948760342e-07,
+      "loss": 1.0171,
+      "step": 2005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.399714367953913e-07,
+      "loss": 1.0029,
+      "step": 2006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.37691006499469e-07,
+      "loss": 1.021,
+      "step": 2007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.354127070986385e-07,
+      "loss": 1.0415,
+      "step": 2008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.331365417003602e-07,
+      "loss": 1.0073,
+      "step": 2009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.308625134091886e-07,
+      "loss": 1.0308,
+      "step": 2010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.285906253267587e-07,
+      "loss": 0.9873,
+      "step": 2011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.263208805517912e-07,
+      "loss": 0.9902,
+      "step": 2012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.24053282180078e-07,
+      "loss": 1.0034,
+      "step": 2013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.21787833304488e-07,
+      "loss": 1.103,
+      "step": 2014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.195245370149555e-07,
+      "loss": 1.0088,
+      "step": 2015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.172633963984818e-07,
+      "loss": 1.0767,
+      "step": 2016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.150044145391237e-07,
+      "loss": 0.9736,
+      "step": 2017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.127475945179982e-07,
+      "loss": 1.0132,
+      "step": 2018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.104929394132706e-07,
+      "loss": 0.958,
+      "step": 2019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.082404523001531e-07,
+      "loss": 0.9673,
+      "step": 2020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.059901362509044e-07,
+      "loss": 1.0376,
+      "step": 2021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.03741994334818e-07,
+      "loss": 1.0054,
+      "step": 2022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.01496029618224e-07,
+      "loss": 0.9878,
+      "step": 2023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.992522451644823e-07,
+      "loss": 1.021,
+      "step": 2024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.970106440339801e-07,
+      "loss": 1.0391,
+      "step": 2025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.947712292841248e-07,
+      "loss": 1.0024,
+      "step": 2026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.925340039693444e-07,
+      "loss": 1.0337,
+      "step": 2027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.902989711410773e-07,
+      "loss": 0.999,
+      "step": 2028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.880661338477753e-07,
+      "loss": 1.0078,
+      "step": 2029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.858354951348924e-07,
+      "loss": 0.9888,
+      "step": 2030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.83607058044885e-07,
+      "loss": 0.9946,
+      "step": 2031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.813808256172063e-07,
+      "loss": 0.9937,
+      "step": 2032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.791568008883039e-07,
+      "loss": 1.0059,
+      "step": 2033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.769349868916119e-07,
+      "loss": 1.022,
+      "step": 2034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.747153866575522e-07,
+      "loss": 1.0469,
+      "step": 2035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.724980032135233e-07,
+      "loss": 0.9766,
+      "step": 2036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.702828395839044e-07,
+      "loss": 1.0771,
+      "step": 2037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.680698987900435e-07,
+      "loss": 1.042,
+      "step": 2038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.658591838502587e-07,
+      "loss": 1.0332,
+      "step": 2039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.636506977798306e-07,
+      "loss": 1.0391,
+      "step": 2040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.614444435910024e-07,
+      "loss": 1.0659,
+      "step": 2041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.592404242929697e-07,
+      "loss": 1.0288,
+      "step": 2042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.57038642891884e-07,
+      "loss": 0.9707,
+      "step": 2043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.548391023908403e-07,
+      "loss": 1.0522,
+      "step": 2044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.526418057898791e-07,
+      "loss": 1.0239,
+      "step": 2045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.504467560859814e-07,
+      "loss": 1.02,
+      "step": 2046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.482539562730607e-07,
+      "loss": 1.04,
+      "step": 2047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.460634093419662e-07,
+      "loss": 1.0156,
+      "step": 2048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.43875118280468e-07,
+      "loss": 1.1001,
+      "step": 2049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.416890860732657e-07,
+      "loss": 0.9956,
+      "step": 2050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.395053157019733e-07,
+      "loss": 0.9927,
+      "step": 2051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.373238101451234e-07,
+      "loss": 1.0181,
+      "step": 2052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.351445723781562e-07,
+      "loss": 1.0181,
+      "step": 2053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.32967605373422e-07,
+      "loss": 0.9766,
+      "step": 2054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.307929121001704e-07,
+      "loss": 0.9927,
+      "step": 2055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.286204955245535e-07,
+      "loss": 1.0151,
+      "step": 2056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.264503586096159e-07,
+      "loss": 1.0312,
+      "step": 2057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.242825043152924e-07,
+      "loss": 1.0542,
+      "step": 2058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.221169355984051e-07,
+      "loss": 1.0605,
+      "step": 2059
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.199536554126603e-07,
+      "loss": 1.0186,
+      "step": 2060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.177926667086399e-07,
+      "loss": 1.042,
+      "step": 2061
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.156339724338036e-07,
+      "loss": 1.0693,
+      "step": 2062
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.134775755324784e-07,
+      "loss": 0.9844,
+      "step": 2063
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.11323478945861e-07,
+      "loss": 1.0176,
+      "step": 2064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.09171685612008e-07,
+      "loss": 0.9912,
+      "step": 2065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.070221984658358e-07,
+      "loss": 1.0278,
+      "step": 2066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.048750204391143e-07,
+      "loss": 1.04,
+      "step": 2067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.027301544604657e-07,
+      "loss": 1.0083,
+      "step": 2068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.005876034553575e-07,
+      "loss": 1.0181,
+      "step": 2069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.984473703460985e-07,
+      "loss": 0.9946,
+      "step": 2070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.963094580518394e-07,
+      "loss": 1.0322,
+      "step": 2071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.941738694885614e-07,
+      "loss": 0.9697,
+      "step": 2072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.920406075690804e-07,
+      "loss": 1.0049,
+      "step": 2073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.899096752030346e-07,
+      "loss": 1.0376,
+      "step": 2074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.877810752968901e-07,
+      "loss": 1.02,
+      "step": 2075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.856548107539247e-07,
+      "loss": 1.0474,
+      "step": 2076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.835308844742376e-07,
+      "loss": 1.0615,
+      "step": 2077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.814092993547342e-07,
+      "loss": 1.0791,
+      "step": 2078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.792900582891303e-07,
+      "loss": 1.0361,
+      "step": 2079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.771731641679406e-07,
+      "loss": 1.0083,
+      "step": 2080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.750586198784829e-07,
+      "loss": 1.0181,
+      "step": 2081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.72946428304866e-07,
+      "loss": 1.0103,
+      "step": 2082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.708365923279931e-07,
+      "loss": 0.9775,
+      "step": 2083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.687291148255527e-07,
+      "loss": 1.0142,
+      "step": 2084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.666239986720162e-07,
+      "loss": 1.0474,
+      "step": 2085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.645212467386346e-07,
+      "loss": 1.0283,
+      "step": 2086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.624208618934356e-07,
+      "loss": 1.0396,
+      "step": 2087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.603228470012162e-07,
+      "loss": 0.9951,
+      "step": 2088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.582272049235431e-07,
+      "loss": 1.0215,
+      "step": 2089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.561339385187449e-07,
+      "loss": 1.0024,
+      "step": 2090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.540430506419099e-07,
+      "loss": 0.9609,
+      "step": 2091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.519545441448842e-07,
+      "loss": 1.0498,
+      "step": 2092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.498684218762639e-07,
+      "loss": 1.0283,
+      "step": 2093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.477846866813934e-07,
+      "loss": 1.0894,
+      "step": 2094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.457033414023613e-07,
+      "loss": 1.0615,
+      "step": 2095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.436243888779982e-07,
+      "loss": 1.0386,
+      "step": 2096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.41547831943868e-07,
+      "loss": 1.0293,
+      "step": 2097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.394736734322705e-07,
+      "loss": 0.9868,
+      "step": 2098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.374019161722315e-07,
+      "loss": 1.0449,
+      "step": 2099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.353325629895039e-07,
+      "loss": 1.0029,
+      "step": 2100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.332656167065591e-07,
+      "loss": 0.9878,
+      "step": 2101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.312010801425892e-07,
+      "loss": 1.0098,
+      "step": 2102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.29138956113494e-07,
+      "loss": 1.0288,
+      "step": 2103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.270792474318889e-07,
+      "loss": 0.9541,
+      "step": 2104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.250219569070904e-07,
+      "loss": 0.9346,
+      "step": 2105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.229670873451197e-07,
+      "loss": 1.0762,
+      "step": 2106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.20914641548694e-07,
+      "loss": 0.9526,
+      "step": 2107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.18864622317226e-07,
+      "loss": 1.0376,
+      "step": 2108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.168170324468171e-07,
+      "loss": 1.0493,
+      "step": 2109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.147718747302577e-07,
+      "loss": 0.9941,
+      "step": 2110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.127291519570184e-07,
+      "loss": 1.0103,
+      "step": 2111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.106888669132497e-07,
+      "loss": 0.978,
+      "step": 2112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.086510223817766e-07,
+      "loss": 1.063,
+      "step": 2113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.066156211420975e-07,
+      "loss": 0.9873,
+      "step": 2114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.045826659703756e-07,
+      "loss": 1.0122,
+      "step": 2115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.025521596394382e-07,
+      "loss": 1.0859,
+      "step": 2116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.005241049187752e-07,
+      "loss": 1.0078,
+      "step": 2117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.98498504574529e-07,
+      "loss": 0.9414,
+      "step": 2118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.964753613694977e-07,
+      "loss": 1.0469,
+      "step": 2119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.944546780631256e-07,
+      "loss": 1.0254,
+      "step": 2120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.924364574115025e-07,
+      "loss": 1.002,
+      "step": 2121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.90420702167359e-07,
+      "loss": 1.042,
+      "step": 2122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.884074150800649e-07,
+      "loss": 1.0161,
+      "step": 2123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.863965988956203e-07,
+      "loss": 1.0181,
+      "step": 2124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.843882563566589e-07,
+      "loss": 0.998,
+      "step": 2125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.82382390202437e-07,
+      "loss": 1.0073,
+      "step": 2126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.803790031688365e-07,
+      "loss": 0.9951,
+      "step": 2127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.783780979883548e-07,
+      "loss": 1.0215,
+      "step": 2128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.763796773901074e-07,
+      "loss": 1.0542,
+      "step": 2129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.743837440998169e-07,
+      "loss": 1.0273,
+      "step": 2130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.723903008398178e-07,
+      "loss": 1.0713,
+      "step": 2131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.703993503290448e-07,
+      "loss": 1.0767,
+      "step": 2132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.684108952830354e-07,
+      "loss": 1.064,
+      "step": 2133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.66424938413921e-07,
+      "loss": 1.0327,
+      "step": 2134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.644414824304282e-07,
+      "loss": 1.0454,
+      "step": 2135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.624605300378703e-07,
+      "loss": 1.0835,
+      "step": 2136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.604820839381459e-07,
+      "loss": 0.9961,
+      "step": 2137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.585061468297377e-07,
+      "loss": 1.002,
+      "step": 2138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.565327214077033e-07,
+      "loss": 0.9863,
+      "step": 2139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.545618103636764e-07,
+      "loss": 1.0127,
+      "step": 2140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.525934163858597e-07,
+      "loss": 1.0117,
+      "step": 2141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.50627542159025e-07,
+      "loss": 1.0298,
+      "step": 2142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.486641903645044e-07,
+      "loss": 1.0098,
+      "step": 2143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.467033636801928e-07,
+      "loss": 1.0469,
+      "step": 2144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.447450647805378e-07,
+      "loss": 1.0576,
+      "step": 2145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.427892963365425e-07,
+      "loss": 1.0161,
+      "step": 2146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.40836061015756e-07,
+      "loss": 1.043,
+      "step": 2147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.388853614822732e-07,
+      "loss": 1.022,
+      "step": 2148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.369372003967297e-07,
+      "loss": 1.0615,
+      "step": 2149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.349915804163012e-07,
+      "loss": 1.0132,
+      "step": 2150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.330485041946943e-07,
+      "loss": 0.9966,
+      "step": 2151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.311079743821489e-07,
+      "loss": 1.0664,
+      "step": 2152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.29169993625429e-07,
+      "loss": 1.0845,
+      "step": 2153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.272345645678249e-07,
+      "loss": 1.0146,
+      "step": 2154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.253016898491435e-07,
+      "loss": 1.0488,
+      "step": 2155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.233713721057108e-07,
+      "loss": 1.0776,
+      "step": 2156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.214436139703614e-07,
+      "loss": 1.0864,
+      "step": 2157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.195184180724429e-07,
+      "loss": 1.0542,
+      "step": 2158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.175957870378043e-07,
+      "loss": 1.0366,
+      "step": 2159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.156757234888006e-07,
+      "loss": 1.0381,
+      "step": 2160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.137582300442807e-07,
+      "loss": 1.0674,
+      "step": 2161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.118433093195897e-07,
+      "loss": 1.0059,
+      "step": 2162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.099309639265652e-07,
+      "loss": 1.0752,
+      "step": 2163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.080211964735292e-07,
+      "loss": 1.0649,
+      "step": 2164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.061140095652906e-07,
+      "loss": 1.0635,
+      "step": 2165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.042094058031367e-07,
+      "loss": 1.0542,
+      "step": 2166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.023073877848314e-07,
+      "loss": 1.0269,
+      "step": 2167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.004079581046123e-07,
+      "loss": 1.0161,
+      "step": 2168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.985111193531878e-07,
+      "loss": 1.0229,
+      "step": 2169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9661687411773e-07,
+      "loss": 0.9775,
+      "step": 2170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.947252249818764e-07,
+      "loss": 1.0293,
+      "step": 2171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.928361745257207e-07,
+      "loss": 1.04,
+      "step": 2172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.909497253258153e-07,
+      "loss": 0.9697,
+      "step": 2173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.890658799551619e-07,
+      "loss": 1.0015,
+      "step": 2174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.871846409832119e-07,
+      "loss": 1.0601,
+      "step": 2175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.853060109758608e-07,
+      "loss": 1.0078,
+      "step": 2176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.834299924954482e-07,
+      "loss": 1.0303,
+      "step": 2177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.815565881007481e-07,
+      "loss": 0.9976,
+      "step": 2178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.796858003469727e-07,
+      "loss": 1.0254,
+      "step": 2179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.778176317857618e-07,
+      "loss": 1.0244,
+      "step": 2180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.759520849651862e-07,
+      "loss": 0.9849,
+      "step": 2181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.740891624297381e-07,
+      "loss": 1.0073,
+      "step": 2182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.722288667203315e-07,
+      "loss": 1.0234,
+      "step": 2183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.703712003742965e-07,
+      "loss": 1.0645,
+      "step": 2184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.685161659253791e-07,
+      "loss": 1.0874,
+      "step": 2185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.666637659037338e-07,
+      "loss": 0.9956,
+      "step": 2186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.648140028359214e-07,
+      "loss": 1.0483,
+      "step": 2187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.629668792449086e-07,
+      "loss": 1.0186,
+      "step": 2188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.611223976500591e-07,
+      "loss": 0.998,
+      "step": 2189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.59280560567135e-07,
+      "loss": 1.0479,
+      "step": 2190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.574413705082904e-07,
+      "loss": 0.9741,
+      "step": 2191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.55604829982071e-07,
+      "loss": 1.1064,
+      "step": 2192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.537709414934045e-07,
+      "loss": 1.0767,
+      "step": 2193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.519397075436058e-07,
+      "loss": 1.0308,
+      "step": 2194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.501111306303666e-07,
+      "loss": 1.0386,
+      "step": 2195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.482852132477562e-07,
+      "loss": 1.0103,
+      "step": 2196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.464619578862143e-07,
+      "loss": 1.0752,
+      "step": 2197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.446413670325529e-07,
+      "loss": 0.98,
+      "step": 2198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.428234431699459e-07,
+      "loss": 0.999,
+      "step": 2199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.410081887779334e-07,
+      "loss": 1.0127,
+      "step": 2200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.391956063324122e-07,
+      "loss": 0.9937,
+      "step": 2201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.373856983056347e-07,
+      "loss": 1.0317,
+      "step": 2202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.355784671662059e-07,
+      "loss": 1.0205,
+      "step": 2203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.337739153790813e-07,
+      "loss": 1.0713,
+      "step": 2204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.31972045405559e-07,
+      "loss": 1.0532,
+      "step": 2205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.301728597032821e-07,
+      "loss": 0.9907,
+      "step": 2206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.283763607262305e-07,
+      "loss": 1.0498,
+      "step": 2207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.265825509247199e-07,
+      "loss": 0.9775,
+      "step": 2208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.247914327453996e-07,
+      "loss": 0.9893,
+      "step": 2209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.23003008631246e-07,
+      "loss": 1.0137,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.212172810215607e-07,
+      "loss": 0.9878,
+      "step": 2211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.194342523519699e-07,
+      "loss": 1.0649,
+      "step": 2212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.176539250544163e-07,
+      "loss": 1.0488,
+      "step": 2213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.158763015571581e-07,
+      "loss": 0.9814,
+      "step": 2214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.141013842847672e-07,
+      "loss": 1.0552,
+      "step": 2215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.123291756581231e-07,
+      "loss": 1.0396,
+      "step": 2216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.105596780944122e-07,
+      "loss": 1.0317,
+      "step": 2217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.087928940071207e-07,
+      "loss": 1.0396,
+      "step": 2218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.07028825806038e-07,
+      "loss": 0.9702,
+      "step": 2219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.052674758972431e-07,
+      "loss": 0.981,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.035088466831134e-07,
+      "loss": 0.9531,
+      "step": 2221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.017529405623115e-07,
+      "loss": 1.0352,
+      "step": 2222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.999997599297888e-07,
+      "loss": 1.0737,
+      "step": 2223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.982493071767758e-07,
+      "loss": 0.9834,
+      "step": 2224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.965015846907865e-07,
+      "loss": 1.0239,
+      "step": 2225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.947565948556066e-07,
+      "loss": 1.0044,
+      "step": 2226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.930143400512988e-07,
+      "loss": 1.0078,
+      "step": 2227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.912748226541924e-07,
+      "loss": 1.0547,
+      "step": 2228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.895380450368841e-07,
+      "loss": 1.0571,
+      "step": 2229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.878040095682335e-07,
+      "loss": 1.0903,
+      "step": 2230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.860727186133607e-07,
+      "loss": 1.0127,
+      "step": 2231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.843441745336419e-07,
+      "loss": 1.0435,
+      "step": 2232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.826183796867059e-07,
+      "loss": 1.0547,
+      "step": 2233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.80895336426434e-07,
+      "loss": 1.0034,
+      "step": 2234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.791750471029519e-07,
+      "loss": 1.0024,
+      "step": 2235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774575140626317e-07,
+      "loss": 1.0684,
+      "step": 2236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.757427396480838e-07,
+      "loss": 1.0527,
+      "step": 2237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7403072619815696e-07,
+      "loss": 1.04,
+      "step": 2238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.723214760479333e-07,
+      "loss": 1.0562,
+      "step": 2239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7061499152872866e-07,
+      "loss": 1.0356,
+      "step": 2240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6891127496808295e-07,
+      "loss": 1.0186,
+      "step": 2241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6721032868976417e-07,
+      "loss": 1.0117,
+      "step": 2242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6551215501375896e-07,
+      "loss": 1.083,
+      "step": 2243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.638167562562751e-07,
+      "loss": 1.022,
+      "step": 2244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6212413472973257e-07,
+      "loss": 1.0049,
+      "step": 2245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6043429274276685e-07,
+      "loss": 0.9932,
+      "step": 2246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5874723260021794e-07,
+      "loss": 1.041,
+      "step": 2247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.570629566031354e-07,
+      "loss": 1.0034,
+      "step": 2248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.553814670487694e-07,
+      "loss": 1.0229,
+      "step": 2249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.537027662305707e-07,
+      "loss": 0.9512,
+      "step": 2250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5202685643818495e-07,
+      "loss": 1.0342,
+      "step": 2251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5035373995745287e-07,
+      "loss": 1.0464,
+      "step": 2252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.48683419070404e-07,
+      "loss": 1.0439,
+      "step": 2253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4701589605525427e-07,
+      "loss": 1.0693,
+      "step": 2254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4535117318640545e-07,
+      "loss": 1.0518,
+      "step": 2255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4368925273443856e-07,
+      "loss": 1.0317,
+      "step": 2256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4203013696611203e-07,
+      "loss": 1.0293,
+      "step": 2257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.403738281443609e-07,
+      "loss": 1.0068,
+      "step": 2258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3872032852828955e-07,
+      "loss": 1.0376,
+      "step": 2259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3706964037317085e-07,
+      "loss": 1.0215,
+      "step": 2260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.354217659304452e-07,
+      "loss": 1.0762,
+      "step": 2261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3377670744771253e-07,
+      "loss": 1.0068,
+      "step": 2262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.321344671687344e-07,
+      "loss": 0.9595,
+      "step": 2263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.304950473334268e-07,
+      "loss": 0.9985,
+      "step": 2264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.288584501778592e-07,
+      "loss": 1.0142,
+      "step": 2265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2722467793425093e-07,
+      "loss": 1.0469,
+      "step": 2266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.255937328309695e-07,
+      "loss": 1.0059,
+      "step": 2267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2396561709252436e-07,
+      "loss": 0.9692,
+      "step": 2268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2234033293956865e-07,
+      "loss": 1.0625,
+      "step": 2269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2071788258889025e-07,
+      "loss": 1.019,
+      "step": 2270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.190982682534145e-07,
+      "loss": 1.021,
+      "step": 2271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.174814921421963e-07,
+      "loss": 0.9858,
+      "step": 2272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.158675564604223e-07,
+      "loss": 1.0645,
+      "step": 2273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.142564634094021e-07,
+      "loss": 1.0137,
+      "step": 2274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.126482151865696e-07,
+      "loss": 0.9854,
+      "step": 2275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1104281398547746e-07,
+      "loss": 1.0342,
+      "step": 2276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.094402619957974e-07,
+      "loss": 0.9956,
+      "step": 2277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.078405614033126e-07,
+      "loss": 1.0142,
+      "step": 2278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.062437143899176e-07,
+      "loss": 0.9961,
+      "step": 2279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.046497231336166e-07,
+      "loss": 1.0923,
+      "step": 2280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0305858980851595e-07,
+      "loss": 1.0625,
+      "step": 2281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.014703165848266e-07,
+      "loss": 1.0444,
+      "step": 2282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9988490562885675e-07,
+      "loss": 0.9663,
+      "step": 2283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.983023591030113e-07,
+      "loss": 0.9785,
+      "step": 2284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9672267916578743e-07,
+      "loss": 1.0137,
+      "step": 2285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.951458679717743e-07,
+      "loss": 1.0454,
+      "step": 2286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.935719276716457e-07,
+      "loss": 0.999,
+      "step": 2287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.920008604121628e-07,
+      "loss": 1.0127,
+      "step": 2288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.904326683361648e-07,
+      "loss": 1.0488,
+      "step": 2289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.888673535825727e-07,
+      "loss": 1.0327,
+      "step": 2290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8730491828637944e-07,
+      "loss": 1.0322,
+      "step": 2291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8574536457865436e-07,
+      "loss": 1.0415,
+      "step": 2292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.841886945865325e-07,
+      "loss": 1.0576,
+      "step": 2293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8263491043321887e-07,
+      "loss": 1.0234,
+      "step": 2294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.810840142379807e-07,
+      "loss": 1.0195,
+      "step": 2295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7953600811614727e-07,
+      "loss": 1.0356,
+      "step": 2296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7799089417910467e-07,
+      "loss": 1.0522,
+      "step": 2297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7644867453429575e-07,
+      "loss": 1.0293,
+      "step": 2298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.749093512852148e-07,
+      "loss": 1.0166,
+      "step": 2299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7337292653140485e-07,
+      "loss": 1.0107,
+      "step": 2300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7183940236845767e-07,
+      "loss": 1.0073,
+      "step": 2301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.703087808880071e-07,
+      "loss": 0.9937,
+      "step": 2302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6878106417772757e-07,
+      "loss": 1.0513,
+      "step": 2303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6725625432133374e-07,
+      "loss": 0.9731,
+      "step": 2304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6573435339857384e-07,
+      "loss": 0.9956,
+      "step": 2305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6421536348522746e-07,
+      "loss": 1.0107,
+      "step": 2306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6269928665310707e-07,
+      "loss": 1.0029,
+      "step": 2307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611861249700482e-07,
+      "loss": 1.0669,
+      "step": 2308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5967588049991317e-07,
+      "loss": 1.0469,
+      "step": 2309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5816855530258376e-07,
+      "loss": 1.0176,
+      "step": 2310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5666415143396054e-07,
+      "loss": 1.0029,
+      "step": 2311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.551626709459588e-07,
+      "loss": 1.0469,
+      "step": 2312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5366411588650866e-07,
+      "loss": 1.0259,
+      "step": 2313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5216848829954714e-07,
+      "loss": 0.9956,
+      "step": 2314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.50675790225021e-07,
+      "loss": 1.0425,
+      "step": 2315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491860236988798e-07,
+      "loss": 1.0049,
+      "step": 2316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.476991907530755e-07,
+      "loss": 0.9854,
+      "step": 2317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4621529341555745e-07,
+      "loss": 1.0923,
+      "step": 2318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4473433371027406e-07,
+      "loss": 1.021,
+      "step": 2319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.432563136571621e-07,
+      "loss": 1.0493,
+      "step": 2320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.417812352721536e-07,
+      "loss": 0.98,
+      "step": 2321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.403091005671655e-07,
+      "loss": 0.9312,
+      "step": 2322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.388399115501012e-07,
+      "loss": 1.0015,
+      "step": 2323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.373736702248451e-07,
+      "loss": 1.0752,
+      "step": 2324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3591037859126266e-07,
+      "loss": 0.9497,
+      "step": 2325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3445003864519486e-07,
+      "loss": 1.062,
+      "step": 2326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.329926523784563e-07,
+      "loss": 1.019,
+      "step": 2327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3153822177883543e-07,
+      "loss": 1.0464,
+      "step": 2328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3008674883008686e-07,
+      "loss": 1.022,
+      "step": 2329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.286382355119319e-07,
+      "loss": 1.0703,
+      "step": 2330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2719268380005496e-07,
+      "loss": 1.0474,
+      "step": 2331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2575009566610193e-07,
+      "loss": 1.0332,
+      "step": 2332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.243104730776753e-07,
+      "loss": 1.0713,
+      "step": 2333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2287381799833427e-07,
+      "loss": 1.0474,
+      "step": 2334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.214401323875882e-07,
+      "loss": 1.0483,
+      "step": 2335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2000941820089893e-07,
+      "loss": 1.0532,
+      "step": 2336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1858167738967383e-07,
+      "loss": 1.0405,
+      "step": 2337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.171569119012649e-07,
+      "loss": 1.0117,
+      "step": 2338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1573512367896545e-07,
+      "loss": 1.0142,
+      "step": 2339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.143163146620104e-07,
+      "loss": 1.0063,
+      "step": 2340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1290048678556786e-07,
+      "loss": 1.0674,
+      "step": 2341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1148764198074304e-07,
+      "loss": 1.0239,
+      "step": 2342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1007778217456956e-07,
+      "loss": 1.0122,
+      "step": 2343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.08670909290012e-07,
+      "loss": 1.0352,
+      "step": 2344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0726702524596003e-07,
+      "loss": 0.9897,
+      "step": 2345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.058661319572259e-07,
+      "loss": 0.9932,
+      "step": 2346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0446823133454346e-07,
+      "loss": 0.9722,
+      "step": 2347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0307332528456577e-07,
+      "loss": 1.0273,
+      "step": 2348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.016814157098588e-07,
+      "loss": 1.0522,
+      "step": 2349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.00292504508905e-07,
+      "loss": 1.0474,
+      "step": 2350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.989065935760943e-07,
+      "loss": 1.001,
+      "step": 2351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.975236848017249e-07,
+      "loss": 1.0259,
+      "step": 2352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.961437800720021e-07,
+      "loss": 1.0107,
+      "step": 2353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.947668812690316e-07,
+      "loss": 1.0146,
+      "step": 2354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.933929902708213e-07,
+      "loss": 1.0132,
+      "step": 2355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9202210895127424e-07,
+      "loss": 0.979,
+      "step": 2356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.906542391801906e-07,
+      "loss": 1.0161,
+      "step": 2357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8928938282326123e-07,
+      "loss": 1.0283,
+      "step": 2358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8792754174206903e-07,
+      "loss": 1.0537,
+      "step": 2359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865687177940818e-07,
+      "loss": 0.9868,
+      "step": 2360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8521291283265417e-07,
+      "loss": 1.0591,
+      "step": 2361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.838601287070214e-07,
+      "loss": 1.0723,
+      "step": 2362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.825103672623003e-07,
+      "loss": 1.0576,
+      "step": 2363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.811636303394835e-07,
+      "loss": 1.0005,
+      "step": 2364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7981991977543865e-07,
+      "loss": 1.0786,
+      "step": 2365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.784792374029055e-07,
+      "loss": 1.0723,
+      "step": 2366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7714158505049437e-07,
+      "loss": 0.9805,
+      "step": 2367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.758069645426817e-07,
+      "loss": 0.9829,
+      "step": 2368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.744753776998102e-07,
+      "loss": 0.9932,
+      "step": 2369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.731468263380827e-07,
+      "loss": 1.0312,
+      "step": 2370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7182131226956427e-07,
+      "loss": 1.042,
+      "step": 2371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7049883730217526e-07,
+      "loss": 1.0874,
+      "step": 2372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.691794032396916e-07,
+      "loss": 1.0752,
+      "step": 2373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.678630118817413e-07,
+      "loss": 0.9507,
+      "step": 2374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6654966502380365e-07,
+      "loss": 1.0298,
+      "step": 2375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6523936445720407e-07,
+      "loss": 1.0508,
+      "step": 2376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6393211196911267e-07,
+      "loss": 0.9697,
+      "step": 2377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.626279093425438e-07,
+      "loss": 0.9995,
+      "step": 2378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6132675835635e-07,
+      "loss": 1.0083,
+      "step": 2379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6002866078522425e-07,
+      "loss": 1.0146,
+      "step": 2380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.587336183996914e-07,
+      "loss": 1.0605,
+      "step": 2381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5744163296611307e-07,
+      "loss": 1.0439,
+      "step": 2382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5615270624667706e-07,
+      "loss": 1.0088,
+      "step": 2383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5486683999940335e-07,
+      "loss": 1.0249,
+      "step": 2384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5358403597813443e-07,
+      "loss": 1.02,
+      "step": 2385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5230429593253893e-07,
+      "loss": 1.0566,
+      "step": 2386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.510276216081037e-07,
+      "loss": 1.0205,
+      "step": 2387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.497540147461361e-07,
+      "loss": 1.0513,
+      "step": 2388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.484834770837585e-07,
+      "loss": 0.9673,
+      "step": 2389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.472160103539084e-07,
+      "loss": 0.9937,
+      "step": 2390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4595161628533315e-07,
+      "loss": 1.0112,
+      "step": 2391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.446902966025902e-07,
+      "loss": 0.9502,
+      "step": 2392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4343205302604254e-07,
+      "loss": 1.085,
+      "step": 2393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.421768872718594e-07,
+      "loss": 1.0283,
+      "step": 2394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4092480105201043e-07,
+      "loss": 1.0137,
+      "step": 2395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.396757960742663e-07,
+      "loss": 1.0186,
+      "step": 2396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.384298740421939e-07,
+      "loss": 1.0522,
+      "step": 2397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3718703665515515e-07,
+      "loss": 1.0171,
+      "step": 2398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3594728560830615e-07,
+      "loss": 1.0117,
+      "step": 2399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3471062259259187e-07,
+      "loss": 1.0171,
+      "step": 2400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3347704929474606e-07,
+      "loss": 0.979,
+      "step": 2401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3224656739728761e-07,
+      "loss": 0.9883,
+      "step": 2402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.310191785785207e-07,
+      "loss": 1.0498,
+      "step": 2403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2979488451252834e-07,
+      "loss": 0.9873,
+      "step": 2404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.285736868691746e-07,
+      "loss": 0.9927,
+      "step": 2405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.273555873140984e-07,
+      "loss": 1.0112,
+      "step": 2406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.26140587508715e-07,
+      "loss": 1.0156,
+      "step": 2407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.249286891102098e-07,
+      "loss": 0.9893,
+      "step": 2408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2371989377154013e-07,
+      "loss": 0.9448,
+      "step": 2409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2251420314142845e-07,
+      "loss": 1.0483,
+      "step": 2410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2131161886436465e-07,
+      "loss": 1.0332,
+      "step": 2411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2011214258060076e-07,
+      "loss": 0.9751,
+      "step": 2412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1891577592615065e-07,
+      "loss": 1.0034,
+      "step": 2413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1772252053278513e-07,
+      "loss": 1.0811,
+      "step": 2414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1653237802803345e-07,
+      "loss": 1.0391,
+      "step": 2415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1534535003517736e-07,
+      "loss": 0.9873,
+      "step": 2416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1416143817325207e-07,
+      "loss": 1.0005,
+      "step": 2417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1298064405704145e-07,
+      "loss": 1.0801,
+      "step": 2418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1180296929707717e-07,
+      "loss": 1.0039,
+      "step": 2419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.106284154996363e-07,
+      "loss": 1.0288,
+      "step": 2420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0945698426673988e-07,
+      "loss": 1.0098,
+      "step": 2421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0828867719614926e-07,
+      "loss": 1.0049,
+      "step": 2422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0712349588136392e-07,
+      "loss": 0.9868,
+      "step": 2423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0596144191162182e-07,
+      "loss": 1.0093,
+      "step": 2424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.048025168718934e-07,
+      "loss": 1.0298,
+      "step": 2425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.036467223428834e-07,
+      "loss": 1.0811,
+      "step": 2426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0249405990102554e-07,
+      "loss": 0.9634,
+      "step": 2427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0134453111848112e-07,
+      "loss": 1.0093,
+      "step": 2428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0019813756313815e-07,
+      "loss": 1.0107,
+      "step": 2429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.990548807986084e-07,
+      "loss": 0.9697,
+      "step": 2430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.979147623842248e-07,
+      "loss": 1.0566,
+      "step": 2431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9677778387504064e-07,
+      "loss": 1.0483,
+      "step": 2432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9564394682182518e-07,
+      "loss": 1.0356,
+      "step": 2433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9451325277106415e-07,
+      "loss": 0.979,
+      "step": 2434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9338570326495555e-07,
+      "loss": 1.0063,
+      "step": 2435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9226129984140945e-07,
+      "loss": 1.0327,
+      "step": 2436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.911400440340433e-07,
+      "loss": 1.084,
+      "step": 2437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9002193737218288e-07,
+      "loss": 1.0742,
+      "step": 2438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889069813808575e-07,
+      "loss": 1.0225,
+      "step": 2439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8779517758080096e-07,
+      "loss": 1.0371,
+      "step": 2440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8668652748844524e-07,
+      "loss": 1.0156,
+      "step": 2441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8558103261592298e-07,
+      "loss": 0.9971,
+      "step": 2442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8447869447106194e-07,
+      "loss": 0.9946,
+      "step": 2443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8337951455738469e-07,
+      "loss": 1.061,
+      "step": 2444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.822834943741067e-07,
+      "loss": 1.002,
+      "step": 2445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8119063541613302e-07,
+      "loss": 1.0156,
+      "step": 2446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8010093917405714e-07,
+      "loss": 0.9839,
+      "step": 2447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7901440713415873e-07,
+      "loss": 1.061,
+      "step": 2448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.779310407784024e-07,
+      "loss": 1.0303,
+      "step": 2449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.768508415844339e-07,
+      "loss": 1.022,
+      "step": 2450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.757738110255802e-07,
+      "loss": 1.0259,
+      "step": 2451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.746999505708452e-07,
+      "loss": 1.0674,
+      "step": 2452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7362926168491057e-07,
+      "loss": 1.0098,
+      "step": 2453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.725617458281309e-07,
+      "loss": 1.02,
+      "step": 2454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7149740445653345e-07,
+      "loss": 1.0654,
+      "step": 2455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7043623902181478e-07,
+      "loss": 1.0288,
+      "step": 2456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6937825097134126e-07,
+      "loss": 1.0459,
+      "step": 2457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6832344174814413e-07,
+      "loss": 0.998,
+      "step": 2458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6727181279092037e-07,
+      "loss": 1.0039,
+      "step": 2459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.662233655340273e-07,
+      "loss": 1.0693,
+      "step": 2460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517810140748436e-07,
+      "loss": 1.0024,
+      "step": 2461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6413602183696808e-07,
+      "loss": 1.0796,
+      "step": 2462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6309712824381318e-07,
+      "loss": 1.0527,
+      "step": 2463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.620614220450062e-07,
+      "loss": 1.042,
+      "step": 2464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6102890465318904e-07,
+      "loss": 1.0562,
+      "step": 2465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5999957747665191e-07,
+      "loss": 0.9609,
+      "step": 2466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5897344191933617e-07,
+      "loss": 1.0127,
+      "step": 2467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5795049938082841e-07,
+      "loss": 0.9526,
+      "step": 2468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5693075125635949e-07,
+      "loss": 1.0396,
+      "step": 2469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5591419893680542e-07,
+      "loss": 1.062,
+      "step": 2470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.549008438086813e-07,
+      "loss": 0.9961,
+      "step": 2471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5389068725414346e-07,
+      "loss": 1.0503,
+      "step": 2472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5288373065098284e-07,
+      "loss": 1.0786,
+      "step": 2473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5187997537262882e-07,
+      "loss": 1.0371,
+      "step": 2474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5087942278814188e-07,
+      "loss": 0.9902,
+      "step": 2475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4988207426221617e-07,
+      "loss": 0.9819,
+      "step": 2476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4888793115517412e-07,
+      "loss": 0.9756,
+      "step": 2477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.478969948229675e-07,
+      "loss": 1.0034,
+      "step": 2478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4690926661717313e-07,
+      "loss": 1.0073,
+      "step": 2479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4592474788499317e-07,
+      "loss": 1.0615,
+      "step": 2480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.449434399692512e-07,
+      "loss": 0.979,
+      "step": 2481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4396534420839214e-07,
+      "loss": 1.0068,
+      "step": 2482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4299046193647914e-07,
+      "loss": 1.0342,
+      "step": 2483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4201879448319356e-07,
+      "loss": 1.0298,
+      "step": 2484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4105034317383e-07,
+      "loss": 1.0645,
+      "step": 2485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4008510932929848e-07,
+      "loss": 0.9995,
+      "step": 2486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3912309426611924e-07,
+      "loss": 1.0591,
+      "step": 2487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.38164299296423e-07,
+      "loss": 0.9956,
+      "step": 2488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.372087257279478e-07,
+      "loss": 0.9932,
+      "step": 2489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.362563748640386e-07,
+      "loss": 1.0444,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.353072480036438e-07,
+      "loss": 1.0303,
+      "step": 2491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3436134644131627e-07,
+      "loss": 1.0664,
+      "step": 2492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3341867146720755e-07,
+      "loss": 1.0273,
+      "step": 2493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3247922436706972e-07,
+      "loss": 1.0142,
+      "step": 2494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3154300642225198e-07,
+      "loss": 1.0796,
+      "step": 2495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.306100189096987e-07,
+      "loss": 1.0122,
+      "step": 2496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2968026310194892e-07,
+      "loss": 0.9663,
+      "step": 2497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2875374026713294e-07,
+      "loss": 1.0693,
+      "step": 2498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2783045166897296e-07,
+      "loss": 1.0288,
+      "step": 2499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2691039856677744e-07,
+      "loss": 1.0083,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.259935822154443e-07,
+      "loss": 1.0503,
+      "step": 2501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2508000386545482e-07,
+      "loss": 0.9917,
+      "step": 2502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2416966476287538e-07,
+      "loss": 1.0308,
+      "step": 2503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2326256614935306e-07,
+      "loss": 1.0386,
+      "step": 2504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.223587092621162e-07,
+      "loss": 1.0693,
+      "step": 2505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2145809533397e-07,
+      "loss": 1.0361,
+      "step": 2506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2056072559329861e-07,
+      "loss": 1.019,
+      "step": 2507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1966660126405934e-07,
+      "loss": 1.0361,
+      "step": 2508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1877572356578409e-07,
+      "loss": 1.0371,
+      "step": 2509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1788809371357568e-07,
+      "loss": 0.9746,
+      "step": 2510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1700371291810842e-07,
+      "loss": 1.0137,
+      "step": 2511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.161225823856238e-07,
+      "loss": 1.0518,
+      "step": 2512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1524470331793075e-07,
+      "loss": 1.0591,
+      "step": 2513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.143700769124037e-07,
+      "loss": 0.9868,
+      "step": 2514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1349870436197924e-07,
+      "loss": 1.042,
+      "step": 2515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1263058685515776e-07,
+      "loss": 1.0444,
+      "step": 2516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.117657255759988e-07,
+      "loss": 1.0273,
+      "step": 2517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1090412170412068e-07,
+      "loss": 1.0342,
+      "step": 2518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.100457764146995e-07,
+      "loss": 1.0806,
+      "step": 2519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0919069087846624e-07,
+      "loss": 1.0332,
+      "step": 2520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0833886626170547e-07,
+      "loss": 1.0171,
+      "step": 2521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0749030372625535e-07,
+      "loss": 1.062,
+      "step": 2522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0664500442950281e-07,
+      "loss": 1.0229,
+      "step": 2523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0580296952438618e-07,
+      "loss": 1.0083,
+      "step": 2524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0496420015938924e-07,
+      "loss": 0.9702,
+      "step": 2525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0412869747854409e-07,
+      "loss": 0.9692,
+      "step": 2526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.032964626214239e-07,
+      "loss": 1.0415,
+      "step": 2527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0246749672314844e-07,
+      "loss": 1.0146,
+      "step": 2528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0164180091437631e-07,
+      "loss": 1.0605,
+      "step": 2529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0081937632130695e-07,
+      "loss": 0.9883,
+      "step": 2530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0000022406567777e-07,
+      "loss": 0.9946,
+      "step": 2531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.918434526476311e-08,
+      "loss": 1.0737,
+      "step": 2532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.837174103137199e-08,
+      "loss": 1.085,
+      "step": 2533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.756241247384807e-08,
+      "loss": 1.0244,
+      "step": 2534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.675636069606642e-08,
+      "loss": 0.9917,
+      "step": 2535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.595358679743261e-08,
+      "loss": 1.0112,
+      "step": 2536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.515409187288188e-08,
+      "loss": 0.9985,
+      "step": 2537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.435787701287724e-08,
+      "loss": 1.0415,
+      "step": 2538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.356494330340749e-08,
+      "loss": 1.0122,
+      "step": 2539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.277529182598638e-08,
+      "loss": 1.0435,
+      "step": 2540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.198892365765072e-08,
+      "loss": 0.9966,
+      "step": 2541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.120583987095921e-08,
+      "loss": 1.0322,
+      "step": 2542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.04260415339911e-08,
+      "loss": 1.0215,
+      "step": 2543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.964952971034418e-08,
+      "loss": 1.0488,
+      "step": 2544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.887630545913323e-08,
+      "loss": 0.98,
+      "step": 2545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.81063698349896e-08,
+      "loss": 1.0576,
+      "step": 2546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.733972388805911e-08,
+      "loss": 0.9971,
+      "step": 2547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.657636866400032e-08,
+      "loss": 1.0005,
+      "step": 2548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.581630520398398e-08,
+      "loss": 1.0708,
+      "step": 2549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.505953454469057e-08,
+      "loss": 0.9536,
+      "step": 2550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.430605771830941e-08,
+      "loss": 1.0186,
+      "step": 2551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.355587575253732e-08,
+      "loss": 0.9834,
+      "step": 2552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.280898967057805e-08,
+      "loss": 0.9927,
+      "step": 2553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.206540049113781e-08,
+      "loss": 1.0493,
+      "step": 2554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.132510922842812e-08,
+      "loss": 0.9897,
+      "step": 2555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.0588116892161e-08,
+      "loss": 1.0137,
+      "step": 2556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.985442448755015e-08,
+      "loss": 1.02,
+      "step": 2557
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.912403301530703e-08,
+      "loss": 1.0171,
+      "step": 2558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.839694347164223e-08,
+      "loss": 1.0571,
+      "step": 2559
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.767315684826138e-08,
+      "loss": 1.0679,
+      "step": 2560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.695267413236562e-08,
+      "loss": 1.0547,
+      "step": 2561
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.623549630665056e-08,
+      "loss": 1.0122,
+      "step": 2562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.552162434930288e-08,
+      "loss": 0.9458,
+      "step": 2563
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.481105923400039e-08,
+      "loss": 1.0293,
+      "step": 2564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.410380192991202e-08,
+      "loss": 0.9995,
+      "step": 2565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.339985340169359e-08,
+      "loss": 1.0171,
+      "step": 2566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.269921460948764e-08,
+      "loss": 1.0171,
+      "step": 2567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.200188650892448e-08,
+      "loss": 1.0117,
+      "step": 2568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.130787005111605e-08,
+      "loss": 1.0059,
+      "step": 2569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.061716618266018e-08,
+      "loss": 1.0488,
+      "step": 2570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.992977584563465e-08,
+      "loss": 0.9976,
+      "step": 2571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.92456999775984e-08,
+      "loss": 0.9805,
+      "step": 2572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.856493951158949e-08,
+      "loss": 1.1016,
+      "step": 2573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.788749537612411e-08,
+      "loss": 1.0166,
+      "step": 2574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.721336849519505e-08,
+      "loss": 1.0098,
+      "step": 2575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.654255978827101e-08,
+      "loss": 1.0347,
+      "step": 2576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.587507017029427e-08,
+      "loss": 0.9604,
+      "step": 2577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.521090055168044e-08,
+      "loss": 0.9731,
+      "step": 2578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.455005183831659e-08,
+      "loss": 1.0361,
+      "step": 2579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.389252493156084e-08,
+      "loss": 1.0073,
+      "step": 2580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.323832072823971e-08,
+      "loss": 1.0244,
+      "step": 2581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.258744012064833e-08,
+      "loss": 0.9639,
+      "step": 2582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.193988399654849e-08,
+      "loss": 1.0171,
+      "step": 2583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.129565323916814e-08,
+      "loss": 0.9917,
+      "step": 2584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.065474872719856e-08,
+      "loss": 1.103,
+      "step": 2585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.001717133479496e-08,
+      "loss": 1.0117,
+      "step": 2586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.938292193157419e-08,
+      "loss": 1.021,
+      "step": 2587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.875200138261428e-08,
+      "loss": 1.0669,
+      "step": 2588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.812441054845325e-08,
+      "loss": 1.0122,
+      "step": 2589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7500150285086376e-08,
+      "loss": 1.0249,
+      "step": 2590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6879221443967016e-08,
+      "loss": 0.979,
+      "step": 2591
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.626162487200465e-08,
+      "loss": 0.9678,
+      "step": 2592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.564736141156407e-08,
+      "loss": 0.9839,
+      "step": 2593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.503643190046315e-08,
+      "loss": 1.0205,
+      "step": 2594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4428837171973114e-08,
+      "loss": 1.042,
+      "step": 2595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.382457805481606e-08,
+      "loss": 0.9409,
+      "step": 2596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.322365537316549e-08,
+      "loss": 1.0264,
+      "step": 2597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2626069946643264e-08,
+      "loss": 1.0366,
+      "step": 2598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2031822590319636e-08,
+      "loss": 1.0552,
+      "step": 2599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.144091411471236e-08,
+      "loss": 1.0391,
+      "step": 2600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0853345325785064e-08,
+      "loss": 1.0376,
+      "step": 2601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.026911702494558e-08,
+      "loss": 1.0283,
+      "step": 2602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.968823000904649e-08,
+      "loss": 0.9492,
+      "step": 2603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.911068507038236e-08,
+      "loss": 1.0137,
+      "step": 2604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8536482996690004e-08,
+      "loss": 1.0508,
+      "step": 2605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.796562457114573e-08,
+      "loss": 1.0205,
+      "step": 2606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.739811057236615e-08,
+      "loss": 1.0347,
+      "step": 2607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6833941774406535e-08,
+      "loss": 1.0107,
+      "step": 2608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.627311894675857e-08,
+      "loss": 1.0088,
+      "step": 2609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5715642854350374e-08,
+      "loss": 0.9668,
+      "step": 2610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5161514257546504e-08,
+      "loss": 1.0352,
+      "step": 2611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.46107339121446e-08,
+      "loss": 1.0474,
+      "step": 2612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.406330256937541e-08,
+      "loss": 1.0366,
+      "step": 2613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3519220975902775e-08,
+      "loss": 1.0381,
+      "step": 2614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.297848987382031e-08,
+      "loss": 1.0034,
+      "step": 2615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2441110000653596e-08,
+      "loss": 1.0098,
+      "step": 2616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.190708208935579e-08,
+      "loss": 1.0288,
+      "step": 2617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1376406868308684e-08,
+      "loss": 1.0146,
+      "step": 2618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.084908506132107e-08,
+      "loss": 0.9351,
+      "step": 2619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0325117387628455e-08,
+      "loss": 0.9795,
+      "step": 2620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9804504561890554e-08,
+      "loss": 1.0723,
+      "step": 2621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.928724729419242e-08,
+      "loss": 1.0889,
+      "step": 2622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.877334629004109e-08,
+      "loss": 1.0122,
+      "step": 2623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.826280225036727e-08,
+      "loss": 1.0381,
+      "step": 2624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7755615871521434e-08,
+      "loss": 0.98,
+      "step": 2625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.725178784527578e-08,
+      "loss": 1.0737,
+      "step": 2626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6751318858820885e-08,
+      "loss": 1.0532,
+      "step": 2627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.625420959476628e-08,
+      "loss": 1.0146,
+      "step": 2628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.576046073113903e-08,
+      "loss": 1.0215,
+      "step": 2629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5270072941382684e-08,
+      "loss": 1.1143,
+      "step": 2630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4783046894356906e-08,
+      "loss": 1.0278,
+      "step": 2631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.429938325433507e-08,
+      "loss": 1.0132,
+      "step": 2632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3819082681006145e-08,
+      "loss": 1.0474,
+      "step": 2633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.334214582946998e-08,
+      "loss": 1.0225,
+      "step": 2634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2868573350240687e-08,
+      "loss": 1.0254,
+      "step": 2635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.239836588924211e-08,
+      "loss": 1.0649,
+      "step": 2636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1931524087808476e-08,
+      "loss": 1.0508,
+      "step": 2637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.146804858268404e-08,
+      "loss": 0.9888,
+      "step": 2638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.100794000602175e-08,
+      "loss": 1.0625,
+      "step": 2639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0551198985381284e-08,
+      "loss": 1.0308,
+      "step": 2640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0097826143730414e-08,
+      "loss": 1.0068,
+      "step": 2641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9647822099442004e-08,
+      "loss": 1.043,
+      "step": 2642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9201187466294246e-08,
+      "loss": 1.0122,
+      "step": 2643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8757922853470123e-08,
+      "loss": 1.0146,
+      "step": 2644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8318028865555736e-08,
+      "loss": 0.9844,
+      "step": 2645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.788150610253948e-08,
+      "loss": 1.0308,
+      "step": 2646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7448355159812313e-08,
+      "loss": 1.0068,
+      "step": 2647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7018576628166095e-08,
+      "loss": 1.0127,
+      "step": 2648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.659217109379275e-08,
+      "loss": 0.9707,
+      "step": 2649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.616913913828373e-08,
+      "loss": 1.0376,
+      "step": 2650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.574948133862887e-08,
+      "loss": 1.0474,
+      "step": 2651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5333198267215862e-08,
+      "loss": 0.9883,
+      "step": 2652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4920290491830257e-08,
+      "loss": 1.0703,
+      "step": 2653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4510758575652937e-08,
+      "loss": 1.0127,
+      "step": 2654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4104603077260703e-08,
+      "loss": 1.0249,
+      "step": 2655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3701824550624864e-08,
+      "loss": 1.0234,
+      "step": 2656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3302423545111807e-08,
+      "loss": 1.0039,
+      "step": 2657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2906400605479663e-08,
+      "loss": 1.0308,
+      "step": 2658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.251375627187996e-08,
+      "loss": 1.0884,
+      "step": 2659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.212449107985598e-08,
+      "loss": 1.0508,
+      "step": 2660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.173860556034163e-08,
+      "loss": 0.9922,
+      "step": 2661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1356100239662002e-08,
+      "loss": 1.0791,
+      "step": 2662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0976975639530606e-08,
+      "loss": 0.9834,
+      "step": 2663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.060123227705102e-08,
+      "loss": 0.9858,
+      "step": 2664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0228870664714128e-08,
+      "loss": 1.0151,
+      "step": 2665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9859891310398948e-08,
+      "loss": 1.04,
+      "step": 2666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9494294717370964e-08,
+      "loss": 1.0615,
+      "step": 2667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9132081384281575e-08,
+      "loss": 1.0225,
+      "step": 2668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8773251805168092e-08,
+      "loss": 1.0356,
+      "step": 2669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8417806469452626e-08,
+      "loss": 1.0586,
+      "step": 2670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.806574586194071e-08,
+      "loss": 1.0811,
+      "step": 2671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7717070462822116e-08,
+      "loss": 0.9658,
+      "step": 2672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7371780747668655e-08,
+      "loss": 1.0117,
+      "step": 2673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7029877187434986e-08,
+      "loss": 1.1084,
+      "step": 2674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6691360248456412e-08,
+      "loss": 1.0889,
+      "step": 2675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6356230392450268e-08,
+      "loss": 0.9692,
+      "step": 2676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6024488076512855e-08,
+      "loss": 1.022,
+      "step": 2677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5696133753121124e-08,
+      "loss": 0.9805,
+      "step": 2678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5371167870130433e-08,
+      "loss": 1.0249,
+      "step": 2679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.504959087077429e-08,
+      "loss": 1.0605,
+      "step": 2680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.473140319366434e-08,
+      "loss": 1.0854,
+      "step": 2681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4416605272789819e-08,
+      "loss": 0.998,
+      "step": 2682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4105197537515602e-08,
+      "loss": 1.0264,
+      "step": 2683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3797180412583322e-08,
+      "loss": 0.9487,
+      "step": 2684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.349255431810942e-08,
+      "loss": 1.0137,
+      "step": 2685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.31913196695857e-08,
+      "loss": 0.9824,
+      "step": 2686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.289347687787823e-08,
+      "loss": 1.0454,
+      "step": 2687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.259902634922594e-08,
+      "loss": 1.0156,
+      "step": 2688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2307968485242572e-08,
+      "loss": 1.0571,
+      "step": 2689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2020303682912237e-08,
+      "loss": 0.9727,
+      "step": 2690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1736032334593306e-08,
+      "loss": 1.0508,
+      "step": 2691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1455154828014515e-08,
+      "loss": 1.0239,
+      "step": 2692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1177671546275526e-08,
+      "loss": 0.979,
+      "step": 2693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.090358286784693e-08,
+      "loss": 1.0034,
+      "step": 2694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0632889166569128e-08,
+      "loss": 1.1143,
+      "step": 2695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.036559081165206e-08,
+      "loss": 0.9917,
+      "step": 2696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0101688167674372e-08,
+      "loss": 0.9907,
+      "step": 2697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.841181594583693e-09,
+      "loss": 1.0381,
+      "step": 2698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.584071447694688e-09,
+      "loss": 1.0747,
+      "step": 2699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.330358077690449e-09,
+      "loss": 0.9819,
+      "step": 2700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.080041830620834e-09,
+      "loss": 1.0171,
+      "step": 2701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.833123047901626e-09,
+      "loss": 1.0112,
+      "step": 2702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.589602066315372e-09,
+      "loss": 1.0132,
+      "step": 2703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.349479218009993e-09,
+      "loss": 1.0112,
+      "step": 2704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.112754830498504e-09,
+      "loss": 1.0278,
+      "step": 2705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.879429226658741e-09,
+      "loss": 1.0117,
+      "step": 2706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.649502724732528e-09,
+      "loss": 1.0044,
+      "step": 2707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.4229756383259465e-09,
+      "loss": 1.0103,
+      "step": 2708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.1998482764082386e-09,
+      "loss": 1.0322,
+      "step": 2709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.980120943311519e-09,
+      "loss": 1.063,
+      "step": 2710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.763793938730778e-09,
+      "loss": 1.0371,
+      "step": 2711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.5508675577227735e-09,
+      "loss": 1.0015,
+      "step": 2712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.341342090706304e-09,
+      "loss": 0.9917,
+      "step": 2713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.1352178234613816e-09,
+      "loss": 0.9702,
+      "step": 2714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9324950371292264e-09,
+      "loss": 1.0659,
+      "step": 2715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.733174008211717e-09,
+      "loss": 1.0391,
+      "step": 2716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.537255008569997e-09,
+      "loss": 0.9941,
+      "step": 2717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3447383054261445e-09,
+      "loss": 1.0166,
+      "step": 2718
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.155624161361505e-09,
+      "loss": 0.9854,
+      "step": 2719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.96991283431586e-09,
+      "loss": 0.9702,
+      "step": 2720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.787604577588534e-09,
+      "loss": 1.0581,
+      "step": 2721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.608699639837288e-09,
+      "loss": 0.9575,
+      "step": 2722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.433198265076932e-09,
+      "loss": 0.9707,
+      "step": 2723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.261100692681264e-09,
+      "loss": 1.0518,
+      "step": 2724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.092407157380851e-09,
+      "loss": 1.0269,
+      "step": 2725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9271178892635875e-09,
+      "loss": 0.9761,
+      "step": 2726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.765233113773858e-09,
+      "loss": 1.0869,
+      "step": 2727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6067530517128192e-09,
+      "loss": 0.9951,
+      "step": 2728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4516779192375616e-09,
+      "loss": 1.0156,
+      "step": 2729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3000079278611154e-09,
+      "loss": 0.9619,
+      "step": 2730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.151743284452724e-09,
+      "loss": 1.0068,
+      "step": 2731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0068841912359035e-09,
+      "loss": 1.0352,
+      "step": 2732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.865430845790107e-09,
+      "loss": 1.0337,
+      "step": 2733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7273834410485033e-09,
+      "loss": 0.9585,
+      "step": 2734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5927421653001995e-09,
+      "loss": 1.0376,
+      "step": 2735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4615072021871855e-09,
+      "loss": 1.0566,
+      "step": 2736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.333678730706279e-09,
+      "loss": 1.0405,
+      "step": 2737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2092569252077366e-09,
+      "loss": 1.0356,
+      "step": 2738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0882419553952537e-09,
+      "loss": 1.0327,
+      "step": 2739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9706339863262424e-09,
+      "loss": 1.0479,
+      "step": 2740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8564331784107214e-09,
+      "loss": 1.0444,
+      "step": 2741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7456396874115933e-09,
+      "loss": 0.9517,
+      "step": 2742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6382536644446445e-09,
+      "loss": 1.0547,
+      "step": 2743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.534275255977713e-09,
+      "loss": 1.0366,
+      "step": 2744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.433704603831243e-09,
+      "loss": 1.0293,
+      "step": 2745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3365418451774526e-09,
+      "loss": 1.0122,
+      "step": 2746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2427871125403334e-09,
+      "loss": 1.1206,
+      "step": 2747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1524405337962063e-09,
+      "loss": 1.0068,
+      "step": 2748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.065502232171778e-09,
+      "loss": 1.0703,
+      "step": 2749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.819723262458057e-10,
+      "loss": 1.0166,
+      "step": 2750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.018509299482669e-10,
+      "loss": 1.0249,
+      "step": 2751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.251381525595237e-10,
+      "loss": 1.0063,
+      "step": 2752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.518340987114347e-10,
+      "loss": 1.0864,
+      "step": 2753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.819388683862449e-10,
+      "loss": 0.9976,
+      "step": 2754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.154525569168623e-10,
+      "loss": 1.0342,
+      "step": 2755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.523752549863037e-10,
+      "loss": 1.0415,
+      "step": 2756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.927070486288043e-10,
+      "loss": 1.0205,
+      "step": 2757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.364480192275977e-10,
+      "loss": 1.0024,
+      "step": 2758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.835982435168584e-10,
+      "loss": 1.0308,
+      "step": 2759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3415779358059174e-10,
+      "loss": 0.9834,
+      "step": 2760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8812673685235657e-10,
+      "loss": 0.9233,
+      "step": 2761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4550513611582007e-10,
+      "loss": 0.9932,
+      "step": 2762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0629304950420258e-10,
+      "loss": 1.0405,
+      "step": 2763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7049053050083308e-10,
+      "loss": 1.0146,
+      "step": 2764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.380976279374835e-10,
+      "loss": 1.0205,
+      "step": 2765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0911438599686686e-10,
+      "loss": 1.02,
+      "step": 2766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.35408442095842e-11,
+      "loss": 1.0254,
+      "step": 2767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.137703745717761e-11,
+      "loss": 0.9971,
+      "step": 2768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.262299596907715e-11,
+      "loss": 1.0059,
+      "step": 2769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7278745325098887e-11,
+      "loss": 1.0249,
+      "step": 2770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.534430645377949e-11,
+      "loss": 1.0352,
+      "step": 2771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.819695632931389e-12,
+      "loss": 0.998,
+      "step": 2772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7049244896427675e-12,
+      "loss": 1.0547,
+      "step": 2773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0215,
+      "step": 2774
+    },
+    {
+      "epoch": 1.0,
+      "step": 2774,
+      "total_flos": 4.525541684217381e+18,
+      "train_loss": 1.030239465516853,
+      "train_runtime": 32801.9753,
+      "train_samples_per_second": 10.823,
+      "train_steps_per_second": 0.085
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2774,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 400,
+  "total_flos": 4.525541684217381e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}