diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,33517 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 5582,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1904761904761906e-07,
+      "loss": 1.3882,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3809523809523811e-07,
+      "loss": 1.5342,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 1.5156,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7619047619047623e-07,
+      "loss": 1.5303,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.952380952380953e-07,
+      "loss": 1.5361,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 1.4883,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.5547,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 1.4629,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 1.4521,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1904761904761906e-06,
+      "loss": 1.5361,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3095238095238096e-06,
+      "loss": 1.4531,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.5342,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5476190476190479e-06,
+      "loss": 1.3955,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.3066,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 1.3047,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 1.4004,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.023809523809524e-06,
+      "loss": 1.3721,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 1.3271,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.261904761904762e-06,
+      "loss": 1.29,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.380952380952381e-06,
+      "loss": 1.1851,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.2256,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6190476190476192e-06,
+      "loss": 1.165,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7380952380952387e-06,
+      "loss": 1.2334,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.1797,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9761904761904763e-06,
+      "loss": 1.166,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0952380952380957e-06,
+      "loss": 1.1309,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 1.1504,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1348,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4523809523809528e-06,
+      "loss": 1.1289,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 1.1729,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.690476190476191e-06,
+      "loss": 1.1157,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 1.1387,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 1.0718,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.047619047619048e-06,
+      "loss": 1.0908,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.0933,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.0913,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.404761904761905e-06,
+      "loss": 1.0913,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.523809523809524e-06,
+      "loss": 1.0654,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 1.0469,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 1.0127,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.880952380952381e-06,
+      "loss": 1.0947,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.0308,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.119047619047619e-06,
+      "loss": 1.042,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2380952380952384e-06,
+      "loss": 1.125,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 1.0464,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.476190476190477e-06,
+      "loss": 1.0293,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.595238095238096e-06,
+      "loss": 1.0146,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.0078,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.0312,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 1.0713,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.071428571428571e-06,
+      "loss": 1.083,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1904761904761914e-06,
+      "loss": 0.9883,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.30952380952381e-06,
+      "loss": 1.04,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 1.0132,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.547619047619048e-06,
+      "loss": 0.3086,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0068,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.785714285714287e-06,
+      "loss": 1.0298,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9047619047619055e-06,
+      "loss": 1.0327,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.023809523809524e-06,
+      "loss": 0.9951,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 1.0684,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.261904761904762e-06,
+      "loss": 1.0293,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.380952380952382e-06,
+      "loss": 0.9531,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.0015,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 1.0273,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.738095238095238e-06,
+      "loss": 1.0068,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.9854,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.976190476190477e-06,
+      "loss": 0.9634,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.095238095238097e-06,
+      "loss": 0.9717,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.9512,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.0737,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.452380952380953e-06,
+      "loss": 1.0107,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.3007,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.690476190476192e-06,
+      "loss": 0.9766,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.80952380952381e-06,
+      "loss": 0.9668,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.9478,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.047619047619049e-06,
+      "loss": 1.04,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.0166,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 1.0532,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.404761904761905e-06,
+      "loss": 0.9941,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.9688,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 0.9766,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.761904761904762e-06,
+      "loss": 1.0073,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.880952380952381e-06,
+      "loss": 0.3411,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9673,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.011904761904762e-05,
+      "loss": 1.0161,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0238095238095238e-05,
+      "loss": 0.9692,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0357142857142859e-05,
+      "loss": 0.9727,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 1.0093,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0595238095238096e-05,
+      "loss": 0.2938,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 1.02,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.9722,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0952380952380955e-05,
+      "loss": 0.9839,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1071428571428572e-05,
+      "loss": 0.9912,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1190476190476192e-05,
+      "loss": 0.9644,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.130952380952381e-05,
+      "loss": 1.0,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.9746,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1547619047619047e-05,
+      "loss": 1.0522,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.981,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1785714285714287e-05,
+      "loss": 0.9531,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.9912,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2023809523809525e-05,
+      "loss": 1.0371,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.9834,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2261904761904763e-05,
+      "loss": 1.0024,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 1.0151,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 1.0142,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.261904761904762e-05,
+      "loss": 0.9502,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2738095238095238e-05,
+      "loss": 0.9956,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.958,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2976190476190478e-05,
+      "loss": 1.0171,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.9556,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3214285714285716e-05,
+      "loss": 0.9648,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.9814,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3452380952380954e-05,
+      "loss": 0.936,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.3151,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3690476190476192e-05,
+      "loss": 1.0112,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3809523809523811e-05,
+      "loss": 0.9478,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3928571428571429e-05,
+      "loss": 0.9536,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4047619047619048e-05,
+      "loss": 0.9365,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 0.9497,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.8906,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4404761904761907e-05,
+      "loss": 0.9966,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4523809523809524e-05,
+      "loss": 0.9458,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4642857142857144e-05,
+      "loss": 0.9136,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4761904761904763e-05,
+      "loss": 0.9883,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4880952380952383e-05,
+      "loss": 0.9683,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9873,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.511904761904762e-05,
+      "loss": 1.0215,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 0.9795,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.535714285714286e-05,
+      "loss": 0.9346,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.9829,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5595238095238098e-05,
+      "loss": 0.9727,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.9517,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.9771,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5952380952380954e-05,
+      "loss": 0.9526,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.9468,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.9082,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.630952380952381e-05,
+      "loss": 0.9336,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 1.0283,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6547619047619046e-05,
+      "loss": 0.9746,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.9438,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.678571428571429e-05,
+      "loss": 0.998,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6904761904761906e-05,
+      "loss": 1.0181,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7023809523809524e-05,
+      "loss": 0.9155,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.9771,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7261904761904763e-05,
+      "loss": 0.959,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7380952380952384e-05,
+      "loss": 0.9473,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.978,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.761904761904762e-05,
+      "loss": 0.9199,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7738095238095237e-05,
+      "loss": 0.9634,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.9688,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.797619047619048e-05,
+      "loss": 1.0181,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 0.9385,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8214285714285715e-05,
+      "loss": 1.0054,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.9556,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8452380952380954e-05,
+      "loss": 0.2932,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.9482,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8690476190476193e-05,
+      "loss": 0.2897,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.880952380952381e-05,
+      "loss": 0.9575,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.892857142857143e-05,
+      "loss": 0.9888,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 0.9697,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 0.9062,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.9927,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9404761904761906e-05,
+      "loss": 0.9146,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9523809523809524e-05,
+      "loss": 1.0034,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.9448,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9761904761904763e-05,
+      "loss": 1.0132,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9880952380952384e-05,
+      "loss": 0.9902,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.936,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998316422032e-05,
+      "loss": 0.9883,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999326568868e-05,
+      "loss": 0.9658,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999984847801658e-05,
+      "loss": 0.9321,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999973062763794e-05,
+      "loss": 0.9561,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999957910579057e-05,
+      "loss": 0.9546,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993939125255e-05,
+      "loss": 0.9658,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999917504790506e-05,
+      "loss": 0.9658,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99998922512003e-05,
+      "loss": 0.9668,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999863630490425e-05,
+      "loss": 0.9111,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999831642670527e-05,
+      "loss": 0.9438,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999796287751377e-05,
+      "loss": 0.979,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999757565744876e-05,
+      "loss": 0.9761,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999715476664063e-05,
+      "loss": 0.9517,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999670020523113e-05,
+      "loss": 0.9917,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999621197337323e-05,
+      "loss": 0.9893,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999569007123143e-05,
+      "loss": 0.9824,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999951344989814e-05,
+      "loss": 0.9844,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999454525681025e-05,
+      "loss": 0.9414,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999392234491637e-05,
+      "loss": 0.9473,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999932657635095e-05,
+      "loss": 1.0005,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999925755128107e-05,
+      "loss": 0.9731,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999918515930524e-05,
+      "loss": 0.9243,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999109400447837e-05,
+      "loss": 0.9536,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999030274734372e-05,
+      "loss": 0.3333,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998947782191485e-05,
+      "loss": 0.9023,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998861922846952e-05,
+      "loss": 0.9072,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998772696729687e-05,
+      "loss": 0.3311,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998680103869727e-05,
+      "loss": 0.9722,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998584144298258e-05,
+      "loss": 1.0205,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999848481804758e-05,
+      "loss": 0.9585,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999838212515115e-05,
+      "loss": 0.9702,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998276065643538e-05,
+      "loss": 0.9146,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999816663956046e-05,
+      "loss": 1.0034,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999805384693876e-05,
+      "loss": 1.0068,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997937687816417e-05,
+      "loss": 0.8926,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999781816223254e-05,
+      "loss": 0.9453,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997695270227383e-05,
+      "loss": 0.322,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997569011842322e-05,
+      "loss": 0.9263,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997439387119867e-05,
+      "loss": 0.9561,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997306396103667e-05,
+      "loss": 0.9126,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997170038838504e-05,
+      "loss": 0.9629,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999703031537029e-05,
+      "loss": 0.9849,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999688722574607e-05,
+      "loss": 0.9443,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999674077001403e-05,
+      "loss": 1.0181,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996590948223478e-05,
+      "loss": 0.9795,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996437760424864e-05,
+      "loss": 0.9873,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996281206669766e-05,
+      "loss": 0.9644,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996121287010904e-05,
+      "loss": 0.9346,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995958001502123e-05,
+      "loss": 0.9429,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99957913501984e-05,
+      "loss": 0.8931,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995621333155852e-05,
+      "loss": 0.9023,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995447950431726e-05,
+      "loss": 0.9463,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995271202084408e-05,
+      "loss": 0.957,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995091088173402e-05,
+      "loss": 0.9434,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999490760875936e-05,
+      "loss": 0.9629,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994720763904065e-05,
+      "loss": 0.9536,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994530553670427e-05,
+      "loss": 0.9629,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999433697812249e-05,
+      "loss": 0.9424,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994140037325444e-05,
+      "loss": 0.9761,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993939731345592e-05,
+      "loss": 0.9238,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993736060250385e-05,
+      "loss": 1.0039,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99935290241084e-05,
+      "loss": 0.9321,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993318622989354e-05,
+      "loss": 0.3062,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993104856964084e-05,
+      "loss": 1.042,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992887726104578e-05,
+      "loss": 0.9844,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999266723048394e-05,
+      "loss": 0.9868,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992443370176416e-05,
+      "loss": 0.3098,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992216145257384e-05,
+      "loss": 0.9453,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991985555803355e-05,
+      "loss": 1.0073,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999175160189197e-05,
+      "loss": 0.9478,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991514283602008e-05,
+      "loss": 0.9414,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991273601013375e-05,
+      "loss": 0.9312,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991029554207113e-05,
+      "loss": 0.9604,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990782143265397e-05,
+      "loss": 0.9761,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990531368271534e-05,
+      "loss": 0.9272,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990277229309963e-05,
+      "loss": 0.9106,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990019726466263e-05,
+      "loss": 0.9648,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989758859827127e-05,
+      "loss": 0.979,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99894946294804e-05,
+      "loss": 0.9082,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989227035515057e-05,
+      "loss": 0.9604,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988956078021195e-05,
+      "loss": 0.9292,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988681757090048e-05,
+      "loss": 0.9102,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988404072813985e-05,
+      "loss": 0.9575,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998812302528651e-05,
+      "loss": 0.9443,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987838614602257e-05,
+      "loss": 0.9482,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987550840856988e-05,
+      "loss": 0.9941,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987259704147602e-05,
+      "loss": 0.9985,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998696520457213e-05,
+      "loss": 0.9541,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998666734222973e-05,
+      "loss": 0.9565,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986366117220702e-05,
+      "loss": 0.9697,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986061529646475e-05,
+      "loss": 0.896,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985753579609605e-05,
+      "loss": 0.8999,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998544226721378e-05,
+      "loss": 1.0137,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998512759256383e-05,
+      "loss": 0.9658,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984809555765706e-05,
+      "loss": 0.9434,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984488156926503e-05,
+      "loss": 0.2877,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984163396154433e-05,
+      "loss": 0.9912,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998383527355885e-05,
+      "loss": 0.9043,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998350378925024e-05,
+      "loss": 0.9927,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983168943340222e-05,
+      "loss": 0.9902,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982830735941537e-05,
+      "loss": 0.9839,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998248916716807e-05,
+      "loss": 0.8643,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982144237134832e-05,
+      "loss": 0.9399,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998179594595796e-05,
+      "loss": 0.9932,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981444293754738e-05,
+      "loss": 0.9673,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981089280643565e-05,
+      "loss": 0.9458,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980730906743985e-05,
+      "loss": 0.8921,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980369172176663e-05,
+      "loss": 0.8804,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998000407706341e-05,
+      "loss": 0.978,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979635621527145e-05,
+      "loss": 0.939,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979263805691945e-05,
+      "loss": 1.0034,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978888629683005e-05,
+      "loss": 1.0098,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978510093626647e-05,
+      "loss": 0.9751,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997812819765033e-05,
+      "loss": 0.9746,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977742941882648e-05,
+      "loss": 0.9243,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977354326453322e-05,
+      "loss": 0.9355,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976962351493206e-05,
+      "loss": 0.9561,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997656701713428e-05,
+      "loss": 0.8511,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976168323509665e-05,
+      "loss": 0.9561,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99757662707536e-05,
+      "loss": 0.8896,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975360859001468e-05,
+      "loss": 0.9824,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974952088389775e-05,
+      "loss": 0.894,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974539959056165e-05,
+      "loss": 0.9619,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99741244711394e-05,
+      "loss": 0.938,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973705624779388e-05,
+      "loss": 0.9556,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997328342011716e-05,
+      "loss": 0.8809,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997285785729488e-05,
+      "loss": 0.9907,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972428936455834e-05,
+      "loss": 0.9336,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997199665774446e-05,
+      "loss": 0.9531,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971561021306298e-05,
+      "loss": 0.9805,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997112202728804e-05,
+      "loss": 0.959,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970679675837507e-05,
+      "loss": 0.9448,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997023396710364e-05,
+      "loss": 0.8887,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969784901236516e-05,
+      "loss": 1.0156,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969332478387342e-05,
+      "loss": 0.9546,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996887669870846e-05,
+      "loss": 0.9868,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968417562353336e-05,
+      "loss": 1.0098,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996795506947657e-05,
+      "loss": 0.8857,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967489220233887e-05,
+      "loss": 0.9028,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967020014782146e-05,
+      "loss": 0.9307,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966547453279336e-05,
+      "loss": 0.9614,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966071535884582e-05,
+      "loss": 0.9624,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965592262758123e-05,
+      "loss": 0.9189,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965109634061347e-05,
+      "loss": 0.9316,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964623649956758e-05,
+      "loss": 0.9497,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964134310607992e-05,
+      "loss": 0.9595,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996364161617982e-05,
+      "loss": 0.9307,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963145566838143e-05,
+      "loss": 0.8555,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996264616274998e-05,
+      "loss": 0.9341,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99621434040835e-05,
+      "loss": 0.9351,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961637291007978e-05,
+      "loss": 0.9692,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961127823693837e-05,
+      "loss": 0.978,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996061500231262e-05,
+      "loss": 0.9102,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960098827037003e-05,
+      "loss": 0.936,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995957929804079e-05,
+      "loss": 0.9839,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959056415498913e-05,
+      "loss": 1.0229,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995853017958744e-05,
+      "loss": 0.9292,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995800059048356e-05,
+      "loss": 0.9399,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957467648365587e-05,
+      "loss": 0.9819,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995693135341298e-05,
+      "loss": 0.9312,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956391705806317e-05,
+      "loss": 0.9438,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99558487057273e-05,
+      "loss": 0.9751,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995530235335877e-05,
+      "loss": 0.9214,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954752648884693e-05,
+      "loss": 0.9766,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954199592490165e-05,
+      "loss": 0.8696,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99536431843614e-05,
+      "loss": 0.9619,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953083424685756e-05,
+      "loss": 0.9209,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952520313651713e-05,
+      "loss": 0.939,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951953851448875e-05,
+      "loss": 0.9648,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951384038267983e-05,
+      "loss": 0.9414,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950810874300902e-05,
+      "loss": 0.8994,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950234359740623e-05,
+      "loss": 0.9648,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949654494781266e-05,
+      "loss": 0.9424,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994907127961808e-05,
+      "loss": 0.999,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948484714447453e-05,
+      "loss": 0.8604,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947894799466876e-05,
+      "loss": 0.9072,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947301534874995e-05,
+      "loss": 0.8945,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994670492087156e-05,
+      "loss": 0.9629,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994610495765747e-05,
+      "loss": 0.9478,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994550164543474e-05,
+      "loss": 0.9434,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994489498440651e-05,
+      "loss": 1.0122,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944284974777053e-05,
+      "loss": 0.9092,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943671616751774e-05,
+      "loss": 0.8608,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943054910537192e-05,
+      "loss": 0.9258,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942434856340972e-05,
+      "loss": 0.918,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941811454371888e-05,
+      "loss": 0.9463,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941184704839857e-05,
+      "loss": 0.894,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940554607955903e-05,
+      "loss": 0.9673,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939921163932198e-05,
+      "loss": 0.9629,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993928437298203e-05,
+      "loss": 0.9478,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938644235319823e-05,
+      "loss": 0.9263,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993800075116111e-05,
+      "loss": 0.9893,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937353920722572e-05,
+      "loss": 0.96,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936703744222e-05,
+      "loss": 0.3096,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993605022187832e-05,
+      "loss": 0.3474,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993539335391159e-05,
+      "loss": 0.98,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934733140542974e-05,
+      "loss": 0.939,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934069581994788e-05,
+      "loss": 0.9473,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993340267849046e-05,
+      "loss": 0.9102,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932732430254543e-05,
+      "loss": 0.8882,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932058837512727e-05,
+      "loss": 0.9346,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931381900491814e-05,
+      "loss": 0.9307,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930701619419743e-05,
+      "loss": 0.9702,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930017994525575e-05,
+      "loss": 0.9229,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929331026039494e-05,
+      "loss": 0.3259,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928640714192818e-05,
+      "loss": 0.9697,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992794705921798e-05,
+      "loss": 0.9463,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992725006134855e-05,
+      "loss": 0.917,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926549720819218e-05,
+      "loss": 0.8882,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925846037865795e-05,
+      "loss": 0.9307,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925139012725223e-05,
+      "loss": 0.9517,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992442864563557e-05,
+      "loss": 0.9575,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992371493683603e-05,
+      "loss": 0.9126,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922997886566913e-05,
+      "loss": 0.9033,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992227749506967e-05,
+      "loss": 0.9263,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992155376258686e-05,
+      "loss": 0.9438,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992082668936218e-05,
+      "loss": 0.335,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920096275640444e-05,
+      "loss": 0.9492,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919362521667596e-05,
+      "loss": 0.9019,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99186254276907e-05,
+      "loss": 0.9219,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991788499395795e-05,
+      "loss": 0.9634,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991714122071866e-05,
+      "loss": 0.9517,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991639410822327e-05,
+      "loss": 0.9297,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915643656723344e-05,
+      "loss": 0.9014,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914889866471572e-05,
+      "loss": 0.9126,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914132737721764e-05,
+      "loss": 0.959,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913372270728862e-05,
+      "loss": 0.8848,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991260846574892e-05,
+      "loss": 0.9502,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911841323039135e-05,
+      "loss": 0.9414,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99110708428578e-05,
+      "loss": 0.9868,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991029702546436e-05,
+      "loss": 0.9341,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909519871119368e-05,
+      "loss": 0.9102,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908739380084498e-05,
+      "loss": 0.9561,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907955552622564e-05,
+      "loss": 0.8906,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907168388997486e-05,
+      "loss": 0.9287,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990637788947432e-05,
+      "loss": 0.9238,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905584054319228e-05,
+      "loss": 0.9077,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904786883799518e-05,
+      "loss": 0.9644,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903986378183605e-05,
+      "loss": 0.3286,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903182537741035e-05,
+      "loss": 0.9771,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990237536274247e-05,
+      "loss": 0.9263,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901564853459702e-05,
+      "loss": 1.0151,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900751010165632e-05,
+      "loss": 0.9487,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989993383313431e-05,
+      "loss": 0.9839,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989911332264088e-05,
+      "loss": 0.9556,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898289478961623e-05,
+      "loss": 0.8652,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897462302373943e-05,
+      "loss": 0.9751,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896631793156363e-05,
+      "loss": 0.9448,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895797951588528e-05,
+      "loss": 0.895,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894960777951202e-05,
+      "loss": 0.9883,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989412027252628e-05,
+      "loss": 0.9541,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893276435596765e-05,
+      "loss": 0.9111,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892429267446796e-05,
+      "loss": 0.9224,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891578768361633e-05,
+      "loss": 0.8901,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890724938627643e-05,
+      "loss": 0.9209,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988986777853233e-05,
+      "loss": 0.9585,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889007288364304e-05,
+      "loss": 0.9136,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888143468413314e-05,
+      "loss": 0.9297,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887276318970222e-05,
+      "loss": 0.334,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886405840327005e-05,
+      "loss": 0.9199,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885532032776772e-05,
+      "loss": 0.9087,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884654896613743e-05,
+      "loss": 0.8979,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988377443213327e-05,
+      "loss": 1.0034,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882890639631812e-05,
+      "loss": 0.937,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988200351940696e-05,
+      "loss": 0.9751,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988111307175742e-05,
+      "loss": 0.9336,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988021929698302e-05,
+      "loss": 0.9487,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987932219538471e-05,
+      "loss": 0.9385,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987842176726455e-05,
+      "loss": 0.9097,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877518012925736e-05,
+      "loss": 0.9087,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876610932672575e-05,
+      "loss": 0.937,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987570052681049e-05,
+      "loss": 0.917,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987478679564604e-05,
+      "loss": 0.9067,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873869739486883e-05,
+      "loss": 0.9512,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872949358641805e-05,
+      "loss": 0.3171,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872025653420718e-05,
+      "loss": 0.957,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987109862413465e-05,
+      "loss": 0.9312,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870168271095735e-05,
+      "loss": 0.9004,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986923459461725e-05,
+      "loss": 0.8994,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868297595013574e-05,
+      "loss": 0.9165,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98673572726002e-05,
+      "loss": 0.9429,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866413627693768e-05,
+      "loss": 0.9038,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865466660612002e-05,
+      "loss": 0.9468,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986451637167377e-05,
+      "loss": 0.9233,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863562761199043e-05,
+      "loss": 0.9219,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986260582950892e-05,
+      "loss": 0.9722,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861645576925612e-05,
+      "loss": 0.3333,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860682003772453e-05,
+      "loss": 0.8936,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859715110373892e-05,
+      "loss": 0.9395,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858744897055498e-05,
+      "loss": 0.9336,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857771364143962e-05,
+      "loss": 0.9653,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856794511967076e-05,
+      "loss": 0.8877,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855814340853768e-05,
+      "loss": 0.8818,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985483085113408e-05,
+      "loss": 0.9688,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853844043139164e-05,
+      "loss": 0.9053,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852853917201295e-05,
+      "loss": 0.9658,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851860473653864e-05,
+      "loss": 0.8877,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985086371283138e-05,
+      "loss": 0.8677,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849863635069464e-05,
+      "loss": 0.9136,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848860240704862e-05,
+      "loss": 0.937,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984785353007543e-05,
+      "loss": 0.9019,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846843503520144e-05,
+      "loss": 0.8574,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845830161379097e-05,
+      "loss": 0.9146,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844813503993493e-05,
+      "loss": 0.8652,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984379353170566e-05,
+      "loss": 0.9189,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842770244859038e-05,
+      "loss": 0.9741,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841743643798184e-05,
+      "loss": 0.937,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840713728868768e-05,
+      "loss": 0.9355,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839680500417583e-05,
+      "loss": 0.894,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983864395879253e-05,
+      "loss": 0.9473,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837604104342627e-05,
+      "loss": 0.9531,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836560937418013e-05,
+      "loss": 0.9204,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835514458369936e-05,
+      "loss": 0.9429,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834464667550765e-05,
+      "loss": 0.9648,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833411565313976e-05,
+      "loss": 0.9419,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832355152014167e-05,
+      "loss": 0.9253,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831295428007053e-05,
+      "loss": 0.981,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830232393649454e-05,
+      "loss": 0.8921,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829166049299317e-05,
+      "loss": 0.3324,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828096395315687e-05,
+      "loss": 0.9536,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827023432058738e-05,
+      "loss": 0.9233,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982594715988976e-05,
+      "loss": 0.8901,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824867579171138e-05,
+      "loss": 0.8979,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982378469026639e-05,
+      "loss": 0.9702,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822698493540146e-05,
+      "loss": 0.9829,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982160898935814e-05,
+      "loss": 0.8945,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820516178087225e-05,
+      "loss": 0.9121,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819420060095368e-05,
+      "loss": 0.9248,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981832063575165e-05,
+      "loss": 0.9644,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817217905426264e-05,
+      "loss": 0.9009,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816111869490516e-05,
+      "loss": 0.9253,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815002528316826e-05,
+      "loss": 0.3167,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813889882278727e-05,
+      "loss": 0.9614,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812773931750862e-05,
+      "loss": 1.0078,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811654677108996e-05,
+      "loss": 0.9795,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981053211872999e-05,
+      "loss": 0.8525,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809406256991833e-05,
+      "loss": 0.3318,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808277092273616e-05,
+      "loss": 0.3223,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980714462495555e-05,
+      "loss": 0.8882,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806008855418955e-05,
+      "loss": 0.8901,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980486978404626e-05,
+      "loss": 0.9082,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803727411221006e-05,
+      "loss": 0.957,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802581737327853e-05,
+      "loss": 0.916,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980143276275256e-05,
+      "loss": 0.3446,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800280487882015e-05,
+      "loss": 0.8687,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97991249131042e-05,
+      "loss": 0.9468,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797966038808213e-05,
+      "loss": 0.8726,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796803865384273e-05,
+      "loss": 0.8965,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795638393223693e-05,
+      "loss": 0.8979,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979446962271891e-05,
+      "loss": 0.9302,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793297554263467e-05,
+      "loss": 0.9497,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792122188252022e-05,
+      "loss": 0.9453,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979094352508033e-05,
+      "loss": 0.9634,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978976156514527e-05,
+      "loss": 0.8848,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978857630884483e-05,
+      "loss": 0.978,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97873877565781e-05,
+      "loss": 0.9917,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978619590874528e-05,
+      "loss": 0.876,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785000765747696e-05,
+      "loss": 0.9414,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978380232798776e-05,
+      "loss": 0.9448,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978260059586901e-05,
+      "loss": 0.9102,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781395569796085e-05,
+      "loss": 0.9019,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780187250174737e-05,
+      "loss": 0.9312,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977897563741183e-05,
+      "loss": 0.9263,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777760731915325e-05,
+      "loss": 0.9287,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977654253409431e-05,
+      "loss": 0.9092,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775321044358957e-05,
+      "loss": 0.8813,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774096263120573e-05,
+      "loss": 0.9731,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977286819079156e-05,
+      "loss": 0.8872,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771636827785422e-05,
+      "loss": 0.9087,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977040217451678e-05,
+      "loss": 0.9219,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976916423140137e-05,
+      "loss": 0.9927,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767922998856015e-05,
+      "loss": 0.9277,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766678477298663e-05,
+      "loss": 0.9268,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765430667148368e-05,
+      "loss": 0.9106,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764179568825276e-05,
+      "loss": 0.9033,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976292518275066e-05,
+      "loss": 0.9312,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976166750934689e-05,
+      "loss": 0.9653,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976040654903744e-05,
+      "loss": 0.9307,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759142302246902e-05,
+      "loss": 0.8867,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975787476940096e-05,
+      "loss": 0.9683,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975660395092642e-05,
+      "loss": 0.8887,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975532984725118e-05,
+      "loss": 0.9883,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754052458804252e-05,
+      "loss": 0.3406,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752771786015756e-05,
+      "loss": 0.9424,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751487829316908e-05,
+      "loss": 0.9424,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750200589140044e-05,
+      "loss": 0.9351,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974891006591859e-05,
+      "loss": 0.9639,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974761626008709e-05,
+      "loss": 0.8706,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974631917208119e-05,
+      "loss": 0.8462,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974501880233764e-05,
+      "loss": 0.978,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743715151294287e-05,
+      "loss": 0.894,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97424082193901e-05,
+      "loss": 0.9927,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741098007065137e-05,
+      "loss": 0.915,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973978451476057e-05,
+      "loss": 0.9097,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738467742918673e-05,
+      "loss": 0.957,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737147691982817e-05,
+      "loss": 0.916,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735824362397497e-05,
+      "loss": 0.9165,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734497754608287e-05,
+      "loss": 0.9536,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973316786906188e-05,
+      "loss": 0.9536,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973183470620607e-05,
+      "loss": 0.8672,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730498266489753e-05,
+      "loss": 0.9102,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972915855036293e-05,
+      "loss": 0.9907,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97278155582767e-05,
+      "loss": 0.9077,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726469290683276e-05,
+      "loss": 0.3199,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725119748035965e-05,
+      "loss": 0.3193,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723766930789177e-05,
+      "loss": 0.9634,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722410839398426e-05,
+      "loss": 0.9082,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721051474320337e-05,
+      "loss": 0.916,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971968883601262e-05,
+      "loss": 0.9038,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97183229249341e-05,
+      "loss": 0.9004,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97169537415447e-05,
+      "loss": 0.9326,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715581286305447e-05,
+      "loss": 0.9097,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971420555967847e-05,
+      "loss": 0.936,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712826562126988e-05,
+      "loss": 0.9399,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971144429411534e-05,
+      "loss": 0.9243,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971005875610896e-05,
+      "loss": 0.8862,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708669948574368e-05,
+      "loss": 0.9517,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970727787197921e-05,
+      "loss": 0.9238,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970588252679221e-05,
+      "loss": 0.8906,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704483913483208e-05,
+      "loss": 0.9468,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970308203252314e-05,
+      "loss": 0.9233,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701676884384036e-05,
+      "loss": 0.9141,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700268469539034e-05,
+      "loss": 0.9238,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698856788462372e-05,
+      "loss": 0.9805,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969744184162938e-05,
+      "loss": 0.9233,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696023629516498e-05,
+      "loss": 0.8921,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969460215260125e-05,
+      "loss": 0.9023,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693177411362283e-05,
+      "loss": 0.9839,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691749406279323e-05,
+      "loss": 0.9048,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690318137833202e-05,
+      "loss": 0.9609,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968888360650585e-05,
+      "loss": 0.9873,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687445812780294e-05,
+      "loss": 0.9375,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686004757140663e-05,
+      "loss": 0.9775,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968456044007219e-05,
+      "loss": 0.9385,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683112862061183e-05,
+      "loss": 0.3047,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968166202359508e-05,
+      "loss": 0.9292,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680207925162396e-05,
+      "loss": 0.9609,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678750567252742e-05,
+      "loss": 0.895,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677289950356844e-05,
+      "loss": 0.873,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675826074966505e-05,
+      "loss": 0.9219,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674358941574642e-05,
+      "loss": 0.9336,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967288855067526e-05,
+      "loss": 0.9434,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671414902763457e-05,
+      "loss": 0.9092,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669937998335438e-05,
+      "loss": 0.9619,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96684578378885e-05,
+      "loss": 0.9497,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666974421921035e-05,
+      "loss": 0.9126,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665487750932534e-05,
+      "loss": 0.9312,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663997825423577e-05,
+      "loss": 0.8682,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662504645895847e-05,
+      "loss": 0.958,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661008212852128e-05,
+      "loss": 0.9375,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659508526796284e-05,
+      "loss": 0.9189,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965800558823329e-05,
+      "loss": 0.9307,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96564993976692e-05,
+      "loss": 0.9253,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654989955611177e-05,
+      "loss": 0.9453,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653477262567472e-05,
+      "loss": 0.9541,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651961319047436e-05,
+      "loss": 0.9731,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965044212556151e-05,
+      "loss": 0.8984,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648919682621223e-05,
+      "loss": 0.8374,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647393990739214e-05,
+      "loss": 0.9565,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645865050429203e-05,
+      "loss": 0.301,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964433286220601e-05,
+      "loss": 0.9175,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964279742658554e-05,
+      "loss": 0.9692,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964125874408481e-05,
+      "loss": 0.918,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963971681522191e-05,
+      "loss": 0.9175,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638171640516037e-05,
+      "loss": 0.9429,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636623220487468e-05,
+      "loss": 0.8901,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635071555657586e-05,
+      "loss": 0.8784,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963351664654886e-05,
+      "loss": 0.9404,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631958493684845e-05,
+      "loss": 0.9082,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963039709759021e-05,
+      "loss": 0.3357,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628832458790685e-05,
+      "loss": 0.9521,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627264577813123e-05,
+      "loss": 0.3092,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962569345518544e-05,
+      "loss": 0.8848,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962411909143667e-05,
+      "loss": 0.366,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962254148709692e-05,
+      "loss": 0.9165,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620960642697394e-05,
+      "loss": 0.8628,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619376558770385e-05,
+      "loss": 0.873,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617789235849284e-05,
+      "loss": 0.9302,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961619867446856e-05,
+      "loss": 0.9878,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961460487516379e-05,
+      "loss": 0.9316,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613007838471622e-05,
+      "loss": 0.9253,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961140756492981e-05,
+      "loss": 0.9683,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609804055077183e-05,
+      "loss": 0.9248,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608197309453673e-05,
+      "loss": 0.9131,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96065873286003e-05,
+      "loss": 0.917,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604974113059164e-05,
+      "loss": 0.9121,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603357663373458e-05,
+      "loss": 0.939,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601737980087468e-05,
+      "loss": 0.9141,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960011506374657e-05,
+      "loss": 0.9556,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598488914897226e-05,
+      "loss": 0.9253,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596859534086982e-05,
+      "loss": 0.3005,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595226921864472e-05,
+      "loss": 0.8608,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959359107877943e-05,
+      "loss": 0.9355,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591952005382668e-05,
+      "loss": 0.8745,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959030970222608e-05,
+      "loss": 0.9097,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588664169862665e-05,
+      "loss": 0.9419,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587015408846495e-05,
+      "loss": 0.8579,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585363419732736e-05,
+      "loss": 0.9082,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583708203077632e-05,
+      "loss": 0.8521,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958204975943853e-05,
+      "loss": 0.9209,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958038808937384e-05,
+      "loss": 0.8975,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957872319344309e-05,
+      "loss": 0.896,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577055072206865e-05,
+      "loss": 0.9067,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575383726226844e-05,
+      "loss": 0.9346,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573709156065807e-05,
+      "loss": 0.9502,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95720313622876e-05,
+      "loss": 0.9414,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570350345457164e-05,
+      "loss": 0.8945,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956866610614052e-05,
+      "loss": 0.8975,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566978644904785e-05,
+      "loss": 0.8965,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956528796231815e-05,
+      "loss": 0.9375,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563594058949895e-05,
+      "loss": 0.9492,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561896935370385e-05,
+      "loss": 0.9448,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956019659215106e-05,
+      "loss": 0.853,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955849302986446e-05,
+      "loss": 0.8984,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556786249084198e-05,
+      "loss": 0.8809,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555076250384975e-05,
+      "loss": 0.9297,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553363034342575e-05,
+      "loss": 0.978,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955164660153386e-05,
+      "loss": 0.9199,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549926952536788e-05,
+      "loss": 0.9111,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954820408793038e-05,
+      "loss": 0.8311,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546478008294763e-05,
+      "loss": 0.8867,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544748714211128e-05,
+      "loss": 0.9482,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543016206261758e-05,
+      "loss": 0.9629,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541280485030013e-05,
+      "loss": 1.0244,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539541551100344e-05,
+      "loss": 0.937,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537799405058266e-05,
+      "loss": 0.2928,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536054047490395e-05,
+      "loss": 0.9087,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953430547898442e-05,
+      "loss": 0.9761,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532553700129107e-05,
+      "loss": 0.8916,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953079871151431e-05,
+      "loss": 0.9048,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952904051373096e-05,
+      "loss": 0.8857,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952727910737107e-05,
+      "loss": 0.875,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525514493027734e-05,
+      "loss": 0.873,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523746671295125e-05,
+      "loss": 0.8838,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521975642768498e-05,
+      "loss": 0.9395,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952020140804418e-05,
+      "loss": 0.9497,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518423967719586e-05,
+      "loss": 0.9136,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951664332239321e-05,
+      "loss": 0.9067,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514859472664625e-05,
+      "loss": 0.9351,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513072419134475e-05,
+      "loss": 0.9258,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951128216240449e-05,
+      "loss": 0.8867,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509488703077483e-05,
+      "loss": 0.876,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507692041757333e-05,
+      "loss": 0.3789,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950589217904901e-05,
+      "loss": 0.9419,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950408911555855e-05,
+      "loss": 0.9302,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502282851893072e-05,
+      "loss": 0.939,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950047338866078e-05,
+      "loss": 0.3101,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498660726470944e-05,
+      "loss": 0.9014,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496844865933915e-05,
+      "loss": 0.8491,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495025807661125e-05,
+      "loss": 0.8584,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493203552265075e-05,
+      "loss": 0.917,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949137810035935e-05,
+      "loss": 0.8965,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948954945255861e-05,
+      "loss": 0.8945,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487717609478586e-05,
+      "loss": 0.8604,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485882571736082e-05,
+      "loss": 0.8691,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484044339948998e-05,
+      "loss": 0.897,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482202914736284e-05,
+      "loss": 0.9448,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480358296717982e-05,
+      "loss": 0.9033,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94785104865152e-05,
+      "loss": 0.9346,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476659484750126e-05,
+      "loss": 0.8906,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474805292046022e-05,
+      "loss": 0.8501,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472947909027224e-05,
+      "loss": 0.939,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947108733631914e-05,
+      "loss": 0.9238,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469223574548255e-05,
+      "loss": 0.9087,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946735662434213e-05,
+      "loss": 0.8662,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465486486329385e-05,
+      "loss": 0.9419,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463613161139738e-05,
+      "loss": 0.9346,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946173664940396e-05,
+      "loss": 0.8989,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94598569517539e-05,
+      "loss": 0.9292,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457974068822492e-05,
+      "loss": 0.9678,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456088001243718e-05,
+      "loss": 0.9189,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454198749652655e-05,
+      "loss": 0.8799,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452306314685442e-05,
+      "loss": 0.8716,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945041069697929e-05,
+      "loss": 0.8853,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448511897172482e-05,
+      "loss": 0.8916,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446609915904375e-05,
+      "loss": 0.9429,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444704753815397e-05,
+      "loss": 0.8784,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442796411547046e-05,
+      "loss": 0.9424,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944088488974189e-05,
+      "loss": 0.8892,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438970189043564e-05,
+      "loss": 0.9126,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437052310096783e-05,
+      "loss": 0.9097,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435131253547327e-05,
+      "loss": 0.8906,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943320702004204e-05,
+      "loss": 0.8618,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431279610228845e-05,
+      "loss": 0.8872,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429349024756735e-05,
+      "loss": 0.9253,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942741526427576e-05,
+      "loss": 0.8594,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425478329437055e-05,
+      "loss": 0.8682,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942353822089281e-05,
+      "loss": 0.877,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421594939296296e-05,
+      "loss": 0.3196,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941964848530184e-05,
+      "loss": 0.9224,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417698859564844e-05,
+      "loss": 0.9072,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415746062741784e-05,
+      "loss": 0.9595,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413790095490188e-05,
+      "loss": 0.9092,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411830958468668e-05,
+      "loss": 0.9238,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409868652336893e-05,
+      "loss": 0.9458,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94079031777556e-05,
+      "loss": 0.9678,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94059345353866e-05,
+      "loss": 0.9092,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403962725892764e-05,
+      "loss": 0.9688,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401987749938027e-05,
+      "loss": 0.9102,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94000096081874e-05,
+      "loss": 0.8989,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939802830130695e-05,
+      "loss": 0.8525,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396043829963817e-05,
+      "loss": 0.9355,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394056194826198e-05,
+      "loss": 0.9629,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392065396563368e-05,
+      "loss": 0.9238,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939007143584565e-05,
+      "loss": 0.8262,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388074313344454e-05,
+      "loss": 0.9448,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386074029732235e-05,
+      "loss": 0.9106,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384070585682518e-05,
+      "loss": 0.2976,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382063981869897e-05,
+      "loss": 0.9468,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938005421897003e-05,
+      "loss": 0.9463,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937804129765963e-05,
+      "loss": 0.9131,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376025218616484e-05,
+      "loss": 0.8994,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937400598251943e-05,
+      "loss": 0.9175,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371983590048383e-05,
+      "loss": 0.9136,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936995804188431e-05,
+      "loss": 0.939,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367929338709245e-05,
+      "loss": 0.8813,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365897481206287e-05,
+      "loss": 0.9546,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363862470059595e-05,
+      "loss": 0.9326,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361824305954383e-05,
+      "loss": 0.8989,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935978298957694e-05,
+      "loss": 0.8696,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357738521614598e-05,
+      "loss": 0.9434,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355690902755776e-05,
+      "loss": 0.9243,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353640133689927e-05,
+      "loss": 0.8701,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935158621510758e-05,
+      "loss": 0.9409,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349529147700328e-05,
+      "loss": 0.9365,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934746893216081e-05,
+      "loss": 0.832,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345405569182735e-05,
+      "loss": 0.9614,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343339059460868e-05,
+      "loss": 0.8896,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934126940369104e-05,
+      "loss": 0.8716,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339196602570133e-05,
+      "loss": 0.9048,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933712065679609e-05,
+      "loss": 0.9292,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335041567067913e-05,
+      "loss": 0.9092,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332959334085672e-05,
+      "loss": 0.9536,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933087395855048e-05,
+      "loss": 0.9121,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328785441164516e-05,
+      "loss": 0.8584,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932669378263102e-05,
+      "loss": 0.8604,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324598983654282e-05,
+      "loss": 0.9497,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322501044939658e-05,
+      "loss": 0.9629,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320399967193556e-05,
+      "loss": 0.9194,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318295751123438e-05,
+      "loss": 0.856,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316188397437824e-05,
+      "loss": 0.8999,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314077906846303e-05,
+      "loss": 0.9126,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311964280059503e-05,
+      "loss": 0.8774,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309847517789114e-05,
+      "loss": 0.8975,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307727620747884e-05,
+      "loss": 0.8887,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930560458964962e-05,
+      "loss": 0.8501,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303478425209173e-05,
+      "loss": 0.8564,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301349128142463e-05,
+      "loss": 0.8984,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299216699166453e-05,
+      "loss": 0.3699,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297081138999164e-05,
+      "loss": 0.9287,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929494244835967e-05,
+      "loss": 0.9131,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292800627968112e-05,
+      "loss": 0.9214,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290655678545665e-05,
+      "loss": 0.3538,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928850760081457e-05,
+      "loss": 0.3383,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928635639549812e-05,
+      "loss": 0.8857,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284202063320652e-05,
+      "loss": 0.9292,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282044605007573e-05,
+      "loss": 0.9404,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279884021285326e-05,
+      "loss": 0.8862,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277720312881416e-05,
+      "loss": 0.9307,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275553480524397e-05,
+      "loss": 0.894,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927338352494388e-05,
+      "loss": 0.9194,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927121044687051e-05,
+      "loss": 0.9146,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269034247036008e-05,
+      "loss": 0.9644,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266854926173134e-05,
+      "loss": 0.9629,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264672485015693e-05,
+      "loss": 0.9009,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926248692429855e-05,
+      "loss": 0.9282,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260298244757618e-05,
+      "loss": 0.9375,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925810644712986e-05,
+      "loss": 0.9434,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255911532153287e-05,
+      "loss": 0.9326,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925371350056696e-05,
+      "loss": 0.8784,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251512353110993e-05,
+      "loss": 0.8892,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249308090526545e-05,
+      "loss": 0.9438,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924710071355583e-05,
+      "loss": 0.9624,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92448902229421e-05,
+      "loss": 0.918,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242676619429664e-05,
+      "loss": 0.8447,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240459903763878e-05,
+      "loss": 0.3374,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923824007669114e-05,
+      "loss": 0.9531,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236017138958913e-05,
+      "loss": 0.8828,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233791091315675e-05,
+      "loss": 0.9346,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923156193451099e-05,
+      "loss": 1.0083,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922932966929544e-05,
+      "loss": 0.916,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227094296420662e-05,
+      "loss": 0.897,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224855816639348e-05,
+      "loss": 0.9033,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222614230705225e-05,
+      "loss": 0.896,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922036953937307e-05,
+      "loss": 0.3291,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218121743398702e-05,
+      "loss": 0.8652,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215870843539e-05,
+      "loss": 0.9443,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213616840551867e-05,
+      "loss": 0.8799,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921135973519626e-05,
+      "loss": 0.8574,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209099528232188e-05,
+      "loss": 0.9258,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206836220420696e-05,
+      "loss": 0.9316,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204569812523874e-05,
+      "loss": 0.9409,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202300305304854e-05,
+      "loss": 0.9307,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200027699527824e-05,
+      "loss": 0.9121,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197751995957992e-05,
+      "loss": 0.9258,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195473195361636e-05,
+      "loss": 0.8584,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919319129850606e-05,
+      "loss": 0.9419,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190906306159607e-05,
+      "loss": 0.9639,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188618219091676e-05,
+      "loss": 0.9009,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91863270380727e-05,
+      "loss": 0.8398,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918403276387416e-05,
+      "loss": 0.8994,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181735397268565e-05,
+      "loss": 0.9087,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917943493902948e-05,
+      "loss": 0.9238,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177131389931503e-05,
+      "loss": 0.9336,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917482475075028e-05,
+      "loss": 0.8789,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172515022262487e-05,
+      "loss": 0.8965,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170202205245845e-05,
+      "loss": 0.918,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167886300479115e-05,
+      "loss": 0.957,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165567308742108e-05,
+      "loss": 0.9092,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916324523081565e-05,
+      "loss": 0.9238,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160920067481637e-05,
+      "loss": 0.9111,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158591819522972e-05,
+      "loss": 0.9058,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156260487723623e-05,
+      "loss": 0.9023,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153926072868582e-05,
+      "loss": 0.9185,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915158857574388e-05,
+      "loss": 0.9448,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149247997136596e-05,
+      "loss": 0.939,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146904337834834e-05,
+      "loss": 0.9243,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914455759862774e-05,
+      "loss": 0.9019,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91422077803055e-05,
+      "loss": 0.8921,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139854883659333e-05,
+      "loss": 0.8931,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91374989094815e-05,
+      "loss": 0.9219,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913513985856529e-05,
+      "loss": 0.8882,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132777731705032e-05,
+      "loss": 0.3357,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913041252969609e-05,
+      "loss": 0.9238,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128044253334874e-05,
+      "loss": 0.8491,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125672903418808e-05,
+      "loss": 0.9355,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123298480746365e-05,
+      "loss": 0.9209,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120920986117058e-05,
+      "loss": 0.895,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118540420331416e-05,
+      "loss": 0.9868,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911615678419102e-05,
+      "loss": 0.9604,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911377007849847e-05,
+      "loss": 0.9307,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111380304057415e-05,
+      "loss": 0.8687,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108987461672527e-05,
+      "loss": 0.9307,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106591552149506e-05,
+      "loss": 0.8516,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104192576295106e-05,
+      "loss": 0.9473,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101790534917088e-05,
+      "loss": 0.8975,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909938542882426e-05,
+      "loss": 0.9316,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096977258826462e-05,
+      "loss": 0.9014,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909456602573456e-05,
+      "loss": 0.9858,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092151730360453e-05,
+      "loss": 0.894,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089734373517072e-05,
+      "loss": 0.9121,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908731395601838e-05,
+      "loss": 0.3171,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908489047867937e-05,
+      "loss": 0.9365,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082463942316057e-05,
+      "loss": 0.8677,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080034347745507e-05,
+      "loss": 0.9087,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077601695785793e-05,
+      "loss": 0.9336,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075165987256027e-05,
+      "loss": 0.8735,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072727222976356e-05,
+      "loss": 0.8584,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070285403767943e-05,
+      "loss": 0.9106,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906784053045299e-05,
+      "loss": 0.8818,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065392603854725e-05,
+      "loss": 0.9463,
+      "step": 919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90629416247974e-05,
+      "loss": 0.8975,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060487594106303e-05,
+      "loss": 0.8804,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058030512607736e-05,
+      "loss": 0.9199,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055570381129046e-05,
+      "loss": 0.8828,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053107200498592e-05,
+      "loss": 0.8804,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050640971545763e-05,
+      "loss": 0.9355,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048171695100985e-05,
+      "loss": 0.8662,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045699371995695e-05,
+      "loss": 0.9321,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904322400306237e-05,
+      "loss": 0.9189,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040745589134493e-05,
+      "loss": 0.9053,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038264131046597e-05,
+      "loss": 0.9341,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903577962963422e-05,
+      "loss": 0.939,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033292085733935e-05,
+      "loss": 0.875,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903080150018334e-05,
+      "loss": 0.8862,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028307873821046e-05,
+      "loss": 0.9082,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025811207486707e-05,
+      "loss": 0.9458,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902331150202098e-05,
+      "loss": 0.894,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020808758265556e-05,
+      "loss": 0.9707,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018302977063153e-05,
+      "loss": 0.8594,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90157941592575e-05,
+      "loss": 0.8701,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013282305693358e-05,
+      "loss": 0.9614,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901076741721651e-05,
+      "loss": 0.896,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008249494673756e-05,
+      "loss": 0.9326,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900572853891292e-05,
+      "loss": 0.9277,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003204550782848e-05,
+      "loss": 0.8638,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000677531133402e-05,
+      "loss": 0.9067,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998147480815473e-05,
+      "loss": 0.9053,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899561440068097e-05,
+      "loss": 0.915,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993078291582814e-05,
+      "loss": 0.8594,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990539154374957e-05,
+      "loss": 0.9702,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987996989912366e-05,
+      "loss": 0.9126,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898545179905103e-05,
+      "loss": 0.9048,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982903582647946e-05,
+      "loss": 0.9316,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980352341561148e-05,
+      "loss": 0.9023,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977798076649673e-05,
+      "loss": 0.9062,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897524078877358e-05,
+      "loss": 0.8433,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972680478793953e-05,
+      "loss": 0.9512,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970117147572885e-05,
+      "loss": 0.8823,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967550795973494e-05,
+      "loss": 0.915,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964981424859905e-05,
+      "loss": 0.8794,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896240903509727e-05,
+      "loss": 0.8779,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959833627551747e-05,
+      "loss": 0.8804,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957255203090517e-05,
+      "loss": 0.9463,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954673762581783e-05,
+      "loss": 0.875,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895208930689475e-05,
+      "loss": 0.8457,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949501836899643e-05,
+      "loss": 0.8857,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894691135346771e-05,
+      "loss": 0.9746,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89443178574712e-05,
+      "loss": 0.9395,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894172134978339e-05,
+      "loss": 0.8711,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893912183127856e-05,
+      "loss": 0.9004,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893651930283201e-05,
+      "loss": 0.8501,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933913765320052e-05,
+      "loss": 0.9077,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893130521962001e-05,
+      "loss": 0.9199,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928693666610225e-05,
+      "loss": 0.8633,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926079107170045e-05,
+      "loss": 0.8882,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923461542179833e-05,
+      "loss": 0.8813,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920840972520963e-05,
+      "loss": 0.875,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8918217399075827e-05,
+      "loss": 0.8857,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915590822727816e-05,
+      "loss": 0.8945,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912961244361345e-05,
+      "loss": 0.8687,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891032866486183e-05,
+      "loss": 0.8687,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907693085115704e-05,
+      "loss": 0.9224,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890505450601041e-05,
+      "loss": 0.915,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890241292843439e-05,
+      "loss": 0.3463,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889976835327711e-05,
+      "loss": 0.9062,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889712078142904e-05,
+      "loss": 0.9536,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889447021378166e-05,
+      "loss": 0.9043,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889181665122746e-05,
+      "loss": 0.9482,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889160094659926e-05,
+      "loss": 0.8086,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888650054497357e-05,
+      "loss": 0.3381,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883838003063902e-05,
+      "loss": 0.9189,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888117246982744e-05,
+      "loss": 0.8545,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887850394616171e-05,
+      "loss": 0.9258,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875832432965252e-05,
+      "loss": 0.9019,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873157931137596e-05,
+      "loss": 0.8301,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88704804415793e-05,
+      "loss": 0.9585,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867799965191905e-05,
+      "loss": 0.9341,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886511650287798e-05,
+      "loss": 0.8564,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862430055541078e-05,
+      "loss": 0.918,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859740624085772e-05,
+      "loss": 0.9331,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885704820941764e-05,
+      "loss": 0.9229,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854352812443255e-05,
+      "loss": 0.8887,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88516544340702e-05,
+      "loss": 0.9106,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848953075207063e-05,
+      "loss": 0.8877,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884624873676343e-05,
+      "loss": 0.8843,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88435414196499e-05,
+      "loss": 0.9033,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840831124778062e-05,
+      "loss": 0.9082,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838117853060517e-05,
+      "loss": 0.8916,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883540160541087e-05,
+      "loss": 0.9072,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832682382743717e-05,
+      "loss": 0.3699,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829960185974667e-05,
+      "loss": 0.8716,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827235016020328e-05,
+      "loss": 0.8901,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88245068737983e-05,
+      "loss": 0.9834,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88217757602272e-05,
+      "loss": 0.9009,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819041676226632e-05,
+      "loss": 0.9126,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816304622717202e-05,
+      "loss": 0.9785,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881356460062052e-05,
+      "loss": 0.8857,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810821610859196e-05,
+      "loss": 0.3352,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880807565435684e-05,
+      "loss": 0.9199,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805326732038052e-05,
+      "loss": 0.8901,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802574844828446e-05,
+      "loss": 0.9414,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799819993654612e-05,
+      "loss": 0.894,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797062179444166e-05,
+      "loss": 0.9077,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794301403125694e-05,
+      "loss": 0.8906,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791537665628803e-05,
+      "loss": 0.9019,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878877096788408e-05,
+      "loss": 0.8989,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878600131082312e-05,
+      "loss": 0.939,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87832286953785e-05,
+      "loss": 0.3258,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780453122483816e-05,
+      "loss": 0.9209,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877767459307364e-05,
+      "loss": 0.8804,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877489310808355e-05,
+      "loss": 0.874,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772108668450106e-05,
+      "loss": 0.9058,
+      "step": 1031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769321275110878e-05,
+      "loss": 0.9019,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766530929004428e-05,
+      "loss": 0.9307,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763737631070306e-05,
+      "loss": 0.8877,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876094138224906e-05,
+      "loss": 0.9092,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758142183482228e-05,
+      "loss": 0.356,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755340035712345e-05,
+      "loss": 0.8926,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752534939882944e-05,
+      "loss": 0.9595,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749726896938535e-05,
+      "loss": 0.8926,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746915907824633e-05,
+      "loss": 0.8711,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874410197348774e-05,
+      "loss": 0.9224,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874128509487536e-05,
+      "loss": 0.958,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738465272935972e-05,
+      "loss": 0.8838,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735642508619055e-05,
+      "loss": 0.3254,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732816802875082e-05,
+      "loss": 0.8862,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8729988156655507e-05,
+      "loss": 0.877,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727156570912782e-05,
+      "loss": 0.9058,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724322046600347e-05,
+      "loss": 0.9224,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721484584672627e-05,
+      "loss": 0.8843,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718644186085045e-05,
+      "loss": 0.7983,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715800851794e-05,
+      "loss": 0.8872,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712954582756895e-05,
+      "loss": 0.8584,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710105379932108e-05,
+      "loss": 0.8818,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870725324427901e-05,
+      "loss": 0.8594,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704398176757963e-05,
+      "loss": 0.8784,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701540178330312e-05,
+      "loss": 0.9009,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698679249958386e-05,
+      "loss": 0.8657,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695815392605506e-05,
+      "loss": 0.8999,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869294860723598e-05,
+      "loss": 0.8813,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690078894815098e-05,
+      "loss": 0.9014,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687206256309133e-05,
+      "loss": 0.877,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684330692685355e-05,
+      "loss": 0.8979,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681452204912004e-05,
+      "loss": 0.8999,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678570793958312e-05,
+      "loss": 0.8984,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675686460794502e-05,
+      "loss": 0.8828,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867279920639177e-05,
+      "loss": 0.365,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669909031722295e-05,
+      "loss": 0.8955,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866701593775925e-05,
+      "loss": 0.9302,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866411992547678e-05,
+      "loss": 0.8467,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661220995850024e-05,
+      "loss": 0.8755,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865831914985509e-05,
+      "loss": 0.877,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865541438846908e-05,
+      "loss": 0.9233,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652506712670067e-05,
+      "loss": 0.8809,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649596123437116e-05,
+      "loss": 0.9409,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646682621750264e-05,
+      "loss": 0.9297,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643766208590536e-05,
+      "loss": 0.937,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864084688493993e-05,
+      "loss": 0.8672,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637924651781433e-05,
+      "loss": 0.939,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634999510099003e-05,
+      "loss": 0.9263,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863207146087758e-05,
+      "loss": 0.9194,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629140505103084e-05,
+      "loss": 0.8994,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626206643762415e-05,
+      "loss": 0.364,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862326987784345e-05,
+      "loss": 0.8794,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620330208335042e-05,
+      "loss": 0.9087,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861738763622703e-05,
+      "loss": 0.9644,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614442162510214e-05,
+      "loss": 0.9102,
+      "step": 1086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611493788176386e-05,
+      "loss": 0.8755,
+      "step": 1087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608542514218308e-05,
+      "loss": 0.8843,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860558834162972e-05,
+      "loss": 0.937,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602631271405343e-05,
+      "loss": 0.8608,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599671304540862e-05,
+      "loss": 0.8525,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859670844203295e-05,
+      "loss": 0.8628,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593742684879243e-05,
+      "loss": 0.874,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590774034078362e-05,
+      "loss": 0.8833,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587802490629897e-05,
+      "loss": 0.8521,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858482805553441e-05,
+      "loss": 0.916,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581850729793446e-05,
+      "loss": 0.9106,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857887051440951e-05,
+      "loss": 0.9331,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857588741038609e-05,
+      "loss": 0.8584,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572901418727647e-05,
+      "loss": 0.8833,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569912540439607e-05,
+      "loss": 0.9106,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856692077652837e-05,
+      "loss": 0.8809,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563926128001316e-05,
+      "loss": 0.8643,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560928595866784e-05,
+      "loss": 0.9756,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8557928181134093e-05,
+      "loss": 0.9092,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554924884813527e-05,
+      "loss": 0.8638,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551918707916344e-05,
+      "loss": 0.8857,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854890965145477e-05,
+      "loss": 0.3789,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545897716442005e-05,
+      "loss": 0.8774,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854288290389221e-05,
+      "loss": 0.9009,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853986521482052e-05,
+      "loss": 0.9087,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536844650243045e-05,
+      "loss": 0.8315,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533821211176843e-05,
+      "loss": 0.8926,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530794898639963e-05,
+      "loss": 0.9019,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527765713651408e-05,
+      "loss": 0.8594,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524733657231152e-05,
+      "loss": 0.9375,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521698730400136e-05,
+      "loss": 0.9756,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851866093418027e-05,
+      "loss": 0.895,
+      "step": 1118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515620269594423e-05,
+      "loss": 0.9282,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851257673766644e-05,
+      "loss": 0.9229,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509530339421116e-05,
+      "loss": 0.897,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850648107588423e-05,
+      "loss": 0.8936,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850342894808251e-05,
+      "loss": 0.8691,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500373957043658e-05,
+      "loss": 0.8647,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849731610379634e-05,
+      "loss": 0.9126,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849425538937018e-05,
+      "loss": 0.8706,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491191814795767e-05,
+      "loss": 0.9536,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488125381104655e-05,
+      "loss": 0.936,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848505608932936e-05,
+      "loss": 0.8843,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481983940503365e-05,
+      "loss": 0.8511,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478908935661107e-05,
+      "loss": 0.8916,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475831075837982e-05,
+      "loss": 0.8794,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472750362070363e-05,
+      "loss": 0.9282,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846966679539557e-05,
+      "loss": 0.9287,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466580376851886e-05,
+      "loss": 0.9326,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463491107478563e-05,
+      "loss": 0.8359,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460398988315798e-05,
+      "loss": 0.875,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457304020404764e-05,
+      "loss": 0.3433,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454206204787576e-05,
+      "loss": 0.3604,
+      "step": 1139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845110554250732e-05,
+      "loss": 0.8857,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448002034608044e-05,
+      "loss": 0.9146,
+      "step": 1141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844489568213474e-05,
+      "loss": 0.8701,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441786486133366e-05,
+      "loss": 0.9414,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843867444765084e-05,
+      "loss": 0.9526,
+      "step": 1144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435559567735032e-05,
+      "loss": 0.9521,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843244184743477e-05,
+      "loss": 0.873,
+      "step": 1146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842932128779984e-05,
+      "loss": 0.916,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426197889880985e-05,
+      "loss": 0.3239,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423071654729896e-05,
+      "loss": 0.8579,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841994258339923e-05,
+      "loss": 0.8828,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416810676942596e-05,
+      "loss": 0.8306,
+      "step": 1151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413675936414552e-05,
+      "loss": 0.8721,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410538362870612e-05,
+      "loss": 0.9409,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407397957367253e-05,
+      "loss": 0.8931,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404254720961888e-05,
+      "loss": 0.335,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401108654712905e-05,
+      "loss": 0.8833,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397959759679625e-05,
+      "loss": 0.9048,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839480803692234e-05,
+      "loss": 0.8823,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839165348750227e-05,
+      "loss": 0.9253,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388496112481612e-05,
+      "loss": 0.8672,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83853359129235e-05,
+      "loss": 0.9336,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382172889892023e-05,
+      "loss": 0.9351,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837900704445222e-05,
+      "loss": 0.9644,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837583837767008e-05,
+      "loss": 0.8594,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837266689061254e-05,
+      "loss": 0.9141,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369492584347494e-05,
+      "loss": 0.8789,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836631545994378e-05,
+      "loss": 0.9014,
+      "step": 1167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363135518471184e-05,
+      "loss": 0.8989,
+      "step": 1168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835995276100044e-05,
+      "loss": 0.8823,
+      "step": 1169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356767188603232e-05,
+      "loss": 0.8667,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353578802352196e-05,
+      "loss": 0.8906,
+      "step": 1171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350387603320908e-05,
+      "loss": 0.9121,
+      "step": 1172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347193592583898e-05,
+      "loss": 0.9009,
+      "step": 1173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343996771216632e-05,
+      "loss": 0.9375,
+      "step": 1174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340797140295538e-05,
+      "loss": 0.9458,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337594700897977e-05,
+      "loss": 0.8735,
+      "step": 1176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833438945410226e-05,
+      "loss": 0.8989,
+      "step": 1177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331181400987647e-05,
+      "loss": 0.9502,
+      "step": 1178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327970542634333e-05,
+      "loss": 0.9429,
+      "step": 1179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324756880123473e-05,
+      "loss": 0.9419,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832154041453715e-05,
+      "loss": 0.8789,
+      "step": 1181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83183211469584e-05,
+      "loss": 0.8872,
+      "step": 1182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315099078471203e-05,
+      "loss": 0.8867,
+      "step": 1183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311874210160477e-05,
+      "loss": 0.8682,
+      "step": 1184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830864654311209e-05,
+      "loss": 0.8916,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830541607841284e-05,
+      "loss": 0.7837,
+      "step": 1186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302182817150485e-05,
+      "loss": 0.8901,
+      "step": 1187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298946760413702e-05,
+      "loss": 0.8276,
+      "step": 1188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295707909292133e-05,
+      "loss": 0.8643,
+      "step": 1189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292466264876342e-05,
+      "loss": 0.8306,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289221828257847e-05,
+      "loss": 0.9492,
+      "step": 1191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285974600529097e-05,
+      "loss": 0.8491,
+      "step": 1192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282724582783485e-05,
+      "loss": 0.9053,
+      "step": 1193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827947177611534e-05,
+      "loss": 0.8667,
+      "step": 1194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827621618161994e-05,
+      "loss": 0.8833,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272957800393485e-05,
+      "loss": 0.9326,
+      "step": 1196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826969663353313e-05,
+      "loss": 0.9175,
+      "step": 1197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266432682136957e-05,
+      "loss": 0.8896,
+      "step": 1198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263165947303992e-05,
+      "loss": 0.8965,
+      "step": 1199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259896430134193e-05,
+      "loss": 0.8765,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825662413172846e-05,
+      "loss": 0.8867,
+      "step": 1201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253349053188625e-05,
+      "loss": 0.8691,
+      "step": 1202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825007119561746e-05,
+      "loss": 0.8774,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246790560118665e-05,
+      "loss": 0.8418,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824350714779689e-05,
+      "loss": 0.877,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240220959757705e-05,
+      "loss": 0.8926,
+      "step": 1206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823693199710762e-05,
+      "loss": 0.8662,
+      "step": 1207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233640260954084e-05,
+      "loss": 0.8955,
+      "step": 1208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230345752405476e-05,
+      "loss": 0.9209,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227048472571104e-05,
+      "loss": 0.9009,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223748422561218e-05,
+      "loss": 0.895,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220445603486993e-05,
+      "loss": 0.8584,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217140016460542e-05,
+      "loss": 0.3457,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213831662594904e-05,
+      "loss": 0.855,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821052054300406e-05,
+      "loss": 0.9287,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820720665880291e-05,
+      "loss": 0.8809,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203890011107295e-05,
+      "loss": 0.9536,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200570601033977e-05,
+      "loss": 0.8975,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197248429700655e-05,
+      "loss": 0.8306,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193923498225957e-05,
+      "loss": 0.9185,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190595807729437e-05,
+      "loss": 0.8945,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187265359331582e-05,
+      "loss": 0.9434,
+      "step": 1222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818393215415381e-05,
+      "loss": 0.8799,
+      "step": 1223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180596193318454e-05,
+      "loss": 0.9023,
+      "step": 1224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817725747794879e-05,
+      "loss": 0.877,
+      "step": 1225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817391600916902e-05,
+      "loss": 0.3151,
+      "step": 1226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170571788104258e-05,
+      "loss": 0.8735,
+      "step": 1227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816722481588056e-05,
+      "loss": 0.8857,
+      "step": 1228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163875093624902e-05,
+      "loss": 0.9355,
+      "step": 1229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160522622465194e-05,
+      "loss": 0.8955,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157167403530263e-05,
+      "loss": 0.8032,
+      "step": 1231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815380943794986e-05,
+      "loss": 0.8916,
+      "step": 1232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150448726854665e-05,
+      "loss": 0.9448,
+      "step": 1233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147085271376285e-05,
+      "loss": 0.9316,
+      "step": 1234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143719072647243e-05,
+      "loss": 0.9058,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140350131800993e-05,
+      "loss": 0.8438,
+      "step": 1236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136978449971917e-05,
+      "loss": 0.958,
+      "step": 1237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81336040282953e-05,
+      "loss": 0.9038,
+      "step": 1238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813022686790737e-05,
+      "loss": 0.8799,
+      "step": 1239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812684696994527e-05,
+      "loss": 0.8809,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123464335547062e-05,
+      "loss": 0.9185,
+      "step": 1241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120078965851733e-05,
+      "loss": 0.856,
+      "step": 1242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116690861999187e-05,
+      "loss": 0.8687,
+      "step": 1243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811330002513026e-05,
+      "loss": 0.3303,
+      "step": 1244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109906456386682e-05,
+      "loss": 0.9297,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810651015691114e-05,
+      "loss": 0.854,
+      "step": 1246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103111127847212e-05,
+      "loss": 0.9214,
+      "step": 1247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099709370339403e-05,
+      "loss": 0.8896,
+      "step": 1248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096304885533142e-05,
+      "loss": 0.9009,
+      "step": 1249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092897674574765e-05,
+      "loss": 0.894,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089487738611542e-05,
+      "loss": 0.3087,
+      "step": 1251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086075078791646e-05,
+      "loss": 0.2983,
+      "step": 1252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082659696264176e-05,
+      "loss": 0.9473,
+      "step": 1253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807924159217914e-05,
+      "loss": 0.8872,
+      "step": 1254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807582076768747e-05,
+      "loss": 0.8911,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8072397223941015e-05,
+      "loss": 0.8125,
+      "step": 1256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068970962092524e-05,
+      "loss": 0.8735,
+      "step": 1257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065541983295687e-05,
+      "loss": 0.8701,
+      "step": 1258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062110288705083e-05,
+      "loss": 0.877,
+      "step": 1259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058675879476225e-05,
+      "loss": 0.8486,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055238756765527e-05,
+      "loss": 0.9253,
+      "step": 1261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051798921730323e-05,
+      "loss": 0.8857,
+      "step": 1262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804835637552886e-05,
+      "loss": 0.9072,
+      "step": 1263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044911119320302e-05,
+      "loss": 0.8638,
+      "step": 1264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041463154264705e-05,
+      "loss": 0.897,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038012481523074e-05,
+      "loss": 0.915,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034559102257286e-05,
+      "loss": 0.9028,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803110301763016e-05,
+      "loss": 0.8682,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027644228805402e-05,
+      "loss": 0.9004,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802418273694765e-05,
+      "loss": 0.8501,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802071854322244e-05,
+      "loss": 0.9004,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017251648796216e-05,
+      "loss": 0.811,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801378205483634e-05,
+      "loss": 0.8755,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801030976251108e-05,
+      "loss": 0.9312,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006834772989604e-05,
+      "loss": 0.9404,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003357087442e-05,
+      "loss": 0.9023,
+      "step": 1276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999876707039256e-05,
+      "loss": 0.8916,
+      "step": 1277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996393632953275e-05,
+      "loss": 0.3589,
+      "step": 1278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992907866356856e-05,
+      "loss": 0.9248,
+      "step": 1279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989419408423718e-05,
+      "loss": 0.938,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985928260328475e-05,
+      "loss": 0.8501,
+      "step": 1281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798243442324665e-05,
+      "loss": 0.8848,
+      "step": 1282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978937898354676e-05,
+      "loss": 0.9224,
+      "step": 1283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975438686829884e-05,
+      "loss": 0.9321,
+      "step": 1284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971936789850514e-05,
+      "loss": 0.8901,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796843220859571e-05,
+      "loss": 0.8901,
+      "step": 1286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964924944245523e-05,
+      "loss": 0.8608,
+      "step": 1287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79614149979809e-05,
+      "loss": 0.9517,
+      "step": 1288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957902370983692e-05,
+      "loss": 0.8896,
+      "step": 1289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795438706443666e-05,
+      "loss": 0.8604,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950869079523454e-05,
+      "loss": 0.8701,
+      "step": 1291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947348417428646e-05,
+      "loss": 0.9253,
+      "step": 1292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794382507933769e-05,
+      "loss": 0.8926,
+      "step": 1293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940299066436954e-05,
+      "loss": 0.8589,
+      "step": 1294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936770379913696e-05,
+      "loss": 0.8667,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933239020956083e-05,
+      "loss": 0.8916,
+      "step": 1296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792970499075318e-05,
+      "loss": 0.8911,
+      "step": 1297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926168290494947e-05,
+      "loss": 0.8799,
+      "step": 1298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792262892137225e-05,
+      "loss": 0.8809,
+      "step": 1299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919086884576843e-05,
+      "loss": 0.8198,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791554218130139e-05,
+      "loss": 0.8896,
+      "step": 1301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791199481273945e-05,
+      "loss": 0.9028,
+      "step": 1302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908444780085472e-05,
+      "loss": 0.8301,
+      "step": 1303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904892084534813e-05,
+      "loss": 0.897,
+      "step": 1304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901336727283714e-05,
+      "loss": 0.9414,
+      "step": 1305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897778709529325e-05,
+      "loss": 0.8594,
+      "step": 1306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894218032469685e-05,
+      "loss": 0.8794,
+      "step": 1307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890654697303725e-05,
+      "loss": 0.8823,
+      "step": 1308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788708870523128e-05,
+      "loss": 0.9141,
+      "step": 1309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788352005745308e-05,
+      "loss": 0.8359,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787994875517073e-05,
+      "loss": 0.9106,
+      "step": 1311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787637479958676e-05,
+      "loss": 0.3723,
+      "step": 1312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872798191904564e-05,
+      "loss": 0.8936,
+      "step": 1313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869218933328448e-05,
+      "loss": 0.894,
+      "step": 1314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865637025063598e-05,
+      "loss": 0.8574,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862052468316105e-05,
+      "loss": 0.9224,
+      "step": 1316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858465264292944e-05,
+      "loss": 0.9146,
+      "step": 1317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785487541420198e-05,
+      "loss": 0.9556,
+      "step": 1318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785128291925197e-05,
+      "loss": 0.3406,
+      "step": 1319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784768778065257e-05,
+      "loss": 0.9355,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844089999614312e-05,
+      "loss": 0.9053,
+      "step": 1321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784048957734863e-05,
+      "loss": 0.8726,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783688651506784e-05,
+      "loss": 0.9331,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833280813985147e-05,
+      "loss": 0.9224,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829672475314656e-05,
+      "loss": 0.8633,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782606150027134e-05,
+      "loss": 0.8711,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822447890071073e-05,
+      "loss": 0.8545,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818831645930616e-05,
+      "loss": 0.8643,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815212769067615e-05,
+      "loss": 0.8442,
+      "step": 1329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811591260700606e-05,
+      "loss": 0.9053,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807967122048997e-05,
+      "loss": 0.8672,
+      "step": 1331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78043403543331e-05,
+      "loss": 0.8647,
+      "step": 1332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800710958774104e-05,
+      "loss": 0.9458,
+      "step": 1333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779707893659408e-05,
+      "loss": 0.8628,
+      "step": 1334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793444289015992e-05,
+      "loss": 0.8994,
+      "step": 1335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789807017263673e-05,
+      "loss": 0.3208,
+      "step": 1336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786167122561858e-05,
+      "loss": 0.8701,
+      "step": 1337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778252460613615e-05,
+      "loss": 0.9277,
+      "step": 1338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777887946921305e-05,
+      "loss": 0.8945,
+      "step": 1339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777523171301992e-05,
+      "loss": 0.9185,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7771581338785026e-05,
+      "loss": 0.9204,
+      "step": 1341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77679283477375e-05,
+      "loss": 0.8481,
+      "step": 1342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776427274110737e-05,
+      "loss": 0.8906,
+      "step": 1343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760614520125523e-05,
+      "loss": 0.8936,
+      "step": 1344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775695368602375e-05,
+      "loss": 0.8828,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77532902400347e-05,
+      "loss": 0.8706,
+      "step": 1346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749624183391926e-05,
+      "loss": 0.8936,
+      "step": 1347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745955517329837e-05,
+      "loss": 0.8926,
+      "step": 1348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742284243083735e-05,
+      "loss": 0.8989,
+      "step": 1349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773861036188979e-05,
+      "loss": 0.832,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734933874985057e-05,
+      "loss": 0.8867,
+      "step": 1351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773125478360747e-05,
+      "loss": 0.9087,
+      "step": 1352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727573088995833e-05,
+      "loss": 0.875,
+      "step": 1353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772388879238983e-05,
+      "loss": 0.8027,
+      "step": 1354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772020189503002e-05,
+      "loss": 0.3474,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716512398157847e-05,
+      "loss": 0.9019,
+      "step": 1356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712820303015607e-05,
+      "loss": 0.3599,
+      "step": 1357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709125610846502e-05,
+      "loss": 0.8794,
+      "step": 1358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705428322894582e-05,
+      "loss": 0.8828,
+      "step": 1359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701728440404786e-05,
+      "loss": 0.8979,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769802596462292e-05,
+      "loss": 0.8379,
+      "step": 1361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7694320896795668e-05,
+      "loss": 0.9092,
+      "step": 1362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690613238170583e-05,
+      "loss": 0.8721,
+      "step": 1363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768690298999609e-05,
+      "loss": 0.8911,
+      "step": 1364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683190153521486e-05,
+      "loss": 0.8403,
+      "step": 1365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679474729996946e-05,
+      "loss": 0.3845,
+      "step": 1366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675756720673508e-05,
+      "loss": 0.8608,
+      "step": 1367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672036126803083e-05,
+      "loss": 0.8457,
+      "step": 1368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668312949638454e-05,
+      "loss": 0.8545,
+      "step": 1369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664587190433272e-05,
+      "loss": 0.9302,
+      "step": 1370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766085885044206e-05,
+      "loss": 0.9419,
+      "step": 1371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765712793092021e-05,
+      "loss": 0.8198,
+      "step": 1372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653394433123975e-05,
+      "loss": 0.8784,
+      "step": 1373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649658358310485e-05,
+      "loss": 0.896,
+      "step": 1374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645919707737734e-05,
+      "loss": 0.877,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642178482664586e-05,
+      "loss": 0.8618,
+      "step": 1376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763843468435077e-05,
+      "loss": 0.8696,
+      "step": 1377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634688314056875e-05,
+      "loss": 0.8364,
+      "step": 1378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630939373044373e-05,
+      "loss": 0.8755,
+      "step": 1379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762718786257558e-05,
+      "loss": 0.959,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623433783913694e-05,
+      "loss": 0.8833,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761967713832277e-05,
+      "loss": 0.8848,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615917927067734e-05,
+      "loss": 0.8174,
+      "step": 1383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612156151414366e-05,
+      "loss": 0.8594,
+      "step": 1384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608391812629313e-05,
+      "loss": 0.8916,
+      "step": 1385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760462491198009e-05,
+      "loss": 0.9185,
+      "step": 1386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760085545073507e-05,
+      "loss": 0.8569,
+      "step": 1387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597083430163488e-05,
+      "loss": 0.8979,
+      "step": 1388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593308851535447e-05,
+      "loss": 0.8882,
+      "step": 1389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75895317161219e-05,
+      "loss": 0.8799,
+      "step": 1390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585752025194675e-05,
+      "loss": 0.897,
+      "step": 1391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581969780026448e-05,
+      "loss": 0.9351,
+      "step": 1392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7578184981890757e-05,
+      "loss": 0.8892,
+      "step": 1393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574397632062007e-05,
+      "loss": 0.9131,
+      "step": 1394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7570607731815457e-05,
+      "loss": 0.8477,
+      "step": 1395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566815282427226e-05,
+      "loss": 0.8975,
+      "step": 1396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756302028517429e-05,
+      "loss": 0.8677,
+      "step": 1397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559222741334484e-05,
+      "loss": 0.8833,
+      "step": 1398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75554226521865e-05,
+      "loss": 0.9155,
+      "step": 1399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551620019009887e-05,
+      "loss": 0.8843,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547814843085055e-05,
+      "loss": 0.9229,
+      "step": 1401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544007125693258e-05,
+      "loss": 0.8848,
+      "step": 1402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540196868116618e-05,
+      "loss": 0.834,
+      "step": 1403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753638407163811e-05,
+      "loss": 0.8657,
+      "step": 1404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753256873754156e-05,
+      "loss": 0.9058,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752875086711165e-05,
+      "loss": 0.9082,
+      "step": 1406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752493046163392e-05,
+      "loss": 0.9302,
+      "step": 1407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521107522394753e-05,
+      "loss": 0.8794,
+      "step": 1408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75172820506814e-05,
+      "loss": 0.386,
+      "step": 1409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513454047781954e-05,
+      "loss": 0.873,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750962351498536e-05,
+      "loss": 0.7925,
+      "step": 1411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505790453581427e-05,
+      "loss": 0.8525,
+      "step": 1412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75019548648608e-05,
+      "loss": 0.8828,
+      "step": 1413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498116750114975e-05,
+      "loss": 0.8872,
+      "step": 1414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494276110636317e-05,
+      "loss": 0.8423,
+      "step": 1415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490432947718024e-05,
+      "loss": 0.8809,
+      "step": 1416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748658726265415e-05,
+      "loss": 0.3125,
+      "step": 1417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482739056739594e-05,
+      "loss": 0.9189,
+      "step": 1418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7478888331270113e-05,
+      "loss": 0.874,
+      "step": 1419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74750350875423e-05,
+      "loss": 0.8765,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471179326853608e-05,
+      "loss": 0.8921,
+      "step": 1421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467321050502322e-05,
+      "loss": 0.9058,
+      "step": 1422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74634602597876e-05,
+      "loss": 0.9238,
+      "step": 1423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459596956009415e-05,
+      "loss": 0.8965,
+      "step": 1424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745573114046861e-05,
+      "loss": 0.9165,
+      "step": 1425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745186281446686e-05,
+      "loss": 0.9307,
+      "step": 1426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447991979306698e-05,
+      "loss": 0.9014,
+      "step": 1427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444118636291486e-05,
+      "loss": 0.8794,
+      "step": 1428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440242786725446e-05,
+      "loss": 0.897,
+      "step": 1429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436364431913636e-05,
+      "loss": 0.8574,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432483573161954e-05,
+      "loss": 0.8696,
+      "step": 1431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742860021177715e-05,
+      "loss": 0.8618,
+      "step": 1432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742471434906681e-05,
+      "loss": 0.8906,
+      "step": 1433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420825986339363e-05,
+      "loss": 0.9731,
+      "step": 1434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416935124904087e-05,
+      "loss": 0.8804,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741304176607109e-05,
+      "loss": 0.9019,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740914591115133e-05,
+      "loss": 0.876,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405247561456605e-05,
+      "loss": 0.323,
+      "step": 1438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401346718299543e-05,
+      "loss": 0.9097,
+      "step": 1439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397443382993623e-05,
+      "loss": 0.9321,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393537556853158e-05,
+      "loss": 0.9688,
+      "step": 1441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389629241193295e-05,
+      "loss": 0.9004,
+      "step": 1442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385718437330037e-05,
+      "loss": 0.8887,
+      "step": 1443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381805146580204e-05,
+      "loss": 0.8511,
+      "step": 1444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377889370261464e-05,
+      "loss": 0.321,
+      "step": 1445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737397110969232e-05,
+      "loss": 0.8359,
+      "step": 1446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737005036619211e-05,
+      "loss": 0.8462,
+      "step": 1447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736612714108101e-05,
+      "loss": 0.3062,
+      "step": 1448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362201435680037e-05,
+      "loss": 0.875,
+      "step": 1449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358273251311032e-05,
+      "loss": 0.8491,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735434258929667e-05,
+      "loss": 0.8726,
+      "step": 1451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350409450960474e-05,
+      "loss": 0.8438,
+      "step": 1452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346473837626796e-05,
+      "loss": 0.8286,
+      "step": 1453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734253575062081e-05,
+      "loss": 0.8701,
+      "step": 1454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338595191268534e-05,
+      "loss": 0.9116,
+      "step": 1455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733465216089682e-05,
+      "loss": 0.8984,
+      "step": 1456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330706660833343e-05,
+      "loss": 0.8853,
+      "step": 1457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732675869240662e-05,
+      "loss": 0.8257,
+      "step": 1458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322808256945982e-05,
+      "loss": 0.8521,
+      "step": 1459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318855355781612e-05,
+      "loss": 0.9189,
+      "step": 1460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314899990244514e-05,
+      "loss": 0.8848,
+      "step": 1461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310942161666516e-05,
+      "loss": 0.8555,
+      "step": 1462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306981871380286e-05,
+      "loss": 0.8765,
+      "step": 1463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303019120719307e-05,
+      "loss": 0.3551,
+      "step": 1464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729905391101791e-05,
+      "loss": 0.9097,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295086243611238e-05,
+      "loss": 0.8823,
+      "step": 1466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291116119835262e-05,
+      "loss": 0.9497,
+      "step": 1467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728714354102679e-05,
+      "loss": 0.8936,
+      "step": 1468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728316850852345e-05,
+      "loss": 0.8291,
+      "step": 1469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279191023663696e-05,
+      "loss": 0.9141,
+      "step": 1470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275211087786813e-05,
+      "loss": 0.8716,
+      "step": 1471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72712287022329e-05,
+      "loss": 0.345,
+      "step": 1472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72672438683429e-05,
+      "loss": 0.8975,
+      "step": 1473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726325658745856e-05,
+      "loss": 0.8779,
+      "step": 1474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259266860922458e-05,
+      "loss": 0.9473,
+      "step": 1475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255274690078002e-05,
+      "loss": 0.9531,
+      "step": 1476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251280076269415e-05,
+      "loss": 0.8477,
+      "step": 1477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724728302084175e-05,
+      "loss": 0.8887,
+      "step": 1478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243283525140876e-05,
+      "loss": 0.9209,
+      "step": 1479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239281590513486e-05,
+      "loss": 0.9263,
+      "step": 1480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235277218307094e-05,
+      "loss": 0.8066,
+      "step": 1481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723127040987003e-05,
+      "loss": 0.8779,
+      "step": 1482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722726116655145e-05,
+      "loss": 0.9023,
+      "step": 1483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223249489701336e-05,
+      "loss": 0.915,
+      "step": 1484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721923538067048e-05,
+      "loss": 0.9033,
+      "step": 1485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215218840810487e-05,
+      "loss": 0.8721,
+      "step": 1486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7211199871473794e-05,
+      "loss": 0.897,
+      "step": 1487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720717847401365e-05,
+      "loss": 0.8882,
+      "step": 1488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203154649784126e-05,
+      "loss": 0.8813,
+      "step": 1489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199128400140098e-05,
+      "loss": 0.9136,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719509972643728e-05,
+      "loss": 0.8608,
+      "step": 1491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191068630032174e-05,
+      "loss": 0.8936,
+      "step": 1492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187035112282124e-05,
+      "loss": 0.9248,
+      "step": 1493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182999174545276e-05,
+      "loss": 0.8628,
+      "step": 1494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178960818180593e-05,
+      "loss": 0.3362,
+      "step": 1495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174920044547846e-05,
+      "loss": 0.8135,
+      "step": 1496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170876855007637e-05,
+      "loss": 0.8306,
+      "step": 1497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166831250921365e-05,
+      "loss": 0.3569,
+      "step": 1498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716278323365125e-05,
+      "loss": 0.9292,
+      "step": 1499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715873280456032e-05,
+      "loss": 0.957,
+      "step": 1500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715467996501242e-05,
+      "loss": 0.8711,
+      "step": 1501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150624716372203e-05,
+      "loss": 0.9272,
+      "step": 1502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714656706000513e-05,
+      "loss": 0.8647,
+      "step": 1503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714250699727749e-05,
+      "loss": 0.8691,
+      "step": 1504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138444529556358e-05,
+      "loss": 0.8896,
+      "step": 1505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134379658209634e-05,
+      "loss": 0.3069,
+      "step": 1506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130312384606023e-05,
+      "loss": 0.9023,
+      "step": 1507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712624271011504e-05,
+      "loss": 0.8955,
+      "step": 1508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712217063610701e-05,
+      "loss": 0.8384,
+      "step": 1509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118096163953056e-05,
+      "loss": 0.8945,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114019295025127e-05,
+      "loss": 0.8735,
+      "step": 1511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109940030695962e-05,
+      "loss": 0.9404,
+      "step": 1512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105858372339112e-05,
+      "loss": 0.9443,
+      "step": 1513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710177432132894e-05,
+      "loss": 0.8979,
+      "step": 1514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097687879040606e-05,
+      "loss": 0.8979,
+      "step": 1515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709359904685008e-05,
+      "loss": 0.8657,
+      "step": 1516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7089507826134136e-05,
+      "loss": 0.9326,
+      "step": 1517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7085414218270347e-05,
+      "loss": 0.8901,
+      "step": 1518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70813182246371e-05,
+      "loss": 0.9312,
+      "step": 1519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707721984661358e-05,
+      "loss": 0.8394,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073119085579773e-05,
+      "loss": 0.9229,
+      "step": 1521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706901594291647e-05,
+      "loss": 0.8755,
+      "step": 1522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7064910420005262e-05,
+      "loss": 0.8589,
+      "step": 1523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706080251822854e-05,
+      "loss": 0.8794,
+      "step": 1524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056692238969506e-05,
+      "loss": 0.8721,
+      "step": 1525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052579583612147e-05,
+      "loss": 0.3706,
+      "step": 1526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048464553541265e-05,
+      "loss": 0.8901,
+      "step": 1527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044347150142452e-05,
+      "loss": 0.8838,
+      "step": 1528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70402273748021e-05,
+      "loss": 0.8301,
+      "step": 1529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036105228907406e-05,
+      "loss": 0.3384,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031980713846358e-05,
+      "loss": 0.9326,
+      "step": 1531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027853831007743e-05,
+      "loss": 0.9038,
+      "step": 1532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702372458178115e-05,
+      "loss": 0.8833,
+      "step": 1533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701959296755696e-05,
+      "loss": 0.8496,
+      "step": 1534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701545898972635e-05,
+      "loss": 0.301,
+      "step": 1535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011322649681302e-05,
+      "loss": 0.896,
+      "step": 1536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007183948814577e-05,
+      "loss": 0.896,
+      "step": 1537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003042888519745e-05,
+      "loss": 0.8335,
+      "step": 1538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998899470191165e-05,
+      "loss": 0.9478,
+      "step": 1539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699475369522399e-05,
+      "loss": 0.9346,
+      "step": 1540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990605565014168e-05,
+      "loss": 0.8857,
+      "step": 1541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6986455080958438e-05,
+      "loss": 0.856,
+      "step": 1542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982302244454332e-05,
+      "loss": 0.8457,
+      "step": 1543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978147056900178e-05,
+      "loss": 0.8936,
+      "step": 1544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697398951969509e-05,
+      "loss": 0.8672,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6969829634238976e-05,
+      "loss": 0.9077,
+      "step": 1546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965667401932535e-05,
+      "loss": 0.8931,
+      "step": 1547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961502824177256e-05,
+      "loss": 0.8745,
+      "step": 1548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957335902375414e-05,
+      "loss": 0.9214,
+      "step": 1549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695316663793008e-05,
+      "loss": 0.8716,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694899503224511e-05,
+      "loss": 0.9014,
+      "step": 1551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944821086725144e-05,
+      "loss": 0.8892,
+      "step": 1552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940644802775624e-05,
+      "loss": 0.8604,
+      "step": 1553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693646618180276e-05,
+      "loss": 0.8975,
+      "step": 1554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932285225213565e-05,
+      "loss": 0.9185,
+      "step": 1555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692810193441583e-05,
+      "loss": 0.8574,
+      "step": 1556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692391631081814e-05,
+      "loss": 0.8813,
+      "step": 1557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919728355829844e-05,
+      "loss": 0.9116,
+      "step": 1558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691553807086111e-05,
+      "loss": 0.8901,
+      "step": 1559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911345457322865e-05,
+      "loss": 0.9224,
+      "step": 1560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907150516626824e-05,
+      "loss": 0.8569,
+      "step": 1561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690295325018549e-05,
+      "loss": 0.8306,
+      "step": 1562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898753659412154e-05,
+      "loss": 0.9038,
+      "step": 1563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894551745720876e-05,
+      "loss": 0.9233,
+      "step": 1564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890347510526512e-05,
+      "loss": 0.7988,
+      "step": 1565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688614095524469e-05,
+      "loss": 0.8921,
+      "step": 1566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688193208129182e-05,
+      "loss": 0.8848,
+      "step": 1567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877720890085105e-05,
+      "loss": 0.8057,
+      "step": 1568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687350738304251e-05,
+      "loss": 0.8511,
+      "step": 1569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6869291561582793e-05,
+      "loss": 0.8633,
+      "step": 1570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865073427125486e-05,
+      "loss": 0.8975,
+      "step": 1571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68608529810909e-05,
+      "loss": 0.8687,
+      "step": 1572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685663022490012e-05,
+      "loss": 0.345,
+      "step": 1573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852405159975025e-05,
+      "loss": 0.8252,
+      "step": 1574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848177787738255e-05,
+      "loss": 0.855,
+      "step": 1575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6843948109613227e-05,
+      "loss": 0.8862,
+      "step": 1576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6839716127024145e-05,
+      "loss": 0.9038,
+      "step": 1577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683548184139598e-05,
+      "loss": 0.8403,
+      "step": 1578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831245254154488e-05,
+      "loss": 0.8833,
+      "step": 1579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827006366726187e-05,
+      "loss": 0.8369,
+      "step": 1580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822765180538386e-05,
+      "loss": 0.9067,
+      "step": 1581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6818521697019146e-05,
+      "loss": 0.853,
+      "step": 1582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681427591759732e-05,
+      "loss": 0.8145,
+      "step": 1583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810027843702533e-05,
+      "loss": 0.8501,
+      "step": 1584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805777476765174e-05,
+      "loss": 0.8955,
+      "step": 1585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6801524818216406e-05,
+      "loss": 0.8916,
+      "step": 1586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797269869488165e-05,
+      "loss": 0.8945,
+      "step": 1587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793012632013162e-05,
+      "loss": 0.8887,
+      "step": 1588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678875310722487e-05,
+      "loss": 0.8618,
+      "step": 1589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784491296557545e-05,
+      "loss": 0.8242,
+      "step": 1590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67802272014462e-05,
+      "loss": 0.8975,
+      "step": 1591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677596082332662e-05,
+      "loss": 0.8765,
+      "step": 1592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6771692163635367e-05,
+      "loss": 0.8843,
+      "step": 1593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767421223809766e-05,
+      "loss": 0.8271,
+      "step": 1594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67631480052879e-05,
+      "loss": 0.8623,
+      "step": 1595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758872509508637e-05,
+      "loss": 0.8618,
+      "step": 1596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6754594737911598e-05,
+      "loss": 0.8687,
+      "step": 1597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675031469193718e-05,
+      "loss": 0.9067,
+      "step": 1598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674603237302654e-05,
+      "loss": 0.8989,
+      "step": 1599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741747782621595e-05,
+      "loss": 0.853,
+      "step": 1600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673746092216504e-05,
+      "loss": 0.8877,
+      "step": 1601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673317179310033e-05,
+      "loss": 0.853,
+      "step": 1602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728880396871677e-05,
+      "loss": 0.9155,
+      "step": 1603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672458673492406e-05,
+      "loss": 0.9131,
+      "step": 1604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6720290808703228e-05,
+      "loss": 0.8525,
+      "step": 1605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671599261965568e-05,
+      "loss": 0.8887,
+      "step": 1606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711692169228685e-05,
+      "loss": 0.8403,
+      "step": 1607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707389458870272e-05,
+      "loss": 0.8599,
+      "step": 1608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703084490029233e-05,
+      "loss": 0.8794,
+      "step": 1609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6698777264155116e-05,
+      "loss": 0.9341,
+      "step": 1610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669446778269823e-05,
+      "loss": 0.8706,
+      "step": 1611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6690156047109646e-05,
+      "loss": 0.9175,
+      "step": 1612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685842058841194e-05,
+      "loss": 0.8926,
+      "step": 1613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681525819345457e-05,
+      "loss": 0.8867,
+      "step": 1614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677207330075783e-05,
+      "loss": 0.8848,
+      "step": 1615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672886592486278e-05,
+      "loss": 0.8887,
+      "step": 1616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6668563608031797e-05,
+      "loss": 0.916,
+      "step": 1617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666423837816796e-05,
+      "loss": 0.9072,
+      "step": 1618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665991090435113e-05,
+      "loss": 0.8862,
+      "step": 1619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655581188038446e-05,
+      "loss": 0.8979,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651249230687788e-05,
+      "loss": 0.8984,
+      "step": 1621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664691503375779e-05,
+      "loss": 0.8247,
+      "step": 1622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664257859870785e-05,
+      "loss": 0.3428,
+      "step": 1623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638239926998105e-05,
+      "loss": 0.8735,
+      "step": 1624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663389902008946e-05,
+      "loss": 0.8887,
+      "step": 1625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629555879443563e-05,
+      "loss": 0.7812,
+      "step": 1626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625210506522814e-05,
+      "loss": 0.8677,
+      "step": 1627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662086290279038e-05,
+      "loss": 0.8374,
+      "step": 1628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616513069710152e-05,
+      "loss": 0.8696,
+      "step": 1629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612161008746798e-05,
+      "loss": 0.8638,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607806721365716e-05,
+      "loss": 0.9937,
+      "step": 1631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603450209033072e-05,
+      "loss": 0.8687,
+      "step": 1632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599091473215766e-05,
+      "loss": 0.8262,
+      "step": 1633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594730515381446e-05,
+      "loss": 0.8657,
+      "step": 1634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659036733699853e-05,
+      "loss": 0.8018,
+      "step": 1635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586001939536155e-05,
+      "loss": 0.8516,
+      "step": 1636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658163432446422e-05,
+      "loss": 0.856,
+      "step": 1637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6577264493253377e-05,
+      "loss": 0.9604,
+      "step": 1638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657289244737501e-05,
+      "loss": 0.8799,
+      "step": 1639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568518188301253e-05,
+      "loss": 0.9033,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564141717504993e-05,
+      "loss": 0.8462,
+      "step": 1641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559763036459855e-05,
+      "loss": 0.9209,
+      "step": 1642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555382146640205e-05,
+      "loss": 0.9131,
+      "step": 1643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6550999049521163e-05,
+      "loss": 0.8774,
+      "step": 1644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6546613746578577e-05,
+      "loss": 0.8696,
+      "step": 1645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542226239289057e-05,
+      "loss": 0.9336,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653783652912994e-05,
+      "loss": 0.9111,
+      "step": 1647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653344461757931e-05,
+      "loss": 0.9326,
+      "step": 1648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529050506115993e-05,
+      "loss": 0.8774,
+      "step": 1649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524654196219555e-05,
+      "loss": 0.9229,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520255689370302e-05,
+      "loss": 0.8965,
+      "step": 1651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651585498704928e-05,
+      "loss": 0.8843,
+      "step": 1652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651145209073827e-05,
+      "loss": 0.8467,
+      "step": 1653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507047001919802e-05,
+      "loss": 0.9375,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502639722077133e-05,
+      "loss": 0.8936,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498230252694268e-05,
+      "loss": 0.8784,
+      "step": 1656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493818595255945e-05,
+      "loss": 0.8916,
+      "step": 1657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648940475124763e-05,
+      "loss": 0.3506,
+      "step": 1658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484988722155537e-05,
+      "loss": 0.8745,
+      "step": 1659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480570509466615e-05,
+      "loss": 0.8838,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647615011466854e-05,
+      "loss": 0.8506,
+      "step": 1661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647172753924973e-05,
+      "loss": 0.8452,
+      "step": 1662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467302784699338e-05,
+      "loss": 0.8545,
+      "step": 1663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646287585250724e-05,
+      "loss": 0.3542,
+      "step": 1664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458446744164063e-05,
+      "loss": 0.8848,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454015461161152e-05,
+      "loss": 0.8667,
+      "step": 1666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644958200499059e-05,
+      "loss": 0.8428,
+      "step": 1667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644514637714519e-05,
+      "loss": 0.8945,
+      "step": 1668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440708579118495e-05,
+      "loss": 0.9077,
+      "step": 1669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436268612404783e-05,
+      "loss": 0.8374,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643182647849906e-05,
+      "loss": 0.9009,
+      "step": 1671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427382178897064e-05,
+      "loss": 0.8887,
+      "step": 1672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422935715095257e-05,
+      "loss": 0.8462,
+      "step": 1673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418487088590834e-05,
+      "loss": 0.916,
+      "step": 1674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414036300881716e-05,
+      "loss": 0.8745,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409583353466554e-05,
+      "loss": 0.9014,
+      "step": 1676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640512824784472e-05,
+      "loss": 0.8345,
+      "step": 1677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400670985516324e-05,
+      "loss": 0.835,
+      "step": 1678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396211567982196e-05,
+      "loss": 0.3156,
+      "step": 1679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391749996743884e-05,
+      "loss": 0.8804,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387286273303674e-05,
+      "loss": 0.8169,
+      "step": 1681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638282039916457e-05,
+      "loss": 0.8447,
+      "step": 1682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6378352375830305e-05,
+      "loss": 0.8276,
+      "step": 1683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373882204805326e-05,
+      "loss": 0.8335,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369409887594815e-05,
+      "loss": 0.855,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364935425704667e-05,
+      "loss": 0.853,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360458820641503e-05,
+      "loss": 0.3369,
+      "step": 1687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635598007391267e-05,
+      "loss": 0.8867,
+      "step": 1688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6351499187026228e-05,
+      "loss": 0.8774,
+      "step": 1689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634701616149096e-05,
+      "loss": 0.8745,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6342530998816376e-05,
+      "loss": 0.8716,
+      "step": 1691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63380437005127e-05,
+      "loss": 0.8452,
+      "step": 1692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333554268090867e-05,
+      "loss": 0.9033,
+      "step": 1693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329062703062546e-05,
+      "loss": 0.8672,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324569006940115e-05,
+      "loss": 0.8916,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320073181236674e-05,
+      "loss": 0.8867,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315575227466034e-05,
+      "loss": 0.9097,
+      "step": 1697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631107514714273e-05,
+      "loss": 0.8804,
+      "step": 1698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306572941782003e-05,
+      "loss": 0.853,
+      "step": 1699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302068612899827e-05,
+      "loss": 0.8345,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297562162012867e-05,
+      "loss": 0.9316,
+      "step": 1701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629305359063852e-05,
+      "loss": 0.8882,
+      "step": 1702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288542900294897e-05,
+      "loss": 0.9277,
+      "step": 1703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284030092500813e-05,
+      "loss": 0.9331,
+      "step": 1704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62795151687758e-05,
+      "loss": 0.7681,
+      "step": 1705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274998130640108e-05,
+      "loss": 0.9248,
+      "step": 1706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627047897961469e-05,
+      "loss": 0.9263,
+      "step": 1707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265957717221213e-05,
+      "loss": 0.9106,
+      "step": 1708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261434344982062e-05,
+      "loss": 0.8315,
+      "step": 1709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256908864420323e-05,
+      "loss": 0.9038,
+      "step": 1710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62523812770598e-05,
+      "loss": 0.8984,
+      "step": 1711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247851584424995e-05,
+      "loss": 0.8706,
+      "step": 1712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243319788041134e-05,
+      "loss": 0.9072,
+      "step": 1713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623878588943414e-05,
+      "loss": 0.8589,
+      "step": 1714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234249890130646e-05,
+      "loss": 0.8071,
+      "step": 1715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6229711791657997e-05,
+      "loss": 0.8022,
+      "step": 1716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622517159554424e-05,
+      "loss": 0.8823,
+      "step": 1717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622062930331813e-05,
+      "loss": 0.9019,
+      "step": 1718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6216084916509122e-05,
+      "loss": 0.9148,
+      "step": 1719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211538436647388e-05,
+      "loss": 0.916,
+      "step": 1720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206989865263802e-05,
+      "loss": 0.9331,
+      "step": 1721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620243920388993e-05,
+      "loss": 0.8628,
+      "step": 1722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619788645405806e-05,
+      "loss": 0.8711,
+      "step": 1723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193331617301167e-05,
+      "loss": 0.8833,
+      "step": 1724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618877469515294e-05,
+      "loss": 0.9336,
+      "step": 1725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618421568914776e-05,
+      "loss": 0.8945,
+      "step": 1726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617965460082072e-05,
+      "loss": 0.8906,
+      "step": 1727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175091431707607e-05,
+      "loss": 0.9092,
+      "step": 1728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170526183344915e-05,
+      "loss": 0.9102,
+      "step": 1729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616595885726983e-05,
+      "loss": 0.3516,
+      "step": 1730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161389455020245e-05,
+      "loss": 0.855,
+      "step": 1731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615681797813475e-05,
+      "loss": 0.917,
+      "step": 1732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615224442815263e-05,
+      "loss": 0.3384,
+      "step": 1733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614766880661387e-05,
+      "loss": 0.9229,
+      "step": 1734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143091115059153e-05,
+      "loss": 0.8384,
+      "step": 1735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6138511355029857e-05,
+      "loss": 0.8965,
+      "step": 1736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613392952806807e-05,
+      "loss": 0.853,
+      "step": 1737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129345635716555e-05,
+      "loss": 0.8735,
+      "step": 1738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124759679518775e-05,
+      "loss": 0.8584,
+      "step": 1739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120171661018904e-05,
+      "loss": 0.8765,
+      "step": 1740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115581581761795e-05,
+      "loss": 0.8491,
+      "step": 1741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110989443292998e-05,
+      "loss": 0.9111,
+      "step": 1742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610639524715876e-05,
+      "loss": 0.8857,
+      "step": 1743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101798994906018e-05,
+      "loss": 0.875,
+      "step": 1744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097200688082396e-05,
+      "loss": 0.9214,
+      "step": 1745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092600328236227e-05,
+      "loss": 0.8218,
+      "step": 1746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087997916916514e-05,
+      "loss": 0.9277,
+      "step": 1747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083393455672965e-05,
+      "loss": 0.8491,
+      "step": 1748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078786946055975e-05,
+      "loss": 0.894,
+      "step": 1749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074178389616625e-05,
+      "loss": 0.8525,
+      "step": 1750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069567787906682e-05,
+      "loss": 0.9141,
+      "step": 1751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064955142478626e-05,
+      "loss": 0.8247,
+      "step": 1752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6060340454885585e-05,
+      "loss": 0.8911,
+      "step": 1753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055723726681413e-05,
+      "loss": 0.8589,
+      "step": 1754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051104959420623e-05,
+      "loss": 0.771,
+      "step": 1755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604648415465843e-05,
+      "loss": 0.9097,
+      "step": 1756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041861313950732e-05,
+      "loss": 0.8999,
+      "step": 1757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037236438854114e-05,
+      "loss": 0.8594,
+      "step": 1758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603260953092584e-05,
+      "loss": 0.8306,
+      "step": 1759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027980591723856e-05,
+      "loss": 0.876,
+      "step": 1760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602334962280681e-05,
+      "loss": 0.8711,
+      "step": 1761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6018716625734016e-05,
+      "loss": 0.9409,
+      "step": 1762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601408160206548e-05,
+      "loss": 0.8657,
+      "step": 1763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009444553361878e-05,
+      "loss": 0.9136,
+      "step": 1764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600480548118458e-05,
+      "loss": 0.8716,
+      "step": 1765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600016438709564e-05,
+      "loss": 0.3276,
+      "step": 1766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995521272657782e-05,
+      "loss": 0.8643,
+      "step": 1767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990876139434414e-05,
+      "loss": 0.8584,
+      "step": 1768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986228988989624e-05,
+      "loss": 0.8354,
+      "step": 1769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981579822888187e-05,
+      "loss": 0.9331,
+      "step": 1770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976928642695542e-05,
+      "loss": 0.9072,
+      "step": 1771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972275449977818e-05,
+      "loss": 0.877,
+      "step": 1772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5967620246301817e-05,
+      "loss": 0.918,
+      "step": 1773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5962963033235016e-05,
+      "loss": 0.9102,
+      "step": 1774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958303812345577e-05,
+      "loss": 0.8994,
+      "step": 1775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953642585202325e-05,
+      "loss": 0.8804,
+      "step": 1776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948979353374775e-05,
+      "loss": 0.8159,
+      "step": 1777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944314118433103e-05,
+      "loss": 0.9653,
+      "step": 1778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939646881948175e-05,
+      "loss": 0.9185,
+      "step": 1779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934977645491513e-05,
+      "loss": 0.8628,
+      "step": 1780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930306410635327e-05,
+      "loss": 0.3792,
+      "step": 1781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592563317895249e-05,
+      "loss": 0.9121,
+      "step": 1782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920957952016563e-05,
+      "loss": 0.8364,
+      "step": 1783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916280731401753e-05,
+      "loss": 0.8989,
+      "step": 1784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5911601518682964e-05,
+      "loss": 0.8657,
+      "step": 1785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906920315435757e-05,
+      "loss": 0.9155,
+      "step": 1786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5902237123236365e-05,
+      "loss": 0.9434,
+      "step": 1787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589755194366169e-05,
+      "loss": 0.3757,
+      "step": 1788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892864778289308e-05,
+      "loss": 0.8789,
+      "step": 1789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888175628697462e-05,
+      "loss": 0.8081,
+      "step": 1790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588348449646506e-05,
+      "loss": 0.8916,
+      "step": 1791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878791383171678e-05,
+      "loss": 0.8716,
+      "step": 1792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874096290397562e-05,
+      "loss": 0.8462,
+      "step": 1793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586939921972362e-05,
+      "loss": 0.9106,
+      "step": 1794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864700172731432e-05,
+      "loss": 0.8384,
+      "step": 1795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859999151003245e-05,
+      "loss": 0.8481,
+      "step": 1796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855296156121958e-05,
+      "loss": 0.8662,
+      "step": 1797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850591189671144e-05,
+      "loss": 0.7773,
+      "step": 1798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584588425323504e-05,
+      "loss": 0.3762,
+      "step": 1799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584117534839855e-05,
+      "loss": 0.8496,
+      "step": 1800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836464476747227e-05,
+      "loss": 0.8613,
+      "step": 1801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5831751639867302e-05,
+      "loss": 0.9399,
+      "step": 1802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827036839345657e-05,
+      "loss": 0.8623,
+      "step": 1803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582232007676984e-05,
+      "loss": 0.8828,
+      "step": 1804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817601353728055e-05,
+      "loss": 0.3235,
+      "step": 1805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812880671809177e-05,
+      "loss": 0.8662,
+      "step": 1806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5808158032602726e-05,
+      "loss": 0.8667,
+      "step": 1807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580343343769889e-05,
+      "loss": 0.915,
+      "step": 1808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798706888688518e-05,
+      "loss": 0.8633,
+      "step": 1809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579397838716311e-05,
+      "loss": 0.8774,
+      "step": 1810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789247934714823e-05,
+      "loss": 0.8438,
+      "step": 1811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578451553293648e-05,
+      "loss": 0.8281,
+      "step": 1812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5779781183421548e-05,
+      "loss": 0.9199,
+      "step": 1813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577504488776416e-05,
+      "loss": 0.8916,
+      "step": 1814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5770306647559107e-05,
+      "loss": 0.9341,
+      "step": 1815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765566464401816e-05,
+      "loss": 0.3506,
+      "step": 1816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760824339888388e-05,
+      "loss": 0.8511,
+      "step": 1817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575608027561557e-05,
+      "loss": 0.8701,
+      "step": 1818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575133427318076e-05,
+      "loss": 0.853,
+      "step": 1819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574658633418201e-05,
+      "loss": 0.9038,
+      "step": 1820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574183646021803e-05,
+      "loss": 0.877,
+      "step": 1821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737084652888173e-05,
+      "loss": 0.894,
+      "step": 1822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573233091379245e-05,
+      "loss": 0.8516,
+      "step": 1823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572757524453151e-05,
+      "loss": 0.8745,
+      "step": 1824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572281764670667e-05,
+      "loss": 0.3271,
+      "step": 1825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571805812191989e-05,
+      "loss": 0.8267,
+      "step": 1826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5713296671773767e-05,
+      "loss": 0.8447,
+      "step": 1827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570853329787156e-05,
+      "loss": 0.8745,
+      "step": 1828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570376800181717e-05,
+      "loss": 0.9111,
+      "step": 1829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569900078521515e-05,
+      "loss": 0.9326,
+      "step": 1830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5694231649670694e-05,
+      "loss": 0.9292,
+      "step": 1831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568946059678964e-05,
+      "loss": 0.8174,
+      "step": 1832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684687628178482e-05,
+      "loss": 0.8774,
+      "step": 1833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5679912745444354e-05,
+      "loss": 0.8813,
+      "step": 1834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675135950195025e-05,
+      "loss": 0.9209,
+      "step": 1835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5670357244038924e-05,
+      "loss": 0.8721,
+      "step": 1836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5665576628585114e-05,
+      "loss": 0.8921,
+      "step": 1837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56607941054433e-05,
+      "loss": 0.8794,
+      "step": 1838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565600967622384e-05,
+      "loss": 0.9575,
+      "step": 1839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5651223342537718e-05,
+      "loss": 0.3691,
+      "step": 1840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646435105996567e-05,
+      "loss": 0.9121,
+      "step": 1841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564164496821267e-05,
+      "loss": 0.9014,
+      "step": 1842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636852930798923e-05,
+      "loss": 0.8643,
+      "step": 1843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632058995368905e-05,
+      "loss": 0.8638,
+      "step": 1844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627263163536784e-05,
+      "loss": 0.8428,
+      "step": 1845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622465436917408e-05,
+      "loss": 0.8867,
+      "step": 1846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561766581712624e-05,
+      "loss": 0.8335,
+      "step": 1847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612864305779388e-05,
+      "loss": 0.8379,
+      "step": 1848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608060904493597e-05,
+      "loss": 0.8491,
+      "step": 1849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603255614886245e-05,
+      "loss": 0.832,
+      "step": 1850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5598448438575348e-05,
+      "loss": 0.9346,
+      "step": 1851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593639377179556e-05,
+      "loss": 0.8604,
+      "step": 1852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588828432318158e-05,
+      "loss": 0.8569,
+      "step": 1853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584015605611074e-05,
+      "loss": 0.8945,
+      "step": 1854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579200898678855e-05,
+      "loss": 0.8901,
+      "step": 1855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557438431314269e-05,
+      "loss": 0.8643,
+      "step": 1856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55695658506244e-05,
+      "loss": 0.343,
+      "step": 1857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556474551274643e-05,
+      "loss": 0.8535,
+      "step": 1858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559923301131874e-05,
+      "loss": 0.8179,
+      "step": 1859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555099217404433e-05,
+      "loss": 0.9028,
+      "step": 1860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550273263188463e-05,
+      "loss": 0.8843,
+      "step": 1861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554544544010893e-05,
+      "loss": 0.9116,
+      "step": 1862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554061574979144e-05,
+      "loss": 0.8647,
+      "step": 1863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535784193862224e-05,
+      "loss": 0.9038,
+      "step": 1864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530950773948147e-05,
+      "loss": 0.8857,
+      "step": 1865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526115491676688e-05,
+      "loss": 0.835,
+      "step": 1866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552127834867597e-05,
+      "loss": 0.8672,
+      "step": 1867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551643934657473e-05,
+      "loss": 0.8628,
+      "step": 1868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551159848700234e-05,
+      "loss": 0.8999,
+      "step": 1869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5506755771588785e-05,
+      "loss": 0.8911,
+      "step": 1870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550191120196469e-05,
+      "loss": 0.8311,
+      "step": 1871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497064779761295e-05,
+      "loss": 0.8926,
+      "step": 1872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492216506610468e-05,
+      "loss": 0.8818,
+      "step": 1873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487366384144692e-05,
+      "loss": 0.8628,
+      "step": 1874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5482514413997085e-05,
+      "loss": 0.9038,
+      "step": 1875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477660597801376e-05,
+      "loss": 0.8354,
+      "step": 1876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472804937191927e-05,
+      "loss": 0.8999,
+      "step": 1877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546794743380371e-05,
+      "loss": 0.8516,
+      "step": 1878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546308808927232e-05,
+      "loss": 0.8447,
+      "step": 1879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5458226905233985e-05,
+      "loss": 0.9272,
+      "step": 1880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5453363883325526e-05,
+      "loss": 0.874,
+      "step": 1881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544849902518441e-05,
+      "loss": 0.9321,
+      "step": 1882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443632332448705e-05,
+      "loss": 0.8862,
+      "step": 1883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438763806757107e-05,
+      "loss": 0.8706,
+      "step": 1884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433893449748917e-05,
+      "loss": 0.8647,
+      "step": 1885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5429021263064064e-05,
+      "loss": 0.9082,
+      "step": 1886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424147248343096e-05,
+      "loss": 0.8672,
+      "step": 1887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541927140722716e-05,
+      "loss": 0.3362,
+      "step": 1888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414393741358032e-05,
+      "loss": 0.9272,
+      "step": 1889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54095142523781e-05,
+      "loss": 0.9067,
+      "step": 1890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404632941930354e-05,
+      "loss": 0.8389,
+      "step": 1891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539974981165842e-05,
+      "loss": 0.9038,
+      "step": 1892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539486486320652e-05,
+      "loss": 0.8555,
+      "step": 1893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538997809821949e-05,
+      "loss": 0.8657,
+      "step": 1894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538508951834278e-05,
+      "loss": 0.8916,
+      "step": 1895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380199125222454e-05,
+      "loss": 0.3396,
+      "step": 1896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375306920505186e-05,
+      "loss": 0.8242,
+      "step": 1897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370412905838244e-05,
+      "loss": 0.8481,
+      "step": 1898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536551708286954e-05,
+      "loss": 0.3323,
+      "step": 1899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360619453247556e-05,
+      "loss": 0.3257,
+      "step": 1900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355720018621403e-05,
+      "loss": 0.8569,
+      "step": 1901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535081878064081e-05,
+      "loss": 0.9111,
+      "step": 1902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5345915740956087e-05,
+      "loss": 0.874,
+      "step": 1903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534101090121817e-05,
+      "loss": 0.8638,
+      "step": 1904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533610426307859e-05,
+      "loss": 0.9448,
+      "step": 1905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331195828189493e-05,
+      "loss": 0.8403,
+      "step": 1906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326285598203624e-05,
+      "loss": 0.8345,
+      "step": 1907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321373574774337e-05,
+      "loss": 0.8867,
+      "step": 1908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316459759555578e-05,
+      "loss": 0.8208,
+      "step": 1909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531154415420192e-05,
+      "loss": 0.8794,
+      "step": 1910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530662676036851e-05,
+      "loss": 0.8457,
+      "step": 1911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530170757971112e-05,
+      "loss": 0.8486,
+      "step": 1912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5296786613886106e-05,
+      "loss": 0.8784,
+      "step": 1913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291863864550443e-05,
+      "loss": 0.3423,
+      "step": 1914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286939333361693e-05,
+      "loss": 0.9277,
+      "step": 1915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282013021978028e-05,
+      "loss": 0.9365,
+      "step": 1916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277084932058205e-05,
+      "loss": 0.8589,
+      "step": 1917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272155065261588e-05,
+      "loss": 0.9175,
+      "step": 1918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526722342324815e-05,
+      "loss": 0.9111,
+      "step": 1919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262290007678446e-05,
+      "loss": 0.8999,
+      "step": 1920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257354820213634e-05,
+      "loss": 0.8516,
+      "step": 1921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252417862515473e-05,
+      "loss": 0.8701,
+      "step": 1922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247479136246305e-05,
+      "loss": 0.8403,
+      "step": 1923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242538643069085e-05,
+      "loss": 0.9028,
+      "step": 1924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237596384647347e-05,
+      "loss": 0.8384,
+      "step": 1925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5232652362645231e-05,
+      "loss": 0.8018,
+      "step": 1926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227706578727466e-05,
+      "loss": 0.8228,
+      "step": 1927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222759034559368e-05,
+      "loss": 0.9326,
+      "step": 1928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217809731806866e-05,
+      "loss": 0.854,
+      "step": 1929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212858672136452e-05,
+      "loss": 0.877,
+      "step": 1930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207905857215238e-05,
+      "loss": 0.8574,
+      "step": 1931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202951288710903e-05,
+      "loss": 0.8545,
+      "step": 1932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197994968291739e-05,
+      "loss": 0.8623,
+      "step": 1933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193036897626603e-05,
+      "loss": 0.8354,
+      "step": 1934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188077078384964e-05,
+      "loss": 0.873,
+      "step": 1935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518311551223687e-05,
+      "loss": 0.8623,
+      "step": 1936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178152200852955e-05,
+      "loss": 0.9585,
+      "step": 1937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173187145904444e-05,
+      "loss": 0.8706,
+      "step": 1938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516822034906315e-05,
+      "loss": 0.834,
+      "step": 1939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516325181200147e-05,
+      "loss": 0.8472,
+      "step": 1940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158281536392382e-05,
+      "loss": 0.9023,
+      "step": 1941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153309523909467e-05,
+      "loss": 0.8833,
+      "step": 1942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148335776226872e-05,
+      "loss": 0.3535,
+      "step": 1943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143360295019333e-05,
+      "loss": 0.877,
+      "step": 1944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138383081962182e-05,
+      "loss": 0.875,
+      "step": 1945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133404138731314e-05,
+      "loss": 0.9102,
+      "step": 1946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128423467003217e-05,
+      "loss": 0.8652,
+      "step": 1947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123441068454967e-05,
+      "loss": 0.8721,
+      "step": 1948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118456944764214e-05,
+      "loss": 0.9189,
+      "step": 1949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511347109760919e-05,
+      "loss": 0.9136,
+      "step": 1950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108483528668705e-05,
+      "loss": 0.938,
+      "step": 1951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103494239622154e-05,
+      "loss": 0.3389,
+      "step": 1952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098503232149504e-05,
+      "loss": 0.8223,
+      "step": 1953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093510507931309e-05,
+      "loss": 0.9238,
+      "step": 1954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088516068648698e-05,
+      "loss": 0.8828,
+      "step": 1955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083519915983373e-05,
+      "loss": 0.8652,
+      "step": 1956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078522051617617e-05,
+      "loss": 0.8901,
+      "step": 1957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073522477234294e-05,
+      "loss": 0.8418,
+      "step": 1958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068521194516833e-05,
+      "loss": 0.8706,
+      "step": 1959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063518205149245e-05,
+      "loss": 0.834,
+      "step": 1960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5058513510816118e-05,
+      "loss": 0.9189,
+      "step": 1961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053507113202606e-05,
+      "loss": 0.3195,
+      "step": 1962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048499013994442e-05,
+      "loss": 0.8867,
+      "step": 1963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043489214877933e-05,
+      "loss": 0.9102,
+      "step": 1964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038477717539955e-05,
+      "loss": 0.8677,
+      "step": 1965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033464523667958e-05,
+      "loss": 0.854,
+      "step": 1966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028449634949964e-05,
+      "loss": 0.8398,
+      "step": 1967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502343305307456e-05,
+      "loss": 0.8687,
+      "step": 1968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018414779730914e-05,
+      "loss": 0.3525,
+      "step": 1969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501339481660875e-05,
+      "loss": 0.8154,
+      "step": 1970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008373165398372e-05,
+      "loss": 0.8848,
+      "step": 1971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003349827790646e-05,
+      "loss": 0.8481,
+      "step": 1972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998324805477008e-05,
+      "loss": 0.894,
+      "step": 1973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993298100149468e-05,
+      "loss": 0.9038,
+      "step": 1974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988269713500585e-05,
+      "loss": 0.3418,
+      "step": 1975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983239647223504e-05,
+      "loss": 0.8252,
+      "step": 1976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497820790301192e-05,
+      "loss": 0.8867,
+      "step": 1977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973174482560106e-05,
+      "loss": 0.8457,
+      "step": 1978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4968139387562895e-05,
+      "loss": 0.8638,
+      "step": 1979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496310261971567e-05,
+      "loss": 0.8599,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49580641807144e-05,
+      "loss": 0.8745,
+      "step": 1981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4953024072255601e-05,
+      "loss": 0.8379,
+      "step": 1982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494798229603636e-05,
+      "loss": 0.8008,
+      "step": 1983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942938853754321e-05,
+      "loss": 0.8677,
+      "step": 1984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937893747107685e-05,
+      "loss": 0.9038,
+      "step": 1985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932846977795223e-05,
+      "loss": 0.8643,
+      "step": 1986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927798547516257e-05,
+      "loss": 0.9028,
+      "step": 1987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922748457970679e-05,
+      "loss": 0.8682,
+      "step": 1988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917696710858925e-05,
+      "loss": 0.8081,
+      "step": 1989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4912643307882e-05,
+      "loss": 0.8911,
+      "step": 1990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907588250741464e-05,
+      "loss": 0.895,
+      "step": 1991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4902531541139436e-05,
+      "loss": 0.8896,
+      "step": 1992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897473180778583e-05,
+      "loss": 0.8745,
+      "step": 1993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892413171362142e-05,
+      "loss": 0.8525,
+      "step": 1994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887351514593888e-05,
+      "loss": 0.8584,
+      "step": 1995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882288212178167e-05,
+      "loss": 0.8145,
+      "step": 1996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877223265819864e-05,
+      "loss": 0.8452,
+      "step": 1997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872156677224437e-05,
+      "loss": 0.8535,
+      "step": 1998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867088448097877e-05,
+      "loss": 0.8711,
+      "step": 1999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862018580146735e-05,
+      "loss": 0.8286,
+      "step": 2000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856947075078119e-05,
+      "loss": 0.8818,
+      "step": 2001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851873934599683e-05,
+      "loss": 0.8901,
+      "step": 2002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846799160419631e-05,
+      "loss": 0.8599,
+      "step": 2003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484172275424672e-05,
+      "loss": 0.8481,
+      "step": 2004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836644717790251e-05,
+      "loss": 0.895,
+      "step": 2005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831565052760084e-05,
+      "loss": 0.8843,
+      "step": 2006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826483760866616e-05,
+      "loss": 0.874,
+      "step": 2007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821400843820805e-05,
+      "loss": 0.8599,
+      "step": 2008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816316303334141e-05,
+      "loss": 0.8442,
+      "step": 2009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481123014111867e-05,
+      "loss": 0.9058,
+      "step": 2010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806142358886984e-05,
+      "loss": 0.8691,
+      "step": 2011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801052958352218e-05,
+      "loss": 0.8389,
+      "step": 2012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479596194122805e-05,
+      "loss": 0.897,
+      "step": 2013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790869309228706e-05,
+      "loss": 0.9097,
+      "step": 2014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785775064068957e-05,
+      "loss": 0.8008,
+      "step": 2015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780679207464113e-05,
+      "loss": 0.8691,
+      "step": 2016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775581741130027e-05,
+      "loss": 0.8945,
+      "step": 2017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770482666783097e-05,
+      "loss": 0.344,
+      "step": 2018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765381986140264e-05,
+      "loss": 0.8315,
+      "step": 2019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760279700918997e-05,
+      "loss": 0.9326,
+      "step": 2020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755175812837326e-05,
+      "loss": 0.8921,
+      "step": 2021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750070323613807e-05,
+      "loss": 0.8628,
+      "step": 2022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744963234967532e-05,
+      "loss": 0.9292,
+      "step": 2023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4739854548618143e-05,
+      "loss": 0.8462,
+      "step": 2024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734744266285812e-05,
+      "loss": 0.8477,
+      "step": 2025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729632389691253e-05,
+      "loss": 0.894,
+      "step": 2026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724518920555712e-05,
+      "loss": 0.7681,
+      "step": 2027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719403860600973e-05,
+      "loss": 0.8682,
+      "step": 2028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714287211549364e-05,
+      "loss": 0.877,
+      "step": 2029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709168975123726e-05,
+      "loss": 0.8774,
+      "step": 2030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704049153047464e-05,
+      "loss": 0.8789,
+      "step": 2031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698927747044499e-05,
+      "loss": 0.8867,
+      "step": 2032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693804758839276e-05,
+      "loss": 0.8574,
+      "step": 2033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688680190156799e-05,
+      "loss": 0.8906,
+      "step": 2034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683554042722585e-05,
+      "loss": 0.8418,
+      "step": 2035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678426318262689e-05,
+      "loss": 0.856,
+      "step": 2036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4673297018503693e-05,
+      "loss": 0.8457,
+      "step": 2037
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668166145172716e-05,
+      "loss": 0.8735,
+      "step": 2038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46630336999974e-05,
+      "loss": 0.8579,
+      "step": 2039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657899684705918e-05,
+      "loss": 0.8838,
+      "step": 2040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465276410102698e-05,
+      "loss": 0.8716,
+      "step": 2041
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464762695068981e-05,
+      "loss": 0.8867,
+      "step": 2042
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642488235424168e-05,
+      "loss": 0.8462,
+      "step": 2043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637347956960344e-05,
+      "loss": 0.8506,
+      "step": 2044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632206117029143e-05,
+      "loss": 0.8877,
+      "step": 2045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627062717361909e-05,
+      "loss": 0.9238,
+      "step": 2046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621917759690502e-05,
+      "loss": 0.8301,
+      "step": 2047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461677124574731e-05,
+      "loss": 0.8809,
+      "step": 2048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611623177265241e-05,
+      "loss": 0.8516,
+      "step": 2049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606473555977737e-05,
+      "loss": 0.8911,
+      "step": 2050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601322383618752e-05,
+      "loss": 0.8608,
+      "step": 2051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596169661922761e-05,
+      "loss": 0.8672,
+      "step": 2052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591015392624775e-05,
+      "loss": 0.8608,
+      "step": 2053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585859577460312e-05,
+      "loss": 0.8833,
+      "step": 2054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580702218165417e-05,
+      "loss": 0.8804,
+      "step": 2055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457554331647665e-05,
+      "loss": 0.8677,
+      "step": 2056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4570382874131098e-05,
+      "loss": 0.8804,
+      "step": 2057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565220892866358e-05,
+      "loss": 0.8638,
+      "step": 2058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560057374420554e-05,
+      "loss": 0.8608,
+      "step": 2059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554892320532319e-05,
+      "loss": 0.8677,
+      "step": 2060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549725732940813e-05,
+      "loss": 0.9175,
+      "step": 2061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4544557613385699e-05,
+      "loss": 0.8452,
+      "step": 2062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539387963607167e-05,
+      "loss": 0.8394,
+      "step": 2063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534216785345922e-05,
+      "loss": 0.8574,
+      "step": 2064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529044080343177e-05,
+      "loss": 0.8369,
+      "step": 2065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523869850340662e-05,
+      "loss": 0.8569,
+      "step": 2066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518694097080622e-05,
+      "loss": 0.8945,
+      "step": 2067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513516822305814e-05,
+      "loss": 0.79,
+      "step": 2068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508338027759503e-05,
+      "loss": 0.8613,
+      "step": 2069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503157715185479e-05,
+      "loss": 0.8555,
+      "step": 2070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4497975886328026e-05,
+      "loss": 0.8066,
+      "step": 2071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492792542931949e-05,
+      "loss": 0.3429,
+      "step": 2072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487607686742563e-05,
+      "loss": 0.8608,
+      "step": 2073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482421319505684e-05,
+      "loss": 0.8657,
+      "step": 2074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477233442967648e-05,
+      "loss": 0.8755,
+      "step": 2075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472044058875297e-05,
+      "loss": 0.834,
+      "step": 2076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466853168975966e-05,
+      "loss": 0.8115,
+      "step": 2077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4461660775017521e-05,
+      "loss": 0.8657,
+      "step": 2078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456466878748312e-05,
+      "loss": 0.8179,
+      "step": 2079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451271481917212e-05,
+      "loss": 0.8848,
+      "step": 2080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446074586273591e-05,
+      "loss": 0.8555,
+      "step": 2081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440876193567319e-05,
+      "loss": 0.8428,
+      "step": 2082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435676305548783e-05,
+      "loss": 0.3474,
+      "step": 2083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430474923968862e-05,
+      "loss": 0.8398,
+      "step": 2084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425272050578947e-05,
+      "loss": 0.9199,
+      "step": 2085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442006768713092e-05,
+      "loss": 0.8193,
+      "step": 2086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414861835377179e-05,
+      "loss": 0.9004,
+      "step": 2087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409654497070609e-05,
+      "loss": 0.8945,
+      "step": 2088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404445673964604e-05,
+      "loss": 0.854,
+      "step": 2089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399235367813056e-05,
+      "loss": 0.8423,
+      "step": 2090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394023580370358e-05,
+      "loss": 0.8022,
+      "step": 2091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388810313391395e-05,
+      "loss": 0.8501,
+      "step": 2092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4383595568631563e-05,
+      "loss": 0.8555,
+      "step": 2093
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378379347846743e-05,
+      "loss": 0.8599,
+      "step": 2094
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373161652793317e-05,
+      "loss": 0.8423,
+      "step": 2095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367942485228169e-05,
+      "loss": 0.812,
+      "step": 2096
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362721846908669e-05,
+      "loss": 0.3516,
+      "step": 2097
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357499739592691e-05,
+      "loss": 0.9033,
+      "step": 2098
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352276165038597e-05,
+      "loss": 0.8237,
+      "step": 2099
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4347051125005249e-05,
+      "loss": 0.8081,
+      "step": 2100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341824621251996e-05,
+      "loss": 0.8462,
+      "step": 2101
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4336596655538686e-05,
+      "loss": 0.8916,
+      "step": 2102
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331367229625656e-05,
+      "loss": 0.855,
+      "step": 2103
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326136345273738e-05,
+      "loss": 0.8975,
+      "step": 2104
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320904004244245e-05,
+      "loss": 0.8169,
+      "step": 2105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315670208298995e-05,
+      "loss": 0.8511,
+      "step": 2106
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310434959200285e-05,
+      "loss": 0.3276,
+      "step": 2107
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305198258710906e-05,
+      "loss": 0.8115,
+      "step": 2108
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299960108594136e-05,
+      "loss": 0.853,
+      "step": 2109
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294720510613745e-05,
+      "loss": 0.8521,
+      "step": 2110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4289479466533982e-05,
+      "loss": 0.3423,
+      "step": 2111
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428423697811959e-05,
+      "loss": 0.8647,
+      "step": 2112
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278993047135802e-05,
+      "loss": 0.8628,
+      "step": 2113
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273747675348326e-05,
+      "loss": 0.8291,
+      "step": 2114
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426850086452336e-05,
+      "loss": 0.8071,
+      "step": 2115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426325261642759e-05,
+      "loss": 0.874,
+      "step": 2116
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425800293282818e-05,
+      "loss": 0.8599,
+      "step": 2117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252751815492782e-05,
+      "loss": 0.8867,
+      "step": 2118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424749926618953e-05,
+      "loss": 0.8779,
+      "step": 2119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424224528668704e-05,
+      "loss": 0.8657,
+      "step": 2120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4236989878754407e-05,
+      "loss": 0.8379,
+      "step": 2121
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231733044161203e-05,
+      "loss": 0.8901,
+      "step": 2122
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226474784677498e-05,
+      "loss": 0.8691,
+      "step": 2123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221215102073823e-05,
+      "loss": 0.8555,
+      "step": 2124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215953998121193e-05,
+      "loss": 0.8174,
+      "step": 2125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421069147459111e-05,
+      "loss": 0.833,
+      "step": 2126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205427533255543e-05,
+      "loss": 0.8857,
+      "step": 2127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4200162175886947e-05,
+      "loss": 0.8906,
+      "step": 2128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194895404258246e-05,
+      "loss": 0.8062,
+      "step": 2129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189627220142848e-05,
+      "loss": 0.8711,
+      "step": 2130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184357625314628e-05,
+      "loss": 0.8887,
+      "step": 2131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179086621547945e-05,
+      "loss": 0.8135,
+      "step": 2132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173814210617627e-05,
+      "loss": 0.873,
+      "step": 2133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168540394298974e-05,
+      "loss": 0.8901,
+      "step": 2134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163265174367766e-05,
+      "loss": 0.8677,
+      "step": 2135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415798855260025e-05,
+      "loss": 0.8232,
+      "step": 2136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152710530773152e-05,
+      "loss": 0.8765,
+      "step": 2137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147431110663654e-05,
+      "loss": 0.8208,
+      "step": 2138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142150294049425e-05,
+      "loss": 0.7803,
+      "step": 2139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41368680827086e-05,
+      "loss": 0.8623,
+      "step": 2140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413158447841978e-05,
+      "loss": 0.8843,
+      "step": 2141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126299482962038e-05,
+      "loss": 0.8262,
+      "step": 2142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121013098114911e-05,
+      "loss": 0.8506,
+      "step": 2143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4115725325658411e-05,
+      "loss": 0.8511,
+      "step": 2144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110436167373009e-05,
+      "loss": 0.8496,
+      "step": 2145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105145625039654e-05,
+      "loss": 0.9292,
+      "step": 2146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409985370043975e-05,
+      "loss": 0.8516,
+      "step": 2147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094560395355169e-05,
+      "loss": 0.8452,
+      "step": 2148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408926571156825e-05,
+      "loss": 0.8711,
+      "step": 2149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083969650861797e-05,
+      "loss": 0.9199,
+      "step": 2150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078672215019076e-05,
+      "loss": 0.8696,
+      "step": 2151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073373405823818e-05,
+      "loss": 0.875,
+      "step": 2152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068073225060213e-05,
+      "loss": 0.8481,
+      "step": 2153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406277167451291e-05,
+      "loss": 0.8306,
+      "step": 2154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057468755967032e-05,
+      "loss": 0.8511,
+      "step": 2155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405216447120815e-05,
+      "loss": 0.9106,
+      "step": 2156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4046858822022298e-05,
+      "loss": 0.8765,
+      "step": 2157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041551810195974e-05,
+      "loss": 0.9048,
+      "step": 2158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036243437516133e-05,
+      "loss": 0.8516,
+      "step": 2159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403093370577018e-05,
+      "loss": 0.8306,
+      "step": 2160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402562261674599e-05,
+      "loss": 0.3196,
+      "step": 2161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402031017223189e-05,
+      "loss": 0.8472,
+      "step": 2162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014996374016659e-05,
+      "loss": 0.3198,
+      "step": 2163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009681223889535e-05,
+      "loss": 0.8926,
+      "step": 2164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004364723640217e-05,
+      "loss": 0.8574,
+      "step": 2165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999046875058857e-05,
+      "loss": 0.8486,
+      "step": 2166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993727679936044e-05,
+      "loss": 0.345,
+      "step": 2167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988407140062845e-05,
+      "loss": 0.8589,
+      "step": 2168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983085257230766e-05,
+      "loss": 0.343,
+      "step": 2169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977762033231765e-05,
+      "loss": 0.8877,
+      "step": 2170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397243746985826e-05,
+      "loss": 0.8633,
+      "step": 2171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967111568903108e-05,
+      "loss": 0.8726,
+      "step": 2172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961784332159632e-05,
+      "loss": 0.8735,
+      "step": 2173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956455761421584e-05,
+      "loss": 0.8301,
+      "step": 2174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951125858483185e-05,
+      "loss": 0.8555,
+      "step": 2175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945794625139097e-05,
+      "loss": 0.9111,
+      "step": 2176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940462063184424e-05,
+      "loss": 0.8511,
+      "step": 2177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935128174414726e-05,
+      "loss": 0.8472,
+      "step": 2178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929792960626005e-05,
+      "loss": 0.9248,
+      "step": 2179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924456423614714e-05,
+      "loss": 0.916,
+      "step": 2180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919118565177745e-05,
+      "loss": 0.835,
+      "step": 2181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3913779387112439e-05,
+      "loss": 0.8169,
+      "step": 2182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390843889121658e-05,
+      "loss": 0.9009,
+      "step": 2183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903097079288397e-05,
+      "loss": 0.8745,
+      "step": 2184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897753953126562e-05,
+      "loss": 0.8457,
+      "step": 2185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389240951453019e-05,
+      "loss": 0.8921,
+      "step": 2186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887063765298832e-05,
+      "loss": 0.8652,
+      "step": 2187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881716707232489e-05,
+      "loss": 0.8828,
+      "step": 2188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876368342131599e-05,
+      "loss": 0.3655,
+      "step": 2189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3871018671797037e-05,
+      "loss": 0.894,
+      "step": 2190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865667698030123e-05,
+      "loss": 0.9106,
+      "step": 2191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860315422632615e-05,
+      "loss": 0.8081,
+      "step": 2192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854961847406703e-05,
+      "loss": 0.3518,
+      "step": 2193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384960697415502e-05,
+      "loss": 0.8779,
+      "step": 2194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844250804680641e-05,
+      "loss": 0.3822,
+      "step": 2195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838893340787065e-05,
+      "loss": 0.8721,
+      "step": 2196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833534584278237e-05,
+      "loss": 0.8662,
+      "step": 2197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828174536958532e-05,
+      "loss": 0.8613,
+      "step": 2198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822813200632764e-05,
+      "loss": 0.8457,
+      "step": 2199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817450577106177e-05,
+      "loss": 0.8613,
+      "step": 2200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381208666818445e-05,
+      "loss": 0.8457,
+      "step": 2201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806721475673697e-05,
+      "loss": 0.8799,
+      "step": 2202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801355001380457e-05,
+      "loss": 0.8979,
+      "step": 2203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795987247111706e-05,
+      "loss": 0.8628,
+      "step": 2204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.379061821467486e-05,
+      "loss": 0.8853,
+      "step": 2205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785247905877746e-05,
+      "loss": 0.8276,
+      "step": 2206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779876322528631e-05,
+      "loss": 0.8735,
+      "step": 2207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774503466436216e-05,
+      "loss": 0.8765,
+      "step": 2208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769129339409623e-05,
+      "loss": 0.7974,
+      "step": 2209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763753943258404e-05,
+      "loss": 0.8477,
+      "step": 2210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375837727979254e-05,
+      "loss": 0.8369,
+      "step": 2211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752999350822438e-05,
+      "loss": 0.3417,
+      "step": 2212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747620158158927e-05,
+      "loss": 0.9126,
+      "step": 2213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742239703613266e-05,
+      "loss": 0.8374,
+      "step": 2214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736857988997143e-05,
+      "loss": 0.8667,
+      "step": 2215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373147501612266e-05,
+      "loss": 0.3325,
+      "step": 2216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726090786802347e-05,
+      "loss": 0.3418,
+      "step": 2217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372070530284916e-05,
+      "loss": 0.8696,
+      "step": 2218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371531856607648e-05,
+      "loss": 0.9111,
+      "step": 2219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709930578298098e-05,
+      "loss": 0.8506,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704541341328236e-05,
+      "loss": 0.874,
+      "step": 2221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699150856981536e-05,
+      "loss": 0.8979,
+      "step": 2222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693759127073057e-05,
+      "loss": 0.8804,
+      "step": 2223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688366153418278e-05,
+      "loss": 0.9121,
+      "step": 2224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682971937833095e-05,
+      "loss": 0.8433,
+      "step": 2225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677576482133832e-05,
+      "loss": 0.8843,
+      "step": 2226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672179788137214e-05,
+      "loss": 0.8789,
+      "step": 2227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666781857660396e-05,
+      "loss": 0.8398,
+      "step": 2228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366138269252095e-05,
+      "loss": 0.8604,
+      "step": 2229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655982294536851e-05,
+      "loss": 0.8799,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36505806655265e-05,
+      "loss": 0.9165,
+      "step": 2231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645177807308715e-05,
+      "loss": 0.8032,
+      "step": 2232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639773721702716e-05,
+      "loss": 0.7749,
+      "step": 2233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634368410528145e-05,
+      "loss": 0.8779,
+      "step": 2234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628961875605055e-05,
+      "loss": 0.8369,
+      "step": 2235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362355411875391e-05,
+      "loss": 0.8428,
+      "step": 2236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618145141795584e-05,
+      "loss": 0.8218,
+      "step": 2237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361273494655137e-05,
+      "loss": 0.8364,
+      "step": 2238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607323534842961e-05,
+      "loss": 0.8975,
+      "step": 2239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601910908492462e-05,
+      "loss": 0.8696,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596497069322392e-05,
+      "loss": 0.8574,
+      "step": 2241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591082019155674e-05,
+      "loss": 0.8291,
+      "step": 2242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358566575981564e-05,
+      "loss": 0.8516,
+      "step": 2243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358024829312603e-05,
+      "loss": 0.8955,
+      "step": 2244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574829620910985e-05,
+      "loss": 0.8511,
+      "step": 2245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569409744995063e-05,
+      "loss": 0.8413,
+      "step": 2246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563988667203214e-05,
+      "loss": 0.8384,
+      "step": 2247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558566389360805e-05,
+      "loss": 0.9175,
+      "step": 2248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35531429132936e-05,
+      "loss": 0.9028,
+      "step": 2249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547718240827764e-05,
+      "loss": 0.8403,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542292373789874e-05,
+      "loss": 0.8638,
+      "step": 2251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536865314006903e-05,
+      "loss": 0.8994,
+      "step": 2252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531437063306223e-05,
+      "loss": 0.8745,
+      "step": 2253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526007623515618e-05,
+      "loss": 0.9077,
+      "step": 2254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520576996463257e-05,
+      "loss": 0.9062,
+      "step": 2255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351514518397772e-05,
+      "loss": 0.8496,
+      "step": 2256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509712187887984e-05,
+      "loss": 0.3406,
+      "step": 2257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504278010023426e-05,
+      "loss": 0.8589,
+      "step": 2258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498842652213811e-05,
+      "loss": 0.9097,
+      "step": 2259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493406116289314e-05,
+      "loss": 0.8335,
+      "step": 2260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3487968404080502e-05,
+      "loss": 0.8643,
+      "step": 2261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3482529517418334e-05,
+      "loss": 0.938,
+      "step": 2262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477089458134171e-05,
+      "loss": 0.8716,
+      "step": 2263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471648228059767e-05,
+      "loss": 0.916,
+      "step": 2264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3466205829027263e-05,
+      "loss": 0.9526,
+      "step": 2265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460762262869202e-05,
+      "loss": 0.8809,
+      "step": 2266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455317531418523e-05,
+      "loss": 0.8755,
+      "step": 2267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449871636508547e-05,
+      "loss": 0.9106,
+      "step": 2268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344442457997299e-05,
+      "loss": 0.8726,
+      "step": 2269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438976363645966e-05,
+      "loss": 0.835,
+      "step": 2270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343352698936197e-05,
+      "loss": 0.8687,
+      "step": 2271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428076458955897e-05,
+      "loss": 0.7944,
+      "step": 2272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422624774263018e-05,
+      "loss": 0.8965,
+      "step": 2273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417171937119004e-05,
+      "loss": 0.8301,
+      "step": 2274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411717949359913e-05,
+      "loss": 0.8652,
+      "step": 2275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406262812822181e-05,
+      "loss": 0.8774,
+      "step": 2276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400806529342648e-05,
+      "loss": 0.8486,
+      "step": 2277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395349100758519e-05,
+      "loss": 0.8398,
+      "step": 2278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389890528907399e-05,
+      "loss": 0.8926,
+      "step": 2279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3384430815627277e-05,
+      "loss": 0.8916,
+      "step": 2280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337896996275652e-05,
+      "loss": 0.8457,
+      "step": 2281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373507972133883e-05,
+      "loss": 0.8354,
+      "step": 2282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3368044845598507e-05,
+      "loss": 0.8647,
+      "step": 2283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362580584989908e-05,
+      "loss": 0.9253,
+      "step": 2284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335711519214799e-05,
+      "loss": 0.8896,
+      "step": 2285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351648668913034e-05,
+      "loss": 0.9443,
+      "step": 2286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346181017125703e-05,
+      "loss": 0.334,
+      "step": 2287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3340712238627042e-05,
+      "loss": 0.8667,
+      "step": 2288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335242335258475e-05,
+      "loss": 0.873,
+      "step": 2289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329771308861804e-05,
+      "loss": 0.8208,
+      "step": 2290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324299161279211e-05,
+      "loss": 0.8462,
+      "step": 2291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318825894353244e-05,
+      "loss": 0.8525,
+      "step": 2292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313351509926846e-05,
+      "loss": 0.8438,
+      "step": 2293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3307876009843325e-05,
+      "loss": 0.896,
+      "step": 2294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302399395946368e-05,
+      "loss": 0.3557,
+      "step": 2295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296921670080037e-05,
+      "loss": 0.8794,
+      "step": 2296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291442834088764e-05,
+      "loss": 0.8799,
+      "step": 2297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285962889817363e-05,
+      "loss": 0.8394,
+      "step": 2298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328048183911101e-05,
+      "loss": 0.7998,
+      "step": 2299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327499968381527e-05,
+      "loss": 0.8286,
+      "step": 2300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269516425776065e-05,
+      "loss": 0.8271,
+      "step": 2301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3264032066839691e-05,
+      "loss": 0.8491,
+      "step": 2302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258546608852823e-05,
+      "loss": 0.8076,
+      "step": 2303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253060053662494e-05,
+      "loss": 0.8936,
+      "step": 2304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247572403116115e-05,
+      "loss": 0.8276,
+      "step": 2305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242083659061469e-05,
+      "loss": 0.8257,
+      "step": 2306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236593823346693e-05,
+      "loss": 0.8608,
+      "step": 2307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3231102897820304e-05,
+      "loss": 0.8364,
+      "step": 2308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225610884331184e-05,
+      "loss": 0.8237,
+      "step": 2309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220117784728577e-05,
+      "loss": 0.8418,
+      "step": 2310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32146236008621e-05,
+      "loss": 0.8242,
+      "step": 2311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209128334581723e-05,
+      "loss": 0.8677,
+      "step": 2312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203631987737791e-05,
+      "loss": 0.8726,
+      "step": 2313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198134562181011e-05,
+      "loss": 0.8516,
+      "step": 2314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192636059762453e-05,
+      "loss": 0.8848,
+      "step": 2315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187136482333547e-05,
+      "loss": 0.8271,
+      "step": 2316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181635831746085e-05,
+      "loss": 0.8604,
+      "step": 2317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176134109852223e-05,
+      "loss": 0.8062,
+      "step": 2318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170631318504475e-05,
+      "loss": 0.8809,
+      "step": 2319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165127459555722e-05,
+      "loss": 0.8677,
+      "step": 2320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159622534859194e-05,
+      "loss": 0.8442,
+      "step": 2321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154116546268482e-05,
+      "loss": 0.9111,
+      "step": 2322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148609495637547e-05,
+      "loss": 0.8105,
+      "step": 2323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3143101384820694e-05,
+      "loss": 0.8423,
+      "step": 2324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313759221567259e-05,
+      "loss": 0.8574,
+      "step": 2325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132081990048257e-05,
+      "loss": 0.835,
+      "step": 2326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126570709803076e-05,
+      "loss": 0.8296,
+      "step": 2327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121058376792783e-05,
+      "loss": 0.7793,
+      "step": 2328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311554499287346e-05,
+      "loss": 0.8853,
+      "step": 2329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110030559901557e-05,
+      "loss": 0.8804,
+      "step": 2330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104515079733863e-05,
+      "loss": 0.8379,
+      "step": 2331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3098998554227529e-05,
+      "loss": 0.894,
+      "step": 2332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093480985240055e-05,
+      "loss": 0.8203,
+      "step": 2333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087962374629293e-05,
+      "loss": 0.854,
+      "step": 2334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082442724253447e-05,
+      "loss": 0.8286,
+      "step": 2335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076922035971062e-05,
+      "loss": 0.8486,
+      "step": 2336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307140031164105e-05,
+      "loss": 0.8804,
+      "step": 2337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065877553122654e-05,
+      "loss": 0.8726,
+      "step": 2338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060353762275477e-05,
+      "loss": 0.7822,
+      "step": 2339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054828940959462e-05,
+      "loss": 0.8735,
+      "step": 2340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049303091034909e-05,
+      "loss": 0.897,
+      "step": 2341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304377621436245e-05,
+      "loss": 0.8335,
+      "step": 2342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038248312803072e-05,
+      "loss": 0.8833,
+      "step": 2343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032719388218114e-05,
+      "loss": 0.3352,
+      "step": 2344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3027189442469238e-05,
+      "loss": 0.8115,
+      "step": 2345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021658477418475e-05,
+      "loss": 0.8979,
+      "step": 2346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016126494928178e-05,
+      "loss": 0.8765,
+      "step": 2347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301059349686106e-05,
+      "loss": 0.8149,
+      "step": 2348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005059485080161e-05,
+      "loss": 0.9297,
+      "step": 2349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999524461448871e-05,
+      "loss": 0.8086,
+      "step": 2350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993988427830919e-05,
+      "loss": 0.8999,
+      "step": 2351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988451386090376e-05,
+      "loss": 0.8413,
+      "step": 2352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982913338091647e-05,
+      "loss": 0.8511,
+      "step": 2353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977374285699483e-05,
+      "loss": 0.8789,
+      "step": 2354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971834230778963e-05,
+      "loss": 0.8145,
+      "step": 2355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966293175195517e-05,
+      "loss": 0.8306,
+      "step": 2356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29607511208149e-05,
+      "loss": 0.8789,
+      "step": 2357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2955208069503208e-05,
+      "loss": 0.8584,
+      "step": 2358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294966402312688e-05,
+      "loss": 0.8359,
+      "step": 2359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944118983552674e-05,
+      "loss": 0.8135,
+      "step": 2360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938572952647695e-05,
+      "loss": 0.9556,
+      "step": 2361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933025932279378e-05,
+      "loss": 0.8286,
+      "step": 2362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927477924315493e-05,
+      "loss": 0.3445,
+      "step": 2363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921928930624134e-05,
+      "loss": 0.8608,
+      "step": 2364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2916378953073743e-05,
+      "loss": 0.7358,
+      "step": 2365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910827993533078e-05,
+      "loss": 0.8481,
+      "step": 2366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905276053871237e-05,
+      "loss": 0.8047,
+      "step": 2367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289972313595764e-05,
+      "loss": 0.9209,
+      "step": 2368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894169241662045e-05,
+      "loss": 0.3562,
+      "step": 2369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888614372854536e-05,
+      "loss": 0.8999,
+      "step": 2370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2883058531405517e-05,
+      "loss": 0.8477,
+      "step": 2371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877501719185732e-05,
+      "loss": 0.8364,
+      "step": 2372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871943938066252e-05,
+      "loss": 0.8105,
+      "step": 2373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2866385189918456e-05,
+      "loss": 0.8398,
+      "step": 2374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860825476614068e-05,
+      "loss": 0.8579,
+      "step": 2375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285526480002513e-05,
+      "loss": 0.8892,
+      "step": 2376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849703162024008e-05,
+      "loss": 0.8711,
+      "step": 2377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844140564483392e-05,
+      "loss": 0.8857,
+      "step": 2378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838577009276296e-05,
+      "loss": 0.9067,
+      "step": 2379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833012498276056e-05,
+      "loss": 0.854,
+      "step": 2380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827447033356326e-05,
+      "loss": 0.9014,
+      "step": 2381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821880616391089e-05,
+      "loss": 0.8506,
+      "step": 2382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816313249254647e-05,
+      "loss": 0.8354,
+      "step": 2383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810744933821611e-05,
+      "loss": 0.3391,
+      "step": 2384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805175671966925e-05,
+      "loss": 0.8403,
+      "step": 2385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799605465565848e-05,
+      "loss": 0.9624,
+      "step": 2386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279403431649395e-05,
+      "loss": 0.8945,
+      "step": 2387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788462226627127e-05,
+      "loss": 0.8916,
+      "step": 2388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278288919784159e-05,
+      "loss": 0.8252,
+      "step": 2389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277731523201386e-05,
+      "loss": 0.8276,
+      "step": 2390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771740331020784e-05,
+      "loss": 0.321,
+      "step": 2391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766164496739515e-05,
+      "loss": 0.8716,
+      "step": 2392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2760587731047522e-05,
+      "loss": 0.8486,
+      "step": 2393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755010035822588e-05,
+      "loss": 0.8438,
+      "step": 2394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749431412942816e-05,
+      "loss": 0.8589,
+      "step": 2395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743851864286611e-05,
+      "loss": 0.894,
+      "step": 2396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738271391732698e-05,
+      "loss": 0.8657,
+      "step": 2397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732689997160102e-05,
+      "loss": 0.8545,
+      "step": 2398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272710768244817e-05,
+      "loss": 0.8657,
+      "step": 2399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721524449476554e-05,
+      "loss": 0.8501,
+      "step": 2400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715940300125216e-05,
+      "loss": 0.8428,
+      "step": 2401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710355236274428e-05,
+      "loss": 0.8198,
+      "step": 2402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704769259804764e-05,
+      "loss": 0.9165,
+      "step": 2403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699182372597108e-05,
+      "loss": 0.8252,
+      "step": 2404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693594576532657e-05,
+      "loss": 0.8696,
+      "step": 2405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268800587349291e-05,
+      "loss": 0.9102,
+      "step": 2406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682416265359665e-05,
+      "loss": 0.8428,
+      "step": 2407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676825754015037e-05,
+      "loss": 0.8677,
+      "step": 2408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267123434134143e-05,
+      "loss": 0.8662,
+      "step": 2409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665642029221565e-05,
+      "loss": 0.8594,
+      "step": 2410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266004881953846e-05,
+      "loss": 0.853,
+      "step": 2411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265445471417544e-05,
+      "loss": 0.8423,
+      "step": 2412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264885971501612e-05,
+      "loss": 0.8984,
+      "step": 2413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643263823944425e-05,
+      "loss": 0.8677,
+      "step": 2414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263766704284458e-05,
+      "loss": 0.8999,
+      "step": 2415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2632069373601114e-05,
+      "loss": 0.8843,
+      "step": 2416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2626470818098842e-05,
+      "loss": 0.856,
+      "step": 2417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620871378222888e-05,
+      "loss": 0.8335,
+      "step": 2418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615271055858669e-05,
+      "loss": 0.7954,
+      "step": 2419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609669852891904e-05,
+      "loss": 0.7998,
+      "step": 2420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26040677712086e-05,
+      "loss": 0.8271,
+      "step": 2421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598464812695073e-05,
+      "loss": 0.8872,
+      "step": 2422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2592860979237922e-05,
+      "loss": 0.7979,
+      "step": 2423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2587256272724042e-05,
+      "loss": 0.853,
+      "step": 2424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258165069504063e-05,
+      "loss": 0.875,
+      "step": 2425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576044248075174e-05,
+      "loss": 0.7959,
+      "step": 2426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570436933715442e-05,
+      "loss": 0.8491,
+      "step": 2427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564828753849511e-05,
+      "loss": 0.8896,
+      "step": 2428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559219710365747e-05,
+      "loss": 0.8315,
+      "step": 2429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553609805152793e-05,
+      "loss": 0.7798,
+      "step": 2430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547999040099597e-05,
+      "loss": 0.8081,
+      "step": 2431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254238741709539e-05,
+      "loss": 0.8525,
+      "step": 2432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536774938029692e-05,
+      "loss": 0.8857,
+      "step": 2433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531161604792312e-05,
+      "loss": 0.9019,
+      "step": 2434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252554741927335e-05,
+      "loss": 0.958,
+      "step": 2435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251993238336319e-05,
+      "loss": 0.8745,
+      "step": 2436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2514316498952495e-05,
+      "loss": 0.8428,
+      "step": 2437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508699767932227e-05,
+      "loss": 0.8828,
+      "step": 2438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2503082192193627e-05,
+      "loss": 0.3418,
+      "step": 2439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249746377362822e-05,
+      "loss": 0.8335,
+      "step": 2440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491844514127813e-05,
+      "loss": 0.8594,
+      "step": 2441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24862244155845e-05,
+      "loss": 0.877,
+      "step": 2442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480603479890653e-05,
+      "loss": 0.9185,
+      "step": 2443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474981708938931e-05,
+      "loss": 0.854,
+      "step": 2444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469359104622276e-05,
+      "loss": 0.8418,
+      "step": 2445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463735668833903e-05,
+      "loss": 0.8901,
+      "step": 2446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458111403467304e-05,
+      "loss": 0.8364,
+      "step": 2447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452486310416267e-05,
+      "loss": 0.7524,
+      "step": 2448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446860391574844e-05,
+      "loss": 0.8359,
+      "step": 2449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244123364883737e-05,
+      "loss": 0.8472,
+      "step": 2450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435606084098457e-05,
+      "loss": 0.8999,
+      "step": 2451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429977699252993e-05,
+      "loss": 0.8257,
+      "step": 2452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2424348496196146e-05,
+      "loss": 0.9209,
+      "step": 2453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241871847682335e-05,
+      "loss": 0.8594,
+      "step": 2454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241308764303033e-05,
+      "loss": 0.8555,
+      "step": 2455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407455996713069e-05,
+      "loss": 0.8013,
+      "step": 2456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401823539767827e-05,
+      "loss": 0.8579,
+      "step": 2457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396190274091147e-05,
+      "loss": 0.8105,
+      "step": 2458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390556201579836e-05,
+      "loss": 0.8267,
+      "step": 2459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384921324130972e-05,
+      "loss": 0.8804,
+      "step": 2460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379285643641908e-05,
+      "loss": 0.8921,
+      "step": 2461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373649162010262e-05,
+      "loss": 0.8979,
+      "step": 2462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368011881133927e-05,
+      "loss": 0.8789,
+      "step": 2463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236237380291107e-05,
+      "loss": 0.8057,
+      "step": 2464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235673492924011e-05,
+      "loss": 0.8379,
+      "step": 2465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351095262019746e-05,
+      "loss": 0.8237,
+      "step": 2466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2345454803148946e-05,
+      "loss": 0.8623,
+      "step": 2467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339813554526938e-05,
+      "loss": 0.8853,
+      "step": 2468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2334171518053218e-05,
+      "loss": 0.8647,
+      "step": 2469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2328528695627548e-05,
+      "loss": 0.853,
+      "step": 2470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322885089149954e-05,
+      "loss": 0.8496,
+      "step": 2471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2317240700520727e-05,
+      "loss": 0.8594,
+      "step": 2472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231159553164042e-05,
+      "loss": 0.9131,
+      "step": 2473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305949584409853e-05,
+      "loss": 0.8408,
+      "step": 2474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300302860730097e-05,
+      "loss": 0.8179,
+      "step": 2475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2294655362502496e-05,
+      "loss": 0.876,
+      "step": 2476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289007091628653e-05,
+      "loss": 0.7563,
+      "step": 2477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283358050010428e-05,
+      "loss": 0.8979,
+      "step": 2478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2277708239549936e-05,
+      "loss": 0.8687,
+      "step": 2479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2272057662149563e-05,
+      "loss": 0.876,
+      "step": 2480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266406319711945e-05,
+      "loss": 0.8955,
+      "step": 2481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260754214139975e-05,
+      "loss": 0.8418,
+      "step": 2482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2255101347336804e-05,
+      "loss": 0.8433,
+      "step": 2483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2249447721205847e-05,
+      "loss": 0.8008,
+      "step": 2484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243793337650762e-05,
+      "loss": 0.8584,
+      "step": 2485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223813819857547e-05,
+      "loss": 0.8281,
+      "step": 2486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232482305884142e-05,
+      "loss": 0.8672,
+      "step": 2487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226825661481211e-05,
+      "loss": 0.8691,
+      "step": 2488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222116826727135e-05,
+      "loss": 0.854,
+      "step": 2489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22155101251595e-05,
+      "loss": 0.853,
+      "step": 2490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209851237050838e-05,
+      "loss": 0.8691,
+      "step": 2491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204191604850806e-05,
+      "loss": 0.8887,
+      "step": 2492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198531230465086e-05,
+      "loss": 0.8838,
+      "step": 2493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192870115799618e-05,
+      "loss": 0.8755,
+      "step": 2494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187208262760585e-05,
+      "loss": 0.8833,
+      "step": 2495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181545673254421e-05,
+      "loss": 0.8701,
+      "step": 2496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2175882349187808e-05,
+      "loss": 0.8208,
+      "step": 2497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170218292467681e-05,
+      "loss": 0.8896,
+      "step": 2498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164553505001206e-05,
+      "loss": 0.8423,
+      "step": 2499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2158887988695815e-05,
+      "loss": 0.8438,
+      "step": 2500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153221745459172e-05,
+      "loss": 0.8389,
+      "step": 2501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147554777199188e-05,
+      "loss": 0.8516,
+      "step": 2502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141887085824022e-05,
+      "loss": 0.8164,
+      "step": 2503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136218673242073e-05,
+      "loss": 0.3398,
+      "step": 2504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130549541361983e-05,
+      "loss": 0.8486,
+      "step": 2505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124879692092634e-05,
+      "loss": 0.8525,
+      "step": 2506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119209127343164e-05,
+      "loss": 0.3218,
+      "step": 2507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113537849022932e-05,
+      "loss": 0.8291,
+      "step": 2508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107865859041542e-05,
+      "loss": 0.813,
+      "step": 2509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210219315930885e-05,
+      "loss": 0.9375,
+      "step": 2510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209651975173494e-05,
+      "loss": 0.8604,
+      "step": 2511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2090845638230133e-05,
+      "loss": 0.8828,
+      "step": 2512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085170820704997e-05,
+      "loss": 0.8687,
+      "step": 2513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079495301070327e-05,
+      "loss": 0.8774,
+      "step": 2514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073819081237161e-05,
+      "loss": 0.3293,
+      "step": 2515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2068142163116769e-05,
+      "loss": 0.8428,
+      "step": 2516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062464548620665e-05,
+      "loss": 0.834,
+      "step": 2517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205678623966058e-05,
+      "loss": 0.3669,
+      "step": 2518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051107238148493e-05,
+      "loss": 0.897,
+      "step": 2519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045427545996617e-05,
+      "loss": 0.3376,
+      "step": 2520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039747165117389e-05,
+      "loss": 0.874,
+      "step": 2521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203406609742348e-05,
+      "loss": 0.2944,
+      "step": 2522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2028384344827797e-05,
+      "loss": 0.8599,
+      "step": 2523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022701909243476e-05,
+      "loss": 0.8916,
+      "step": 2524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2017018792583875e-05,
+      "loss": 0.8926,
+      "step": 2525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2011334996762593e-05,
+      "loss": 0.8457,
+      "step": 2526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005650523693457e-05,
+      "loss": 0.3398,
+      "step": 2527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199996537529051e-05,
+      "loss": 0.9097,
+      "step": 2528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994279553468032e-05,
+      "loss": 0.8838,
+      "step": 2529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198859306014053e-05,
+      "loss": 0.3499,
+      "step": 2530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982905897222735e-05,
+      "loss": 0.8535,
+      "step": 2531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977218066629599e-05,
+      "loss": 0.8472,
+      "step": 2532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197152957027631e-05,
+      "loss": 0.8569,
+      "step": 2533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965840410078268e-05,
+      "loss": 0.7905,
+      "step": 2534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960150587951108e-05,
+      "loss": 0.8872,
+      "step": 2535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195446010581067e-05,
+      "loss": 0.8579,
+      "step": 2536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948768965573044e-05,
+      "loss": 0.876,
+      "step": 2537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1943077169154514e-05,
+      "loss": 0.3333,
+      "step": 2538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19373847184716e-05,
+      "loss": 0.834,
+      "step": 2539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.193169161544104e-05,
+      "loss": 0.8789,
+      "step": 2540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192599786197979e-05,
+      "loss": 0.8911,
+      "step": 2541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920303460005023e-05,
+      "loss": 0.8188,
+      "step": 2542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914608411434136e-05,
+      "loss": 0.3274,
+      "step": 2543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908912718184741e-05,
+      "loss": 0.8374,
+      "step": 2544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1903216382174664e-05,
+      "loss": 0.8262,
+      "step": 2545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897519405321956e-05,
+      "loss": 0.8691,
+      "step": 2546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891821789544871e-05,
+      "loss": 0.853,
+      "step": 2547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886123536761886e-05,
+      "loss": 0.8901,
+      "step": 2548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1880424648891693e-05,
+      "loss": 0.8154,
+      "step": 2549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18747251278532e-05,
+      "loss": 0.7856,
+      "step": 2550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869024975565521e-05,
+      "loss": 0.8823,
+      "step": 2551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863324193947984e-05,
+      "loss": 0.9175,
+      "step": 2552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857622784920138e-05,
+      "loss": 0.8589,
+      "step": 2553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851920750401728e-05,
+      "loss": 0.8848,
+      "step": 2554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846218092312725e-05,
+      "loss": 0.8911,
+      "step": 2555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1840514812573297e-05,
+      "loss": 0.8276,
+      "step": 2556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183481091310383e-05,
+      "loss": 0.8975,
+      "step": 2557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1829106395824918e-05,
+      "loss": 0.8442,
+      "step": 2558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823401262657357e-05,
+      "loss": 0.8154,
+      "step": 2559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817695515522156e-05,
+      "loss": 0.8535,
+      "step": 2560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811989156340533e-05,
+      "loss": 0.8843,
+      "step": 2561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806282187033899e-05,
+      "loss": 0.8389,
+      "step": 2562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180057460952389e-05,
+      "loss": 0.8535,
+      "step": 2563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794866425732327e-05,
+      "loss": 0.8647,
+      "step": 2564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789157637581252e-05,
+      "loss": 0.8809,
+      "step": 2565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17834482469929e-05,
+      "loss": 0.873,
+      "step": 2566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777738255889708e-05,
+      "loss": 0.855,
+      "step": 2567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772027666194324e-05,
+      "loss": 0.8516,
+      "step": 2568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1766316479829592e-05,
+      "loss": 0.8306,
+      "step": 2569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760604698718559e-05,
+      "loss": 0.8013,
+      "step": 2570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754892324784463e-05,
+      "loss": 0.8936,
+      "step": 2571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174917935995076e-05,
+      "loss": 0.8477,
+      "step": 2572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743465806141086e-05,
+      "loss": 0.8579,
+      "step": 2573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173775166527929e-05,
+      "loss": 0.8638,
+      "step": 2574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732036939289407e-05,
+      "loss": 0.9043,
+      "step": 2575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726321630095676e-05,
+      "loss": 0.7783,
+      "step": 2576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720605739622532e-05,
+      "loss": 0.8149,
+      "step": 2577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714889269794605e-05,
+      "loss": 0.856,
+      "step": 2578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709172222536717e-05,
+      "loss": 0.8096,
+      "step": 2579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170345459977389e-05,
+      "loss": 0.8823,
+      "step": 2580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697736403431332e-05,
+      "loss": 0.8506,
+      "step": 2581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692017635434456e-05,
+      "loss": 0.8706,
+      "step": 2582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1686298297708853e-05,
+      "loss": 0.8228,
+      "step": 2583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680578392180318e-05,
+      "loss": 0.8955,
+      "step": 2584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167485792077483e-05,
+      "loss": 0.7803,
+      "step": 2585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669136885418564e-05,
+      "loss": 0.8379,
+      "step": 2586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663415288037879e-05,
+      "loss": 0.8486,
+      "step": 2587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657693130559322e-05,
+      "loss": 0.8481,
+      "step": 2588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651970414909646e-05,
+      "loss": 0.8281,
+      "step": 2589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646247143015762e-05,
+      "loss": 0.8369,
+      "step": 2590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640523316804794e-05,
+      "loss": 0.8491,
+      "step": 2591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634798938204045e-05,
+      "loss": 0.8462,
+      "step": 2592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1629074009141e-05,
+      "loss": 0.8838,
+      "step": 2593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623348531543333e-05,
+      "loss": 0.8193,
+      "step": 2594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1617622507338899e-05,
+      "loss": 0.8633,
+      "step": 2595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.161189593845574e-05,
+      "loss": 0.8525,
+      "step": 2596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606168826822084e-05,
+      "loss": 0.8335,
+      "step": 2597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1600441174366332e-05,
+      "loss": 0.7563,
+      "step": 2598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594712983017088e-05,
+      "loss": 0.8525,
+      "step": 2599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588984254703107e-05,
+      "loss": 0.3496,
+      "step": 2600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583254991353347e-05,
+      "loss": 0.939,
+      "step": 2601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577525194896945e-05,
+      "loss": 0.8413,
+      "step": 2602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1571794867263211e-05,
+      "loss": 0.875,
+      "step": 2603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566064010381633e-05,
+      "loss": 0.8394,
+      "step": 2604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156033262618188e-05,
+      "loss": 0.8096,
+      "step": 2605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15546007165938e-05,
+      "loss": 0.8555,
+      "step": 2606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548868283547416e-05,
+      "loss": 0.8989,
+      "step": 2607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543135328972923e-05,
+      "loss": 0.8013,
+      "step": 2608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537401854800709e-05,
+      "loss": 0.8911,
+      "step": 2609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1531667862961312e-05,
+      "loss": 0.7759,
+      "step": 2610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1525933355385458e-05,
+      "loss": 0.8281,
+      "step": 2611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1520198334004047e-05,
+      "loss": 0.8657,
+      "step": 2612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514462800748156e-05,
+      "loss": 0.7837,
+      "step": 2613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508726757549015e-05,
+      "loss": 0.7725,
+      "step": 2614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502990206338053e-05,
+      "loss": 0.8843,
+      "step": 2615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149725314904685e-05,
+      "loss": 0.8179,
+      "step": 2616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491515587607163e-05,
+      "loss": 0.8799,
+      "step": 2617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485777523950917e-05,
+      "loss": 0.8457,
+      "step": 2618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1480038960010211e-05,
+      "loss": 0.8501,
+      "step": 2619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474299897717305e-05,
+      "loss": 0.8403,
+      "step": 2620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146856033900463e-05,
+      "loss": 0.811,
+      "step": 2621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462820285804793e-05,
+      "loss": 0.835,
+      "step": 2622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457079740050553e-05,
+      "loss": 0.8804,
+      "step": 2623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145133870367484e-05,
+      "loss": 0.897,
+      "step": 2624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445597178610757e-05,
+      "loss": 0.8052,
+      "step": 2625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439855166791559e-05,
+      "loss": 0.8135,
+      "step": 2626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143411267015067e-05,
+      "loss": 0.8076,
+      "step": 2627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428369690621684e-05,
+      "loss": 0.7842,
+      "step": 2628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422626230138348e-05,
+      "loss": 0.8159,
+      "step": 2629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416882290634575e-05,
+      "loss": 0.8013,
+      "step": 2630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411137874044438e-05,
+      "loss": 0.8193,
+      "step": 2631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405392982302175e-05,
+      "loss": 0.8345,
+      "step": 2632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139964761734218e-05,
+      "loss": 0.8159,
+      "step": 2633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393901781099004e-05,
+      "loss": 0.792,
+      "step": 2634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138815547550736e-05,
+      "loss": 0.8047,
+      "step": 2635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382408702502121e-05,
+      "loss": 0.3613,
+      "step": 2636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376661464018315e-05,
+      "loss": 0.7671,
+      "step": 2637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1370913761991128e-05,
+      "loss": 0.8872,
+      "step": 2638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365165598355897e-05,
+      "loss": 0.8247,
+      "step": 2639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359416975048119e-05,
+      "loss": 0.8555,
+      "step": 2640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353667894003446e-05,
+      "loss": 0.8232,
+      "step": 2641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347918357157686e-05,
+      "loss": 0.8188,
+      "step": 2642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342168366446796e-05,
+      "loss": 0.9033,
+      "step": 2643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1336417923806882e-05,
+      "loss": 0.8545,
+      "step": 2644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330667031174215e-05,
+      "loss": 0.8633,
+      "step": 2645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324915690485207e-05,
+      "loss": 0.8184,
+      "step": 2646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319163903676426e-05,
+      "loss": 0.8027,
+      "step": 2647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313411672684589e-05,
+      "loss": 0.8467,
+      "step": 2648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307658999446559e-05,
+      "loss": 0.9023,
+      "step": 2649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301905885899352e-05,
+      "loss": 0.8574,
+      "step": 2650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296152333980128e-05,
+      "loss": 0.8501,
+      "step": 2651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290398345626203e-05,
+      "loss": 0.7603,
+      "step": 2652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284643922775035e-05,
+      "loss": 0.8213,
+      "step": 2653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278889067364218e-05,
+      "loss": 0.8726,
+      "step": 2654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1273133781331513e-05,
+      "loss": 0.8896,
+      "step": 2655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126737806661481e-05,
+      "loss": 0.3391,
+      "step": 2656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261621925152148e-05,
+      "loss": 0.9141,
+      "step": 2657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125586535888171e-05,
+      "loss": 0.8564,
+      "step": 2658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125010836974182e-05,
+      "loss": 0.8721,
+      "step": 2659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244350959670948e-05,
+      "loss": 0.814,
+      "step": 2660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238593130607702e-05,
+      "loss": 0.3346,
+      "step": 2661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232834884490838e-05,
+      "loss": 0.8545,
+      "step": 2662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1227076223259239e-05,
+      "loss": 0.8716,
+      "step": 2663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122131714885194e-05,
+      "loss": 0.8643,
+      "step": 2664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215557663208113e-05,
+      "loss": 0.8828,
+      "step": 2665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209797768267068e-05,
+      "loss": 0.3645,
+      "step": 2666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204037465968243e-05,
+      "loss": 0.8623,
+      "step": 2667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119827675825123e-05,
+      "loss": 0.8613,
+      "step": 2668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192515647055744e-05,
+      "loss": 0.8652,
+      "step": 2669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1186754134321645e-05,
+      "loss": 0.7671,
+      "step": 2670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180992221988923e-05,
+      "loss": 0.8364,
+      "step": 2671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175229911997703e-05,
+      "loss": 0.877,
+      "step": 2672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1169467206288242e-05,
+      "loss": 0.8447,
+      "step": 2673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163704106800936e-05,
+      "loss": 0.8623,
+      "step": 2674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157940615476313e-05,
+      "loss": 0.8765,
+      "step": 2675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152176734255027e-05,
+      "loss": 0.834,
+      "step": 2676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146412465077866e-05,
+      "loss": 0.8218,
+      "step": 2677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140647809885751e-05,
+      "loss": 0.8247,
+      "step": 2678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113488277061973e-05,
+      "loss": 0.8062,
+      "step": 2679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112911734922098e-05,
+      "loss": 0.853,
+      "step": 2680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123351547630819e-05,
+      "loss": 0.8149,
+      "step": 2681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1117585367790668e-05,
+      "loss": 0.7979,
+      "step": 2682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1111818811642095e-05,
+      "loss": 0.9092,
+      "step": 2683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1106051881126792e-05,
+      "loss": 0.8501,
+      "step": 2684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1100284578186571e-05,
+      "loss": 0.8135,
+      "step": 2685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094516904763378e-05,
+      "loss": 0.8179,
+      "step": 2686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108874886279927e-05,
+      "loss": 0.8804,
+      "step": 2687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082980454236445e-05,
+      "loss": 0.8623,
+      "step": 2688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077211681017213e-05,
+      "loss": 0.8848,
+      "step": 2689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071442545084008e-05,
+      "loss": 0.8408,
+      "step": 2690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106567304837939e-05,
+      "loss": 0.7939,
+      "step": 2691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059903192846038e-05,
+      "loss": 0.7983,
+      "step": 2692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105413298042675e-05,
+      "loss": 0.8525,
+      "step": 2693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048362413064453e-05,
+      "loss": 0.8823,
+      "step": 2694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042591492702184e-05,
+      "loss": 0.8486,
+      "step": 2695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1036820221283096e-05,
+      "loss": 0.8608,
+      "step": 2696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031048600750474e-05,
+      "loss": 0.8574,
+      "step": 2697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102527663304771e-05,
+      "loss": 0.8271,
+      "step": 2698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1019504320118314e-05,
+      "loss": 0.7988,
+      "step": 2699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013731663905916e-05,
+      "loss": 0.8823,
+      "step": 2700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007958666354257e-05,
+      "loss": 0.8457,
+      "step": 2701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002185329407198e-05,
+      "loss": 0.8105,
+      "step": 2702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996411655008708e-05,
+      "loss": 0.8599,
+      "step": 2703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990637645102877e-05,
+      "loss": 0.8784,
+      "step": 2704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984863301633905e-05,
+      "loss": 0.3589,
+      "step": 2705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979088626546097e-05,
+      "loss": 0.8691,
+      "step": 2706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097331362178388e-05,
+      "loss": 0.8325,
+      "step": 2707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967538289291793e-05,
+      "loss": 0.8501,
+      "step": 2708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961762631014472e-05,
+      "loss": 0.8516,
+      "step": 2709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955986648896677e-05,
+      "loss": 0.8232,
+      "step": 2710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950210344883269e-05,
+      "loss": 0.3242,
+      "step": 2711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944433720919218e-05,
+      "loss": 0.7754,
+      "step": 2712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938656778949605e-05,
+      "loss": 0.9277,
+      "step": 2713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0932879520919624e-05,
+      "loss": 0.8188,
+      "step": 2714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0927101948774554e-05,
+      "loss": 0.8481,
+      "step": 2715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921324064459799e-05,
+      "loss": 0.8193,
+      "step": 2716
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915545869920869e-05,
+      "loss": 0.8677,
+      "step": 2717
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909767367103366e-05,
+      "loss": 0.8203,
+      "step": 2718
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903988557953003e-05,
+      "loss": 0.8252,
+      "step": 2719
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898209444415593e-05,
+      "loss": 0.8828,
+      "step": 2720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089243002843706e-05,
+      "loss": 0.7876,
+      "step": 2721
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886650311963419e-05,
+      "loss": 0.855,
+      "step": 2722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880870296940787e-05,
+      "loss": 0.856,
+      "step": 2723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875089985315394e-05,
+      "loss": 0.7617,
+      "step": 2724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0869309379033555e-05,
+      "loss": 0.8418,
+      "step": 2725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086352848004169e-05,
+      "loss": 0.8516,
+      "step": 2726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857747290286322e-05,
+      "loss": 0.8633,
+      "step": 2727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851965811714065e-05,
+      "loss": 0.8247,
+      "step": 2728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846184046271632e-05,
+      "loss": 0.8354,
+      "step": 2729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840401995905837e-05,
+      "loss": 0.8188,
+      "step": 2730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834619662563584e-05,
+      "loss": 0.8687,
+      "step": 2731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828837048191875e-05,
+      "loss": 0.8232,
+      "step": 2732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823054154737804e-05,
+      "loss": 0.8779,
+      "step": 2733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0817270984148569e-05,
+      "loss": 0.8184,
+      "step": 2734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811487538371449e-05,
+      "loss": 0.3594,
+      "step": 2735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805703819353815e-05,
+      "loss": 0.8384,
+      "step": 2736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0799919829043143e-05,
+      "loss": 0.8569,
+      "step": 2737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794135569386994e-05,
+      "loss": 0.875,
+      "step": 2738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078835104233301e-05,
+      "loss": 0.8608,
+      "step": 2739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782566249828937e-05,
+      "loss": 0.8408,
+      "step": 2740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776781193822605e-05,
+      "loss": 0.8599,
+      "step": 2741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770995876261932e-05,
+      "loss": 0.3528,
+      "step": 2742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765210299094924e-05,
+      "loss": 0.832,
+      "step": 2743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759424464269676e-05,
+      "loss": 0.8042,
+      "step": 2744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753638373734367e-05,
+      "loss": 0.856,
+      "step": 2745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747852029437264e-05,
+      "loss": 0.9102,
+      "step": 2746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742065433326722e-05,
+      "loss": 0.896,
+      "step": 2747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073627858735118e-05,
+      "loss": 0.3518,
+      "step": 2748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0730491493459149e-05,
+      "loss": 0.853,
+      "step": 2749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724704153599245e-05,
+      "loss": 0.8799,
+      "step": 2750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718916569720153e-05,
+      "loss": 0.8306,
+      "step": 2751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0713128743770638e-05,
+      "loss": 0.9023,
+      "step": 2752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707340677699554e-05,
+      "loss": 0.8389,
+      "step": 2753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701552373455835e-05,
+      "loss": 0.8218,
+      "step": 2754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695763832988491e-05,
+      "loss": 0.8408,
+      "step": 2755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689975058246611e-05,
+      "loss": 0.8374,
+      "step": 2756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0684186051179372e-05,
+      "loss": 0.8252,
+      "step": 2757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678396813736023e-05,
+      "loss": 0.8706,
+      "step": 2758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0672607347865883e-05,
+      "loss": 0.3508,
+      "step": 2759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0666817655518362e-05,
+      "loss": 0.8779,
+      "step": 2760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066102773864294e-05,
+      "loss": 0.9141,
+      "step": 2761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655237599189168e-05,
+      "loss": 0.8789,
+      "step": 2762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649447239106678e-05,
+      "loss": 0.8438,
+      "step": 2763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643656660345177e-05,
+      "loss": 0.8789,
+      "step": 2764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063786586485444e-05,
+      "loss": 0.8604,
+      "step": 2765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632074854584315e-05,
+      "loss": 0.8633,
+      "step": 2766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626283631484736e-05,
+      "loss": 0.3313,
+      "step": 2767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062049219750569e-05,
+      "loss": 0.769,
+      "step": 2768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614700554597243e-05,
+      "loss": 0.8433,
+      "step": 2769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608908704709535e-05,
+      "loss": 0.8389,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0603116649792771e-05,
+      "loss": 0.8745,
+      "step": 2771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0597324391797224e-05,
+      "loss": 0.8486,
+      "step": 2772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591531932673242e-05,
+      "loss": 0.8535,
+      "step": 2773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585739274371232e-05,
+      "loss": 0.8613,
+      "step": 2774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579946418841671e-05,
+      "loss": 0.8477,
+      "step": 2775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574153368035112e-05,
+      "loss": 0.8662,
+      "step": 2776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0568360123902161e-05,
+      "loss": 0.3425,
+      "step": 2777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056256668839349e-05,
+      "loss": 0.8193,
+      "step": 2778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0556773063459842e-05,
+      "loss": 0.8135,
+      "step": 2779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0550979251052023e-05,
+      "loss": 0.7803,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545185253120898e-05,
+      "loss": 0.8013,
+      "step": 2781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0539391071617394e-05,
+      "loss": 0.8389,
+      "step": 2782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533596708492509e-05,
+      "loss": 0.8311,
+      "step": 2783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527802165697288e-05,
+      "loss": 0.8237,
+      "step": 2784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522007445182844e-05,
+      "loss": 0.9062,
+      "step": 2785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0516212548900357e-05,
+      "loss": 0.7856,
+      "step": 2786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510417478801059e-05,
+      "loss": 0.3228,
+      "step": 2787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504622236836229e-05,
+      "loss": 0.7812,
+      "step": 2788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498826824957225e-05,
+      "loss": 0.8306,
+      "step": 2789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049303124511545e-05,
+      "loss": 0.8687,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487235499262367e-05,
+      "loss": 0.8813,
+      "step": 2791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481439589349493e-05,
+      "loss": 0.8594,
+      "step": 2792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475643517328403e-05,
+      "loss": 0.8618,
+      "step": 2793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046984728515072e-05,
+      "loss": 0.8394,
+      "step": 2794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046405089476813e-05,
+      "loss": 0.8726,
+      "step": 2795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458254348132371e-05,
+      "loss": 0.8447,
+      "step": 2796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452457647195225e-05,
+      "loss": 0.876,
+      "step": 2797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044666079390853e-05,
+      "loss": 0.877,
+      "step": 2798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440863790224183e-05,
+      "loss": 0.8022,
+      "step": 2799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435066638094122e-05,
+      "loss": 0.8813,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042926933947034e-05,
+      "loss": 0.9307,
+      "step": 2801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423471896304878e-05,
+      "loss": 0.8394,
+      "step": 2802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417674310549826e-05,
+      "loss": 0.3403,
+      "step": 2803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411876584157316e-05,
+      "loss": 0.8628,
+      "step": 2804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040607871907954e-05,
+      "loss": 0.9102,
+      "step": 2805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400280717268728e-05,
+      "loss": 0.8379,
+      "step": 2806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0394482580677155e-05,
+      "loss": 0.7944,
+      "step": 2807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388684311257143e-05,
+      "loss": 0.7524,
+      "step": 2808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382885910961064e-05,
+      "loss": 0.835,
+      "step": 2809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037708738174133e-05,
+      "loss": 0.3289,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371288725550392e-05,
+      "loss": 0.8232,
+      "step": 2811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036548994434075e-05,
+      "loss": 0.8921,
+      "step": 2812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359691040064945e-05,
+      "loss": 0.8228,
+      "step": 2813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353892014675558e-05,
+      "loss": 0.7832,
+      "step": 2814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034809287012521e-05,
+      "loss": 0.8726,
+      "step": 2815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342293608366564e-05,
+      "loss": 0.8887,
+      "step": 2816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336494231352323e-05,
+      "loss": 0.834,
+      "step": 2817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330694741035225e-05,
+      "loss": 0.833,
+      "step": 2818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0324895139368055e-05,
+      "loss": 0.8301,
+      "step": 2819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319095428303623e-05,
+      "loss": 0.8213,
+      "step": 2820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0313295609794782e-05,
+      "loss": 0.8164,
+      "step": 2821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307495685794426e-05,
+      "loss": 0.7983,
+      "step": 2822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301695658255478e-05,
+      "loss": 0.8462,
+      "step": 2823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295895529130895e-05,
+      "loss": 0.3269,
+      "step": 2824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290095300373675e-05,
+      "loss": 0.8384,
+      "step": 2825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028429497393684e-05,
+      "loss": 0.8926,
+      "step": 2826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278494551773459e-05,
+      "loss": 0.3533,
+      "step": 2827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0272694035836615e-05,
+      "loss": 0.8438,
+      "step": 2828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026689342807944e-05,
+      "loss": 0.8745,
+      "step": 2829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261092730455085e-05,
+      "loss": 0.3406,
+      "step": 2830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255291944916734e-05,
+      "loss": 0.8623,
+      "step": 2831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249491073417605e-05,
+      "loss": 0.8496,
+      "step": 2832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243690117910941e-05,
+      "loss": 0.856,
+      "step": 2833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237889080350014e-05,
+      "loss": 0.8604,
+      "step": 2834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232087962688126e-05,
+      "loss": 0.8169,
+      "step": 2835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02262867668786e-05,
+      "loss": 0.8516,
+      "step": 2836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220485494874792e-05,
+      "loss": 0.8184,
+      "step": 2837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214684148630076e-05,
+      "loss": 0.8281,
+      "step": 2838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208882730097864e-05,
+      "loss": 0.8037,
+      "step": 2839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203081241231579e-05,
+      "loss": 0.3276,
+      "step": 2840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197279683984669e-05,
+      "loss": 0.8457,
+      "step": 2841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191478060310613e-05,
+      "loss": 0.8066,
+      "step": 2842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185676372162912e-05,
+      "loss": 0.8472,
+      "step": 2843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179874621495081e-05,
+      "loss": 0.8501,
+      "step": 2844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174072810260658e-05,
+      "loss": 0.8687,
+      "step": 2845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168270940413205e-05,
+      "loss": 0.8335,
+      "step": 2846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162469013906302e-05,
+      "loss": 0.8198,
+      "step": 2847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156667032693548e-05,
+      "loss": 0.9023,
+      "step": 2848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150864998728565e-05,
+      "loss": 0.8667,
+      "step": 2849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145062913964982e-05,
+      "loss": 0.7876,
+      "step": 2850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013926078035645e-05,
+      "loss": 0.8184,
+      "step": 2851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133458599856647e-05,
+      "loss": 0.8086,
+      "step": 2852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012765637441925e-05,
+      "loss": 0.8677,
+      "step": 2853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0121854105997962e-05,
+      "loss": 0.8608,
+      "step": 2854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116051796546496e-05,
+      "loss": 0.8271,
+      "step": 2855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110249448018578e-05,
+      "loss": 0.8286,
+      "step": 2856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104447062367955e-05,
+      "loss": 0.8311,
+      "step": 2857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098644641548372e-05,
+      "loss": 0.9004,
+      "step": 2858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0092842187513607e-05,
+      "loss": 0.3184,
+      "step": 2859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087039702217427e-05,
+      "loss": 0.8413,
+      "step": 2860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081237187613619e-05,
+      "loss": 0.8667,
+      "step": 2861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0075434645655987e-05,
+      "loss": 0.3596,
+      "step": 2862
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069632078298333e-05,
+      "loss": 0.3472,
+      "step": 2863
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063829487494467e-05,
+      "loss": 0.834,
+      "step": 2864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058026875198223e-05,
+      "loss": 0.9111,
+      "step": 2865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052224243363424e-05,
+      "loss": 0.8174,
+      "step": 2866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046421593943907e-05,
+      "loss": 0.8975,
+      "step": 2867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040618928893516e-05,
+      "loss": 0.814,
+      "step": 2868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034816250166097e-05,
+      "loss": 0.8462,
+      "step": 2869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029013559715506e-05,
+      "loss": 0.3591,
+      "step": 2870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023210859495594e-05,
+      "loss": 0.8105,
+      "step": 2871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0017408151460223e-05,
+      "loss": 0.8301,
+      "step": 2872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011605437563263e-05,
+      "loss": 0.833,
+      "step": 2873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005802719758563e-05,
+      "loss": 0.8516,
+      "step": 2874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.854,
+      "step": 2875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994197280241442e-06,
+      "loss": 0.834,
+      "step": 2876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98839456243674e-06,
+      "loss": 0.8955,
+      "step": 2877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982591848539779e-06,
+      "loss": 0.8511,
+      "step": 2878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976789140504406e-06,
+      "loss": 0.8511,
+      "step": 2879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970986440284497e-06,
+      "loss": 0.8379,
+      "step": 2880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965183749833907e-06,
+      "loss": 0.8447,
+      "step": 2881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959381071106486e-06,
+      "loss": 0.8389,
+      "step": 2882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953578406056096e-06,
+      "loss": 0.8735,
+      "step": 2883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947775756636578e-06,
+      "loss": 0.7192,
+      "step": 2884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94197312480178e-06,
+      "loss": 0.8145,
+      "step": 2885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936170512505536e-06,
+      "loss": 0.8413,
+      "step": 2886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930367921701672e-06,
+      "loss": 0.8237,
+      "step": 2887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.924565354344018e-06,
+      "loss": 0.8501,
+      "step": 2888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918762812386383e-06,
+      "loss": 0.8433,
+      "step": 2889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912960297782576e-06,
+      "loss": 0.8071,
+      "step": 2890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907157812486398e-06,
+      "loss": 0.8237,
+      "step": 2891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901355358451628e-06,
+      "loss": 0.8735,
+      "step": 2892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89555293763205e-06,
+      "loss": 0.8394,
+      "step": 2893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889750551981422e-06,
+      "loss": 0.8145,
+      "step": 2894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883948203453509e-06,
+      "loss": 0.8784,
+      "step": 2895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878145894002042e-06,
+      "loss": 0.8325,
+      "step": 2896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872343625580752e-06,
+      "loss": 0.3281,
+      "step": 2897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866541400143357e-06,
+      "loss": 0.9097,
+      "step": 2898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86073921964355e-06,
+      "loss": 0.8379,
+      "step": 2899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854937086035021e-06,
+      "loss": 0.7612,
+      "step": 2900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84913500127144e-06,
+      "loss": 0.8452,
+      "step": 2901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843332967306452e-06,
+      "loss": 0.8521,
+      "step": 2902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8375309860937e-06,
+      "loss": 0.814,
+      "step": 2903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8317290595868e-06,
+      "loss": 0.7959,
+      "step": 2904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825927189739347e-06,
+      "loss": 0.8052,
+      "step": 2905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820125378504924e-06,
+      "loss": 0.855,
+      "step": 2906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.814323627837088e-06,
+      "loss": 0.8369,
+      "step": 2907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808521939689389e-06,
+      "loss": 0.8833,
+      "step": 2908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802720316015334e-06,
+      "loss": 0.7412,
+      "step": 2909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.796918758768426e-06,
+      "loss": 0.7993,
+      "step": 2910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791117269902141e-06,
+      "loss": 0.8574,
+      "step": 2911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785315851369926e-06,
+      "loss": 0.8618,
+      "step": 2912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779514505125212e-06,
+      "loss": 0.8848,
+      "step": 2913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773713233121405e-06,
+      "loss": 0.8149,
+      "step": 2914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767912037311877e-06,
+      "loss": 0.8315,
+      "step": 2915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762110919649988e-06,
+      "loss": 0.8545,
+      "step": 2916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756309882089059e-06,
+      "loss": 0.8589,
+      "step": 2917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750508926582398e-06,
+      "loss": 0.9067,
+      "step": 2918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74470805508327e-06,
+      "loss": 0.7646,
+      "step": 2919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738907269544918e-06,
+      "loss": 0.8125,
+      "step": 2920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733106571920564e-06,
+      "loss": 0.8208,
+      "step": 2921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.727305964163387e-06,
+      "loss": 0.291,
+      "step": 2922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721505448226544e-06,
+      "loss": 0.8335,
+      "step": 2923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715705026063164e-06,
+      "loss": 0.832,
+      "step": 2924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.709904699626329e-06,
+      "loss": 0.9072,
+      "step": 2925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704104470869109e-06,
+      "loss": 0.7964,
+      "step": 2926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698304341744524e-06,
+      "loss": 0.8438,
+      "step": 2927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692504314205576e-06,
+      "loss": 0.8291,
+      "step": 2928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68670439020522e-06,
+      "loss": 0.9146,
+      "step": 2929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.680904571696378e-06,
+      "loss": 0.8896,
+      "step": 2930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.67510486063195e-06,
+      "loss": 0.8784,
+      "step": 2931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669305258964775e-06,
+      "loss": 0.8618,
+      "step": 2932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663505768647678e-06,
+      "loss": 0.8394,
+      "step": 2933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65770639163344e-06,
+      "loss": 0.8486,
+      "step": 2934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651907129874792e-06,
+      "loss": 0.8599,
+      "step": 2935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646107985324445e-06,
+      "loss": 0.7271,
+      "step": 2936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640308959935055e-06,
+      "loss": 0.8213,
+      "step": 2937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634510055659253e-06,
+      "loss": 0.8032,
+      "step": 2938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628711274449612e-06,
+      "loss": 0.8242,
+      "step": 2939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622912618258672e-06,
+      "loss": 0.8291,
+      "step": 2940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617114089038938e-06,
+      "loss": 0.8833,
+      "step": 2941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611315688742859e-06,
+      "loss": 0.8872,
+      "step": 2942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605517419322849e-06,
+      "loss": 0.8608,
+      "step": 2943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599719282731277e-06,
+      "loss": 0.8066,
+      "step": 2944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.593921280920461e-06,
+      "loss": 0.7905,
+      "step": 2945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588123415842686e-06,
+      "loss": 0.7881,
+      "step": 2946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582325689450175e-06,
+      "loss": 0.8584,
+      "step": 2947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576528103695124e-06,
+      "loss": 0.8716,
+      "step": 2948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570730660529662e-06,
+      "loss": 0.8745,
+      "step": 2949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564933361905878e-06,
+      "loss": 0.8472,
+      "step": 2950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55913620977582e-06,
+      "loss": 0.3582,
+      "step": 2951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553339206091473e-06,
+      "loss": 0.8501,
+      "step": 2952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547542352804779e-06,
+      "loss": 0.8179,
+      "step": 2953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541745651867634e-06,
+      "loss": 0.8838,
+      "step": 2954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53594910523187e-06,
+      "loss": 0.8867,
+      "step": 2955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530152714849281e-06,
+      "loss": 0.8701,
+      "step": 2956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524356482671602e-06,
+      "loss": 0.873,
+      "step": 2957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518560410650509e-06,
+      "loss": 0.8066,
+      "step": 2958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512764500737636e-06,
+      "loss": 0.342,
+      "step": 2959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50696875488455e-06,
+      "loss": 0.3252,
+      "step": 2960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501173175042778e-06,
+      "loss": 0.7637,
+      "step": 2961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495377763163774e-06,
+      "loss": 0.8325,
+      "step": 2962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489582521198946e-06,
+      "loss": 0.8613,
+      "step": 2963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483787451099646e-06,
+      "loss": 0.8735,
+      "step": 2964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477992554817156e-06,
+      "loss": 0.8691,
+      "step": 2965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472197834302715e-06,
+      "loss": 0.7627,
+      "step": 2966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466403291507498e-06,
+      "loss": 0.8467,
+      "step": 2967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460608928382609e-06,
+      "loss": 0.8472,
+      "step": 2968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454814746879105e-06,
+      "loss": 0.8369,
+      "step": 2969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449020748947977e-06,
+      "loss": 0.8555,
+      "step": 2970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443226936540161e-06,
+      "loss": 0.8228,
+      "step": 2971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437433311606514e-06,
+      "loss": 0.2963,
+      "step": 2972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431639876097842e-06,
+      "loss": 0.8237,
+      "step": 2973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425846631964892e-06,
+      "loss": 0.7627,
+      "step": 2974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420053581158329e-06,
+      "loss": 0.8818,
+      "step": 2975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414260725628772e-06,
+      "loss": 0.8105,
+      "step": 2976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408468067326763e-06,
+      "loss": 0.895,
+      "step": 2977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402675608202777e-06,
+      "loss": 0.8618,
+      "step": 2978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.396883350207234e-06,
+      "loss": 0.8477,
+      "step": 2979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391091295290467e-06,
+      "loss": 0.7896,
+      "step": 2980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38529944540276e-06,
+      "loss": 0.8599,
+      "step": 2981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379507802494315e-06,
+      "loss": 0.8921,
+      "step": 2982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373716368515267e-06,
+      "loss": 0.8765,
+      "step": 2983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367925145415688e-06,
+      "loss": 0.856,
+      "step": 2984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362134135145565e-06,
+      "loss": 0.8169,
+      "step": 2985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356343339654826e-06,
+      "loss": 0.8032,
+      "step": 2986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350552760893327e-06,
+      "loss": 0.3396,
+      "step": 2987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.344762400810836e-06,
+      "loss": 0.322,
+      "step": 2988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338972261357062e-06,
+      "loss": 0.9297,
+      "step": 2989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333182344481637e-06,
+      "loss": 0.811,
+      "step": 2990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32739265213412e-06,
+      "loss": 0.813,
+      "step": 2991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32160318626398e-06,
+      "loss": 0.7998,
+      "step": 2992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315813948820628e-06,
+      "loss": 0.7583,
+      "step": 2993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310024941753392e-06,
+      "loss": 0.8188,
+      "step": 2994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304236167011514e-06,
+      "loss": 0.7798,
+      "step": 2995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298447626544168e-06,
+      "loss": 0.8569,
+      "step": 2996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29265932230045e-06,
+      "loss": 0.8237,
+      "step": 2997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286871256229365e-06,
+      "loss": 0.8276,
+      "step": 2998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28108343027985e-06,
+      "loss": 0.8413,
+      "step": 2999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275295846400754e-06,
+      "loss": 0.3176,
+      "step": 3000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.269508506540853e-06,
+      "loss": 0.79,
+      "step": 3001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263721412648826e-06,
+      "loss": 0.8623,
+      "step": 3002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257934566673278e-06,
+      "loss": 0.8662,
+      "step": 3003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252147970562737e-06,
+      "loss": 0.8467,
+      "step": 3004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246361626265638e-06,
+      "loss": 0.8477,
+      "step": 3005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240575535730327e-06,
+      "loss": 0.8403,
+      "step": 3006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23478970090508e-06,
+      "loss": 0.8511,
+      "step": 3007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22900412373807e-06,
+      "loss": 0.8979,
+      "step": 3008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223218806177398e-06,
+      "loss": 0.855,
+      "step": 3009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217433750171068e-06,
+      "loss": 0.7905,
+      "step": 3010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211648957666995e-06,
+      "loss": 0.7988,
+      "step": 3011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205864430613011e-06,
+      "loss": 0.8813,
+      "step": 3012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200080170956858e-06,
+      "loss": 0.8208,
+      "step": 3013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194296180646188e-06,
+      "loss": 0.8027,
+      "step": 3014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188512461628558e-06,
+      "loss": 0.814,
+      "step": 3015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.182729015851433e-06,
+      "loss": 0.8711,
+      "step": 3016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176945845262198e-06,
+      "loss": 0.8135,
+      "step": 3017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171162951808128e-06,
+      "loss": 0.8311,
+      "step": 3018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165380337436419e-06,
+      "loss": 0.7803,
+      "step": 3019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159598004094168e-06,
+      "loss": 0.8003,
+      "step": 3020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15381595372837e-06,
+      "loss": 0.8774,
+      "step": 3021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148034188285938e-06,
+      "loss": 0.8931,
+      "step": 3022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14225270971368e-06,
+      "loss": 0.8628,
+      "step": 3023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136471519958314e-06,
+      "loss": 0.8252,
+      "step": 3024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13069062096645e-06,
+      "loss": 0.8643,
+      "step": 3025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124910014684609e-06,
+      "loss": 0.8501,
+      "step": 3026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119129703059216e-06,
+      "loss": 0.8066,
+      "step": 3027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113349688036586e-06,
+      "loss": 0.8301,
+      "step": 3028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107569971562944e-06,
+      "loss": 0.3308,
+      "step": 3029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101790555584412e-06,
+      "loss": 0.853,
+      "step": 3030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096011442047e-06,
+      "loss": 0.8252,
+      "step": 3031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090232632896639e-06,
+      "loss": 0.917,
+      "step": 3032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084454130079133e-06,
+      "loss": 0.916,
+      "step": 3033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078675935540203e-06,
+      "loss": 0.8638,
+      "step": 3034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072898051225451e-06,
+      "loss": 0.8408,
+      "step": 3035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06712047908038e-06,
+      "loss": 0.8252,
+      "step": 3036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061343221050397e-06,
+      "loss": 0.8311,
+      "step": 3037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055566279080784e-06,
+      "loss": 0.7969,
+      "step": 3038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049789655116734e-06,
+      "loss": 0.7637,
+      "step": 3039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044013351103328e-06,
+      "loss": 0.8374,
+      "step": 3040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.038237368985531e-06,
+      "loss": 0.8521,
+      "step": 3041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03246171070821e-06,
+      "loss": 0.8335,
+      "step": 3042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.026686378216118e-06,
+      "loss": 0.8867,
+      "step": 3043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020911373453906e-06,
+      "loss": 0.8809,
+      "step": 3044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.0151366983661e-06,
+      "loss": 0.8496,
+      "step": 3045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009362354897123e-06,
+      "loss": 0.8174,
+      "step": 3046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003588344991293e-06,
+      "loss": 0.8477,
+      "step": 3047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997814670592803e-06,
+      "loss": 0.8394,
+      "step": 3048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992041333645744e-06,
+      "loss": 0.7886,
+      "step": 3049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98626833609409e-06,
+      "loss": 0.8906,
+      "step": 3050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98049567988169e-06,
+      "loss": 0.8315,
+      "step": 3051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.974723366952293e-06,
+      "loss": 0.8345,
+      "step": 3052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96895139924953e-06,
+      "loss": 0.833,
+      "step": 3053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963179778716906e-06,
+      "loss": 0.7812,
+      "step": 3054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95740850729782e-06,
+      "loss": 0.7949,
+      "step": 3055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951637586935548e-06,
+      "loss": 0.8018,
+      "step": 3056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945867019573253e-06,
+      "loss": 0.3601,
+      "step": 3057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940096807153967e-06,
+      "loss": 0.7876,
+      "step": 3058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934326951620613e-06,
+      "loss": 0.7881,
+      "step": 3059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928557454915997e-06,
+      "loss": 0.8145,
+      "step": 3060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922788318982792e-06,
+      "loss": 0.8647,
+      "step": 3061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917019545763558e-06,
+      "loss": 0.8652,
+      "step": 3062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911251137200733e-06,
+      "loss": 0.8286,
+      "step": 3063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905483095236627e-06,
+      "loss": 0.3413,
+      "step": 3064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89971542181343e-06,
+      "loss": 0.3217,
+      "step": 3065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89394811887321e-06,
+      "loss": 0.8511,
+      "step": 3066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888181188357909e-06,
+      "loss": 0.8403,
+      "step": 3067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882414632209337e-06,
+      "loss": 0.8281,
+      "step": 3068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876648452369184e-06,
+      "loss": 0.8647,
+      "step": 3069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870882650779021e-06,
+      "loss": 0.834,
+      "step": 3070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865117229380272e-06,
+      "loss": 0.7959,
+      "step": 3071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859352190114252e-06,
+      "loss": 0.8311,
+      "step": 3072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853587534922139e-06,
+      "loss": 0.7925,
+      "step": 3073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847823265744976e-06,
+      "loss": 0.8369,
+      "step": 3074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842059384523689e-06,
+      "loss": 0.8408,
+      "step": 3075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836295893199064e-06,
+      "loss": 0.8022,
+      "step": 3076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830532793711761e-06,
+      "loss": 0.8145,
+      "step": 3077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824770088002302e-06,
+      "loss": 0.8105,
+      "step": 3078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819007778011079e-06,
+      "loss": 0.8604,
+      "step": 3079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.813245865678359e-06,
+      "loss": 0.8472,
+      "step": 3080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807484352944256e-06,
+      "loss": 0.7671,
+      "step": 3081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801723241748774e-06,
+      "loss": 0.8774,
+      "step": 3082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795962534031762e-06,
+      "loss": 0.8579,
+      "step": 3083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790202231732935e-06,
+      "loss": 0.7458,
+      "step": 3084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784442336791889e-06,
+      "loss": 0.8516,
+      "step": 3085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778682851148062e-06,
+      "loss": 0.8184,
+      "step": 3086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772923776740764e-06,
+      "loss": 0.8369,
+      "step": 3087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767165115509169e-06,
+      "loss": 0.7954,
+      "step": 3088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7614068693923e-06,
+      "loss": 0.7969,
+      "step": 3089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755649040329055e-06,
+      "loss": 0.3547,
+      "step": 3090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749891630258181e-06,
+      "loss": 0.8447,
+      "step": 3091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744134641118294e-06,
+      "loss": 0.8345,
+      "step": 3092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738378074847855e-06,
+      "loss": 0.3455,
+      "step": 3093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73262193338519e-06,
+      "loss": 0.8008,
+      "step": 3094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72686621866849e-06,
+      "loss": 0.8442,
+      "step": 3095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721110932635782e-06,
+      "loss": 0.8335,
+      "step": 3096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71535607722497e-06,
+      "loss": 0.9043,
+      "step": 3097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7096016543738e-06,
+      "loss": 0.8989,
+      "step": 3098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703847666019873e-06,
+      "loss": 0.8574,
+      "step": 3099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698094114100651e-06,
+      "loss": 0.8101,
+      "step": 3100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692341000553447e-06,
+      "loss": 0.9092,
+      "step": 3101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686588327315414e-06,
+      "loss": 0.7979,
+      "step": 3102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680836096323576e-06,
+      "loss": 0.8452,
+      "step": 3103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675084309514793e-06,
+      "loss": 0.8159,
+      "step": 3104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669332968825788e-06,
+      "loss": 0.772,
+      "step": 3105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663582076193121e-06,
+      "loss": 0.8628,
+      "step": 3106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65783163355321e-06,
+      "loss": 0.832,
+      "step": 3107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652081642842319e-06,
+      "loss": 0.9053,
+      "step": 3108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.646332105996556e-06,
+      "loss": 0.8638,
+      "step": 3109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640583024951885e-06,
+      "loss": 0.8887,
+      "step": 3110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634834401644108e-06,
+      "loss": 0.8716,
+      "step": 3111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.629086238008876e-06,
+      "loss": 0.875,
+      "step": 3112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623338535981686e-06,
+      "loss": 0.8813,
+      "step": 3113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.617591297497879e-06,
+      "loss": 0.8892,
+      "step": 3114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611844524492643e-06,
+      "loss": 0.8262,
+      "step": 3115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606098218901e-06,
+      "loss": 0.832,
+      "step": 3116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600352382657822e-06,
+      "loss": 0.7725,
+      "step": 3117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594607017697828e-06,
+      "loss": 0.8389,
+      "step": 3118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588862125955562e-06,
+      "loss": 0.833,
+      "step": 3119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583117709365428e-06,
+      "loss": 0.9058,
+      "step": 3120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577373769861657e-06,
+      "loss": 0.3701,
+      "step": 3121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57163030937832e-06,
+      "loss": 0.833,
+      "step": 3122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565887329849332e-06,
+      "loss": 0.7715,
+      "step": 3123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560144833208443e-06,
+      "loss": 0.8213,
+      "step": 3124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554402821389248e-06,
+      "loss": 0.793,
+      "step": 3125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548661296325163e-06,
+      "loss": 0.7676,
+      "step": 3126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542920259949448e-06,
+      "loss": 0.7905,
+      "step": 3127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537179714195209e-06,
+      "loss": 0.7651,
+      "step": 3128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.531439660995368e-06,
+      "loss": 0.875,
+      "step": 3129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525700102282698e-06,
+      "loss": 0.7891,
+      "step": 3130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.519961039989794e-06,
+      "loss": 0.7837,
+      "step": 3131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514222476049085e-06,
+      "loss": 0.8174,
+      "step": 3132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50848441239284e-06,
+      "loss": 0.8159,
+      "step": 3133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502746850953151e-06,
+      "loss": 0.7886,
+      "step": 3134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497009793661948e-06,
+      "loss": 0.8398,
+      "step": 3135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491273242450986e-06,
+      "loss": 0.3134,
+      "step": 3136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485537199251848e-06,
+      "loss": 0.8491,
+      "step": 3137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479801665995954e-06,
+      "loss": 0.9399,
+      "step": 3138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474066644614544e-06,
+      "loss": 0.8486,
+      "step": 3139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.468332137038692e-06,
+      "loss": 0.7832,
+      "step": 3140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462598145199296e-06,
+      "loss": 0.8037,
+      "step": 3141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456864671027077e-06,
+      "loss": 0.7988,
+      "step": 3142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451131716452587e-06,
+      "loss": 0.8496,
+      "step": 3143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445399283406202e-06,
+      "loss": 0.8398,
+      "step": 3144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439667373818123e-06,
+      "loss": 0.832,
+      "step": 3145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433935989618372e-06,
+      "loss": 0.8208,
+      "step": 3146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42820513273679e-06,
+      "loss": 0.7388,
+      "step": 3147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422474805103056e-06,
+      "loss": 0.8384,
+      "step": 3148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.416745008646651e-06,
+      "loss": 0.8125,
+      "step": 3149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411015745296895e-06,
+      "loss": 0.7856,
+      "step": 3150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40528701698292e-06,
+      "loss": 0.8477,
+      "step": 3151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399558825633666e-06,
+      "loss": 0.7622,
+      "step": 3152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39383117317792e-06,
+      "loss": 0.3003,
+      "step": 3153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.388104061544265e-06,
+      "loss": 0.8237,
+      "step": 3154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382377492661105e-06,
+      "loss": 0.877,
+      "step": 3155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37665146845667e-06,
+      "loss": 0.8896,
+      "step": 3156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370925990859e-06,
+      "loss": 0.8428,
+      "step": 3157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365201061795956e-06,
+      "loss": 0.8276,
+      "step": 3158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359476683195208e-06,
+      "loss": 0.7944,
+      "step": 3159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35375285698424e-06,
+      "loss": 0.8179,
+      "step": 3160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348029585090361e-06,
+      "loss": 0.8408,
+      "step": 3161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342306869440676e-06,
+      "loss": 0.3408,
+      "step": 3162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336584711962124e-06,
+      "loss": 0.8481,
+      "step": 3163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33086311458144e-06,
+      "loss": 0.79,
+      "step": 3164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325142079225171e-06,
+      "loss": 0.8477,
+      "step": 3165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319421607819684e-06,
+      "loss": 0.8672,
+      "step": 3166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313701702291147e-06,
+      "loss": 0.8281,
+      "step": 3167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307982364565549e-06,
+      "loss": 0.3398,
+      "step": 3168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30226359656867e-06,
+      "loss": 0.811,
+      "step": 3169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296545400226113e-06,
+      "loss": 0.8706,
+      "step": 3170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290827777463286e-06,
+      "loss": 0.8555,
+      "step": 3171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285110730205396e-06,
+      "loss": 0.8311,
+      "step": 3172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27939426037747e-06,
+      "loss": 0.7764,
+      "step": 3173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273678369904328e-06,
+      "loss": 0.7974,
+      "step": 3174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267963060710597e-06,
+      "loss": 0.8374,
+      "step": 3175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262248334720714e-06,
+      "loss": 0.8398,
+      "step": 3176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256534193858914e-06,
+      "loss": 0.8857,
+      "step": 3177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250820640049244e-06,
+      "loss": 0.8145,
+      "step": 3178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245107675215539e-06,
+      "loss": 0.8628,
+      "step": 3179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239395301281445e-06,
+      "loss": 0.8569,
+      "step": 3180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.233683520170411e-06,
+      "loss": 0.8584,
+      "step": 3181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227972333805678e-06,
+      "loss": 0.8354,
+      "step": 3182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222261744110293e-06,
+      "loss": 0.8662,
+      "step": 3183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216551753007107e-06,
+      "loss": 0.8374,
+      "step": 3184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21084236241875e-06,
+      "loss": 0.8242,
+      "step": 3185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205133574267675e-06,
+      "loss": 0.8613,
+      "step": 3186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199425390476112e-06,
+      "loss": 0.8394,
+      "step": 3187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193717812966103e-06,
+      "loss": 0.8496,
+      "step": 3188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18801084365947e-06,
+      "loss": 0.7964,
+      "step": 3189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182304484477844e-06,
+      "loss": 0.8408,
+      "step": 3190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176598737342647e-06,
+      "loss": 0.877,
+      "step": 3191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170893604175083e-06,
+      "loss": 0.8594,
+      "step": 3192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16518908689617e-06,
+      "loss": 0.8252,
+      "step": 3193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159485187426708e-06,
+      "loss": 0.8438,
+      "step": 3194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153781907687279e-06,
+      "loss": 0.8306,
+      "step": 3195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148079249598275e-06,
+      "loss": 0.8262,
+      "step": 3196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142377215079864e-06,
+      "loss": 0.7944,
+      "step": 3197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136675806052017e-06,
+      "loss": 0.8174,
+      "step": 3198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130975024434482e-06,
+      "loss": 0.8076,
+      "step": 3199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1252748721468e-06,
+      "loss": 0.854,
+      "step": 3200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119575351108309e-06,
+      "loss": 0.8296,
+      "step": 3201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113876463238117e-06,
+      "loss": 0.8623,
+      "step": 3202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108178210455134e-06,
+      "loss": 0.855,
+      "step": 3203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102480594678049e-06,
+      "loss": 0.8638,
+      "step": 3204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096783617825337e-06,
+      "loss": 0.8584,
+      "step": 3205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09108728181526e-06,
+      "loss": 0.8315,
+      "step": 3206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085391588565869e-06,
+      "loss": 0.8931,
+      "step": 3207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07969653999498e-06,
+      "loss": 0.9038,
+      "step": 3208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074002138020215e-06,
+      "loss": 0.8838,
+      "step": 3209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068308384558962e-06,
+      "loss": 0.8354,
+      "step": 3210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062615281528404e-06,
+      "loss": 0.8774,
+      "step": 3211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056922830845491e-06,
+      "loss": 0.3284,
+      "step": 3212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051231034426958e-06,
+      "loss": 0.8442,
+      "step": 3213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045539894189331e-06,
+      "loss": 0.8569,
+      "step": 3214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039849412048897e-06,
+      "loss": 0.8281,
+      "step": 3215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034159589921734e-06,
+      "loss": 0.8433,
+      "step": 3216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028470429723694e-06,
+      "loss": 0.8335,
+      "step": 3217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022781933370403e-06,
+      "loss": 0.3286,
+      "step": 3218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01709410277727e-06,
+      "loss": 0.8823,
+      "step": 3219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011406939859472e-06,
+      "loss": 0.833,
+      "step": 3220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005720446531972e-06,
+      "loss": 0.8296,
+      "step": 3221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000034624709495e-06,
+      "loss": 0.8906,
+      "step": 3222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994349476306546e-06,
+      "loss": 0.813,
+      "step": 3223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.988665003237409e-06,
+      "loss": 0.8076,
+      "step": 3224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.982981207416127e-06,
+      "loss": 0.8501,
+      "step": 3225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97729809075653e-06,
+      "loss": 0.8208,
+      "step": 3226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971615655172208e-06,
+      "loss": 0.7715,
+      "step": 3227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965933902576523e-06,
+      "loss": 0.8521,
+      "step": 3228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960252834882615e-06,
+      "loss": 0.8374,
+      "step": 3229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954572454003385e-06,
+      "loss": 0.9131,
+      "step": 3230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948892761851508e-06,
+      "loss": 0.9048,
+      "step": 3231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943213760339423e-06,
+      "loss": 0.8237,
+      "step": 3232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937535451379338e-06,
+      "loss": 0.834,
+      "step": 3233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931857836883233e-06,
+      "loss": 0.8491,
+      "step": 3234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926180918762842e-06,
+      "loss": 0.8823,
+      "step": 3235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920504698929676e-06,
+      "loss": 0.8047,
+      "step": 3236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914829179295008e-06,
+      "loss": 0.7812,
+      "step": 3237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90915436176987e-06,
+      "loss": 0.833,
+      "step": 3238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903480248265064e-06,
+      "loss": 0.8315,
+      "step": 3239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897806840691151e-06,
+      "loss": 0.8359,
+      "step": 3240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892134140958461e-06,
+      "loss": 0.8901,
+      "step": 3241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886462150977074e-06,
+      "loss": 0.772,
+      "step": 3242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880790872656838e-06,
+      "loss": 0.8618,
+      "step": 3243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875120307907367e-06,
+      "loss": 0.8433,
+      "step": 3244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86945045863802e-06,
+      "loss": 0.8716,
+      "step": 3245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86378132675793e-06,
+      "loss": 0.7817,
+      "step": 3246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858112914175983e-06,
+      "loss": 0.8403,
+      "step": 3247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852445222800815e-06,
+      "loss": 0.8135,
+      "step": 3248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846778254540831e-06,
+      "loss": 0.7808,
+      "step": 3249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841112011304185e-06,
+      "loss": 0.7529,
+      "step": 3250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835446494998796e-06,
+      "loss": 0.8433,
+      "step": 3251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829781707532325e-06,
+      "loss": 0.7939,
+      "step": 3252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824117650812193e-06,
+      "loss": 0.3374,
+      "step": 3253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818454326745584e-06,
+      "loss": 0.8252,
+      "step": 3254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81279173723942e-06,
+      "loss": 0.8818,
+      "step": 3255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807129884200385e-06,
+      "loss": 0.8691,
+      "step": 3256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801468769534919e-06,
+      "loss": 0.7803,
+      "step": 3257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795808395149197e-06,
+      "loss": 0.8921,
+      "step": 3258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790148762949164e-06,
+      "loss": 0.8643,
+      "step": 3259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784489874840507e-06,
+      "loss": 0.8706,
+      "step": 3260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778831732728652e-06,
+      "loss": 0.8687,
+      "step": 3261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773174338518794e-06,
+      "loss": 0.8281,
+      "step": 3262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767517694115858e-06,
+      "loss": 0.813,
+      "step": 3263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761861801424535e-06,
+      "loss": 0.8477,
+      "step": 3264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756206662349241e-06,
+      "loss": 0.8262,
+      "step": 3265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750552278794154e-06,
+      "loss": 0.7979,
+      "step": 3266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.744898652663198e-06,
+      "loss": 0.8145,
+      "step": 3267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.739245785860028e-06,
+      "loss": 0.8369,
+      "step": 3268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733593680288058e-06,
+      "loss": 0.8647,
+      "step": 3269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72794233785044e-06,
+      "loss": 0.9248,
+      "step": 3270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722291760450065e-06,
+      "loss": 0.8359,
+      "step": 3271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716641949989575e-06,
+      "loss": 0.811,
+      "step": 3272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710992908371347e-06,
+      "loss": 0.8276,
+      "step": 3273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705344637497507e-06,
+      "loss": 0.8589,
+      "step": 3274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699697139269908e-06,
+      "loss": 0.7886,
+      "step": 3275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69405041559015e-06,
+      "loss": 0.7959,
+      "step": 3276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688404468359584e-06,
+      "loss": 0.7944,
+      "step": 3277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682759299479274e-06,
+      "loss": 0.833,
+      "step": 3278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677114910850049e-06,
+      "loss": 0.772,
+      "step": 3279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.671471304372457e-06,
+      "loss": 0.8438,
+      "step": 3280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.665828481946785e-06,
+      "loss": 0.7915,
+      "step": 3281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660186445473065e-06,
+      "loss": 0.8384,
+      "step": 3282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654545196851056e-06,
+      "loss": 0.8999,
+      "step": 3283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648904737980255e-06,
+      "loss": 0.8457,
+      "step": 3284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643265070759895e-06,
+      "loss": 0.8452,
+      "step": 3285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637626197088933e-06,
+      "loss": 0.7964,
+      "step": 3286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631988118866074e-06,
+      "loss": 0.8638,
+      "step": 3287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6263508379897375e-06,
+      "loss": 0.3501,
+      "step": 3288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.620714356358096e-06,
+      "loss": 0.3223,
+      "step": 3289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615078675869033e-06,
+      "loss": 0.873,
+      "step": 3290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.609443798420164e-06,
+      "loss": 0.9067,
+      "step": 3291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603809725908855e-06,
+      "loss": 0.8296,
+      "step": 3292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5981764602321735e-06,
+      "loss": 0.8286,
+      "step": 3293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592544003286934e-06,
+      "loss": 0.8062,
+      "step": 3294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5869123569696745e-06,
+      "loss": 0.8809,
+      "step": 3295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58128152317665e-06,
+      "loss": 0.812,
+      "step": 3296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575651503803858e-06,
+      "loss": 0.8784,
+      "step": 3297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.570022300747007e-06,
+      "loss": 0.7666,
+      "step": 3298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564393915901546e-06,
+      "loss": 0.8423,
+      "step": 3299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.558766351162635e-06,
+      "loss": 0.8037,
+      "step": 3300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5531396084251574e-06,
+      "loss": 0.8545,
+      "step": 3301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547513689583736e-06,
+      "loss": 0.7974,
+      "step": 3302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5418885965327e-06,
+      "loss": 0.8618,
+      "step": 3303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536264331166101e-06,
+      "loss": 0.3054,
+      "step": 3304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530640895377728e-06,
+      "loss": 0.8188,
+      "step": 3305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525018291061068e-06,
+      "loss": 0.8193,
+      "step": 3306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519396520109349e-06,
+      "loss": 0.8477,
+      "step": 3307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5137755844155056e-06,
+      "loss": 0.8389,
+      "step": 3308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50815548587219e-06,
+      "loss": 0.8364,
+      "step": 3309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502536226371784e-06,
+      "loss": 0.8311,
+      "step": 3310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.496917807806373e-06,
+      "loss": 0.832,
+      "step": 3311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491300232067775e-06,
+      "loss": 0.8384,
+      "step": 3312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48568350104751e-06,
+      "loss": 0.8682,
+      "step": 3313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480067616636815e-06,
+      "loss": 0.8252,
+      "step": 3314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474452580726654e-06,
+      "loss": 0.8374,
+      "step": 3315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468838395207689e-06,
+      "loss": 0.8086,
+      "step": 3316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463225061970312e-06,
+      "loss": 0.8535,
+      "step": 3317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4576125829046164e-06,
+      "loss": 0.3362,
+      "step": 3318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.452000959900408e-06,
+      "loss": 0.8306,
+      "step": 3319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446390194847211e-06,
+      "loss": 0.3242,
+      "step": 3320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.440780289634255e-06,
+      "loss": 0.8228,
+      "step": 3321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43517124615049e-06,
+      "loss": 0.8164,
+      "step": 3322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429563066284562e-06,
+      "loss": 0.8511,
+      "step": 3323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423955751924829e-06,
+      "loss": 0.8794,
+      "step": 3324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418349304959372e-06,
+      "loss": 0.8325,
+      "step": 3325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.412743727275959e-06,
+      "loss": 0.8569,
+      "step": 3326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407139020762082e-06,
+      "loss": 0.8525,
+      "step": 3327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401535187304931e-06,
+      "loss": 0.8765,
+      "step": 3328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3959322287914e-06,
+      "loss": 0.8633,
+      "step": 3329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3903301471081004e-06,
+      "loss": 0.8564,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3847289441413314e-06,
+      "loss": 0.8193,
+      "step": 3331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379128621777116e-06,
+      "loss": 0.874,
+      "step": 3332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.373529181901162e-06,
+      "loss": 0.8447,
+      "step": 3333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367930626398887e-06,
+      "loss": 0.8403,
+      "step": 3334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362332957155421e-06,
+      "loss": 0.7969,
+      "step": 3335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3567361760555765e-06,
+      "loss": 0.874,
+      "step": 3336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3511402849838845e-06,
+      "loss": 0.7744,
+      "step": 3337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345545285824567e-06,
+      "loss": 0.8242,
+      "step": 3338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339951180461541e-06,
+      "loss": 0.3143,
+      "step": 3339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334357970778436e-06,
+      "loss": 0.8413,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3287656586585696e-06,
+      "loss": 0.8579,
+      "step": 3341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.323174245984967e-06,
+      "loss": 0.7803,
+      "step": 3342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.317583734640336e-06,
+      "loss": 0.8413,
+      "step": 3343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311994126507091e-06,
+      "loss": 0.3247,
+      "step": 3344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306405423467344e-06,
+      "loss": 0.8818,
+      "step": 3345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300817627402893e-06,
+      "loss": 0.876,
+      "step": 3346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29523074019524e-06,
+      "loss": 0.8286,
+      "step": 3347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289644763725579e-06,
+      "loss": 0.8569,
+      "step": 3348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284059699874786e-06,
+      "loss": 0.8013,
+      "step": 3349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278475550523447e-06,
+      "loss": 0.7803,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27289231755183e-06,
+      "loss": 0.3259,
+      "step": 3351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267310002839901e-06,
+      "loss": 0.7808,
+      "step": 3352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2617286082673065e-06,
+      "loss": 0.8516,
+      "step": 3353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256148135713388e-06,
+      "loss": 0.8398,
+      "step": 3354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.250568587057185e-06,
+      "loss": 0.7476,
+      "step": 3355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244989964177413e-06,
+      "loss": 0.7817,
+      "step": 3356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239412268952481e-06,
+      "loss": 0.8203,
+      "step": 3357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23383550326049e-06,
+      "loss": 0.8252,
+      "step": 3358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228259668979218e-06,
+      "loss": 0.8018,
+      "step": 3359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222684767986141e-06,
+      "loss": 0.853,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217110802158415e-06,
+      "loss": 0.325,
+      "step": 3361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211537773372874e-06,
+      "loss": 0.8315,
+      "step": 3362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.205965683506053e-06,
+      "loss": 0.8618,
+      "step": 3363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200394534434154e-06,
+      "loss": 0.8271,
+      "step": 3364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194824328033077e-06,
+      "loss": 0.8389,
+      "step": 3365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.189255066178392e-06,
+      "loss": 0.832,
+      "step": 3366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183686750745356e-06,
+      "loss": 0.8564,
+      "step": 3367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178119383608912e-06,
+      "loss": 0.3525,
+      "step": 3368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172552966643675e-06,
+      "loss": 0.8291,
+      "step": 3369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166987501723948e-06,
+      "loss": 0.8042,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161422990723709e-06,
+      "loss": 0.7603,
+      "step": 3371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15585943551661e-06,
+      "loss": 0.897,
+      "step": 3372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150296837975995e-06,
+      "loss": 0.8252,
+      "step": 3373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144735199974871e-06,
+      "loss": 0.8315,
+      "step": 3374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1391745233859355e-06,
+      "loss": 0.8071,
+      "step": 3375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.133614810081548e-06,
+      "loss": 0.8291,
+      "step": 3376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128056061933752e-06,
+      "loss": 0.9014,
+      "step": 3377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122498280814269e-06,
+      "loss": 0.7627,
+      "step": 3378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116941468594485e-06,
+      "loss": 0.8135,
+      "step": 3379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1113856271454684e-06,
+      "loss": 0.8462,
+      "step": 3380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105830758337959e-06,
+      "loss": 0.8003,
+      "step": 3381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100276864042364e-06,
+      "loss": 0.8291,
+      "step": 3382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094723946128768e-06,
+      "loss": 0.8135,
+      "step": 3383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089172006466922e-06,
+      "loss": 0.8032,
+      "step": 3384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083621046926262e-06,
+      "loss": 0.8442,
+      "step": 3385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078071069375868e-06,
+      "loss": 0.7573,
+      "step": 3386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07252207568451e-06,
+      "loss": 0.8726,
+      "step": 3387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0669740677206255e-06,
+      "loss": 0.814,
+      "step": 3388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0614270473523075e-06,
+      "loss": 0.8398,
+      "step": 3389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055881016447328e-06,
+      "loss": 0.8394,
+      "step": 3390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050335976873124e-06,
+      "loss": 0.8779,
+      "step": 3391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.044791930496792e-06,
+      "loss": 0.7886,
+      "step": 3392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039248879185102e-06,
+      "loss": 0.8071,
+      "step": 3393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033706824804483e-06,
+      "loss": 0.8896,
+      "step": 3394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028165769221039e-06,
+      "loss": 0.814,
+      "step": 3395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0226257143005216e-06,
+      "loss": 0.8525,
+      "step": 3396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017086661908355e-06,
+      "loss": 0.8491,
+      "step": 3397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011548613909628e-06,
+      "loss": 0.8062,
+      "step": 3398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0060115721690826e-06,
+      "loss": 0.8677,
+      "step": 3399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0004755385511314e-06,
+      "loss": 0.8101,
+      "step": 3400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994940514919845e-06,
+      "loss": 0.8032,
+      "step": 3401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.989406503138945e-06,
+      "loss": 0.8052,
+      "step": 3402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983873505071823e-06,
+      "loss": 0.8086,
+      "step": 3403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978341522581531e-06,
+      "loss": 0.8623,
+      "step": 3404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972810557530765e-06,
+      "loss": 0.8779,
+      "step": 3405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967280611781891e-06,
+      "loss": 0.8691,
+      "step": 3406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961751687196928e-06,
+      "loss": 0.8076,
+      "step": 3407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956223785637555e-06,
+      "loss": 0.8125,
+      "step": 3408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950696908965096e-06,
+      "loss": 0.8257,
+      "step": 3409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.945171059040538e-06,
+      "loss": 0.8042,
+      "step": 3410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939646237724528e-06,
+      "loss": 0.8511,
+      "step": 3411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934122446877348e-06,
+      "loss": 0.8159,
+      "step": 3412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928599688358952e-06,
+      "loss": 0.3391,
+      "step": 3413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923077964028941e-06,
+      "loss": 0.8438,
+      "step": 3414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9175572757465585e-06,
+      "loss": 0.8081,
+      "step": 3415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912037625370709e-06,
+      "loss": 0.8062,
+      "step": 3416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906519014759946e-06,
+      "loss": 0.8169,
+      "step": 3417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.901001445772475e-06,
+      "loss": 0.8594,
+      "step": 3418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895484920266141e-06,
+      "loss": 0.8179,
+      "step": 3419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889969440098445e-06,
+      "loss": 0.8115,
+      "step": 3420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884455007126542e-06,
+      "loss": 0.8545,
+      "step": 3421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87894162320722e-06,
+      "loss": 0.323,
+      "step": 3422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8734292901969244e-06,
+      "loss": 0.8589,
+      "step": 3423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867918009951746e-06,
+      "loss": 0.8003,
+      "step": 3424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862407784327413e-06,
+      "loss": 0.8203,
+      "step": 3425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8568986151793085e-06,
+      "loss": 0.8472,
+      "step": 3426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851390504362454e-06,
+      "loss": 0.7935,
+      "step": 3427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84588345373152e-06,
+      "loss": 0.8164,
+      "step": 3428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.840377465140813e-06,
+      "loss": 0.8564,
+      "step": 3429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834872540444282e-06,
+      "loss": 0.8335,
+      "step": 3430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829368681495528e-06,
+      "loss": 0.3269,
+      "step": 3431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823865890147781e-06,
+      "loss": 0.8208,
+      "step": 3432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.818364168253918e-06,
+      "loss": 0.8262,
+      "step": 3433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812863517666459e-06,
+      "loss": 0.8643,
+      "step": 3434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80736394023755e-06,
+      "loss": 0.7764,
+      "step": 3435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801865437818991e-06,
+      "loss": 0.7651,
+      "step": 3436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796368012262209e-06,
+      "loss": 0.8203,
+      "step": 3437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7908716654182815e-06,
+      "loss": 0.8359,
+      "step": 3438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785376399137905e-06,
+      "loss": 0.8506,
+      "step": 3439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779882215271423e-06,
+      "loss": 0.8062,
+      "step": 3440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77438911566882e-06,
+      "loss": 0.8354,
+      "step": 3441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768897102179697e-06,
+      "loss": 0.8169,
+      "step": 3442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76340617665331e-06,
+      "loss": 0.856,
+      "step": 3443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757916340938538e-06,
+      "loss": 0.8169,
+      "step": 3444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752427596883886e-06,
+      "loss": 0.8242,
+      "step": 3445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746939946337509e-06,
+      "loss": 0.8013,
+      "step": 3446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741453391147179e-06,
+      "loss": 0.8364,
+      "step": 3447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.735967933160312e-06,
+      "loss": 0.8511,
+      "step": 3448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730483574223939e-06,
+      "loss": 0.8169,
+      "step": 3449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725000316184731e-06,
+      "loss": 0.8413,
+      "step": 3450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719518160888991e-06,
+      "loss": 0.7632,
+      "step": 3451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714037110182643e-06,
+      "loss": 0.8433,
+      "step": 3452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708557165911238e-06,
+      "loss": 0.8315,
+      "step": 3453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703078329919968e-06,
+      "loss": 0.8584,
+      "step": 3454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697600604053634e-06,
+      "loss": 0.8169,
+      "step": 3455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.692123990156678e-06,
+      "loss": 0.7524,
+      "step": 3456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686648490073159e-06,
+      "loss": 0.8887,
+      "step": 3457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68117410564676e-06,
+      "loss": 0.7998,
+      "step": 3458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6757008387207956e-06,
+      "loss": 0.8252,
+      "step": 3459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670228691138196e-06,
+      "loss": 0.832,
+      "step": 3460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664757664741528e-06,
+      "loss": 0.854,
+      "step": 3461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.659287761372961e-06,
+      "loss": 0.856,
+      "step": 3462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.653818982874299e-06,
+      "loss": 0.9165,
+      "step": 3463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648351331086972e-06,
+      "loss": 0.8677,
+      "step": 3464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6428848078520135e-06,
+      "loss": 0.814,
+      "step": 3465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637419415010095e-06,
+      "loss": 0.8477,
+      "step": 3466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631955154401498e-06,
+      "loss": 0.8281,
+      "step": 3467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626492027866118e-06,
+      "loss": 0.8345,
+      "step": 3468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621030037243483e-06,
+      "loss": 0.8384,
+      "step": 3469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615569184372726e-06,
+      "loss": 0.8408,
+      "step": 3470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610109471092605e-06,
+      "loss": 0.8062,
+      "step": 3471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604650899241486e-06,
+      "loss": 0.8188,
+      "step": 3472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599193470657356e-06,
+      "loss": 0.8257,
+      "step": 3473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593737187177821e-06,
+      "loss": 0.877,
+      "step": 3474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.588282050640089e-06,
+      "loss": 0.8345,
+      "step": 3475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582828062880997e-06,
+      "loss": 0.8506,
+      "step": 3476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577375225736987e-06,
+      "loss": 0.834,
+      "step": 3477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571923541044107e-06,
+      "loss": 0.3523,
+      "step": 3478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5664730106380325e-06,
+      "loss": 0.8267,
+      "step": 3479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561023636354036e-06,
+      "loss": 0.8149,
+      "step": 3480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5555754200270125e-06,
+      "loss": 0.7446,
+      "step": 3481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550128363491457e-06,
+      "loss": 0.8989,
+      "step": 3482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544682468581478e-06,
+      "loss": 0.8418,
+      "step": 3483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5392377371308005e-06,
+      "loss": 0.3296,
+      "step": 3484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53379417097274e-06,
+      "loss": 0.8267,
+      "step": 3485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.528351771940238e-06,
+      "loss": 0.8389,
+      "step": 3486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522910541865833e-06,
+      "loss": 0.7988,
+      "step": 3487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517470482581667e-06,
+      "loss": 0.3342,
+      "step": 3488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.512031595919501e-06,
+      "loss": 0.8335,
+      "step": 3489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506593883710686e-06,
+      "loss": 0.8433,
+      "step": 3490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5011573477861915e-06,
+      "loss": 0.8706,
+      "step": 3491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495721989976578e-06,
+      "loss": 0.8457,
+      "step": 3492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4902878121120165e-06,
+      "loss": 0.8564,
+      "step": 3493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484854816022284e-06,
+      "loss": 0.8545,
+      "step": 3494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4794230035367435e-06,
+      "loss": 0.8042,
+      "step": 3495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473992376484386e-06,
+      "loss": 0.7729,
+      "step": 3496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4685629366937806e-06,
+      "loss": 0.7891,
+      "step": 3497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463134685993097e-06,
+      "loss": 0.8286,
+      "step": 3498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457707626210129e-06,
+      "loss": 0.8633,
+      "step": 3499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452281759172236e-06,
+      "loss": 0.8276,
+      "step": 3500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446857086706404e-06,
+      "loss": 0.366,
+      "step": 3501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4414336106391995e-06,
+      "loss": 0.7905,
+      "step": 3502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436011332796788e-06,
+      "loss": 0.7769,
+      "step": 3503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4305902550049405e-06,
+      "loss": 0.7837,
+      "step": 3504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425170379089018e-06,
+      "loss": 0.8276,
+      "step": 3505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4197517068739735e-06,
+      "loss": 0.8032,
+      "step": 3506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414334240184365e-06,
+      "loss": 0.8428,
+      "step": 3507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408917980844326e-06,
+      "loss": 0.7573,
+      "step": 3508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4035029306776094e-06,
+      "loss": 0.8535,
+      "step": 3509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398089091507542e-06,
+      "loss": 0.8472,
+      "step": 3510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39267646515704e-06,
+      "loss": 0.7754,
+      "step": 3511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387265053448633e-06,
+      "loss": 0.8438,
+      "step": 3512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381854858204416e-06,
+      "loss": 0.8833,
+      "step": 3513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376445881246093e-06,
+      "loss": 0.813,
+      "step": 3514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371038124394951e-06,
+      "loss": 0.8857,
+      "step": 3515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365631589471858e-06,
+      "loss": 0.7988,
+      "step": 3516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360226278297287e-06,
+      "loss": 0.9048,
+      "step": 3517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354822192691285e-06,
+      "loss": 0.8413,
+      "step": 3518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349419334473501e-06,
+      "loss": 0.8159,
+      "step": 3519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3440177054631524e-06,
+      "loss": 0.7993,
+      "step": 3520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338617307479052e-06,
+      "loss": 0.7915,
+      "step": 3521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333218142339606e-06,
+      "loss": 0.8057,
+      "step": 3522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327820211862787e-06,
+      "loss": 0.854,
+      "step": 3523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322423517866172e-06,
+      "loss": 0.8447,
+      "step": 3524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317028062166908e-06,
+      "loss": 0.7944,
+      "step": 3525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311633846581726e-06,
+      "loss": 0.8413,
+      "step": 3526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306240872926946e-06,
+      "loss": 0.8652,
+      "step": 3527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300849143018464e-06,
+      "loss": 0.7964,
+      "step": 3528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295458658671767e-06,
+      "loss": 0.8389,
+      "step": 3529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290069421701906e-06,
+      "loss": 0.7656,
+      "step": 3530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284681433923523e-06,
+      "loss": 0.8354,
+      "step": 3531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279294697150842e-06,
+      "loss": 0.8208,
+      "step": 3532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273909213197656e-06,
+      "loss": 0.8481,
+      "step": 3533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268524983877343e-06,
+      "loss": 0.7954,
+      "step": 3534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2631420110028615e-06,
+      "loss": 0.8232,
+      "step": 3535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2577602963867345e-06,
+      "loss": 0.8521,
+      "step": 3536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252379841841076e-06,
+      "loss": 0.8125,
+      "step": 3537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247000649177563e-06,
+      "loss": 0.8188,
+      "step": 3538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241622720207462e-06,
+      "loss": 0.8237,
+      "step": 3539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2362460567415976e-06,
+      "loss": 0.3181,
+      "step": 3540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2308706605903775e-06,
+      "loss": 0.833,
+      "step": 3541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225496533563786e-06,
+      "loss": 0.7998,
+      "step": 3542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220123677471371e-06,
+      "loss": 0.8066,
+      "step": 3543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214752094122258e-06,
+      "loss": 0.8599,
+      "step": 3544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209381785325146e-06,
+      "loss": 0.8247,
+      "step": 3545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2040127528882935e-06,
+      "loss": 0.8276,
+      "step": 3546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198644998619546e-06,
+      "loss": 0.7776,
+      "step": 3547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193278524326305e-06,
+      "loss": 0.3328,
+      "step": 3548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187913331815553e-06,
+      "loss": 0.7246,
+      "step": 3549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182549422893827e-06,
+      "loss": 0.8281,
+      "step": 3550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177186799367236e-06,
+      "loss": 0.8442,
+      "step": 3551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1718254630414695e-06,
+      "loss": 0.8257,
+      "step": 3552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1664654157217675e-06,
+      "loss": 0.3354,
+      "step": 3553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161106659212936e-06,
+      "loss": 0.7896,
+      "step": 3554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155749195319363e-06,
+      "loss": 0.8013,
+      "step": 3555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150393025844981e-06,
+      "loss": 0.8311,
+      "step": 3556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145038152593299e-06,
+      "loss": 0.8584,
+      "step": 3557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1396845773673905e-06,
+      "loss": 0.8379,
+      "step": 3558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134332301969879e-06,
+      "loss": 0.8022,
+      "step": 3559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128981328202965e-06,
+      "loss": 0.3453,
+      "step": 3560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123631657868403e-06,
+      "loss": 0.8052,
+      "step": 3561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118283292767514e-06,
+      "loss": 0.8003,
+      "step": 3562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1129362347011725e-06,
+      "loss": 0.8115,
+      "step": 3563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1075904854698135e-06,
+      "loss": 0.8145,
+      "step": 3564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102246046873441e-06,
+      "loss": 0.3259,
+      "step": 3565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096902920711605e-06,
+      "loss": 0.7949,
+      "step": 3566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091561108783423e-06,
+      "loss": 0.8203,
+      "step": 3567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.086220612887565e-06,
+      "loss": 0.832,
+      "step": 3568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080881434822259e-06,
+      "loss": 0.8003,
+      "step": 3569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07554357638529e-06,
+      "loss": 0.8013,
+      "step": 3570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070207039373995e-06,
+      "loss": 0.8052,
+      "step": 3571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064871825585278e-06,
+      "loss": 0.8408,
+      "step": 3572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059537936815581e-06,
+      "loss": 0.7935,
+      "step": 3573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054205374860906e-06,
+      "loss": 0.8027,
+      "step": 3574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048874141516819e-06,
+      "loss": 0.7612,
+      "step": 3575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043544238578418e-06,
+      "loss": 0.7939,
+      "step": 3576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038215667840372e-06,
+      "loss": 0.8623,
+      "step": 3577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032888431096895e-06,
+      "loss": 0.8164,
+      "step": 3578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027562530141743e-06,
+      "loss": 0.8262,
+      "step": 3579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0222379667682364e-06,
+      "loss": 0.8267,
+      "step": 3580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016914742769236e-06,
+      "loss": 0.812,
+      "step": 3581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011592859937158e-06,
+      "loss": 0.8154,
+      "step": 3582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0062723200639586e-06,
+      "loss": 0.8364,
+      "step": 3583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000953124941146e-06,
+      "loss": 0.8237,
+      "step": 3584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9956352763597835e-06,
+      "loss": 0.8472,
+      "step": 3585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9903187761104645e-06,
+      "loss": 0.834,
+      "step": 3586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985003625983344e-06,
+      "loss": 0.351,
+      "step": 3587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979689827768116e-06,
+      "loss": 0.7783,
+      "step": 3588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974377383254012e-06,
+      "loss": 0.8501,
+      "step": 3589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.969066294229821e-06,
+      "loss": 0.8521,
+      "step": 3590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963756562483869e-06,
+      "loss": 0.3309,
+      "step": 3591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958448189804027e-06,
+      "loss": 0.7695,
+      "step": 3592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9531411779777034e-06,
+      "loss": 0.8149,
+      "step": 3593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947835528791851e-06,
+      "loss": 0.7915,
+      "step": 3594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9425312440329704e-06,
+      "loss": 0.8188,
+      "step": 3595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93722832548709e-06,
+      "loss": 0.8589,
+      "step": 3596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9319267749397905e-06,
+      "loss": 0.7827,
+      "step": 3597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926626594176186e-06,
+      "loss": 0.8071,
+      "step": 3598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9213277849809255e-06,
+      "loss": 0.8418,
+      "step": 3599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916030349138204e-06,
+      "loss": 0.833,
+      "step": 3600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.910734288431751e-06,
+      "loss": 0.9033,
+      "step": 3601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9054396046448355e-06,
+      "loss": 0.3389,
+      "step": 3602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900146299560255e-06,
+      "loss": 0.7832,
+      "step": 3603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894854374960348e-06,
+      "loss": 0.7993,
+      "step": 3604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.889563832626994e-06,
+      "loss": 0.3104,
+      "step": 3605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884274674341593e-06,
+      "loss": 0.8164,
+      "step": 3606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878986901885091e-06,
+      "loss": 0.834,
+      "step": 3607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873700517037967e-06,
+      "loss": 0.8711,
+      "step": 3608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868415521580222e-06,
+      "loss": 0.8555,
+      "step": 3609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863131917291402e-06,
+      "loss": 0.8506,
+      "step": 3610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857849705950578e-06,
+      "loss": 0.8174,
+      "step": 3611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85256888933635e-06,
+      "loss": 0.7769,
+      "step": 3612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847289469226854e-06,
+      "loss": 0.8643,
+      "step": 3613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842011447399749e-06,
+      "loss": 0.8281,
+      "step": 3614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8367348256322355e-06,
+      "loss": 0.833,
+      "step": 3615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831459605701029e-06,
+      "loss": 0.8433,
+      "step": 3616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826185789382377e-06,
+      "loss": 0.8467,
+      "step": 3617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82091337845206e-06,
+      "loss": 0.7192,
+      "step": 3618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815642374685374e-06,
+      "loss": 0.8306,
+      "step": 3619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8103727798571575e-06,
+      "loss": 0.812,
+      "step": 3620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8051045957417564e-06,
+      "loss": 0.3337,
+      "step": 3621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799837824113053e-06,
+      "loss": 0.8047,
+      "step": 3622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794572466744459e-06,
+      "loss": 0.8125,
+      "step": 3623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789308525408891e-06,
+      "loss": 0.7871,
+      "step": 3624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784046001878809e-06,
+      "loss": 0.8643,
+      "step": 3625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778784897926184e-06,
+      "loss": 0.8135,
+      "step": 3626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7735252153225065e-06,
+      "loss": 0.8398,
+      "step": 3627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768266955838797e-06,
+      "loss": 0.8174,
+      "step": 3628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.763010121245596e-06,
+      "loss": 0.9072,
+      "step": 3629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757754713312964e-06,
+      "loss": 0.7583,
+      "step": 3630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752500733810473e-06,
+      "loss": 0.8403,
+      "step": 3631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7472481845072195e-06,
+      "loss": 0.7646,
+      "step": 3632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741997067171825e-06,
+      "loss": 0.8306,
+      "step": 3633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736747383572414e-06,
+      "loss": 0.7988,
+      "step": 3634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7314991354766416e-06,
+      "loss": 0.8325,
+      "step": 3635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7262523246516775e-06,
+      "loss": 0.7979,
+      "step": 3636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7210069528641995e-06,
+      "loss": 0.79,
+      "step": 3637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715763021880412e-06,
+      "loss": 0.8433,
+      "step": 3638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710520533466021e-06,
+      "loss": 0.8062,
+      "step": 3639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7052794893862604e-06,
+      "loss": 0.8057,
+      "step": 3640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7000398914058665e-06,
+      "loss": 0.8135,
+      "step": 3641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694801741289095e-06,
+      "loss": 0.3169,
+      "step": 3642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6895650407997185e-06,
+      "loss": 0.9243,
+      "step": 3643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684329791701006e-06,
+      "loss": 0.8794,
+      "step": 3644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679095995755758e-06,
+      "loss": 0.8237,
+      "step": 3645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6738636547262694e-06,
+      "loss": 0.8892,
+      "step": 3646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668632770374347e-06,
+      "loss": 0.7783,
+      "step": 3647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6634033444613154e-06,
+      "loss": 0.7603,
+      "step": 3648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6581753787480035e-06,
+      "loss": 0.8232,
+      "step": 3649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.652948874994754e-06,
+      "loss": 0.7627,
+      "step": 3650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647723834961407e-06,
+      "loss": 0.7837,
+      "step": 3651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642500260407313e-06,
+      "loss": 0.8159,
+      "step": 3652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637278153091336e-06,
+      "loss": 0.7778,
+      "step": 3653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.632057514771835e-06,
+      "loss": 0.8379,
+      "step": 3654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626838347206683e-06,
+      "loss": 0.8936,
+      "step": 3655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621620652153261e-06,
+      "loss": 0.8252,
+      "step": 3656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616404431368439e-06,
+      "loss": 0.8184,
+      "step": 3657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611189686608608e-06,
+      "loss": 0.8374,
+      "step": 3658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605976419629651e-06,
+      "loss": 0.791,
+      "step": 3659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.600764632186949e-06,
+      "loss": 0.8037,
+      "step": 3660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595554326035399e-06,
+      "loss": 0.8491,
+      "step": 3661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590345502929393e-06,
+      "loss": 0.874,
+      "step": 3662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585138164622825e-06,
+      "loss": 0.7974,
+      "step": 3663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579932312869083e-06,
+      "loss": 0.813,
+      "step": 3664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574727949421057e-06,
+      "loss": 0.8164,
+      "step": 3665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569525076031142e-06,
+      "loss": 0.8062,
+      "step": 3666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56432369445122e-06,
+      "loss": 0.8384,
+      "step": 3667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559123806432682e-06,
+      "loss": 0.7979,
+      "step": 3668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553925413726415e-06,
+      "loss": 0.7656,
+      "step": 3669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.54872851808279e-06,
+      "loss": 0.8452,
+      "step": 3670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543533121251692e-06,
+      "loss": 0.7847,
+      "step": 3671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.538339224982483e-06,
+      "loss": 0.8374,
+      "step": 3672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.533146831024037e-06,
+      "loss": 0.7881,
+      "step": 3673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527955941124707e-06,
+      "loss": 0.8325,
+      "step": 3674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522766557032351e-06,
+      "loss": 0.874,
+      "step": 3675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5175786804943175e-06,
+      "loss": 0.8267,
+      "step": 3676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51239231325744e-06,
+      "loss": 0.8066,
+      "step": 3677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507207457068053e-06,
+      "loss": 0.8301,
+      "step": 3678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.50202411367198e-06,
+      "loss": 0.8091,
+      "step": 3679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496842284814526e-06,
+      "loss": 0.7925,
+      "step": 3680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491661972240497e-06,
+      "loss": 0.8184,
+      "step": 3681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4864831776941886e-06,
+      "loss": 0.7988,
+      "step": 3682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4813059029193805e-06,
+      "loss": 0.8062,
+      "step": 3683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476130149659343e-06,
+      "loss": 0.8276,
+      "step": 3684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470955919656826e-06,
+      "loss": 0.8506,
+      "step": 3685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465783214654084e-06,
+      "loss": 0.8428,
+      "step": 3686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460612036392834e-06,
+      "loss": 0.8271,
+      "step": 3687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455442386614303e-06,
+      "loss": 0.835,
+      "step": 3688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4502742670591924e-06,
+      "loss": 0.311,
+      "step": 3689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445107679467683e-06,
+      "loss": 0.8047,
+      "step": 3690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439942625579451e-06,
+      "loss": 0.8765,
+      "step": 3691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434779107133643e-06,
+      "loss": 0.8037,
+      "step": 3692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4296171258689066e-06,
+      "loss": 0.8433,
+      "step": 3693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424456683523351e-06,
+      "loss": 0.7827,
+      "step": 3694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419297781834585e-06,
+      "loss": 0.7939,
+      "step": 3695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414140422539691e-06,
+      "loss": 0.8574,
+      "step": 3696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408984607375225e-06,
+      "loss": 0.8232,
+      "step": 3697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4038303380772405e-06,
+      "loss": 0.8379,
+      "step": 3698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398677616381255e-06,
+      "loss": 0.3374,
+      "step": 3699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393526444022263e-06,
+      "loss": 0.9229,
+      "step": 3700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3883768227347595e-06,
+      "loss": 0.8428,
+      "step": 3701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383228754252691e-06,
+      "loss": 0.8013,
+      "step": 3702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378082240309501e-06,
+      "loss": 0.835,
+      "step": 3703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3729372826380955e-06,
+      "loss": 0.8223,
+      "step": 3704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367793882970858e-06,
+      "loss": 0.8057,
+      "step": 3705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3626520430396624e-06,
+      "loss": 0.834,
+      "step": 3706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357511764575833e-06,
+      "loss": 0.8623,
+      "step": 3707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352373049310191e-06,
+      "loss": 0.8315,
+      "step": 3708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347235898973025e-06,
+      "loss": 0.8491,
+      "step": 3709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342100315294082e-06,
+      "loss": 0.8008,
+      "step": 3710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336966300002604e-06,
+      "loss": 0.813,
+      "step": 3711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331833854827291e-06,
+      "loss": 0.8218,
+      "step": 3712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326702981496307e-06,
+      "loss": 0.7578,
+      "step": 3713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321573681737313e-06,
+      "loss": 0.812,
+      "step": 3714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316445957277415e-06,
+      "loss": 0.833,
+      "step": 3715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311319809843203e-06,
+      "loss": 0.8125,
+      "step": 3716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306195241160728e-06,
+      "loss": 0.8574,
+      "step": 3717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301072252955508e-06,
+      "loss": 0.75,
+      "step": 3718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295950846952536e-06,
+      "loss": 0.8716,
+      "step": 3719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290831024876272e-06,
+      "loss": 0.8438,
+      "step": 3720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285712788450641e-06,
+      "loss": 0.8149,
+      "step": 3721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28059613939903e-06,
+      "loss": 0.7935,
+      "step": 3722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275481079444291e-06,
+      "loss": 0.3384,
+      "step": 3723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270367610308753e-06,
+      "loss": 0.7832,
+      "step": 3724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265255733714191e-06,
+      "loss": 0.8301,
+      "step": 3725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.260145451381858e-06,
+      "loss": 0.8169,
+      "step": 3726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255036765032471e-06,
+      "loss": 0.7993,
+      "step": 3727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249929676386196e-06,
+      "loss": 0.7847,
+      "step": 3728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244824187162676e-06,
+      "loss": 0.8706,
+      "step": 3729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2397202990810035e-06,
+      "loss": 0.8262,
+      "step": 3730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2346180138597425e-06,
+      "loss": 0.8315,
+      "step": 3731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2295173332169045e-06,
+      "loss": 0.8247,
+      "step": 3732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224418258869974e-06,
+      "loss": 0.8071,
+      "step": 3733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219320792535892e-06,
+      "loss": 0.7959,
+      "step": 3734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2142249359310445e-06,
+      "loss": 0.8237,
+      "step": 3735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2091306907712955e-06,
+      "loss": 0.8604,
+      "step": 3736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2040380587719564e-06,
+      "loss": 0.812,
+      "step": 3737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198947041647786e-06,
+      "loss": 0.8213,
+      "step": 3738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193857641113017e-06,
+      "loss": 0.8457,
+      "step": 3739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188769858881328e-06,
+      "loss": 0.3616,
+      "step": 3740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1836836966658595e-06,
+      "loss": 0.7891,
+      "step": 3741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178599156179198e-06,
+      "loss": 0.8818,
+      "step": 3742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173516239133383e-06,
+      "loss": 0.8281,
+      "step": 3743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16843494723992e-06,
+      "loss": 0.7686,
+      "step": 3744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16335528220975e-06,
+      "loss": 0.79,
+      "step": 3745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.158277245753282e-06,
+      "loss": 0.8413,
+      "step": 3746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153200839580372e-06,
+      "loss": 0.8213,
+      "step": 3747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148126065400319e-06,
+      "loss": 0.8706,
+      "step": 3748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143052924921883e-06,
+      "loss": 0.7856,
+      "step": 3749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1379814198532665e-06,
+      "loss": 0.8447,
+      "step": 3750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132911551902128e-06,
+      "loss": 0.8672,
+      "step": 3751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127843322775565e-06,
+      "loss": 0.7686,
+      "step": 3752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122776734180133e-06,
+      "loss": 0.8506,
+      "step": 3753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117711787821836e-06,
+      "loss": 0.8765,
+      "step": 3754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112648485406116e-06,
+      "loss": 0.835,
+      "step": 3755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1075868286378625e-06,
+      "loss": 0.8262,
+      "step": 3756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10252681922142e-06,
+      "loss": 0.8066,
+      "step": 3757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097468458860568e-06,
+      "loss": 0.7935,
+      "step": 3758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092411749258536e-06,
+      "loss": 0.3093,
+      "step": 3759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087356692118004e-06,
+      "loss": 0.8398,
+      "step": 3760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0823032891410776e-06,
+      "loss": 0.8066,
+      "step": 3761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077251542029327e-06,
+      "loss": 0.8101,
+      "step": 3762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072201452483744e-06,
+      "loss": 0.8027,
+      "step": 3763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067153022204782e-06,
+      "loss": 0.811,
+      "step": 3764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0621062528923175e-06,
+      "loss": 0.8535,
+      "step": 3765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057061146245681e-06,
+      "loss": 0.8223,
+      "step": 3766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052017703963641e-06,
+      "loss": 0.8047,
+      "step": 3767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.046975927744398e-06,
+      "loss": 0.833,
+      "step": 3768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041935819285603e-06,
+      "loss": 0.8237,
+      "step": 3769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.036897380284335e-06,
+      "loss": 0.8105,
+      "step": 3770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03186061243711e-06,
+      "loss": 0.8745,
+      "step": 3771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026825517439893e-06,
+      "loss": 0.8291,
+      "step": 3772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0217920969880775e-06,
+      "loss": 0.8047,
+      "step": 3773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016760352776498e-06,
+      "loss": 0.7949,
+      "step": 3774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011730286499419e-06,
+      "loss": 0.8296,
+      "step": 3775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0067018998505355e-06,
+      "loss": 0.8765,
+      "step": 3776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001675194522995e-06,
+      "loss": 0.8042,
+      "step": 3777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996650172209356e-06,
+      "loss": 0.771,
+      "step": 3778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991626834601629e-06,
+      "loss": 0.8208,
+      "step": 3779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9866051833912535e-06,
+      "loss": 0.8257,
+      "step": 3780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981585220269088e-06,
+      "loss": 0.7974,
+      "step": 3781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976566946925442e-06,
+      "loss": 0.8911,
+      "step": 3782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971550365050039e-06,
+      "loss": 0.8032,
+      "step": 3783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966535476332045e-06,
+      "loss": 0.7646,
+      "step": 3784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961522282460047e-06,
+      "loss": 0.8403,
+      "step": 3785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956510785122068e-06,
+      "loss": 0.8354,
+      "step": 3786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951500986005561e-06,
+      "loss": 0.3271,
+      "step": 3787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946492886797396e-06,
+      "loss": 0.8003,
+      "step": 3788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.941486489183886e-06,
+      "loss": 0.8447,
+      "step": 3789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.936481794850759e-06,
+      "loss": 0.7988,
+      "step": 3790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93147880548317e-06,
+      "loss": 0.7871,
+      "step": 3791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9264775227657065e-06,
+      "loss": 0.7915,
+      "step": 3792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921477948382381e-06,
+      "loss": 0.7822,
+      "step": 3793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9164800840166286e-06,
+      "loss": 0.8638,
+      "step": 3794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911483931351307e-06,
+      "loss": 0.7979,
+      "step": 3795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906489492068693e-06,
+      "loss": 0.8203,
+      "step": 3796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9014967678505e-06,
+      "loss": 0.812,
+      "step": 3797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89650576037785e-06,
+      "loss": 0.7959,
+      "step": 3798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8915164713312956e-06,
+      "loss": 0.8311,
+      "step": 3799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886528902390812e-06,
+      "loss": 0.771,
+      "step": 3800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881543055235786e-06,
+      "loss": 0.834,
+      "step": 3801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876558931545035e-06,
+      "loss": 0.8228,
+      "step": 3802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871576532996787e-06,
+      "loss": 0.7949,
+      "step": 3803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866595861268692e-06,
+      "loss": 0.8022,
+      "step": 3804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861616918037822e-06,
+      "loss": 0.8643,
+      "step": 3805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8566397049806645e-06,
+      "loss": 0.8242,
+      "step": 3806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85166422377313e-06,
+      "loss": 0.7588,
+      "step": 3807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846690476090536e-06,
+      "loss": 0.8008,
+      "step": 3808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841718463607618e-06,
+      "loss": 0.8325,
+      "step": 3809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836748187998537e-06,
+      "loss": 0.8218,
+      "step": 3810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831779650936853e-06,
+      "loss": 0.7471,
+      "step": 3811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826812854095556e-06,
+      "loss": 0.8555,
+      "step": 3812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821847799147048e-06,
+      "loss": 0.8296,
+      "step": 3813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8168844877631315e-06,
+      "loss": 0.8481,
+      "step": 3814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811922921615038e-06,
+      "loss": 0.8228,
+      "step": 3815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8069631023733985e-06,
+      "loss": 0.7959,
+      "step": 3816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802005031708268e-06,
+      "loss": 0.8374,
+      "step": 3817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7970487112890985e-06,
+      "loss": 0.813,
+      "step": 3818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792094142784764e-06,
+      "loss": 0.7954,
+      "step": 3819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787141327863549e-06,
+      "loss": 0.7744,
+      "step": 3820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782190268193137e-06,
+      "loss": 0.7725,
+      "step": 3821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7772409654406325e-06,
+      "loss": 0.8325,
+      "step": 3822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77229342127254e-06,
+      "loss": 0.7939,
+      "step": 3823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.767347637354772e-06,
+      "loss": 0.8521,
+      "step": 3824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762403615352655e-06,
+      "loss": 0.8022,
+      "step": 3825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757461356930916e-06,
+      "loss": 0.8208,
+      "step": 3826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7525208637536965e-06,
+      "loss": 0.7656,
+      "step": 3827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7475821374845325e-06,
+      "loss": 0.8711,
+      "step": 3828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742645179786367e-06,
+      "loss": 0.8281,
+      "step": 3829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7377099923215585e-06,
+      "loss": 0.8135,
+      "step": 3830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732776576751853e-06,
+      "loss": 0.8325,
+      "step": 3831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7278449347384124e-06,
+      "loss": 0.8062,
+      "step": 3832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722915067941802e-06,
+      "loss": 0.8115,
+      "step": 3833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7179869780219775e-06,
+      "loss": 0.7998,
+      "step": 3834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7130606666383105e-06,
+      "loss": 0.8125,
+      "step": 3835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7081361354495595e-06,
+      "loss": 0.8369,
+      "step": 3836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.703213386113897e-06,
+      "loss": 0.8018,
+      "step": 3837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698292420288884e-06,
+      "loss": 0.8369,
+      "step": 3838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693373239631491e-06,
+      "loss": 0.8374,
+      "step": 3839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6884558457980845e-06,
+      "loss": 0.854,
+      "step": 3840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68354024044442e-06,
+      "loss": 0.8647,
+      "step": 3841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678626425225667e-06,
+      "loss": 0.8159,
+      "step": 3842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673714401796381e-06,
+      "loss": 0.8232,
+      "step": 3843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668804171810511e-06,
+      "loss": 0.3005,
+      "step": 3844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663895736921412e-06,
+      "loss": 0.7451,
+      "step": 3845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658989098781832e-06,
+      "loss": 0.8599,
+      "step": 3846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654084259043915e-06,
+      "loss": 0.3481,
+      "step": 3847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6491812193591945e-06,
+      "loss": 0.8169,
+      "step": 3848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644279981378596e-06,
+      "loss": 0.8149,
+      "step": 3849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63938054675245e-06,
+      "loss": 0.8594,
+      "step": 3850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634482917130465e-06,
+      "loss": 0.7954,
+      "step": 3851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629587094161754e-06,
+      "loss": 0.8154,
+      "step": 3852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62469307949482e-06,
+      "loss": 0.7773,
+      "step": 3853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6198008747775455e-06,
+      "loss": 0.8286,
+      "step": 3854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614910481657222e-06,
+      "loss": 0.7891,
+      "step": 3855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610021901780516e-06,
+      "loss": 0.79,
+      "step": 3856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605135136793484e-06,
+      "loss": 0.7896,
+      "step": 3857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60025018834158e-06,
+      "loss": 0.8438,
+      "step": 3858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595367058069645e-06,
+      "loss": 0.7837,
+      "step": 3859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5904857476219056e-06,
+      "loss": 0.8062,
+      "step": 3860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585606258641973e-06,
+      "loss": 0.8462,
+      "step": 3861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580728592772843e-06,
+      "loss": 0.8096,
+      "step": 3862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5758527516569095e-06,
+      "loss": 0.8315,
+      "step": 3863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570978736935936e-06,
+      "loss": 0.8574,
+      "step": 3864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566106550251085e-06,
+      "loss": 0.7959,
+      "step": 3865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5612361932428985e-06,
+      "loss": 0.8325,
+      "step": 3866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556367667551297e-06,
+      "loss": 0.8101,
+      "step": 3867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5515009748155935e-06,
+      "loss": 0.8232,
+      "step": 3868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546636116674475e-06,
+      "loss": 0.7998,
+      "step": 3869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541773094766021e-06,
+      "loss": 0.8555,
+      "step": 3870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536911910727679e-06,
+      "loss": 0.8066,
+      "step": 3871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532052566196291e-06,
+      "loss": 0.751,
+      "step": 3872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527195062808075e-06,
+      "loss": 0.8242,
+      "step": 3873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522339402198624e-06,
+      "loss": 0.8657,
+      "step": 3874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517485586002918e-06,
+      "loss": 0.8179,
+      "step": 3875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512633615855313e-06,
+      "loss": 0.8242,
+      "step": 3876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507783493389537e-06,
+      "loss": 0.3215,
+      "step": 3877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5029352202387066e-06,
+      "loss": 0.8291,
+      "step": 3878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498088798035309e-06,
+      "loss": 0.8013,
+      "step": 3879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493244228411216e-06,
+      "loss": 0.8081,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488401512997666e-06,
+      "loss": 0.8223,
+      "step": 3881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483560653425273e-06,
+      "loss": 0.8516,
+      "step": 3882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4787216513240366e-06,
+      "loss": 0.7974,
+      "step": 3883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473884508323315e-06,
+      "loss": 0.7573,
+      "step": 3884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469049226051857e-06,
+      "loss": 0.7988,
+      "step": 3885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4642158061377796e-06,
+      "loss": 0.7056,
+      "step": 3886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459384250208563e-06,
+      "loss": 0.8247,
+      "step": 3887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454554559891074e-06,
+      "loss": 0.8184,
+      "step": 3888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44972673681154e-06,
+      "loss": 0.8462,
+      "step": 3889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444900782595569e-06,
+      "loss": 0.7866,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44007669886813e-06,
+      "loss": 0.8574,
+      "step": 3891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435254487253569e-06,
+      "loss": 0.7891,
+      "step": 3892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430434149375604e-06,
+      "loss": 0.8579,
+      "step": 3893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425615686857311e-06,
+      "loss": 0.8213,
+      "step": 3894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420799101321148e-06,
+      "loss": 0.8267,
+      "step": 3895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415984394388933e-06,
+      "loss": 0.8081,
+      "step": 3896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4111715676818455e-06,
+      "loss": 0.8457,
+      "step": 3897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406360622820446e-06,
+      "loss": 0.833,
+      "step": 3898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401551561424654e-06,
+      "loss": 0.854,
+      "step": 3899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396744385113758e-06,
+      "loss": 0.8184,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391939095506408e-06,
+      "loss": 0.8208,
+      "step": 3901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387135694220613e-06,
+      "loss": 0.8247,
+      "step": 3902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382334182873763e-06,
+      "loss": 0.8477,
+      "step": 3903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3775345630825935e-06,
+      "loss": 0.812,
+      "step": 3904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372736836463215e-06,
+      "loss": 0.8423,
+      "step": 3905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367941004631101e-06,
+      "loss": 0.8193,
+      "step": 3906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363147069201075e-06,
+      "loss": 0.8052,
+      "step": 3907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358355031787337e-06,
+      "loss": 0.7725,
+      "step": 3908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353564894003437e-06,
+      "loss": 0.832,
+      "step": 3909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348776657462287e-06,
+      "loss": 0.8052,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343990323776162e-06,
+      "loss": 0.8179,
+      "step": 3911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3392058945566975e-06,
+      "loss": 0.7998,
+      "step": 3912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3344233714148875e-06,
+      "loss": 0.8335,
+      "step": 3913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32964275596108e-06,
+      "loss": 0.8228,
+      "step": 3914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324864049804976e-06,
+      "loss": 0.8203,
+      "step": 3915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320087254555653e-06,
+      "loss": 0.792,
+      "step": 3916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315312371821516e-06,
+      "loss": 0.8262,
+      "step": 3917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310539403210361e-06,
+      "loss": 0.8291,
+      "step": 3918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305768350329311e-06,
+      "loss": 0.8325,
+      "step": 3919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300999214784852e-06,
+      "loss": 0.7578,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296231998182833e-06,
+      "loss": 0.8535,
+      "step": 3921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291466702128443e-06,
+      "loss": 0.8076,
+      "step": 3922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286703328226234e-06,
+      "loss": 0.7881,
+      "step": 3923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281941878080113e-06,
+      "loss": 0.8228,
+      "step": 3924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2771823532933285e-06,
+      "loss": 0.7871,
+      "step": 3925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272424755468493e-06,
+      "loss": 0.7979,
+      "step": 3926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267669086207556e-06,
+      "loss": 0.8257,
+      "step": 3927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262915347111831e-06,
+      "loss": 0.8374,
+      "step": 3928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258163539781972e-06,
+      "loss": 0.8013,
+      "step": 3929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2534136658179904e-06,
+      "loss": 0.8896,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2486657268192435e-06,
+      "loss": 0.8159,
+      "step": 3931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243919724384432e-06,
+      "loss": 0.8423,
+      "step": 3932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239175660111614e-06,
+      "loss": 0.3274,
+      "step": 3933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234433535598188e-06,
+      "loss": 0.769,
+      "step": 3934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229693352440897e-06,
+      "loss": 0.7832,
+      "step": 3935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224955112235838e-06,
+      "loss": 0.835,
+      "step": 3936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220218816578451e-06,
+      "loss": 0.7954,
+      "step": 3937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2154844670635235e-06,
+      "loss": 0.7759,
+      "step": 3938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21075206528518e-06,
+      "loss": 0.7773,
+      "step": 3939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206021612836893e-06,
+      "loss": 0.7983,
+      "step": 3940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201293111311486e-06,
+      "loss": 0.8384,
+      "step": 3941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196566562301111e-06,
+      "loss": 0.3672,
+      "step": 3942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191841967397275e-06,
+      "loss": 0.8477,
+      "step": 3943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187119328190827e-06,
+      "loss": 0.3245,
+      "step": 3944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182398646271945e-06,
+      "loss": 0.8496,
+      "step": 3945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177679923230164e-06,
+      "loss": 0.7905,
+      "step": 3946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1729631606543455e-06,
+      "loss": 0.8882,
+      "step": 3947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1682483601327015e-06,
+      "loss": 0.2975,
+      "step": 3948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163535523252774e-06,
+      "loss": 0.8335,
+      "step": 3949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158824651601451e-06,
+      "loss": 0.8105,
+      "step": 3950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1541157467649604e-06,
+      "loss": 0.3091,
+      "step": 3951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149408810328856e-06,
+      "loss": 0.7793,
+      "step": 3952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144703843878046e-06,
+      "loss": 0.7642,
+      "step": 3953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140000848996761e-06,
+      "loss": 0.8647,
+      "step": 3954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135299827268568e-06,
+      "loss": 0.8081,
+      "step": 3955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130600780276381e-06,
+      "loss": 0.7876,
+      "step": 3956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125903709602442e-06,
+      "loss": 0.7773,
+      "step": 3957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.121208616828325e-06,
+      "loss": 0.8608,
+      "step": 3958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116515503534945e-06,
+      "loss": 0.8555,
+      "step": 3959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11182437130254e-06,
+      "loss": 0.793,
+      "step": 3960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107135221710694e-06,
+      "loss": 0.8384,
+      "step": 3961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1024480563383115e-06,
+      "loss": 0.877,
+      "step": 3962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097762876763637e-06,
+      "loss": 0.835,
+      "step": 3963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093079684564246e-06,
+      "loss": 0.7617,
+      "step": 3964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088398481317036e-06,
+      "loss": 0.8677,
+      "step": 3965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083719268598249e-06,
+      "loss": 0.8384,
+      "step": 3966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079042047983443e-06,
+      "loss": 0.8154,
+      "step": 3967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074366821047511e-06,
+      "loss": 0.8145,
+      "step": 3968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069693589364676e-06,
+      "loss": 0.772,
+      "step": 3969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065022354508488e-06,
+      "loss": 0.894,
+      "step": 3970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060353118051829e-06,
+      "loss": 0.8218,
+      "step": 3971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0556858815668996e-06,
+      "loss": 0.811,
+      "step": 3972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051020646625228e-06,
+      "loss": 0.7783,
+      "step": 3973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046357414797678e-06,
+      "loss": 0.8809,
+      "step": 3974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041696187654426e-06,
+      "loss": 0.7593,
+      "step": 3975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0370369667649835e-06,
+      "loss": 0.8164,
+      "step": 3976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032379753698187e-06,
+      "loss": 0.7964,
+      "step": 3977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027724550022183e-06,
+      "loss": 0.8086,
+      "step": 3978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023071357304461e-06,
+      "loss": 0.8506,
+      "step": 3979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018420177111816e-06,
+      "loss": 0.8623,
+      "step": 3980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013771011010377e-06,
+      "loss": 0.8184,
+      "step": 3981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009123860565589e-06,
+      "loss": 0.8057,
+      "step": 3982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00447872734222e-06,
+      "loss": 0.8145,
+      "step": 3983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999835612904363e-06,
+      "loss": 0.781,
+      "step": 3984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995194518815419e-06,
+      "loss": 0.8213,
+      "step": 3985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990555446638127e-06,
+      "loss": 0.8447,
+      "step": 3986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985918397934527e-06,
+      "loss": 0.8369,
+      "step": 3987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9812833742659865e-06,
+      "loss": 0.8184,
+      "step": 3988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97665037719319e-06,
+      "loss": 0.8022,
+      "step": 3989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972019408276141e-06,
+      "loss": 0.8267,
+      "step": 3990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967390469074164e-06,
+      "loss": 0.8091,
+      "step": 3991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96276356114589e-06,
+      "loss": 0.8374,
+      "step": 3992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958138686049269e-06,
+      "loss": 0.854,
+      "step": 3993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953515845341574e-06,
+      "loss": 0.7974,
+      "step": 3994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94889504057938e-06,
+      "loss": 0.8511,
+      "step": 3995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9442762733185906e-06,
+      "loss": 0.8237,
+      "step": 3996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939659545114416e-06,
+      "loss": 0.8179,
+      "step": 3997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935044857521378e-06,
+      "loss": 0.7964,
+      "step": 3998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930432212093317e-06,
+      "loss": 0.813,
+      "step": 3999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925821610383379e-06,
+      "loss": 0.75,
+      "step": 4000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92121305394403e-06,
+      "loss": 0.7866,
+      "step": 4001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916606544327036e-06,
+      "loss": 0.8413,
+      "step": 4002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912002083083486e-06,
+      "loss": 0.7998,
+      "step": 4003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907399671763777e-06,
+      "loss": 0.8125,
+      "step": 4004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902799311917607e-06,
+      "loss": 0.8062,
+      "step": 4005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898201005093986e-06,
+      "loss": 0.8564,
+      "step": 4006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893604752841246e-06,
+      "loss": 0.8135,
+      "step": 4007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889010556707005e-06,
+      "loss": 0.8071,
+      "step": 4008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884418418238207e-06,
+      "loss": 0.772,
+      "step": 4009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879828338981099e-06,
+      "loss": 0.8618,
+      "step": 4010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875240320481226e-06,
+      "loss": 0.3535,
+      "step": 4011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870654364283453e-06,
+      "loss": 0.7695,
+      "step": 4012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866070471931933e-06,
+      "loss": 0.7925,
+      "step": 4013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861488644970144e-06,
+      "loss": 0.8135,
+      "step": 4014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856908884940851e-06,
+      "loss": 0.8506,
+      "step": 4015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.852331193386133e-06,
+      "loss": 0.8052,
+      "step": 4016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847755571847375e-06,
+      "loss": 0.7729,
+      "step": 4017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8431820218652516e-06,
+      "loss": 0.7876,
+      "step": 4018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838610544979756e-06,
+      "loss": 0.7998,
+      "step": 4019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.834041142730174e-06,
+      "loss": 0.811,
+      "step": 4020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829473816655088e-06,
+      "loss": 0.8735,
+      "step": 4021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824908568292393e-06,
+      "loss": 0.8105,
+      "step": 4022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820345399179282e-06,
+      "loss": 0.8691,
+      "step": 4023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815784310852243e-06,
+      "loss": 0.8174,
+      "step": 4024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811225304847066e-06,
+      "loss": 0.793,
+      "step": 4025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806668382698836e-06,
+      "loss": 0.3381,
+      "step": 4026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8021135459419446e-06,
+      "loss": 0.8711,
+      "step": 4027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7975607961100715e-06,
+      "loss": 0.856,
+      "step": 4028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7930101347362004e-06,
+      "loss": 0.8091,
+      "step": 4029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7884615633526146e-06,
+      "loss": 0.9097,
+      "step": 4030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.783915083490881e-06,
+      "loss": 0.8979,
+      "step": 4031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779370696681878e-06,
+      "loss": 0.7935,
+      "step": 4032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774828404455764e-06,
+      "loss": 0.8193,
+      "step": 4033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.770288208342008e-06,
+      "loss": 0.8047,
+      "step": 4034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7657501098693563e-06,
+      "loss": 0.8232,
+      "step": 4035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7612141105658606e-06,
+      "loss": 0.8174,
+      "step": 4036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7566802119588674e-06,
+      "loss": 0.8638,
+      "step": 4037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7521484155750045e-06,
+      "loss": 0.8662,
+      "step": 4038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.747618722940204e-06,
+      "loss": 0.3096,
+      "step": 4039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7430911355796805e-06,
+      "loss": 0.7627,
+      "step": 4040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7385656550179407e-06,
+      "loss": 0.7666,
+      "step": 4041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.734042282778788e-06,
+      "loss": 0.3077,
+      "step": 4042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7295210203853115e-06,
+      "loss": 0.7754,
+      "step": 4043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7250018693598944e-06,
+      "loss": 0.8477,
+      "step": 4044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7204848312242028e-06,
+      "loss": 0.7432,
+      "step": 4045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7159699074991894e-06,
+      "loss": 0.7773,
+      "step": 4046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.711457099705107e-06,
+      "loss": 0.3479,
+      "step": 4047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7069464093614803e-06,
+      "loss": 0.3145,
+      "step": 4048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702437837987134e-06,
+      "loss": 0.8018,
+      "step": 4049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6979313871001776e-06,
+      "loss": 0.854,
+      "step": 4050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6934270582179956e-06,
+      "loss": 0.856,
+      "step": 4051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.688924852857274e-06,
+      "loss": 0.7637,
+      "step": 4052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6844247725339665e-06,
+      "loss": 0.8184,
+      "step": 4053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6799268187633298e-06,
+      "loss": 0.7578,
+      "step": 4054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6754309930598866e-06,
+      "loss": 0.8184,
+      "step": 4055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6709372969374546e-06,
+      "loss": 0.3354,
+      "step": 4056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6664457319091363e-06,
+      "loss": 0.8071,
+      "step": 4057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661956299487307e-06,
+      "loss": 0.8003,
+      "step": 4058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6574690011836255e-06,
+      "loss": 0.8374,
+      "step": 4059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.652983838509043e-06,
+      "loss": 0.7603,
+      "step": 4060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.648500812973775e-06,
+      "loss": 0.8379,
+      "step": 4061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.644019926087331e-06,
+      "loss": 0.7793,
+      "step": 4062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.639541179358499e-06,
+      "loss": 0.8433,
+      "step": 4063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.635064574295335e-06,
+      "loss": 0.8359,
+      "step": 4064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630590112405189e-06,
+      "loss": 0.7949,
+      "step": 4065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6261177951946746e-06,
+      "loss": 0.8691,
+      "step": 4066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6216476241696985e-06,
+      "loss": 0.7773,
+      "step": 4067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617179600835431e-06,
+      "loss": 0.8115,
+      "step": 4068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6127137266963265e-06,
+      "loss": 0.853,
+      "step": 4069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6082500032561185e-06,
+      "loss": 0.7905,
+      "step": 4070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6037884320178073e-06,
+      "loss": 0.8535,
+      "step": 4071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.599329014483678e-06,
+      "loss": 0.8359,
+      "step": 4072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5948717521552845e-06,
+      "loss": 0.8135,
+      "step": 4073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590416646533451e-06,
+      "loss": 0.749,
+      "step": 4074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5859636991182867e-06,
+      "loss": 0.7759,
+      "step": 4075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5815129114091663e-06,
+      "loss": 0.8232,
+      "step": 4076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5770642849047443e-06,
+      "loss": 0.3511,
+      "step": 4077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5726178211029396e-06,
+      "loss": 0.322,
+      "step": 4078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5681735215009403e-06,
+      "loss": 0.8193,
+      "step": 4079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5637313875952205e-06,
+      "loss": 0.8232,
+      "step": 4080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5592914208815086e-06,
+      "loss": 0.7993,
+      "step": 4081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5548536228548127e-06,
+      "loss": 0.8477,
+      "step": 4082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550417995009414e-06,
+      "loss": 0.8428,
+      "step": 4083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5459845388388494e-06,
+      "loss": 0.7749,
+      "step": 4084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.541553255835939e-06,
+      "loss": 0.752,
+      "step": 4085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5371241474927597e-06,
+      "loss": 0.8042,
+      "step": 4086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5326972153006677e-06,
+      "loss": 0.7974,
+      "step": 4087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5282724607502728e-06,
+      "loss": 0.8257,
+      "step": 4088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.523849885331462e-06,
+      "loss": 0.3601,
+      "step": 4089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5194294905333904e-06,
+      "loss": 0.8325,
+      "step": 4090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5150112778444647e-06,
+      "loss": 0.8398,
+      "step": 4091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510595248752374e-06,
+      "loss": 0.8311,
+      "step": 4092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5061814047440624e-06,
+      "loss": 0.7983,
+      "step": 4093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5017697473057346e-06,
+      "loss": 0.8271,
+      "step": 4094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4973602779228665e-06,
+      "loss": 0.7974,
+      "step": 4095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.492952998080199e-06,
+      "loss": 0.8184,
+      "step": 4096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4885479092617325e-06,
+      "loss": 0.8218,
+      "step": 4097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484145012950727e-06,
+      "loss": 0.8003,
+      "step": 4098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479744310629701e-06,
+      "loss": 0.8115,
+      "step": 4099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.47534580378045e-06,
+      "loss": 0.8511,
+      "step": 4100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4709494938840093e-06,
+      "loss": 0.7788,
+      "step": 4101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.466555382420691e-06,
+      "loss": 0.772,
+      "step": 4102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4621634708700637e-06,
+      "loss": 0.812,
+      "step": 4103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4577737607109442e-06,
+      "loss": 0.8062,
+      "step": 4104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4533862534214246e-06,
+      "loss": 0.7764,
+      "step": 4105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4490009504788444e-06,
+      "loss": 0.7783,
+      "step": 4106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4446178533597983e-06,
+      "loss": 0.8223,
+      "step": 4107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440236963540148e-06,
+      "loss": 0.7563,
+      "step": 4108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.435858282495006e-06,
+      "loss": 0.8828,
+      "step": 4109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4314818116987483e-06,
+      "loss": 0.8076,
+      "step": 4110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427107552624995e-06,
+      "loss": 0.8271,
+      "step": 4111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422735506746626e-06,
+      "loss": 0.7661,
+      "step": 4112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.418365675535783e-06,
+      "loss": 0.8252,
+      "step": 4113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413998060463849e-06,
+      "loss": 0.7983,
+      "step": 4114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409632663001473e-06,
+      "loss": 0.8389,
+      "step": 4115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.405269484618554e-06,
+      "loss": 0.8467,
+      "step": 4116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4009085267842366e-06,
+      "loss": 0.8691,
+      "step": 4117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396549790966931e-06,
+      "loss": 0.8574,
+      "step": 4118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3921932786342827e-06,
+      "loss": 0.7935,
+      "step": 4119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387838991253205e-06,
+      "loss": 0.792,
+      "step": 4120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.383486930289849e-06,
+      "loss": 0.8237,
+      "step": 4121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3791370972096218e-06,
+      "loss": 0.8096,
+      "step": 4122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374789493477185e-06,
+      "loss": 0.7808,
+      "step": 4123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3704441205564396e-06,
+      "loss": 0.8237,
+      "step": 4124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366100979910545e-06,
+      "loss": 0.3889,
+      "step": 4125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.361760073001901e-06,
+      "loss": 0.8037,
+      "step": 4126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3574214012921525e-06,
+      "loss": 0.7227,
+      "step": 4127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3530849662422116e-06,
+      "loss": 0.8394,
+      "step": 4128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.348750769312212e-06,
+      "loss": 0.7886,
+      "step": 4129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.344418811961555e-06,
+      "loss": 0.8223,
+      "step": 4130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340089095648873e-06,
+      "loss": 0.7944,
+      "step": 4131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3357616218320455e-06,
+      "loss": 0.8115,
+      "step": 4132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3314363919682035e-06,
+      "loss": 0.8184,
+      "step": 4133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.327113407513721e-06,
+      "loss": 0.8306,
+      "step": 4134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3227926699242165e-06,
+      "loss": 0.8101,
+      "step": 4135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.318474180654545e-06,
+      "loss": 0.8013,
+      "step": 4136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3141579411588087e-06,
+      "loss": 0.853,
+      "step": 4137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.309843952890358e-06,
+      "loss": 0.8296,
+      "step": 4138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3055322173017735e-06,
+      "loss": 0.8896,
+      "step": 4139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.301222735844887e-06,
+      "loss": 0.8457,
+      "step": 4140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2969155099707705e-06,
+      "loss": 0.8345,
+      "step": 4141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2926105411297295e-06,
+      "loss": 0.8066,
+      "step": 4142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288307830771319e-06,
+      "loss": 0.7979,
+      "step": 4143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284007380344323e-06,
+      "loss": 0.7974,
+      "step": 4144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.279709191296777e-06,
+      "loss": 0.7783,
+      "step": 4145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.275413265075941e-06,
+      "loss": 0.7959,
+      "step": 4146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2711196031283233e-06,
+      "loss": 0.7896,
+      "step": 4147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2668282068996716e-06,
+      "loss": 0.8154,
+      "step": 4148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2625390778349588e-06,
+      "loss": 0.7427,
+      "step": 4149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.258252217378407e-06,
+      "loss": 0.7793,
+      "step": 4150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2539676269734668e-06,
+      "loss": 0.8008,
+      "step": 4151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2496853080628232e-06,
+      "loss": 0.8618,
+      "step": 4152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2454052620884024e-06,
+      "loss": 0.7993,
+      "step": 4153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2411274904913657e-06,
+      "loss": 0.7866,
+      "step": 4154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2368519947121015e-06,
+      "loss": 0.7861,
+      "step": 4155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2325787761902392e-06,
+      "loss": 0.729,
+      "step": 4156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228307836364634e-06,
+      "loss": 0.8433,
+      "step": 4157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2240391766733825e-06,
+      "loss": 0.7964,
+      "step": 4158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219772798553803e-06,
+      "loss": 0.8105,
+      "step": 4159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215508703442456e-06,
+      "loss": 0.7783,
+      "step": 4160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.211246892775132e-06,
+      "loss": 0.8208,
+      "step": 4161
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2069873679868404e-06,
+      "loss": 0.8032,
+      "step": 4162
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2027301305118384e-06,
+      "loss": 0.7871,
+      "step": 4163
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.198475181783599e-06,
+      "loss": 0.8477,
+      "step": 4164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1942225232348288e-06,
+      "loss": 0.3181,
+      "step": 4165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1899721562974663e-06,
+      "loss": 0.7886,
+      "step": 4166
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185724082402677e-06,
+      "loss": 0.8062,
+      "step": 4167
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.181478302980856e-06,
+      "loss": 0.8281,
+      "step": 4168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.17723481946162e-06,
+      "loss": 0.8809,
+      "step": 4169
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.172993633273813e-06,
+      "loss": 0.8604,
+      "step": 4170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168754745845516e-06,
+      "loss": 0.8149,
+      "step": 4171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1645181586040217e-06,
+      "loss": 0.7554,
+      "step": 4172
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1602838729758578e-06,
+      "loss": 0.8496,
+      "step": 4173
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1560518903867776e-06,
+      "loss": 0.8232,
+      "step": 4174
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1518222122617494e-06,
+      "loss": 0.8237,
+      "step": 4175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.147594840024978e-06,
+      "loss": 0.8428,
+      "step": 4176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143369775099879e-06,
+      "loss": 0.7432,
+      "step": 4177
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1391470189091046e-06,
+      "loss": 0.8745,
+      "step": 4178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.134926572874516e-06,
+      "loss": 0.8481,
+      "step": 4179
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.130708438417207e-06,
+      "loss": 0.8164,
+      "step": 4180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1264926169574906e-06,
+      "loss": 0.8267,
+      "step": 4181
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122279109914895e-06,
+      "loss": 0.8359,
+      "step": 4182
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11806791870818e-06,
+      "loss": 0.3276,
+      "step": 4183
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.113859044755315e-06,
+      "loss": 0.8325,
+      "step": 4184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1096524894734927e-06,
+      "loss": 0.8403,
+      "step": 4185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105448254279125e-06,
+      "loss": 0.8491,
+      "step": 4186
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101246340587848e-06,
+      "loss": 0.8076,
+      "step": 4187
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0970467498145118e-06,
+      "loss": 0.7583,
+      "step": 4188
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0928494833731814e-06,
+      "loss": 0.7842,
+      "step": 4189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0886545426771387e-06,
+      "loss": 0.8169,
+      "step": 4190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0844619291388932e-06,
+      "loss": 0.8555,
+      "step": 4191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0802716441701554e-06,
+      "loss": 0.7739,
+      "step": 4192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0760836891818636e-06,
+      "loss": 0.7568,
+      "step": 4193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0718980655841714e-06,
+      "loss": 0.8325,
+      "step": 4194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0677147747864344e-06,
+      "loss": 0.8232,
+      "step": 4195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063533818197242e-06,
+      "loss": 0.3215,
+      "step": 4196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0593551972243772e-06,
+      "loss": 0.8052,
+      "step": 4197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0551789132748567e-06,
+      "loss": 0.7656,
+      "step": 4198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0510049677548927e-06,
+      "loss": 0.853,
+      "step": 4199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0468333620699208e-06,
+      "loss": 0.7666,
+      "step": 4200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.042664097624588e-06,
+      "loss": 0.8081,
+      "step": 4201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0384971758227454e-06,
+      "loss": 0.8447,
+      "step": 4202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0343325980674676e-06,
+      "loss": 0.8208,
+      "step": 4203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0301703657610284e-06,
+      "loss": 0.8262,
+      "step": 4204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.026010480304913e-06,
+      "loss": 0.769,
+      "step": 4205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021852943099823e-06,
+      "loss": 0.8145,
+      "step": 4206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017697755545671e-06,
+      "loss": 0.7412,
+      "step": 4207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0135449190415656e-06,
+      "loss": 0.8174,
+      "step": 4208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0093944349858374e-06,
+      "loss": 0.7793,
+      "step": 4209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0052463047760138e-06,
+      "loss": 0.8716,
+      "step": 4210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0011005298088403e-06,
+      "loss": 0.7886,
+      "step": 4211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9969571114802586e-06,
+      "loss": 0.8335,
+      "step": 4212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9928160511854255e-06,
+      "loss": 0.7832,
+      "step": 4213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9886773503187026e-06,
+      "loss": 0.8662,
+      "step": 4214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9845410102736495e-06,
+      "loss": 0.8198,
+      "step": 4215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9804070324430435e-06,
+      "loss": 0.854,
+      "step": 4216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.976275418218855e-06,
+      "loss": 0.7998,
+      "step": 4217
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9721461689922594e-06,
+      "loss": 0.9033,
+      "step": 4218
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968019286153644e-06,
+      "loss": 0.8472,
+      "step": 4219
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9638947710925937e-06,
+      "loss": 0.7739,
+      "step": 4220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9597726251978996e-06,
+      "loss": 0.835,
+      "step": 4221
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.955652849857551e-06,
+      "loss": 0.7744,
+      "step": 4222
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9515354464587353e-06,
+      "loss": 0.7759,
+      "step": 4223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.947420416387855e-06,
+      "loss": 0.8218,
+      "step": 4224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9433077610304974e-06,
+      "loss": 0.8101,
+      "step": 4225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939197481771461e-06,
+      "loss": 0.8477,
+      "step": 4226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9350895799947433e-06,
+      "loss": 0.8408,
+      "step": 4227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9309840570835336e-06,
+      "loss": 0.8174,
+      "step": 4228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9268809144202303e-06,
+      "loss": 0.7383,
+      "step": 4229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.922780153386422e-06,
+      "loss": 0.8203,
+      "step": 4230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9186817753629025e-06,
+      "loss": 0.8213,
+      "step": 4231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9145857817296554e-06,
+      "loss": 0.79,
+      "step": 4232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9104921738658674e-06,
+      "loss": 0.8232,
+      "step": 4233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9064009531499225e-06,
+      "loss": 0.7998,
+      "step": 4234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9023121209593953e-06,
+      "loss": 0.7773,
+      "step": 4235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8982256786710617e-06,
+      "loss": 0.8203,
+      "step": 4236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89414162766089e-06,
+      "loss": 0.75,
+      "step": 4237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890059969304041e-06,
+      "loss": 0.7803,
+      "step": 4238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8859807049748734e-06,
+      "loss": 0.7905,
+      "step": 4239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.881903836046942e-06,
+      "loss": 0.8022,
+      "step": 4240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8778293638929935e-06,
+      "loss": 0.7979,
+      "step": 4241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.873757289884963e-06,
+      "loss": 0.741,
+      "step": 4242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.869687615393979e-06,
+      "loss": 0.3345,
+      "step": 4243
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8656203417903704e-06,
+      "loss": 0.7832,
+      "step": 4244
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8615554704436452e-06,
+      "loss": 0.7725,
+      "step": 4245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8574930027225123e-06,
+      "loss": 0.3158,
+      "step": 4246
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85343293999487e-06,
+      "loss": 0.7842,
+      "step": 4247
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.849375283627801e-06,
+      "loss": 0.8174,
+      "step": 4248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8453200349875855e-06,
+      "loss": 0.8403,
+      "step": 4249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8412671954396833e-06,
+      "loss": 0.873,
+      "step": 4250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.837216766348755e-06,
+      "loss": 0.7969,
+      "step": 4251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8331687490786376e-06,
+      "loss": 0.7866,
+      "step": 4252
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829123144992364e-06,
+      "loss": 0.8442,
+      "step": 4253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8250799554521547e-06,
+      "loss": 0.855,
+      "step": 4254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821039181819413e-06,
+      "loss": 0.7578,
+      "step": 4255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8170008254547265e-06,
+      "loss": 0.832,
+      "step": 4256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8129648877178784e-06,
+      "loss": 0.8169,
+      "step": 4257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8089313699678267e-06,
+      "loss": 0.8647,
+      "step": 4258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804900273562723e-06,
+      "loss": 0.8374,
+      "step": 4259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8008715998599025e-06,
+      "loss": 0.7402,
+      "step": 4260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7968453502158765e-06,
+      "loss": 0.8604,
+      "step": 4261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792821525986353e-06,
+      "loss": 0.8306,
+      "step": 4262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.788800128526209e-06,
+      "loss": 0.7925,
+      "step": 4263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.784781159189519e-06,
+      "loss": 0.7695,
+      "step": 4264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7807646193295257e-06,
+      "loss": 0.8491,
+      "step": 4265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7767505102986635e-06,
+      "loss": 0.748,
+      "step": 4266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7727388334485496e-06,
+      "loss": 0.811,
+      "step": 4267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7687295901299715e-06,
+      "loss": 0.8052,
+      "step": 4268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76472278169291e-06,
+      "loss": 0.8438,
+      "step": 4269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7607184094865182e-06,
+      "loss": 0.8135,
+      "step": 4270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756716474859125e-06,
+      "loss": 0.8208,
+      "step": 4271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7527169791582497e-06,
+      "loss": 0.7866,
+      "step": 4272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.748719923730583e-06,
+      "loss": 0.7646,
+      "step": 4273
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7447253099220004e-06,
+      "loss": 0.7939,
+      "step": 4274
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740733139077546e-06,
+      "loss": 0.8008,
+      "step": 4275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.736743412541445e-06,
+      "loss": 0.3452,
+      "step": 4276
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.732756131657105e-06,
+      "loss": 0.8257,
+      "step": 4277
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7287712977671e-06,
+      "loss": 0.855,
+      "step": 4278
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7247889122131886e-06,
+      "loss": 0.8057,
+      "step": 4279
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7208089763363066e-06,
+      "loss": 0.6953,
+      "step": 4280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7168314914765526e-06,
+      "loss": 0.8716,
+      "step": 4281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712856458973213e-06,
+      "loss": 0.7998,
+      "step": 4282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.70888388016474e-06,
+      "loss": 0.8506,
+      "step": 4283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7049137563887674e-06,
+      "loss": 0.3127,
+      "step": 4284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7009460889820917e-06,
+      "loss": 0.8818,
+      "step": 4285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696980879280692e-06,
+      "loss": 0.7915,
+      "step": 4286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6930181286197177e-06,
+      "loss": 0.792,
+      "step": 4287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6890578383334853e-06,
+      "loss": 0.7375,
+      "step": 4288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6851000097554893e-06,
+      "loss": 0.8047,
+      "step": 4289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.681144644218392e-06,
+      "loss": 0.8433,
+      "step": 4290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677191743054022e-06,
+      "loss": 0.8413,
+      "step": 4291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.673241307593386e-06,
+      "loss": 0.77,
+      "step": 4292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6692933391666576e-06,
+      "loss": 0.8345,
+      "step": 4293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665347839103182e-06,
+      "loss": 0.8833,
+      "step": 4294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661404808731468e-06,
+      "loss": 0.8794,
+      "step": 4295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6574642493791923e-06,
+      "loss": 0.7822,
+      "step": 4296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.653526162373209e-06,
+      "loss": 0.7842,
+      "step": 4297
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6495905490395267e-06,
+      "loss": 0.8506,
+      "step": 4298
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6456574107033306e-06,
+      "loss": 0.877,
+      "step": 4299
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641726748688973e-06,
+      "loss": 0.8081,
+      "step": 4300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6377985643199644e-06,
+      "loss": 0.3193,
+      "step": 4301
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.63387285891899e-06,
+      "loss": 0.8682,
+      "step": 4302
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629949633807891e-06,
+      "loss": 0.7983,
+      "step": 4303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6260288903076838e-06,
+      "loss": 0.751,
+      "step": 4304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622110629738538e-06,
+      "loss": 0.7944,
+      "step": 4305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6181948534197964e-06,
+      "loss": 0.3259,
+      "step": 4306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6142815626699657e-06,
+      "loss": 0.8003,
+      "step": 4307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6103707588067074e-06,
+      "loss": 0.8213,
+      "step": 4308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6064624431468477e-06,
+      "loss": 0.812,
+      "step": 4309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602556617006383e-06,
+      "loss": 0.8242,
+      "step": 4310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5986532817004607e-06,
+      "loss": 0.792,
+      "step": 4311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5947524385433976e-06,
+      "loss": 0.8711,
+      "step": 4312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.590854088848671e-06,
+      "loss": 0.854,
+      "step": 4313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58695823392891e-06,
+      "loss": 0.8613,
+      "step": 4314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5830648750959164e-06,
+      "loss": 0.8369,
+      "step": 4315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579174013660638e-06,
+      "loss": 0.311,
+      "step": 4316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575285650933195e-06,
+      "loss": 0.7969,
+      "step": 4317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5713997882228537e-06,
+      "loss": 0.8657,
+      "step": 4318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.567516426838047e-06,
+      "loss": 0.8081,
+      "step": 4319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563635568086368e-06,
+      "loss": 0.3313,
+      "step": 4320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559757213274554e-06,
+      "loss": 0.8086,
+      "step": 4321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5558813637085158e-06,
+      "loss": 0.7559,
+      "step": 4322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5520080206933072e-06,
+      "loss": 0.7969,
+      "step": 4323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5481371855331415e-06,
+      "loss": 0.7852,
+      "step": 4324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5442688595313926e-06,
+      "loss": 0.7881,
+      "step": 4325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.540403043990586e-06,
+      "loss": 0.7988,
+      "step": 4326
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5365397402124036e-06,
+      "loss": 0.7944,
+      "step": 4327
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532678949497679e-06,
+      "loss": 0.7881,
+      "step": 4328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.528820673146396e-06,
+      "loss": 0.7729,
+      "step": 4329
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5249649124577037e-06,
+      "loss": 0.8667,
+      "step": 4330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5211116687298877e-06,
+      "loss": 0.8423,
+      "step": 4331
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5172609432604066e-06,
+      "loss": 0.811,
+      "step": 4332
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.513412737345854e-06,
+      "loss": 0.8223,
+      "step": 4333
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.509567052281978e-06,
+      "loss": 0.7373,
+      "step": 4334
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505723889363687e-06,
+      "loss": 0.8501,
+      "step": 4335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5018832498850277e-06,
+      "loss": 0.7539,
+      "step": 4336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4980451351392056e-06,
+      "loss": 0.8086,
+      "step": 4337
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494209546418577e-06,
+      "loss": 0.8594,
+      "step": 4338
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4903764850146385e-06,
+      "loss": 0.8252,
+      "step": 4339
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4865459522180492e-06,
+      "loss": 0.8647,
+      "step": 4340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4827179493186006e-06,
+      "loss": 0.8408,
+      "step": 4341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478892477605248e-06,
+      "loss": 0.7964,
+      "step": 4342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4750695383660826e-06,
+      "loss": 0.7954,
+      "step": 4343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4712491328883505e-06,
+      "loss": 0.8423,
+      "step": 4344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.467431262458443e-06,
+      "loss": 0.8623,
+      "step": 4345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.463615928361891e-06,
+      "loss": 0.8623,
+      "step": 4346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459803131883385e-06,
+      "loss": 0.8003,
+      "step": 4347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4559928743067475e-06,
+      "loss": 0.8286,
+      "step": 4348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45218515691495e-06,
+      "loss": 0.7822,
+      "step": 4349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4483799809901133e-06,
+      "loss": 0.8027,
+      "step": 4350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4445773478135006e-06,
+      "loss": 0.8125,
+      "step": 4351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4407772586655175e-06,
+      "loss": 0.7778,
+      "step": 4352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4369797148257136e-06,
+      "loss": 0.8174,
+      "step": 4353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4331847175727753e-06,
+      "loss": 0.7944,
+      "step": 4354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429392268184546e-06,
+      "loss": 0.8335,
+      "step": 4355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.425602367937995e-06,
+      "loss": 0.7163,
+      "step": 4356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4218150181092447e-06,
+      "loss": 0.7842,
+      "step": 4357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4180302199735572e-06,
+      "loss": 0.8027,
+      "step": 4358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.414247974805327e-06,
+      "loss": 0.8379,
+      "step": 4359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.410468283878101e-06,
+      "loss": 0.7861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406691148464557e-06,
+      "loss": 0.3345,
+      "step": 4361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4029165698365142e-06,
+      "loss": 0.8066,
+      "step": 4362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.399144549264932e-06,
+      "loss": 0.7729,
+      "step": 4363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.395375088019911e-06,
+      "loss": 0.835,
+      "step": 4364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39160818737069e-06,
+      "loss": 0.833,
+      "step": 4365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387843848585639e-06,
+      "loss": 0.8584,
+      "step": 4366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3840820729322677e-06,
+      "loss": 0.8022,
+      "step": 4367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3803228616772313e-06,
+      "loss": 0.7749,
+      "step": 4368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376566216086307e-06,
+      "loss": 0.9219,
+      "step": 4369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3728121374244205e-06,
+      "loss": 0.8315,
+      "step": 4370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3690606269556306e-06,
+      "loss": 0.8423,
+      "step": 4371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3653116859431246e-06,
+      "loss": 0.833,
+      "step": 4372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3615653156492334e-06,
+      "loss": 0.8164,
+      "step": 4373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3578215173354146e-06,
+      "loss": 0.8711,
+      "step": 4374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3540802922622686e-06,
+      "loss": 0.8354,
+      "step": 4375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3503416416895174e-06,
+      "loss": 0.7837,
+      "step": 4376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3466055668760266e-06,
+      "loss": 0.8076,
+      "step": 4377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3428720690797933e-06,
+      "loss": 0.814,
+      "step": 4378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3391411495579396e-06,
+      "loss": 0.7905,
+      "step": 4379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3354128095667293e-06,
+      "loss": 0.7285,
+      "step": 4380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331687050361551e-06,
+      "loss": 0.8057,
+      "step": 4381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.32796387319692e-06,
+      "loss": 0.8403,
+      "step": 4382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3242432793264945e-06,
+      "loss": 0.8442,
+      "step": 4383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3205252700030544e-06,
+      "loss": 0.7749,
+      "step": 4384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.316809846478516e-06,
+      "loss": 0.8184,
+      "step": 4385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3130970100039153e-06,
+      "loss": 0.832,
+      "step": 4386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3093867618294198e-06,
+      "loss": 0.8613,
+      "step": 4387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3056791032043356e-06,
+      "loss": 0.7896,
+      "step": 4388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3019740353770814e-06,
+      "loss": 0.8452,
+      "step": 4389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2982715595952142e-06,
+      "loss": 0.8682,
+      "step": 4390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2945716771054205e-06,
+      "loss": 0.8301,
+      "step": 4391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2908743891535003e-06,
+      "loss": 0.772,
+      "step": 4392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287179696984394e-06,
+      "loss": 0.811,
+      "step": 4393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283487601842157e-06,
+      "loss": 0.7764,
+      "step": 4394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279798104969981e-06,
+      "loss": 0.8071,
+      "step": 4395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.276111207610173e-06,
+      "loss": 0.8208,
+      "step": 4396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2724269110041686e-06,
+      "loss": 0.7686,
+      "step": 4397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268745216392533e-06,
+      "loss": 0.8228,
+      "step": 4398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265066125014943e-06,
+      "loss": 0.8076,
+      "step": 4399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.261389638110213e-06,
+      "loss": 0.8286,
+      "step": 4400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25771575691627e-06,
+      "loss": 0.793,
+      "step": 4401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2540444826701646e-06,
+      "loss": 0.7666,
+      "step": 4402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.250375816608075e-06,
+      "loss": 0.813,
+      "step": 4403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2467097599653e-06,
+      "loss": 0.8228,
+      "step": 4404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243046313976254e-06,
+      "loss": 0.8159,
+      "step": 4405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.23938547987448e-06,
+      "loss": 0.7847,
+      "step": 4406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2357272588926347e-06,
+      "loss": 0.7749,
+      "step": 4407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2320716522625017e-06,
+      "loss": 0.7773,
+      "step": 4408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2284186612149773e-06,
+      "loss": 0.8296,
+      "step": 4409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22476828698008e-06,
+      "loss": 0.3257,
+      "step": 4410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2211205307869545e-06,
+      "loss": 0.3372,
+      "step": 4411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2174753938638495e-06,
+      "loss": 0.7861,
+      "step": 4412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2138328774381448e-06,
+      "loss": 0.8174,
+      "step": 4413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.21019298273633e-06,
+      "loss": 0.8257,
+      "step": 4414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206555710984013e-06,
+      "loss": 0.8154,
+      "step": 4415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2029210634059206e-06,
+      "loss": 0.8442,
+      "step": 4416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.199289041225896e-06,
+      "loss": 0.8369,
+      "step": 4417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195659645666901e-06,
+      "loss": 0.7168,
+      "step": 4418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192032877951006e-06,
+      "loss": 0.8521,
+      "step": 4419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188408739299399e-06,
+      "loss": 0.7466,
+      "step": 4420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184787230932388e-06,
+      "loss": 0.8218,
+      "step": 4421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181168354069386e-06,
+      "loss": 0.8281,
+      "step": 4422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17755210992893e-06,
+      "loss": 0.8506,
+      "step": 4423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173938499728666e-06,
+      "loss": 0.8091,
+      "step": 4424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1703275246853484e-06,
+      "loss": 0.8105,
+      "step": 4425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166719186014854e-06,
+      "loss": 0.8252,
+      "step": 4426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163113484932162e-06,
+      "loss": 0.8105,
+      "step": 4427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159510422651373e-06,
+      "loss": 0.3406,
+      "step": 4428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.155910000385689e-06,
+      "loss": 0.834,
+      "step": 4429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152312219347431e-06,
+      "loss": 0.8384,
+      "step": 4430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1487170807480304e-06,
+      "loss": 0.8364,
+      "step": 4431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1451245857980217e-06,
+      "loss": 0.8047,
+      "step": 4432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.141534735707059e-06,
+      "loss": 0.8267,
+      "step": 4433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1379475316838995e-06,
+      "loss": 0.8218,
+      "step": 4434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1343629749364047e-06,
+      "loss": 0.7896,
+      "step": 4435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1307810666715567e-06,
+      "loss": 0.8384,
+      "step": 4436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1272018080954373e-06,
+      "loss": 0.3362,
+      "step": 4437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.123625200413243e-06,
+      "loss": 0.8589,
+      "step": 4438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.120051244829271e-06,
+      "loss": 0.7891,
+      "step": 4439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116479942546924e-06,
+      "loss": 0.8545,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112911294768721e-06,
+      "loss": 0.8018,
+      "step": 4441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1093453026962764e-06,
+      "loss": 0.8237,
+      "step": 4442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.105781967530318e-06,
+      "loss": 0.813,
+      "step": 4443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1022212904706783e-06,
+      "loss": 0.8369,
+      "step": 4444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098663272716288e-06,
+      "loss": 0.7739,
+      "step": 4445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0951079154651923e-06,
+      "loss": 0.7881,
+      "step": 4446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0915552199145295e-06,
+      "loss": 0.853,
+      "step": 4447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0880051872605533e-06,
+      "loss": 0.8306,
+      "step": 4448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.084457818698611e-06,
+      "loss": 0.3207,
+      "step": 4449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0809131154231587e-06,
+      "loss": 0.8081,
+      "step": 4450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0773710786277545e-06,
+      "loss": 0.7744,
+      "step": 4451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.073831709505054e-06,
+      "loss": 0.8374,
+      "step": 4452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070295009246823e-06,
+      "loss": 0.8174,
+      "step": 4453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0667609790439215e-06,
+      "loss": 0.7842,
+      "step": 4454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063229620086308e-06,
+      "loss": 0.813,
+      "step": 4455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0597009335630493e-06,
+      "loss": 0.8398,
+      "step": 4456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0561749206623137e-06,
+      "loss": 0.7778,
+      "step": 4457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0526515825713565e-06,
+      "loss": 0.833,
+      "step": 4458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0491309204765487e-06,
+      "loss": 0.7573,
+      "step": 4459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045612935563346e-06,
+      "loss": 0.769,
+      "step": 4460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042097629016313e-06,
+      "loss": 0.811,
+      "step": 4461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0385850020191034e-06,
+      "loss": 0.7969,
+      "step": 4462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0350750557544773e-06,
+      "loss": 0.7578,
+      "step": 4463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0315677914042887e-06,
+      "loss": 0.8086,
+      "step": 4464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0280632101494855e-06,
+      "loss": 0.8159,
+      "step": 4465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.024561313170118e-06,
+      "loss": 0.8184,
+      "step": 4466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021062101645329e-06,
+      "loss": 0.8667,
+      "step": 4467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0175655767533532e-06,
+      "loss": 0.7715,
+      "step": 4468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.014071739671528e-06,
+      "loss": 0.8105,
+      "step": 4469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0105805915762834e-06,
+      "loss": 0.8564,
+      "step": 4470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.007092133643145e-06,
+      "loss": 0.7603,
+      "step": 4471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00360636704673e-06,
+      "loss": 0.8677,
+      "step": 4472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000123292960746e-06,
+      "loss": 0.8315,
+      "step": 4473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9966429125580057e-06,
+      "loss": 0.8105,
+      "step": 4474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9931652270103994e-06,
+      "loss": 0.8345,
+      "step": 4475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9896902374889225e-06,
+      "loss": 0.3002,
+      "step": 4476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986217945163661e-06,
+      "loss": 0.8228,
+      "step": 4477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982748351203785e-06,
+      "loss": 0.8101,
+      "step": 4478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9792814567775643e-06,
+      "loss": 0.8188,
+      "step": 4479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9758172630523507e-06,
+      "loss": 0.8267,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9723557711946004e-06,
+      "loss": 0.8223,
+      "step": 4481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9688969823698457e-06,
+      "loss": 0.2936,
+      "step": 4482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.965440897742715e-06,
+      "loss": 0.8306,
+      "step": 4483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961987518476931e-06,
+      "loss": 0.7534,
+      "step": 4484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9585368457352925e-06,
+      "loss": 0.7793,
+      "step": 4485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.955088880679704e-06,
+      "loss": 0.7715,
+      "step": 4486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951643624471142e-06,
+      "loss": 0.8066,
+      "step": 4487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9482010782696793e-06,
+      "loss": 0.8511,
+      "step": 4488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9447612432344754e-06,
+      "loss": 0.8447,
+      "step": 4489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9413241205237766e-06,
+      "loss": 0.8281,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937889711294918e-06,
+      "loss": 0.8159,
+      "step": 4491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934458016704317e-06,
+      "loss": 0.8423,
+      "step": 4492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931029037907476e-06,
+      "loss": 0.7812,
+      "step": 4493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927602776058991e-06,
+      "loss": 0.7793,
+      "step": 4494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9241792323125308e-06,
+      "loss": 0.8564,
+      "step": 4495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9207584078208608e-06,
+      "loss": 0.8457,
+      "step": 4496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9173403037358273e-06,
+      "loss": 0.7891,
+      "step": 4497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9139249212083554e-06,
+      "loss": 0.7759,
+      "step": 4498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910512261388461e-06,
+      "loss": 0.8169,
+      "step": 4499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9071023254252353e-06,
+      "loss": 0.8447,
+      "step": 4500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9036951144668625e-06,
+      "loss": 0.8706,
+      "step": 4501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9002906296605993e-06,
+      "loss": 0.8286,
+      "step": 4502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.896888872152789e-06,
+      "loss": 0.8154,
+      "step": 4503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8934898430888614e-06,
+      "loss": 0.854,
+      "step": 4504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8900935436133195e-06,
+      "loss": 0.8408,
+      "step": 4505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8866999748697457e-06,
+      "loss": 0.7476,
+      "step": 4506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8833091380008163e-06,
+      "loss": 0.854,
+      "step": 4507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8799210341482698e-06,
+      "loss": 0.8506,
+      "step": 4508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8765356644529387e-06,
+      "loss": 0.7666,
+      "step": 4509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8731530300547319e-06,
+      "loss": 0.8232,
+      "step": 4510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8697731320926305e-06,
+      "loss": 0.8311,
+      "step": 4511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8663959717047032e-06,
+      "loss": 0.8062,
+      "step": 4512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8630215500280868e-06,
+      "loss": 0.8535,
+      "step": 4513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8596498681990083e-06,
+      "loss": 0.8379,
+      "step": 4514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8562809273527592e-06,
+      "loss": 0.7954,
+      "step": 4515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8529147286237181e-06,
+      "loss": 0.8276,
+      "step": 4516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.849551273145338e-06,
+      "loss": 0.8506,
+      "step": 4517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846190562050143e-06,
+      "loss": 0.7954,
+      "step": 4518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8428325964697402e-06,
+      "loss": 0.7773,
+      "step": 4519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8394773775348085e-06,
+      "loss": 0.8379,
+      "step": 4520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8361249063750986e-06,
+      "loss": 0.835,
+      "step": 4521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.832775184119443e-06,
+      "loss": 0.7268,
+      "step": 4522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8294282118957451e-06,
+      "loss": 0.7935,
+      "step": 4523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8260839908309846e-06,
+      "loss": 0.8159,
+      "step": 4524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8227425220512118e-06,
+      "loss": 0.7856,
+      "step": 4525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8194038066815478e-06,
+      "loss": 0.8091,
+      "step": 4526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8160678458461944e-06,
+      "loss": 0.8242,
+      "step": 4527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8127346406684188e-06,
+      "loss": 0.7861,
+      "step": 4528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809404192270564e-06,
+      "loss": 0.8384,
+      "step": 4529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8060765017740466e-06,
+      "loss": 0.8066,
+      "step": 4530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.802751570299347e-06,
+      "loss": 0.8394,
+      "step": 4531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799429398966027e-06,
+      "loss": 0.8013,
+      "step": 4532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.796109988892708e-06,
+      "loss": 0.7759,
+      "step": 4533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7927933411970922e-06,
+      "loss": 0.7935,
+      "step": 4534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7894794569959418e-06,
+      "loss": 0.8682,
+      "step": 4535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7861683374050954e-06,
+      "loss": 0.8311,
+      "step": 4536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782859983539461e-06,
+      "loss": 0.8535,
+      "step": 4537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7795543965130091e-06,
+      "loss": 0.8457,
+      "step": 4538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7762515774387856e-06,
+      "loss": 0.8364,
+      "step": 4539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.772951527428901e-06,
+      "loss": 0.7935,
+      "step": 4540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7696542475945256e-06,
+      "loss": 0.7974,
+      "step": 4541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7663597390459176e-06,
+      "loss": 0.8735,
+      "step": 4542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7630680028923808e-06,
+      "loss": 0.8481,
+      "step": 4543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7597790402422987e-06,
+      "loss": 0.296,
+      "step": 4544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7564928522031155e-06,
+      "loss": 0.7822,
+      "step": 4545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.753209439881337e-06,
+      "loss": 0.7847,
+      "step": 4546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7499288043825434e-06,
+      "loss": 0.8369,
+      "step": 4547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7466509468113757e-06,
+      "loss": 0.8496,
+      "step": 4548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7433758682715419e-06,
+      "loss": 0.8101,
+      "step": 4549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7401035698658098e-06,
+      "loss": 0.7925,
+      "step": 4550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7368340526960103e-06,
+      "loss": 0.8413,
+      "step": 4551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7335673178630463e-06,
+      "loss": 0.7769,
+      "step": 4552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7303033664668735e-06,
+      "loss": 0.8198,
+      "step": 4553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7270421996065167e-06,
+      "loss": 0.7642,
+      "step": 4554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7237838183800648e-06,
+      "loss": 0.3245,
+      "step": 4555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.720528223884661e-06,
+      "loss": 0.8164,
+      "step": 4556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7172754172165195e-06,
+      "loss": 0.3384,
+      "step": 4557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7140253994709077e-06,
+      "loss": 0.8242,
+      "step": 4558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7107781717421556e-06,
+      "loss": 0.7549,
+      "step": 4559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075337351236588e-06,
+      "loss": 0.7876,
+      "step": 4560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042920907078676e-06,
+      "loss": 0.8345,
+      "step": 4561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7010532395862988e-06,
+      "loss": 0.7852,
+      "step": 4562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6978171828495193e-06,
+      "loss": 0.7788,
+      "step": 4563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6945839215871595e-06,
+      "loss": 0.8218,
+      "step": 4564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.691353456887913e-06,
+      "loss": 0.3291,
+      "step": 4565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.688125789839522e-06,
+      "loss": 0.792,
+      "step": 4566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684900921528796e-06,
+      "loss": 0.8052,
+      "step": 4567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6816788530416005e-06,
+      "loss": 0.8345,
+      "step": 4568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6784595854628516e-06,
+      "loss": 0.8086,
+      "step": 4569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6752431198765308e-06,
+      "loss": 0.8359,
+      "step": 4570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672029457365667e-06,
+      "loss": 0.8091,
+      "step": 4571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6688185990123574e-06,
+      "loss": 0.7754,
+      "step": 4572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6656105458977423e-06,
+      "loss": 0.8979,
+      "step": 4573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6624052991020245e-06,
+      "loss": 0.8638,
+      "step": 4574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.659202859704465e-06,
+      "loss": 0.8066,
+      "step": 4575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6560032287833682e-06,
+      "loss": 0.8149,
+      "step": 4576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6528064074161066e-06,
+      "loss": 0.771,
+      "step": 4577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6496123966790957e-06,
+      "loss": 0.7734,
+      "step": 4578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6464211976478074e-06,
+      "loss": 0.8164,
+      "step": 4579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6432328113967698e-06,
+      "loss": 0.7925,
+      "step": 4580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6400472389995626e-06,
+      "loss": 0.8096,
+      "step": 4581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6368644815288194e-06,
+      "loss": 0.7568,
+      "step": 4582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.633684540056224e-06,
+      "loss": 0.8154,
+      "step": 4583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.630507415652507e-06,
+      "loss": 0.8047,
+      "step": 4584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6273331093874634e-06,
+      "loss": 0.8242,
+      "step": 4585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.624161622329924e-06,
+      "loss": 0.8442,
+      "step": 4586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6209929555477832e-06,
+      "loss": 0.8184,
+      "step": 4587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.61782711010798e-06,
+      "loss": 0.7744,
+      "step": 4588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6146640870765017e-06,
+      "loss": 0.8062,
+      "step": 4589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.611503887518391e-06,
+      "loss": 0.8916,
+      "step": 4590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6083465124977315e-06,
+      "loss": 0.7822,
+      "step": 4591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6051919630776658e-06,
+      "loss": 0.8032,
+      "step": 4592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.602040240320375e-06,
+      "loss": 0.834,
+      "step": 4593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5988913452870958e-06,
+      "loss": 0.8208,
+      "step": 4594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5957452790381123e-06,
+      "loss": 0.7935,
+      "step": 4595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5926020426327503e-06,
+      "loss": 0.8525,
+      "step": 4596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5894616371293892e-06,
+      "loss": 0.7656,
+      "step": 4597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5863240635854526e-06,
+      "loss": 0.7993,
+      "step": 4598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5831893230574058e-06,
+      "loss": 0.7642,
+      "step": 4599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5800574166007687e-06,
+      "loss": 0.8286,
+      "step": 4600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5769283452701035e-06,
+      "loss": 0.8027,
+      "step": 4601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5738021101190171e-06,
+      "loss": 0.8481,
+      "step": 4602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.570678712200162e-06,
+      "loss": 0.8545,
+      "step": 4603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5675581525652317e-06,
+      "loss": 0.8276,
+      "step": 4604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5644404322649709e-06,
+      "loss": 0.7881,
+      "step": 4605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5613255523491622e-06,
+      "loss": 0.8384,
+      "step": 4606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5582135138666343e-06,
+      "loss": 0.3389,
+      "step": 4607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5551043178652624e-06,
+      "loss": 0.8379,
+      "step": 4608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5519979653919582e-06,
+      "loss": 0.7925,
+      "step": 4609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5488944574926801e-06,
+      "loss": 0.8223,
+      "step": 4610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5457937952124292e-06,
+      "loss": 0.8389,
+      "step": 4611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5426959795952412e-06,
+      "loss": 0.7759,
+      "step": 4612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5396010116842032e-06,
+      "loss": 0.8027,
+      "step": 4613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5365088925214389e-06,
+      "loss": 0.3069,
+      "step": 4614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5334196231481147e-06,
+      "loss": 0.8423,
+      "step": 4615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.530333204604434e-06,
+      "loss": 0.8403,
+      "step": 4616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5272496379296387e-06,
+      "loss": 0.8179,
+      "step": 4617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5241689241620206e-06,
+      "loss": 0.8057,
+      "step": 4618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5210910643388976e-06,
+      "loss": 0.728,
+      "step": 4619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5180160594966364e-06,
+      "loss": 0.8203,
+      "step": 4620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5149439106706397e-06,
+      "loss": 0.7754,
+      "step": 4621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.511874618895346e-06,
+      "loss": 0.7729,
+      "step": 4622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5088081852042357e-06,
+      "loss": 0.8223,
+      "step": 4623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5057446106298224e-06,
+      "loss": 0.8457,
+      "step": 4624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5026838962036628e-06,
+      "loss": 0.8579,
+      "step": 4625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4996260429563426e-06,
+      "loss": 0.7939,
+      "step": 4626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4965710519174914e-06,
+      "loss": 0.8384,
+      "step": 4627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4935189241157744e-06,
+      "loss": 0.7939,
+      "step": 4628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.490469660578886e-06,
+      "loss": 0.8291,
+      "step": 4629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.487423262333565e-06,
+      "loss": 0.7847,
+      "step": 4630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4843797304055796e-06,
+      "loss": 0.3542,
+      "step": 4631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4813390658197325e-06,
+      "loss": 0.7671,
+      "step": 4632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4783012695998634e-06,
+      "loss": 0.8188,
+      "step": 4633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4752663427688473e-06,
+      "loss": 0.8008,
+      "step": 4634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4722342863485938e-06,
+      "loss": 0.7729,
+      "step": 4635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4692051013600394e-06,
+      "loss": 0.8403,
+      "step": 4636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4661787888231582e-06,
+      "loss": 0.7969,
+      "step": 4637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4631553497569594e-06,
+      "loss": 0.7568,
+      "step": 4638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.460134785179479e-06,
+      "loss": 0.8008,
+      "step": 4639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4571170961077896e-06,
+      "loss": 0.7964,
+      "step": 4640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4541022835579953e-06,
+      "loss": 0.7651,
+      "step": 4641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4510903485452289e-06,
+      "loss": 0.8369,
+      "step": 4642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4480812920836573e-06,
+      "loss": 0.8398,
+      "step": 4643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4450751151864751e-06,
+      "loss": 0.8403,
+      "step": 4644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4420718188659112e-06,
+      "loss": 0.7642,
+      "step": 4645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4390714041332187e-06,
+      "loss": 0.7627,
+      "step": 4646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4360738719986856e-06,
+      "loss": 0.7832,
+      "step": 4647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4330792234716318e-06,
+      "loss": 0.8467,
+      "step": 4648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4300874595603953e-06,
+      "loss": 0.8589,
+      "step": 4649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.427098581272356e-06,
+      "loss": 0.8242,
+      "step": 4650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4241125896139118e-06,
+      "loss": 0.7734,
+      "step": 4651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4211294855904922e-06,
+      "loss": 0.7261,
+      "step": 4652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4181492702065568e-06,
+      "loss": 0.8462,
+      "step": 4653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4151719444655898e-06,
+      "loss": 0.834,
+      "step": 4654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4121975093701057e-06,
+      "loss": 0.8413,
+      "step": 4655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4092259659216411e-06,
+      "loss": 0.8159,
+      "step": 4656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4062573151207581e-06,
+      "loss": 0.7861,
+      "step": 4657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4032915579670536e-06,
+      "loss": 0.8062,
+      "step": 4658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4003286954591388e-06,
+      "loss": 0.8076,
+      "step": 4659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.397368728594658e-06,
+      "loss": 0.8145,
+      "step": 4660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3944116583702816e-06,
+      "loss": 0.8872,
+      "step": 4661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3914574857816943e-06,
+      "loss": 0.8262,
+      "step": 4662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3885062118236192e-06,
+      "loss": 0.7568,
+      "step": 4663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3855578374897927e-06,
+      "loss": 0.7876,
+      "step": 4664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.382612363772976e-06,
+      "loss": 0.8398,
+      "step": 4665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3796697916649582e-06,
+      "loss": 0.8423,
+      "step": 4666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.37673012215655e-06,
+      "loss": 0.8018,
+      "step": 4667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373793356237586e-06,
+      "loss": 0.8213,
+      "step": 4668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3708594948969188e-06,
+      "loss": 0.8281,
+      "step": 4669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3679285391224229e-06,
+      "loss": 0.7915,
+      "step": 4670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3650004899010006e-06,
+      "loss": 0.8037,
+      "step": 4671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.362075348218569e-06,
+      "loss": 0.8403,
+      "step": 4672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359153115060069e-06,
+      "loss": 0.7925,
+      "step": 4673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3562337914094658e-06,
+      "loss": 0.7529,
+      "step": 4674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3533173782497366e-06,
+      "loss": 0.8184,
+      "step": 4675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.350403876562888e-06,
+      "loss": 0.793,
+      "step": 4676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3474932873299351e-06,
+      "loss": 0.8022,
+      "step": 4677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3445856115309253e-06,
+      "loss": 0.8018,
+      "step": 4678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.341680850144913e-06,
+      "loss": 0.7539,
+      "step": 4679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3387790041499793e-06,
+      "loss": 0.7578,
+      "step": 4680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3358800745232225e-06,
+      "loss": 0.7949,
+      "step": 4681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332984062240753e-06,
+      "loss": 0.8047,
+      "step": 4682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3300909682777085e-06,
+      "loss": 0.792,
+      "step": 4683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3272007936082353e-06,
+      "loss": 0.8335,
+      "step": 4684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3243135392055006e-06,
+      "loss": 0.7764,
+      "step": 4685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3214292060416878e-06,
+      "loss": 0.8545,
+      "step": 4686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3185477950879966e-06,
+      "loss": 0.8364,
+      "step": 4687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3156693073146476e-06,
+      "loss": 0.8091,
+      "step": 4688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3127937436908689e-06,
+      "loss": 0.7964,
+      "step": 4689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3099211051849047e-06,
+      "loss": 0.791,
+      "step": 4690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3070513927640227e-06,
+      "loss": 0.7837,
+      "step": 4691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3041846073944953e-06,
+      "loss": 0.855,
+      "step": 4692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3013207500416158e-06,
+      "loss": 0.7812,
+      "step": 4693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.298459821669692e-06,
+      "loss": 0.8184,
+      "step": 4694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2956018232420387e-06,
+      "loss": 0.8135,
+      "step": 4695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2927467557209917e-06,
+      "loss": 0.7822,
+      "step": 4696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.289894620067894e-06,
+      "loss": 0.7964,
+      "step": 4697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.287045417243108e-06,
+      "loss": 0.79,
+      "step": 4698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2841991482060012e-06,
+      "loss": 0.7881,
+      "step": 4699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2813558139149574e-06,
+      "loss": 0.8076,
+      "step": 4700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2785154153273738e-06,
+      "loss": 0.8052,
+      "step": 4701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2756779533996544e-06,
+      "loss": 0.833,
+      "step": 4702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2728434290872194e-06,
+      "loss": 0.8188,
+      "step": 4703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2700118433444953e-06,
+      "loss": 0.8232,
+      "step": 4704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2671831971249205e-06,
+      "loss": 0.8042,
+      "step": 4705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2643574913809442e-06,
+      "loss": 0.8076,
+      "step": 4706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.26153472706403e-06,
+      "loss": 0.7969,
+      "step": 4707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2587149051246416e-06,
+      "loss": 0.7788,
+      "step": 4708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.25589802651226e-06,
+      "loss": 0.7939,
+      "step": 4709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2530840921753706e-06,
+      "loss": 0.8345,
+      "step": 4710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.25027310306147e-06,
+      "loss": 0.7725,
+      "step": 4711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2474650601170602e-06,
+      "loss": 0.8291,
+      "step": 4712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2446599642876545e-06,
+      "loss": 0.8086,
+      "step": 4713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.241857816517774e-06,
+      "loss": 0.7734,
+      "step": 4714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2390586177509411e-06,
+      "loss": 0.833,
+      "step": 4715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.236262368929696e-06,
+      "loss": 0.8203,
+      "step": 4716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2334690709955754e-06,
+      "loss": 0.8145,
+      "step": 4717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.230678724889124e-06,
+      "loss": 0.7803,
+      "step": 4718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2278913315498975e-06,
+      "loss": 0.8149,
+      "step": 4719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2251068919164543e-06,
+      "loss": 0.8203,
+      "step": 4720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2223254069263612e-06,
+      "loss": 0.8765,
+      "step": 4721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2195468775161855e-06,
+      "loss": 0.8159,
+      "step": 4722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2167713046215e-06,
+      "loss": 0.8667,
+      "step": 4723
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2139986891768852e-06,
+      "loss": 0.7979,
+      "step": 4724
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2112290321159226e-06,
+      "loss": 0.7949,
+      "step": 4725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2084623343711987e-06,
+      "loss": 0.8105,
+      "step": 4726
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2056985968743073e-06,
+      "loss": 0.8545,
+      "step": 4727
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2029378205558373e-06,
+      "loss": 0.811,
+      "step": 4728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2001800063453894e-06,
+      "loss": 0.8643,
+      "step": 4729
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1974251551715587e-06,
+      "loss": 0.8247,
+      "step": 4730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.19467326796195e-06,
+      "loss": 0.8105,
+      "step": 4731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1919243456431628e-06,
+      "loss": 0.8232,
+      "step": 4732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1891783891408038e-06,
+      "loss": 0.7842,
+      "step": 4733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.186435399379483e-06,
+      "loss": 0.7959,
+      "step": 4734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183695377282802e-06,
+      "loss": 0.7407,
+      "step": 4735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1809583237733736e-06,
+      "loss": 0.7686,
+      "step": 4736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.178224239772805e-06,
+      "loss": 0.7866,
+      "step": 4737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.175493126201701e-06,
+      "loss": 0.8237,
+      "step": 4738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.172764983979675e-06,
+      "loss": 0.8096,
+      "step": 4739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1700398140253322e-06,
+      "loss": 0.7739,
+      "step": 4740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1673176172562839e-06,
+      "loss": 0.8418,
+      "step": 4741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1645983945891338e-06,
+      "loss": 0.7871,
+      "step": 4742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1618821469394837e-06,
+      "loss": 0.792,
+      "step": 4743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159168875221941e-06,
+      "loss": 0.8584,
+      "step": 4744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1564585803501015e-06,
+      "loss": 0.8262,
+      "step": 4745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.15375126323657e-06,
+      "loss": 0.8516,
+      "step": 4746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1510469247929402e-06,
+      "loss": 0.7886,
+      "step": 4747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1483455659298004e-06,
+      "loss": 0.7524,
+      "step": 4748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1456471875567476e-06,
+      "loss": 0.7769,
+      "step": 4749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1429517905823618e-06,
+      "loss": 0.7524,
+      "step": 4750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1402593759142277e-06,
+      "loss": 0.8364,
+      "step": 4751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.137569944458925e-06,
+      "loss": 0.8203,
+      "step": 4752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1348834971220234e-06,
+      "loss": 0.3225,
+      "step": 4753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1322000348080964e-06,
+      "loss": 0.7861,
+      "step": 4754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.129519558420704e-06,
+      "loss": 0.8223,
+      "step": 4755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1268420688624037e-06,
+      "loss": 0.8218,
+      "step": 4756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1241675670347518e-06,
+      "loss": 0.7178,
+      "step": 4757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.121496053838288e-06,
+      "loss": 0.8306,
+      "step": 4758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1188275301725615e-06,
+      "loss": 0.8208,
+      "step": 4759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1161619969361014e-06,
+      "loss": 0.7944,
+      "step": 4760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.113499455026431e-06,
+      "loss": 0.7632,
+      "step": 4761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1108399053400766e-06,
+      "loss": 0.8115,
+      "step": 4762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1081833487725436e-06,
+      "loss": 0.7876,
+      "step": 4763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.105529786218339e-06,
+      "loss": 0.7739,
+      "step": 4764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1028792185709602e-06,
+      "loss": 0.8403,
+      "step": 4765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1002316467228913e-06,
+      "loss": 0.8271,
+      "step": 4766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0975870715656146e-06,
+      "loss": 0.8047,
+      "step": 4767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0949454939895954e-06,
+      "loss": 0.8271,
+      "step": 4768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092306914884299e-06,
+      "loss": 0.8223,
+      "step": 4769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0896713351381716e-06,
+      "loss": 0.8057,
+      "step": 4770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0870387556386563e-06,
+      "loss": 0.8315,
+      "step": 4771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0844091772721854e-06,
+      "loss": 0.8477,
+      "step": 4772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.081782600924175e-06,
+      "loss": 0.79,
+      "step": 4773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0791590274790375e-06,
+      "loss": 0.8066,
+      "step": 4774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.076538457820171e-06,
+      "loss": 0.7686,
+      "step": 4775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0739208928299582e-06,
+      "loss": 0.814,
+      "step": 4776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0713063333897766e-06,
+      "loss": 0.7534,
+      "step": 4777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0686947803799897e-06,
+      "loss": 0.8052,
+      "step": 4778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0660862346799494e-06,
+      "loss": 0.7896,
+      "step": 4779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0634806971679933e-06,
+      "loss": 0.7949,
+      "step": 4780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.060878168721442e-06,
+      "loss": 0.853,
+      "step": 4781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0582786502166132e-06,
+      "loss": 0.8433,
+      "step": 4782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0556821425288011e-06,
+      "loss": 0.8003,
+      "step": 4783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.053088646532292e-06,
+      "loss": 0.8062,
+      "step": 4784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0504981631003575e-06,
+      "loss": 0.7612,
+      "step": 4785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0479106931052518e-06,
+      "loss": 0.7842,
+      "step": 4786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0453262374182193e-06,
+      "loss": 0.8115,
+      "step": 4787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0427447969094828e-06,
+      "loss": 0.8477,
+      "step": 4788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0401663724482568e-06,
+      "loss": 0.7974,
+      "step": 4789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.037590964902734e-06,
+      "loss": 0.8057,
+      "step": 4790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0350185751400954e-06,
+      "loss": 0.7725,
+      "step": 4791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0324492040265077e-06,
+      "loss": 0.7793,
+      "step": 4792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0298828524271144e-06,
+      "loss": 0.7607,
+      "step": 4793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0273195212060483e-06,
+      "loss": 0.7676,
+      "step": 4794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0247592112264226e-06,
+      "loss": 0.8193,
+      "step": 4795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0222019233503312e-06,
+      "loss": 0.8521,
+      "step": 4796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0196476584388548e-06,
+      "loss": 0.8174,
+      "step": 4797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0170964173520536e-06,
+      "loss": 0.7964,
+      "step": 4798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0145482009489737e-06,
+      "loss": 0.772,
+      "step": 4799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0120030100876354e-06,
+      "loss": 0.7622,
+      "step": 4800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0094608456250442e-06,
+      "loss": 0.7661,
+      "step": 4801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006921708417189e-06,
+      "loss": 0.7876,
+      "step": 4802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0043855993190333e-06,
+      "loss": 0.8081,
+      "step": 4803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0018525191845273e-06,
+      "loss": 0.7944,
+      "step": 4804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.993224688666003e-07,
+      "loss": 0.8384,
+      "step": 4805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.967954492171539e-07,
+      "loss": 0.8154,
+      "step": 4806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.942714610870818e-07,
+      "loss": 0.7915,
+      "step": 4807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.917505053262467e-07,
+      "loss": 0.7695,
+      "step": 4808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.892325827834915e-07,
+      "loss": 0.8486,
+      "step": 4809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.867176943066425e-07,
+      "loss": 0.8184,
+      "step": 4810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.842058407425015e-07,
+      "loss": 0.7964,
+      "step": 4811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.816970229368505e-07,
+      "loss": 0.8896,
+      "step": 4812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.791912417344474e-07,
+      "loss": 0.8872,
+      "step": 4813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.766884979790238e-07,
+      "loss": 0.8013,
+      "step": 4814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.741887925132977e-07,
+      "loss": 0.8145,
+      "step": 4815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.716921261789537e-07,
+      "loss": 0.769,
+      "step": 4816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.691984998166614e-07,
+      "loss": 0.8081,
+      "step": 4817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.667079142660652e-07,
+      "loss": 0.8379,
+      "step": 4818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.642203703657804e-07,
+      "loss": 0.791,
+      "step": 4819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.617358689534052e-07,
+      "loss": 0.7964,
+      "step": 4820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.592544108655066e-07,
+      "loss": 0.813,
+      "step": 4821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.567759969376344e-07,
+      "loss": 0.7637,
+      "step": 4822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.543006280043043e-07,
+      "loss": 0.8545,
+      "step": 4823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.518283048990152e-07,
+      "loss": 0.8159,
+      "step": 4824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.493590284542364e-07,
+      "loss": 0.8096,
+      "step": 4825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.468927995014099e-07,
+      "loss": 0.8374,
+      "step": 4826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.444296188709567e-07,
+      "loss": 0.8052,
+      "step": 4827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.419694873922658e-07,
+      "loss": 0.8105,
+      "step": 4828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.39512405893701e-07,
+      "loss": 0.7808,
+      "step": 4829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.370583752026008e-07,
+      "loss": 0.3103,
+      "step": 4830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.346073961452762e-07,
+      "loss": 0.7964,
+      "step": 4831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.321594695470115e-07,
+      "loss": 0.3555,
+      "step": 4832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.297145962320608e-07,
+      "loss": 0.8096,
+      "step": 4833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.272727770236478e-07,
+      "loss": 0.8193,
+      "step": 4834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.248340127439759e-07,
+      "loss": 0.7969,
+      "step": 4835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.223983042142104e-07,
+      "loss": 0.8203,
+      "step": 4836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.199656522544953e-07,
+      "loss": 0.7881,
+      "step": 4837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.175360576839431e-07,
+      "loss": 0.8159,
+      "step": 4838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.15109521320634e-07,
+      "loss": 0.8794,
+      "step": 4839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.126860439816231e-07,
+      "loss": 0.8188,
+      "step": 4840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.102656264829301e-07,
+      "loss": 0.814,
+      "step": 4841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.078482696395497e-07,
+      "loss": 0.853,
+      "step": 4842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.054339742654416e-07,
+      "loss": 0.8213,
+      "step": 4843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.030227411735382e-07,
+      "loss": 0.7881,
+      "step": 4844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.006145711757408e-07,
+      "loss": 0.8228,
+      "step": 4845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.982094650829132e-07,
+      "loss": 0.792,
+      "step": 4846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.958074237048975e-07,
+      "loss": 0.7615,
+      "step": 4847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.934084478504945e-07,
+      "loss": 0.8271,
+      "step": 4848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.91012538327477e-07,
+      "loss": 0.7881,
+      "step": 4849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.886196959425875e-07,
+      "loss": 0.7656,
+      "step": 4850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.862299215015302e-07,
+      "loss": 0.75,
+      "step": 4851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.838432158089838e-07,
+      "loss": 0.7988,
+      "step": 4852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.814595796685866e-07,
+      "loss": 0.2887,
+      "step": 4853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.790790138829452e-07,
+      "loss": 0.7578,
+      "step": 4854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.767015192536355e-07,
+      "loss": 0.7212,
+      "step": 4855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.743270965811945e-07,
+      "loss": 0.8169,
+      "step": 4856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.719557466651274e-07,
+      "loss": 0.7915,
+      "step": 4857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.695874703039086e-07,
+      "loss": 0.8218,
+      "step": 4858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.672222682949694e-07,
+      "loss": 0.8198,
+      "step": 4859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.648601414347135e-07,
+      "loss": 0.8081,
+      "step": 4860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.625010905185038e-07,
+      "loss": 0.814,
+      "step": 4861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601451163406682e-07,
+      "loss": 0.8486,
+      "step": 4862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.577922196945021e-07,
+      "loss": 0.7822,
+      "step": 4863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.554424013722617e-07,
+      "loss": 0.3295,
+      "step": 4864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.530956621651698e-07,
+      "loss": 0.7397,
+      "step": 4865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.507520028634075e-07,
+      "loss": 0.813,
+      "step": 4866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.484114242561204e-07,
+      "loss": 0.7852,
+      "step": 4867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.460739271314223e-07,
+      "loss": 0.7983,
+      "step": 4868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.437395122763792e-07,
+      "loss": 0.8345,
+      "step": 4869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.41408180477028e-07,
+      "loss": 0.8174,
+      "step": 4870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.390799325183663e-07,
+      "loss": 0.7915,
+      "step": 4871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.36754769184348e-07,
+      "loss": 0.8721,
+      "step": 4872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.344326912578949e-07,
+      "loss": 0.7568,
+      "step": 4873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.321136995208834e-07,
+      "loss": 0.8555,
+      "step": 4874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.29797794754158e-07,
+      "loss": 0.7798,
+      "step": 4875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.27484977737516e-07,
+      "loss": 0.7891,
+      "step": 4876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.251752492497222e-07,
+      "loss": 0.8452,
+      "step": 4877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.228686100684979e-07,
+      "loss": 0.77,
+      "step": 4878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.205650609705218e-07,
+      "loss": 0.812,
+      "step": 4879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.182646027314378e-07,
+      "loss": 0.7676,
+      "step": 4880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.159672361258464e-07,
+      "loss": 0.8101,
+      "step": 4881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.13672961927302e-07,
+      "loss": 0.8335,
+      "step": 4882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.113817809083268e-07,
+      "loss": 0.7686,
+      "step": 4883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.090936938403959e-07,
+      "loss": 0.7632,
+      "step": 4884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.068087014939452e-07,
+      "loss": 0.7881,
+      "step": 4885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.045268046383669e-07,
+      "loss": 0.7705,
+      "step": 4886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.022480040420089e-07,
+      "loss": 0.7383,
+      "step": 4887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.999723004721815e-07,
+      "loss": 0.7705,
+      "step": 4888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.976996946951476e-07,
+      "loss": 0.8296,
+      "step": 4889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.954301874761295e-07,
+      "loss": 0.8271,
+      "step": 4890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.93163779579308e-07,
+      "loss": 0.7544,
+      "step": 4891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.909004717678137e-07,
+      "loss": 0.8271,
+      "step": 4892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.886402648037428e-07,
+      "loss": 0.7925,
+      "step": 4893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.863831594481375e-07,
+      "loss": 0.7793,
+      "step": 4894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.841291564610043e-07,
+      "loss": 0.7944,
+      "step": 4895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.818782566012983e-07,
+      "loss": 0.791,
+      "step": 4896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.796304606269323e-07,
+      "loss": 0.8369,
+      "step": 4897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.773857692947773e-07,
+      "loss": 0.7681,
+      "step": 4898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.751441833606533e-07,
+      "loss": 0.7754,
+      "step": 4899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.729057035793385e-07,
+      "loss": 0.7217,
+      "step": 4900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.706703307045637e-07,
+      "loss": 0.7983,
+      "step": 4901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.684380654890123e-07,
+      "loss": 0.7847,
+      "step": 4902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.662089086843238e-07,
+      "loss": 0.7871,
+      "step": 4903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.639828610410916e-07,
+      "loss": 0.7505,
+      "step": 4904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.617599233088591e-07,
+      "loss": 0.7754,
+      "step": 4905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.595400962361254e-07,
+      "loss": 0.8379,
+      "step": 4906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.573233805703383e-07,
+      "loss": 0.812,
+      "step": 4907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.551097770579041e-07,
+      "loss": 0.8008,
+      "step": 4908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.52899286444173e-07,
+      "loss": 0.7615,
+      "step": 4909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.506919094734544e-07,
+      "loss": 0.8047,
+      "step": 4910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.48487646889009e-07,
+      "loss": 0.835,
+      "step": 4911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46286499433041e-07,
+      "loss": 0.814,
+      "step": 4912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.440884678467164e-07,
+      "loss": 0.7612,
+      "step": 4913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.418935528701443e-07,
+      "loss": 0.8237,
+      "step": 4914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.397017552423846e-07,
+      "loss": 0.8433,
+      "step": 4915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.375130757014515e-07,
+      "loss": 0.3699,
+      "step": 4916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.353275149843087e-07,
+      "loss": 0.7393,
+      "step": 4917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.331450738268686e-07,
+      "loss": 0.7705,
+      "step": 4918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.309657529639935e-07,
+      "loss": 0.8345,
+      "step": 4919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.287895531294908e-07,
+      "loss": 0.7759,
+      "step": 4920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.266164750561266e-07,
+      "loss": 0.3304,
+      "step": 4921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.244465194756045e-07,
+      "loss": 0.8052,
+      "step": 4922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.222796871185855e-07,
+      "loss": 0.8228,
+      "step": 4923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.201159787146772e-07,
+      "loss": 0.8022,
+      "step": 4924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179553949924301e-07,
+      "loss": 0.8052,
+      "step": 4925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1579793667935e-07,
+      "loss": 0.7759,
+      "step": 4926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.136436045018846e-07,
+      "loss": 0.7334,
+      "step": 4927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.114923991854328e-07,
+      "loss": 0.7891,
+      "step": 4928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.093443214543372e-07,
+      "loss": 0.8013,
+      "step": 4929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.071993720318892e-07,
+      "loss": 0.8003,
+      "step": 4930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.050575516403291e-07,
+      "loss": 0.3132,
+      "step": 4931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.029188610008387e-07,
+      "loss": 0.7495,
+      "step": 4932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.007833008335507e-07,
+      "loss": 0.8188,
+      "step": 4933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.986508718575402e-07,
+      "loss": 0.8193,
+      "step": 4934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.965215747908272e-07,
+      "loss": 0.8545,
+      "step": 4935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.943954103503814e-07,
+      "loss": 0.8042,
+      "step": 4936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.922723792521147e-07,
+      "loss": 0.8047,
+      "step": 4937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.901524822108885e-07,
+      "loss": 0.8096,
+      "step": 4938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.880357199405008e-07,
+      "loss": 0.8184,
+      "step": 4939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.859220931536992e-07,
+      "loss": 0.7959,
+      "step": 4940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.838116025621766e-07,
+      "loss": 0.876,
+      "step": 4941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.817042488765657e-07,
+      "loss": 0.8516,
+      "step": 4942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.796000328064467e-07,
+      "loss": 0.8237,
+      "step": 4943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.774989550603428e-07,
+      "loss": 0.8311,
+      "step": 4944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.754010163457181e-07,
+      "loss": 0.7725,
+      "step": 4945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.733062173689819e-07,
+      "loss": 0.6987,
+      "step": 4946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.712145588354857e-07,
+      "loss": 0.7964,
+      "step": 4947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.691260414495238e-07,
+      "loss": 0.7969,
+      "step": 4948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.670406659143314e-07,
+      "loss": 0.8037,
+      "step": 4949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.649584329320879e-07,
+      "loss": 0.8423,
+      "step": 4950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.628793432039149e-07,
+      "loss": 0.8115,
+      "step": 4951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.60803397429871e-07,
+      "loss": 0.875,
+      "step": 4952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.587305963089641e-07,
+      "loss": 0.7827,
+      "step": 4953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.566609405391355e-07,
+      "loss": 0.7773,
+      "step": 4954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.545944308172681e-07,
+      "loss": 0.8125,
+      "step": 4955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.525310678391938e-07,
+      "loss": 0.3635,
+      "step": 4956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.504708522996761e-07,
+      "loss": 0.7681,
+      "step": 4957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.48413784892421e-07,
+      "loss": 0.3066,
+      "step": 4958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.463598663100767e-07,
+      "loss": 0.769,
+      "step": 4959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.443090972442279e-07,
+      "loss": 0.7783,
+      "step": 4960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.42261478385402e-07,
+      "loss": 0.769,
+      "step": 4961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.402170104230643e-07,
+      "loss": 0.8359,
+      "step": 4962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.381756940456186e-07,
+      "loss": 0.875,
+      "step": 4963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.361375299404082e-07,
+      "loss": 0.7607,
+      "step": 4964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.341025187937133e-07,
+      "loss": 0.8579,
+      "step": 4965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.320706612907556e-07,
+      "loss": 0.7979,
+      "step": 4966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.300419581156935e-07,
+      "loss": 0.8159,
+      "step": 4967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.280164099516195e-07,
+      "loss": 0.8433,
+      "step": 4968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.259940174805723e-07,
+      "loss": 0.7456,
+      "step": 4969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.239747813835196e-07,
+      "loss": 0.7783,
+      "step": 4970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.219587023403717e-07,
+      "loss": 0.7891,
+      "step": 4971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.199457810299725e-07,
+      "loss": 0.8188,
+      "step": 4972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.17936018130103e-07,
+      "loss": 0.8135,
+      "step": 4973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.159294143174832e-07,
+      "loss": 0.7944,
+      "step": 4974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.13925970267768e-07,
+      "loss": 0.855,
+      "step": 4975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.11925686655549e-07,
+      "loss": 0.835,
+      "step": 4976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.099285641543518e-07,
+      "loss": 0.8022,
+      "step": 4977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.079346034366373e-07,
+      "loss": 0.7705,
+      "step": 4978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.059438051738053e-07,
+      "loss": 0.855,
+      "step": 4979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.039561700361874e-07,
+      "loss": 0.8247,
+      "step": 4980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.019716986930513e-07,
+      "loss": 0.7964,
+      "step": 4981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.999903918126027e-07,
+      "loss": 0.8735,
+      "step": 4982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.980122500619734e-07,
+      "loss": 0.8564,
+      "step": 4983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.960372741072384e-07,
+      "loss": 0.7739,
+      "step": 4984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940654646134003e-07,
+      "loss": 0.8462,
+      "step": 4985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.920968222444002e-07,
+      "loss": 0.8086,
+      "step": 4986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.901313476631087e-07,
+      "loss": 0.8369,
+      "step": 4987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.881690415313335e-07,
+      "loss": 0.8452,
+      "step": 4988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.862099045098135e-07,
+      "loss": 0.7578,
+      "step": 4989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.84253937258219e-07,
+      "loss": 0.7979,
+      "step": 4990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.823011404351575e-07,
+      "loss": 0.7983,
+      "step": 4991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.803515146981642e-07,
+      "loss": 0.7852,
+      "step": 4992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.784050607037084e-07,
+      "loss": 0.8198,
+      "step": 4993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.764617791071913e-07,
+      "loss": 0.8276,
+      "step": 4994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.745216705629464e-07,
+      "loss": 0.7544,
+      "step": 4995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.725847357242398e-07,
+      "loss": 0.8042,
+      "step": 4996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.706509752432676e-07,
+      "loss": 0.7339,
+      "step": 4997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.687203897711546e-07,
+      "loss": 0.7769,
+      "step": 4998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667929799579619e-07,
+      "loss": 0.7876,
+      "step": 4999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648687464526759e-07,
+      "loss": 0.791,
+      "step": 5000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.629476899032171e-07,
+      "loss": 0.7676,
+      "step": 5001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.610298109564371e-07,
+      "loss": 0.8398,
+      "step": 5002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.591151102581127e-07,
+      "loss": 0.8047,
+      "step": 5003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.572035884529559e-07,
+      "loss": 0.7974,
+      "step": 5004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.55295246184604e-07,
+      "loss": 0.8506,
+      "step": 5005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533900840956275e-07,
+      "loss": 0.8242,
+      "step": 5006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.51488102827521e-07,
+      "loss": 0.8018,
+      "step": 5007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.495893030207133e-07,
+      "loss": 0.7959,
+      "step": 5008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.47693685314562e-07,
+      "loss": 0.7539,
+      "step": 5009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.458012503473486e-07,
+      "loss": 0.812,
+      "step": 5010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.439119987562835e-07,
+      "loss": 0.7183,
+      "step": 5011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.420259311775123e-07,
+      "loss": 0.8682,
+      "step": 5012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.40143048246099e-07,
+      "loss": 0.8218,
+      "step": 5013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.38263350596041e-07,
+      "loss": 0.8408,
+      "step": 5014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.363868388602644e-07,
+      "loss": 0.8403,
+      "step": 5015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.345135136706147e-07,
+      "loss": 0.7842,
+      "step": 5016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.326433756578752e-07,
+      "loss": 0.8354,
+      "step": 5017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.307764254517467e-07,
+      "loss": 0.7603,
+      "step": 5018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.289126636808628e-07,
+      "loss": 0.7837,
+      "step": 5019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.270520909727783e-07,
+      "loss": 0.8438,
+      "step": 5020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.251947079539787e-07,
+      "loss": 0.7559,
+      "step": 5021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.23340515249876e-07,
+      "loss": 0.748,
+      "step": 5022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.21489513484803e-07,
+      "loss": 0.7344,
+      "step": 5023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.196417032820222e-07,
+      "loss": 0.7979,
+      "step": 5024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.177970852637193e-07,
+      "loss": 0.7129,
+      "step": 5025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.159556600510051e-07,
+      "loss": 0.8359,
+      "step": 5026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.141174282639172e-07,
+      "loss": 0.8066,
+      "step": 5027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.122823905214169e-07,
+      "loss": 0.8228,
+      "step": 5028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.104505474413912e-07,
+      "loss": 0.8467,
+      "step": 5029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0862189964065e-07,
+      "loss": 0.8003,
+      "step": 5030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.067964477349252e-07,
+      "loss": 0.8247,
+      "step": 5031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.049741923388774e-07,
+      "loss": 0.7495,
+      "step": 5032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.031551340660856e-07,
+      "loss": 0.7783,
+      "step": 5033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.013392735290579e-07,
+      "loss": 0.8281,
+      "step": 5034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.995266113392227e-07,
+      "loss": 0.7563,
+      "step": 5035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.977171481069288e-07,
+      "loss": 0.7744,
+      "step": 5036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.959108844414551e-07,
+      "loss": 0.7842,
+      "step": 5037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.941078209509942e-07,
+      "loss": 0.8633,
+      "step": 5038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.923079582426693e-07,
+      "loss": 0.7539,
+      "step": 5039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.905112969225201e-07,
+      "loss": 0.8291,
+      "step": 5040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.887178375955104e-07,
+      "loss": 0.8955,
+      "step": 5041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.869275808655272e-07,
+      "loss": 0.8071,
+      "step": 5042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.85140527335377e-07,
+      "loss": 0.8071,
+      "step": 5043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833566776067899e-07,
+      "loss": 0.7744,
+      "step": 5044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.815760322804153e-07,
+      "loss": 0.8408,
+      "step": 5045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.797985919558224e-07,
+      "loss": 0.7896,
+      "step": 5046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.780243572315047e-07,
+      "loss": 0.7881,
+      "step": 5047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.762533287048743e-07,
+      "loss": 0.8218,
+      "step": 5048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7448550697226646e-07,
+      "loss": 0.7998,
+      "step": 5049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.727208926289306e-07,
+      "loss": 0.7422,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.709594862690414e-07,
+      "loss": 0.7974,
+      "step": 5051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6920128848569336e-07,
+      "loss": 0.8389,
+      "step": 5052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.674462998708962e-07,
+      "loss": 0.7651,
+      "step": 5053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.656945210155828e-07,
+      "loss": 0.7944,
+      "step": 5054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6394595250960684e-07,
+      "loss": 0.769,
+      "step": 5055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.622005949417363e-07,
+      "loss": 0.7729,
+      "step": 5056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.604584488996611e-07,
+      "loss": 0.7729,
+      "step": 5057
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5871951496998857e-07,
+      "loss": 0.8403,
+      "step": 5058
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5698379373824485e-07,
+      "loss": 0.7974,
+      "step": 5059
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5525128578887355e-07,
+      "loss": 0.7793,
+      "step": 5060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.53521991705238e-07,
+      "loss": 0.8745,
+      "step": 5061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5179591206962023e-07,
+      "loss": 0.8203,
+      "step": 5062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.500730474632164e-07,
+      "loss": 0.8188,
+      "step": 5063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4835339846614145e-07,
+      "loss": 0.7939,
+      "step": 5064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4663696565742873e-07,
+      "loss": 0.7974,
+      "step": 5065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4492374961502717e-07,
+      "loss": 0.835,
+      "step": 5066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432137509158041e-07,
+      "loss": 0.8496,
+      "step": 5067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4150697013554344e-07,
+      "loss": 0.7319,
+      "step": 5068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3980340784894214e-07,
+      "loss": 0.8013,
+      "step": 5069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3810306462962027e-07,
+      "loss": 0.7993,
+      "step": 5070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3640594105010647e-07,
+      "loss": 0.8506,
+      "step": 5071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3471203768185034e-07,
+      "loss": 0.7983,
+      "step": 5072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.330213550952145e-07,
+      "loss": 0.8086,
+      "step": 5073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.313338938594791e-07,
+      "loss": 0.8281,
+      "step": 5074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.296496545428397e-07,
+      "loss": 0.3531,
+      "step": 5075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.279686377124026e-07,
+      "loss": 0.8242,
+      "step": 5076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2629084393419616e-07,
+      "loss": 0.8579,
+      "step": 5077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.246162737731574e-07,
+      "loss": 0.8188,
+      "step": 5078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2294492779313966e-07,
+      "loss": 0.8257,
+      "step": 5079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.212768065569128e-07,
+      "loss": 0.812,
+      "step": 5080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.196119106261587e-07,
+      "loss": 0.8564,
+      "step": 5081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1795024056147435e-07,
+      "loss": 0.8281,
+      "step": 5082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.162917969223701e-07,
+      "loss": 0.7534,
+      "step": 5083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1463658026726806e-07,
+      "loss": 0.8267,
+      "step": 5084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1298459115350684e-07,
+      "loss": 0.8423,
+      "step": 5085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.11335830137336e-07,
+      "loss": 0.8271,
+      "step": 5086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.096902977739203e-07,
+      "loss": 0.3376,
+      "step": 5087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.080479946173366e-07,
+      "loss": 0.3672,
+      "step": 5088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.064089212205713e-07,
+      "loss": 0.8203,
+      "step": 5089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.047730781355286e-07,
+      "loss": 0.8872,
+      "step": 5090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0314046591302114e-07,
+      "loss": 0.3014,
+      "step": 5091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.015110851027759e-07,
+      "loss": 0.8252,
+      "step": 5092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.998849362534285e-07,
+      "loss": 0.8066,
+      "step": 5093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9826201991253087e-07,
+      "loss": 0.7729,
+      "step": 5094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9664233662654375e-07,
+      "loss": 0.812,
+      "step": 5095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9502588694083964e-07,
+      "loss": 0.8057,
+      "step": 5096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.934126713997033e-07,
+      "loss": 0.8354,
+      "step": 5097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.918026905463279e-07,
+      "loss": 0.8115,
+      "step": 5098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.901959449228188e-07,
+      "loss": 0.8774,
+      "step": 5099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8859243507019416e-07,
+      "loss": 0.7295,
+      "step": 5100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8699216152837894e-07,
+      "loss": 0.7603,
+      "step": 5101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.853951248362131e-07,
+      "loss": 0.7944,
+      "step": 5102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.83801325531441e-07,
+      "loss": 0.8018,
+      "step": 5103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8221076415072e-07,
+      "loss": 0.7612,
+      "step": 5104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8062344122961837e-07,
+      "loss": 0.8423,
+      "step": 5105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7903935730260965e-07,
+      "loss": 0.7754,
+      "step": 5106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7745851290308275e-07,
+      "loss": 0.7783,
+      "step": 5107
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7588090856333193e-07,
+      "loss": 0.8335,
+      "step": 5108
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7430654481455996e-07,
+      "loss": 0.8364,
+      "step": 5109
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7273542218688173e-07,
+      "loss": 0.7744,
+      "step": 5110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7116754120931744e-07,
+      "loss": 0.8477,
+      "step": 5111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6960290240979467e-07,
+      "loss": 0.7944,
+      "step": 5112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6804150631515546e-07,
+      "loss": 0.7778,
+      "step": 5113
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.664833534511436e-07,
+      "loss": 0.7827,
+      "step": 5114
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6492844434241724e-07,
+      "loss": 0.8149,
+      "step": 5115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6337677951253427e-07,
+      "loss": 0.7979,
+      "step": 5116
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6182835948396666e-07,
+      "loss": 0.7969,
+      "step": 5117
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6028318477809077e-07,
+      "loss": 0.8008,
+      "step": 5118
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.587412559151904e-07,
+      "loss": 0.8237,
+      "step": 5119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5720257341445906e-07,
+      "loss": 0.8071,
+      "step": 5120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5566713779399354e-07,
+      "loss": 0.8354,
+      "step": 5121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.541349495707991e-07,
+      "loss": 0.8018,
+      "step": 5122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.526060092607897e-07,
+      "loss": 0.7866,
+      "step": 5123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5108031737877913e-07,
+      "loss": 0.7622,
+      "step": 5124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4955787443849533e-07,
+      "loss": 0.8452,
+      "step": 5125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.48038680952566e-07,
+      "loss": 0.8315,
+      "step": 5126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465227374325286e-07,
+      "loss": 0.8936,
+      "step": 5127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.45010044388826e-07,
+      "loss": 0.7939,
+      "step": 5128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4350060233080293e-07,
+      "loss": 0.3298,
+      "step": 5129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4199441176671513e-07,
+      "loss": 0.7974,
+      "step": 5130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.404914732037179e-07,
+      "loss": 0.7827,
+      "step": 5131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.389917871478743e-07,
+      "loss": 0.7939,
+      "step": 5132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.374953541041526e-07,
+      "loss": 0.8555,
+      "step": 5133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3600217457642523e-07,
+      "loss": 0.8311,
+      "step": 5134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3451224906747126e-07,
+      "loss": 0.792,
+      "step": 5135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.330255780789682e-07,
+      "loss": 0.8066,
+      "step": 5136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.315421621115023e-07,
+      "loss": 0.3169,
+      "step": 5137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3006200166456504e-07,
+      "loss": 0.7549,
+      "step": 5138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2858509723654566e-07,
+      "loss": 0.8037,
+      "step": 5139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2711144932474294e-07,
+      "loss": 0.7905,
+      "step": 5140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2564105842535885e-07,
+      "loss": 0.8252,
+      "step": 5141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.241739250334952e-07,
+      "loss": 0.7925,
+      "step": 5142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.22710049643159e-07,
+      "loss": 0.8218,
+      "step": 5143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.212494327472582e-07,
+      "loss": 0.7969,
+      "step": 5144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.197920748376082e-07,
+      "loss": 0.8179,
+      "step": 5145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1833797640492214e-07,
+      "loss": 0.8354,
+      "step": 5146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168871379388172e-07,
+      "loss": 0.3052,
+      "step": 5147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.154395599278159e-07,
+      "loss": 0.8071,
+      "step": 5148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1399524285933846e-07,
+      "loss": 0.8159,
+      "step": 5149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.125541872197102e-07,
+      "loss": 0.8257,
+      "step": 5150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1111639349415525e-07,
+      "loss": 0.833,
+      "step": 5151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0968186216680184e-07,
+      "loss": 0.3348,
+      "step": 5152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0825059372067814e-07,
+      "loss": 0.8105,
+      "step": 5153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.068225886377163e-07,
+      "loss": 0.7725,
+      "step": 5154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0539784739874954e-07,
+      "loss": 0.8325,
+      "step": 5155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.039763704835064e-07,
+      "loss": 0.8291,
+      "step": 5156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.025581583706216e-07,
+      "loss": 0.8125,
+      "step": 5157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0114321153763114e-07,
+      "loss": 0.7266,
+      "step": 5158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.997315304609671e-07,
+      "loss": 0.749,
+      "step": 5159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.983231156159661e-07,
+      "loss": 0.7681,
+      "step": 5160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9691796747686295e-07,
+      "loss": 0.8267,
+      "step": 5161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.955160865167928e-07,
+      "loss": 0.3282,
+      "step": 5162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9411747320779225e-07,
+      "loss": 0.8027,
+      "step": 5163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.927221280207948e-07,
+      "loss": 0.7803,
+      "step": 5164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.913300514256334e-07,
+      "loss": 0.8203,
+      "step": 5165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.899412438910454e-07,
+      "loss": 0.8052,
+      "step": 5166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.885557058846611e-07,
+      "loss": 0.8511,
+      "step": 5167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.87173437873014e-07,
+      "loss": 0.8491,
+      "step": 5168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8579444032153623e-07,
+      "loss": 0.7759,
+      "step": 5169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.844187136945564e-07,
+      "loss": 0.8096,
+      "step": 5170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.830462584553029e-07,
+      "loss": 0.7769,
+      "step": 5171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.816770750659026e-07,
+      "loss": 0.8257,
+      "step": 5172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8031116398738343e-07,
+      "loss": 0.8037,
+      "step": 5173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.789485256796665e-07,
+      "loss": 0.7603,
+      "step": 5174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.775891606015735e-07,
+      "loss": 0.832,
+      "step": 5175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7623306921082507e-07,
+      "loss": 0.7827,
+      "step": 5176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.74880251964037e-07,
+      "loss": 0.7534,
+      "step": 5177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7353070931672496e-07,
+      "loss": 0.8384,
+      "step": 5178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7218444172330103e-07,
+      "loss": 0.8252,
+      "step": 5179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.708414496370726e-07,
+      "loss": 0.8213,
+      "step": 5180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.695017335102501e-07,
+      "loss": 0.7891,
+      "step": 5181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6816529379393274e-07,
+      "loss": 0.8716,
+      "step": 5182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.668321309381228e-07,
+      "loss": 0.7681,
+      "step": 5183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6550224539171555e-07,
+      "loss": 0.3179,
+      "step": 5184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.641756376025051e-07,
+      "loss": 0.7515,
+      "step": 5185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6285230801718186e-07,
+      "loss": 0.8345,
+      "step": 5186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.615322570813295e-07,
+      "loss": 0.8623,
+      "step": 5187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.602154852394323e-07,
+      "loss": 0.8325,
+      "step": 5188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.589019929348657e-07,
+      "loss": 0.8628,
+      "step": 5189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5759178060990265e-07,
+      "loss": 0.8398,
+      "step": 5190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.562848487057146e-07,
+      "loss": 0.7778,
+      "step": 5191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5498119766236285e-07,
+      "loss": 0.8228,
+      "step": 5192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5368082791881075e-07,
+      "loss": 0.8091,
+      "step": 5193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.523837399129103e-07,
+      "loss": 0.7554,
+      "step": 5194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5108993408141213e-07,
+      "loss": 0.8276,
+      "step": 5195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4979941085995994e-07,
+      "loss": 0.7729,
+      "step": 5196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.48512170683094e-07,
+      "loss": 0.7778,
+      "step": 5197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4722821398424747e-07,
+      "loss": 0.3503,
+      "step": 5198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4594754119575014e-07,
+      "loss": 0.8462,
+      "step": 5199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4467015274882266e-07,
+      "loss": 0.8672,
+      "step": 5200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4339604907358317e-07,
+      "loss": 0.8682,
+      "step": 5201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4212523059903956e-07,
+      "loss": 0.8047,
+      "step": 5202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4085769775310076e-07,
+      "loss": 0.8613,
+      "step": 5203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3959345096255973e-07,
+      "loss": 0.8423,
+      "step": 5204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3833249065311147e-07,
+      "loss": 0.832,
+      "step": 5205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.370748172493409e-07,
+      "loss": 0.8428,
+      "step": 5206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3582043117472474e-07,
+      "loss": 0.8052,
+      "step": 5207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.345693328516352e-07,
+      "loss": 0.8047,
+      "step": 5208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3332152270133633e-07,
+      "loss": 0.7656,
+      "step": 5209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3207700114398546e-07,
+      "loss": 0.8418,
+      "step": 5210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3083576859863177e-07,
+      "loss": 0.8652,
+      "step": 5211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2959782548321873e-07,
+      "loss": 0.8086,
+      "step": 5212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.283631722145796e-07,
+      "loss": 0.3107,
+      "step": 5213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2713180920844403e-07,
+      "loss": 0.8223,
+      "step": 5214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2590373687942812e-07,
+      "loss": 0.7734,
+      "step": 5215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2467895564104448e-07,
+      "loss": 0.7705,
+      "step": 5216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2345746590569539e-07,
+      "loss": 0.835,
+      "step": 5217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2223926808467632e-07,
+      "loss": 0.7856,
+      "step": 5218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2102436258817361e-07,
+      "loss": 0.833,
+      "step": 5219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1981274982526447e-07,
+      "loss": 0.7861,
+      "step": 5220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1860443020391698e-07,
+      "loss": 0.7793,
+      "step": 5221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1739940413099347e-07,
+      "loss": 0.8086,
+      "step": 5222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1619767201224163e-07,
+      "loss": 0.8408,
+      "step": 5223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.149992342523055e-07,
+      "loss": 0.7305,
+      "step": 5224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.138040912547179e-07,
+      "loss": 0.8667,
+      "step": 5225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1261224342190135e-07,
+      "loss": 0.7568,
+      "step": 5226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1142369115517148e-07,
+      "loss": 0.8242,
+      "step": 5227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1023843485472927e-07,
+      "loss": 0.8291,
+      "step": 5228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0905647491967218e-07,
+      "loss": 0.8018,
+      "step": 5229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.078778117479807e-07,
+      "loss": 0.8164,
+      "step": 5230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0670244573653297e-07,
+      "loss": 0.8623,
+      "step": 5231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0553037728109128e-07,
+      "loss": 0.751,
+      "step": 5232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0436160677630878e-07,
+      "loss": 0.2949,
+      "step": 5233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0319613461573072e-07,
+      "loss": 0.8325,
+      "step": 5234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0203396119178765e-07,
+      "loss": 0.8784,
+      "step": 5235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0087508689580314e-07,
+      "loss": 0.7627,
+      "step": 5236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9971951211798623e-07,
+      "loss": 0.8726,
+      "step": 5237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9856723724743788e-07,
+      "loss": 0.8008,
+      "step": 5238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.974182626721488e-07,
+      "loss": 0.8159,
+      "step": 5239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9627258877899403e-07,
+      "loss": 0.7764,
+      "step": 5240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.951302159537427e-07,
+      "loss": 0.7651,
+      "step": 5241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9399114458104718e-07,
+      "loss": 0.8701,
+      "step": 5242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9285537504445062e-07,
+      "loss": 0.8149,
+      "step": 5243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.917229077263838e-07,
+      "loss": 0.8472,
+      "step": 5244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9059374300816835e-07,
+      "loss": 0.7788,
+      "step": 5245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8946788127001126e-07,
+      "loss": 0.7642,
+      "step": 5246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8834532289100705e-07,
+      "loss": 0.7739,
+      "step": 5247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8722606824913668e-07,
+      "loss": 0.8013,
+      "step": 5248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.861101177212743e-07,
+      "loss": 0.8066,
+      "step": 5249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8499747168317595e-07,
+      "loss": 0.8174,
+      "step": 5250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.838881305094864e-07,
+      "loss": 0.7871,
+      "step": 5251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.827820945737391e-07,
+      "loss": 0.8335,
+      "step": 5252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8167936424835275e-07,
+      "loss": 0.835,
+      "step": 5253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8057993990463596e-07,
+      "loss": 0.8247,
+      "step": 5254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7948382191277925e-07,
+      "loss": 0.7725,
+      "step": 5255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7839101064186403e-07,
+      "loss": 0.8359,
+      "step": 5256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.77301506459856e-07,
+      "loss": 0.7944,
+      "step": 5257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7621530973360944e-07,
+      "loss": 0.8711,
+      "step": 5258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7513242082886406e-07,
+      "loss": 0.7549,
+      "step": 5259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.740528401102448e-07,
+      "loss": 0.7529,
+      "step": 5260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7297656794126205e-07,
+      "loss": 0.7383,
+      "step": 5261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7190360468431476e-07,
+      "loss": 0.8604,
+      "step": 5262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.708339507006862e-07,
+      "loss": 0.8369,
+      "step": 5263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.697676063505449e-07,
+      "loss": 0.7827,
+      "step": 5264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6870457199294698e-07,
+      "loss": 0.7725,
+      "step": 5265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.676448479858317e-07,
+      "loss": 0.3331,
+      "step": 5266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6658843468602582e-07,
+      "loss": 0.8208,
+      "step": 5267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6553533244923702e-07,
+      "loss": 0.855,
+      "step": 5268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6448554163006613e-07,
+      "loss": 0.8462,
+      "step": 5269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6343906258198927e-07,
+      "loss": 0.8179,
+      "step": 5270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6239589565737458e-07,
+      "loss": 0.8237,
+      "step": 5271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6135604120747328e-07,
+      "loss": 0.8369,
+      "step": 5272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6031949958241978e-07,
+      "loss": 0.3007,
+      "step": 5273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5928627113123374e-07,
+      "loss": 0.834,
+      "step": 5274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.582563562018191e-07,
+      "loss": 0.8101,
+      "step": 5275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5722975514096518e-07,
+      "loss": 0.8257,
+      "step": 5276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5620646829434206e-07,
+      "loss": 0.834,
+      "step": 5277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.551864960065097e-07,
+      "loss": 0.7964,
+      "step": 5278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5416983862090673e-07,
+      "loss": 0.8066,
+      "step": 5279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.531564964798593e-07,
+      "loss": 0.7778,
+      "step": 5280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5214646992457227e-07,
+      "loss": 0.8018,
+      "step": 5281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5113975929514025e-07,
+      "loss": 0.8203,
+      "step": 5282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5013636493053762e-07,
+      "loss": 0.7812,
+      "step": 5283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4913628716862195e-07,
+      "loss": 0.8325,
+      "step": 5284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4813952634613716e-07,
+      "loss": 0.8604,
+      "step": 5285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.471460827987048e-07,
+      "loss": 0.8032,
+      "step": 5286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4615595686083617e-07,
+      "loss": 0.7983,
+      "step": 5287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4516914886592014e-07,
+      "loss": 0.8428,
+      "step": 5288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4418565914623206e-07,
+      "loss": 0.8379,
+      "step": 5289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4320548803292588e-07,
+      "loss": 0.7607,
+      "step": 5290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4222863585604206e-07,
+      "loss": 0.7974,
+      "step": 5291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4125510294450307e-07,
+      "loss": 0.7622,
+      "step": 5292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4028488962610887e-07,
+      "loss": 0.8511,
+      "step": 5293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3931799622754928e-07,
+      "loss": 0.8203,
+      "step": 5294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3835442307439162e-07,
+      "loss": 0.8232,
+      "step": 5295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3739417049108416e-07,
+      "loss": 0.8193,
+      "step": 5296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.364372388009594e-07,
+      "loss": 0.8716,
+      "step": 5297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3548362832623175e-07,
+      "loss": 0.79,
+      "step": 5298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3453333938799773e-07,
+      "loss": 0.811,
+      "step": 5299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.335863723062336e-07,
+      "loss": 0.8301,
+      "step": 5300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3264272739979767e-07,
+      "loss": 0.8247,
+      "step": 5301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3170240498643017e-07,
+      "loss": 0.8164,
+      "step": 5302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3076540538275117e-07,
+      "loss": 0.8555,
+      "step": 5303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.29831728904265e-07,
+      "loss": 0.7837,
+      "step": 5304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2890137586535346e-07,
+      "loss": 0.7866,
+      "step": 5305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2797434657928153e-07,
+      "loss": 0.876,
+      "step": 5306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2705064135819622e-07,
+      "loss": 0.8179,
+      "step": 5307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2613026051312093e-07,
+      "loss": 0.7788,
+      "step": 5308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2521320435396223e-07,
+      "loss": 0.7998,
+      "step": 5309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2429947318950863e-07,
+      "loss": 0.8096,
+      "step": 5310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2338906732742628e-07,
+      "loss": 0.7759,
+      "step": 5311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.224819870742655e-07,
+      "loss": 0.8115,
+      "step": 5312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.21578232735452e-07,
+      "loss": 0.8311,
+      "step": 5313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2067780461529455e-07,
+      "loss": 0.8359,
+      "step": 5314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1978070301698175e-07,
+      "loss": 0.7632,
+      "step": 5315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1888692824258085e-07,
+      "loss": 0.8315,
+      "step": 5316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1799648059304003e-07,
+      "loss": 0.8208,
+      "step": 5317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1710936036818832e-07,
+      "loss": 0.7886,
+      "step": 5318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1622556786673123e-07,
+      "loss": 0.8389,
+      "step": 5319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1534510338625626e-07,
+      "loss": 0.8213,
+      "step": 5320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1446796722322962e-07,
+      "loss": 0.7559,
+      "step": 5321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1359415967299614e-07,
+      "loss": 0.8047,
+      "step": 5322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1272368102978048e-07,
+      "loss": 0.7861,
+      "step": 5323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1185653158668707e-07,
+      "loss": 0.8096,
+      "step": 5324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1099271163569791e-07,
+      "loss": 0.7808,
+      "step": 5325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1013222146767477e-07,
+      "loss": 0.7988,
+      "step": 5326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0927506137235922e-07,
+      "loss": 0.8574,
+      "step": 5327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0842123163836926e-07,
+      "loss": 0.833,
+      "step": 5328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0757073255320272e-07,
+      "loss": 0.8545,
+      "step": 5329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0672356440323606e-07,
+      "loss": 0.8418,
+      "step": 5330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0587972747372333e-07,
+      "loss": 0.7949,
+      "step": 5331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0503922204879946e-07,
+      "loss": 0.8433,
+      "step": 5332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0420204841147475e-07,
+      "loss": 0.7769,
+      "step": 5333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0336820684363813e-07,
+      "loss": 0.8433,
+      "step": 5334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0253769762605726e-07,
+      "loss": 0.8501,
+      "step": 5335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.017105210383773e-07,
+      "loss": 0.8384,
+      "step": 5336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0088667735912216e-07,
+      "loss": 0.8027,
+      "step": 5337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0006616686569215e-07,
+      "loss": 0.8135,
+      "step": 5338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.924898983436737e-08,
+      "loss": 0.8442,
+      "step": 5339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.843514654030217e-08,
+      "loss": 0.8403,
+      "step": 5340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.76246372575318e-08,
+      "loss": 0.7939,
+      "step": 5341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.681746225896682e-08,
+      "loss": 0.7856,
+      "step": 5342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.60136218163954e-08,
+      "loss": 0.7871,
+      "step": 5343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.521311620048213e-08,
+      "loss": 0.8091,
+      "step": 5344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.44159456807725e-08,
+      "loss": 0.7861,
+      "step": 5345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.362211052568404e-08,
+      "loss": 0.8091,
+      "step": 5346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.283161100251514e-08,
+      "loss": 0.7622,
+      "step": 5347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.204444737743735e-08,
+      "loss": 0.8511,
+      "step": 5348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.126061991550195e-08,
+      "loss": 0.8027,
+      "step": 5349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.048012888063562e-08,
+      "loss": 0.8154,
+      "step": 5350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.970297453564147e-08,
+      "loss": 0.7666,
+      "step": 5351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.89291571422013e-08,
+      "loss": 0.8599,
+      "step": 5352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.815867696087e-08,
+      "loss": 0.7759,
+      "step": 5353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.739153425107893e-08,
+      "loss": 0.7817,
+      "step": 5354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.662772927114038e-08,
+      "loss": 0.811,
+      "step": 5355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.586726227823638e-08,
+      "loss": 0.7891,
+      "step": 5356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.51101335284299e-08,
+      "loss": 0.8638,
+      "step": 5357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.435634327665698e-08,
+      "loss": 0.832,
+      "step": 5358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.360589177673128e-08,
+      "loss": 0.813,
+      "step": 5359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.285877928134068e-08,
+      "loss": 0.7671,
+      "step": 5360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.211500604205058e-08,
+      "loss": 0.7549,
+      "step": 5361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.137457230929957e-08,
+      "loss": 0.8208,
+      "step": 5362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.063747833240487e-08,
+      "loss": 0.8179,
+      "step": 5363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.990372435955685e-08,
+      "loss": 0.8188,
+      "step": 5364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.91733106378223e-08,
+      "loss": 0.7666,
+      "step": 5365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.844623741314228e-08,
+      "loss": 0.769,
+      "step": 5366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.772250493033317e-08,
+      "loss": 0.8354,
+      "step": 5367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.700211343308783e-08,
+      "loss": 0.8442,
+      "step": 5368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.628506316397221e-08,
+      "loss": 0.8379,
+      "step": 5369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.557135436443097e-08,
+      "loss": 0.8262,
+      "step": 5370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.486098727477964e-08,
+      "loss": 0.8574,
+      "step": 5371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.4153962134208e-08,
+      "loss": 0.8203,
+      "step": 5372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.345027918078562e-08,
+      "loss": 0.7881,
+      "step": 5373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.274993865145075e-08,
+      "loss": 0.8213,
+      "step": 5374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.20529407820203e-08,
+      "loss": 0.3201,
+      "step": 5375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.135928580718432e-08,
+      "loss": 0.3479,
+      "step": 5376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.06689739605071e-08,
+      "loss": 0.8125,
+      "step": 5377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.998200547442824e-08,
+      "loss": 0.8149,
+      "step": 5378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.929838058025828e-08,
+      "loss": 0.8184,
+      "step": 5379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.861809950818754e-08,
+      "loss": 0.7578,
+      "step": 5380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.7941162487275e-08,
+      "loss": 0.8853,
+      "step": 5381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.726756974545612e-08,
+      "loss": 0.834,
+      "step": 5382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.659732150954168e-08,
+      "loss": 0.7656,
+      "step": 5383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.593041800521227e-08,
+      "loss": 0.8115,
+      "step": 5384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.526685945702715e-08,
+      "loss": 0.8267,
+      "step": 5385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.460664608841538e-08,
+      "loss": 0.75,
+      "step": 5386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.394977812168135e-08,
+      "loss": 0.8477,
+      "step": 5387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.329625577800369e-08,
+      "loss": 0.8169,
+      "step": 5388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.264607927743083e-08,
+      "loss": 0.8418,
+      "step": 5389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.199924883889097e-08,
+      "loss": 0.8184,
+      "step": 5390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.13557646801799e-08,
+      "loss": 0.7651,
+      "step": 5391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.071562701796985e-08,
+      "loss": 0.8345,
+      "step": 5392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.007883606780396e-08,
+      "loss": 0.8506,
+      "step": 5393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9445392044099606e-08,
+      "loss": 0.3237,
+      "step": 5394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.881529516014728e-08,
+      "loss": 0.8037,
+      "step": 5395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.818854562811171e-08,
+      "loss": 0.8145,
+      "step": 5396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.756514365902854e-08,
+      "loss": 0.3503,
+      "step": 5397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6945089462807635e-08,
+      "loss": 0.8188,
+      "step": 5398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.632838324822865e-08,
+      "loss": 0.7856,
+      "step": 5399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5715025222948804e-08,
+      "loss": 0.8501,
+      "step": 5400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5105015593494014e-08,
+      "loss": 0.7896,
+      "step": 5401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.449835456526442e-08,
+      "loss": 0.835,
+      "step": 5402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3895042342531065e-08,
+      "loss": 0.8462,
+      "step": 5403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.329507912844034e-08,
+      "loss": 0.7803,
+      "step": 5404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2698465125008424e-08,
+      "loss": 0.8638,
+      "step": 5405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.210520053312573e-08,
+      "loss": 0.7939,
+      "step": 5406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1515285552551365e-08,
+      "loss": 0.8174,
+      "step": 5407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.092872038191976e-08,
+      "loss": 0.8022,
+      "step": 5408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0345505218737375e-08,
+      "loss": 0.8706,
+      "step": 5409
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.976564025938047e-08,
+      "loss": 0.7798,
+      "step": 5410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9189125699100616e-08,
+      "loss": 0.7793,
+      "step": 5411
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.86159617320181e-08,
+      "loss": 0.8408,
+      "step": 5412
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.804614855112633e-08,
+      "loss": 0.3148,
+      "step": 5413
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.747968634828959e-08,
+      "loss": 0.8345,
+      "step": 5414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6916575314245315e-08,
+      "loss": 0.7983,
+      "step": 5415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6356815638601836e-08,
+      "loss": 0.7471,
+      "step": 5416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.580040750983838e-08,
+      "loss": 0.8477,
+      "step": 5417
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.52473511153062e-08,
+      "loss": 0.7983,
+      "step": 5418
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4697646641229664e-08,
+      "loss": 0.812,
+      "step": 5419
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4151294272700706e-08,
+      "loss": 0.8457,
+      "step": 5420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.360829419368551e-08,
+      "loss": 0.811,
+      "step": 5421
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.306864658702004e-08,
+      "loss": 0.793,
+      "step": 5422
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.253235163441449e-08,
+      "loss": 0.7725,
+      "step": 5423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.199940951644443e-08,
+      "loss": 0.8442,
+      "step": 5424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1469820412561865e-08,
+      "loss": 0.7715,
+      "step": 5425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0943584501087485e-08,
+      "loss": 0.772,
+      "step": 5426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.042070195921177e-08,
+      "loss": 0.8403,
+      "step": 5427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9901172962999444e-08,
+      "loss": 0.832,
+      "step": 5428
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9384997687382795e-08,
+      "loss": 0.7944,
+      "step": 5429
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.887217630616613e-08,
+      "loss": 0.7935,
+      "step": 5430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8362708992024656e-08,
+      "loss": 0.7842,
+      "step": 5431
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.785659591650448e-08,
+      "loss": 0.8091,
+      "step": 5432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.73538372500204e-08,
+      "loss": 0.7783,
+      "step": 5433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.685443316186032e-08,
+      "loss": 0.8398,
+      "step": 5434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6358383820180865e-08,
+      "loss": 0.7891,
+      "step": 5435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5865689392009517e-08,
+      "loss": 0.8203,
+      "step": 5436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.537635004324469e-08,
+      "loss": 0.8262,
+      "step": 5437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.489036593865458e-08,
+      "loss": 0.7705,
+      "step": 5438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.440773724187607e-08,
+      "loss": 0.8091,
+      "step": 5439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3928464115420276e-08,
+      "loss": 0.855,
+      "step": 5440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.345254672066367e-08,
+      "loss": 0.791,
+      "step": 5441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2979985217855835e-08,
+      "loss": 0.7769,
+      "step": 5442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2510779766116166e-08,
+      "loss": 0.8579,
+      "step": 5443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.204493052343383e-08,
+      "loss": 0.793,
+      "step": 5444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.158243764666558e-08,
+      "loss": 0.7671,
+      "step": 5445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.112330129154128e-08,
+      "loss": 0.7603,
+      "step": 5446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.066752161265951e-08,
+      "loss": 0.8374,
+      "step": 5447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0215098763487495e-08,
+      "loss": 0.8628,
+      "step": 5448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9766032896364484e-08,
+      "loss": 0.8291,
+      "step": 5449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9320324162496195e-08,
+      "loss": 0.7866,
+      "step": 5450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8877972711961467e-08,
+      "loss": 0.8164,
+      "step": 5451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8438978693705598e-08,
+      "loss": 0.8286,
+      "step": 5452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8003342255545905e-08,
+      "loss": 0.8154,
+      "step": 5453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7571063544166165e-08,
+      "loss": 0.7954,
+      "step": 5454
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7142142705123275e-08,
+      "loss": 0.8271,
+      "step": 5455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.67165798828406e-08,
+      "loss": 0.3048,
+      "step": 5456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.62943752206124e-08,
+      "loss": 0.8184,
+      "step": 5457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5875528860600517e-08,
+      "loss": 0.8203,
+      "step": 5458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5460040943837693e-08,
+      "loss": 0.8115,
+      "step": 5459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.504791161022535e-08,
+      "loss": 0.833,
+      "step": 5460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4639140998532484e-08,
+      "loss": 0.7554,
+      "step": 5461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4233729246401216e-08,
+      "loss": 0.8262,
+      "step": 5462
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3831676490337907e-08,
+      "loss": 0.8052,
+      "step": 5463
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3432982865720932e-08,
+      "loss": 0.8374,
+      "step": 5464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.303764850679513e-08,
+      "loss": 0.7949,
+      "step": 5465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.264567354667846e-08,
+      "loss": 0.7637,
+      "step": 5466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2257058117352015e-08,
+      "loss": 0.8311,
+      "step": 5467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1871802349671124e-08,
+      "loss": 0.3767,
+      "step": 5468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1489906373356462e-08,
+      "loss": 0.814,
+      "step": 5469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1111370316997394e-08,
+      "loss": 0.8398,
+      "step": 5470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.073619430805418e-08,
+      "loss": 0.8003,
+      "step": 5471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.036437847285466e-08,
+      "loss": 0.7573,
+      "step": 5472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995922936593136e-08,
+      "loss": 0.8291,
+      "step": 5473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.963082782333703e-08,
+      "loss": 0.8394,
+      "step": 5474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.926909325601689e-08,
+      "loss": 0.3303,
+      "step": 5475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8910719356436404e-08,
+      "loss": 0.314,
+      "step": 5476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8555706245265702e-08,
+      "loss": 0.7837,
+      "step": 5477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8204054042041398e-08,
+      "loss": 0.3311,
+      "step": 5478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7855762865170988e-08,
+      "loss": 0.7632,
+      "step": 5479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7510832831930667e-08,
+      "loss": 0.7793,
+      "step": 5480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7169264058461977e-08,
+      "loss": 0.3049,
+      "step": 5481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6831056659778465e-08,
+      "loss": 0.8232,
+      "step": 5482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6496210749757935e-08,
+      "loss": 0.8013,
+      "step": 5483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6164726441150192e-08,
+      "loss": 0.7817,
+      "step": 5484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5836603845569288e-08,
+      "loss": 0.8247,
+      "step": 5485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5511843073500175e-08,
+      "loss": 0.7979,
+      "step": 5486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5190444234294278e-08,
+      "loss": 0.7778,
+      "step": 5487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4872407436171688e-08,
+      "loss": 0.8071,
+      "step": 5488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.45577327862223e-08,
+      "loss": 0.7817,
+      "step": 5489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.424642039039914e-08,
+      "loss": 0.7935,
+      "step": 5490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3938470353527244e-08,
+      "loss": 0.8335,
+      "step": 5491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3633882779298113e-08,
+      "loss": 0.8218,
+      "step": 5492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3332657770270819e-08,
+      "loss": 0.8843,
+      "step": 5493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3034795427873115e-08,
+      "loss": 0.8008,
+      "step": 5494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.274029585240033e-08,
+      "loss": 0.7446,
+      "step": 5495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2449159143013146e-08,
+      "loss": 0.7812,
+      "step": 5496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2161385397744251e-08,
+      "loss": 0.7729,
+      "step": 5497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1876974713489476e-08,
+      "loss": 0.3347,
+      "step": 5498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1595927186015542e-08,
+      "loss": 0.7988,
+      "step": 5499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1318242909955645e-08,
+      "loss": 0.79,
+      "step": 5500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1043921978808325e-08,
+      "loss": 0.3212,
+      "step": 5501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0772964484945247e-08,
+      "loss": 0.8281,
+      "step": 5502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.050537051959899e-08,
+      "loss": 0.7656,
+      "step": 5503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0241140172874143e-08,
+      "loss": 0.8501,
+      "step": 5504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.980273533740648e-09,
+      "loss": 0.7817,
+      "step": 5505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.72277069003602e-09,
+      "loss": 0.8169,
+      "step": 5506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.46863172846646e-09,
+      "loss": 0.8242,
+      "step": 5507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.217856734603514e-09,
+      "loss": 0.7944,
+      "step": 5508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.970445792888527e-09,
+      "loss": 0.8555,
+      "step": 5509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.726398986627083e-09,
+      "loss": 0.2982,
+      "step": 5510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.485716397994558e-09,
+      "loss": 0.8848,
+      "step": 5511
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.248398108031686e-09,
+      "loss": 0.7683,
+      "step": 5512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.014444196647875e-09,
+      "loss": 0.8154,
+      "step": 5513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.78385474261789e-09,
+      "loss": 0.8242,
+      "step": 5514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.556629823586293e-09,
+      "loss": 0.8052,
+      "step": 5515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.332769516062988e-09,
+      "loss": 0.7681,
+      "step": 5516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.1122738954243486e-09,
+      "loss": 0.7778,
+      "step": 5517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.89514303591543e-09,
+      "loss": 0.8359,
+      "step": 5518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.681377010647749e-09,
+      "loss": 0.7827,
+      "step": 5519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.470975891599285e-09,
+      "loss": 0.8398,
+      "step": 5520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.263939749615588e-09,
+      "loss": 0.8052,
+      "step": 5521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.060268654408674e-09,
+      "loss": 0.8174,
+      "step": 5522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8599626745570184e-09,
+      "loss": 0.77,
+      "step": 5523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.66302187750889e-09,
+      "loss": 0.8433,
+      "step": 5524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.46944632957569e-09,
+      "loss": 0.8296,
+      "step": 5525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2792360959375014e-09,
+      "loss": 0.8232,
+      "step": 5526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.092391240640871e-09,
+      "loss": 0.8086,
+      "step": 5527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.908911826599916e-09,
+      "loss": 0.7588,
+      "step": 5528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.728797915595218e-09,
+      "loss": 0.812,
+      "step": 5529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.552049568273819e-09,
+      "loss": 0.8267,
+      "step": 5530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.378666844149226e-09,
+      "loss": 0.7778,
+      "step": 5531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.208649801602516e-09,
+      "loss": 0.8184,
+      "step": 5532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0419984978801175e-09,
+      "loss": 0.8096,
+      "step": 5533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.878712989097144e-09,
+      "loss": 0.8384,
+      "step": 5534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.718793330235171e-09,
+      "loss": 0.7871,
+      "step": 5535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5622395751389036e-09,
+      "loss": 0.8472,
+      "step": 5536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.409051776525063e-09,
+      "loss": 0.7925,
+      "step": 5537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2592299859735e-09,
+      "loss": 0.7759,
+      "step": 5538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1127742539316386e-09,
+      "loss": 0.8276,
+      "step": 5539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.969684629712255e-09,
+      "loss": 0.7988,
+      "step": 5540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8299611614979184e-09,
+      "loss": 0.8564,
+      "step": 5541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.693603896333219e-09,
+      "loss": 0.8452,
+      "step": 5542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.560612880134761e-09,
+      "loss": 0.769,
+      "step": 5543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.430988157680059e-09,
+      "loss": 0.8213,
+      "step": 5544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.304729772617531e-09,
+      "loss": 0.3372,
+      "step": 5545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1818377674587276e-09,
+      "loss": 0.8184,
+      "step": 5546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.062312183584991e-09,
+      "loss": 0.7866,
+      "step": 5547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9461530612419066e-09,
+      "loss": 0.7886,
+      "step": 5548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8333604395404102e-09,
+      "loss": 0.7959,
+      "step": 5549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7239343564623423e-09,
+      "loss": 0.7964,
+      "step": 5550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.617874848851564e-09,
+      "loss": 0.8467,
+      "step": 5551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5151819524195088e-09,
+      "loss": 0.7866,
+      "step": 5552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4158557017451836e-09,
+      "loss": 0.8042,
+      "step": 5553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3198961302740564e-09,
+      "loss": 0.8398,
+      "step": 5554
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2273032703158382e-09,
+      "loss": 0.8228,
+      "step": 5555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1380771530489222e-09,
+      "loss": 0.8174,
+      "step": 5556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.052217808515943e-09,
+      "loss": 0.3264,
+      "step": 5557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.697252656293288e-10,
+      "loss": 0.8306,
+      "step": 5558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.905995521624189e-10,
+      "loss": 0.7368,
+      "step": 5559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.148406947616761e-10,
+      "loss": 0.7529,
+      "step": 5560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.42448718933364e-10,
+      "loss": 0.7837,
+      "step": 5561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.734236490535395e-10,
+      "loss": 0.3121,
+      "step": 5562
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.077655083658318e-10,
+      "loss": 0.8242,
+      "step": 5563
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.454743189758916e-10,
+      "loss": 0.2959,
+      "step": 5564
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.865501018602725e-10,
+      "loss": 0.8574,
+      "step": 5565
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.309928768586602e-10,
+      "loss": 0.7466,
+      "step": 5566
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7880266267720235e-10,
+      "loss": 0.8667,
+      "step": 5567
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2997947689183963e-10,
+      "loss": 0.7842,
+      "step": 5568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.845233359383137e-10,
+      "loss": 0.8584,
+      "step": 5569
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.424342551254899e-10,
+      "loss": 0.8418,
+      "step": 5570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0371224862425487e-10,
+      "loss": 0.7495,
+      "step": 5571
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6835732947306783e-10,
+      "loss": 0.7163,
+      "step": 5572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3636950957574002e-10,
+      "loss": 0.8184,
+      "step": 5573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0774879970476548e-10,
+      "loss": 0.7949,
+      "step": 5574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.249520949576984e-11,
+      "loss": 0.8599,
+      "step": 5575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.060874745195123e-11,
+      "loss": 0.7773,
+      "step": 5576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2089420944080305e-11,
+      "loss": 0.8511,
+      "step": 5577
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.693723620716959e-11,
+      "loss": 0.7944,
+      "step": 5578
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.515219834269388e-11,
+      "loss": 0.8267,
+      "step": 5579
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.734311319700481e-12,
+      "loss": 0.8486,
+      "step": 5580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.683577971478556e-12,
+      "loss": 0.7998,
+      "step": 5581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3887,
+      "step": 5582
+    },
+    {
+      "epoch": 1.0,
+      "step": 5582,
+      "total_flos": 2.2874903160750604e+19,
+      "train_loss": 0.8285522898308738,
+      "train_runtime": 107499.3725,
+      "train_samples_per_second": 6.646,
+      "train_steps_per_second": 0.052
+    }
+  ],
+  "max_steps": 5582,
+  "num_train_epochs": 1,
+  "total_flos": 2.2874903160750604e+19,
+  "trial_name": null,
+  "trial_params": null
+}