diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,55969 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.999396900066341,
+  "global_step": 9324,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857144e-08,
+      "loss": 0.8121,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714287e-07,
+      "loss": 0.7395,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.142857142857143e-07,
+      "loss": 0.7592,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 0.7375,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 0.7514,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.285714285714286e-07,
+      "loss": 0.7513,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.7281,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 0.7293,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.428571428571428e-07,
+      "loss": 0.8903,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 0.7312,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.857142857142857e-07,
+      "loss": 0.7163,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 0.7149,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.285714285714287e-07,
+      "loss": 0.7133,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6977,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 0.6685,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 0.6675,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2142857142857144e-06,
+      "loss": 0.6095,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2857142857142856e-06,
+      "loss": 0.5569,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3571428571428572e-06,
+      "loss": 0.6008,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.5878,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 0.503,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5714285714285714e-06,
+      "loss": 0.475,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.642857142857143e-06,
+      "loss": 0.4418,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 0.3964,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 0.4181,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8571428571428573e-06,
+      "loss": 0.4335,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.928571428571429e-06,
+      "loss": 0.3693,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.3931,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0714285714285717e-06,
+      "loss": 0.3669,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.3427,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2142857142857146e-06,
+      "loss": 0.3502,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.3406,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3571428571428574e-06,
+      "loss": 0.3514,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.428571428571429e-06,
+      "loss": 0.3166,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 0.3152,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 0.3146,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.642857142857143e-06,
+      "loss": 0.3663,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7142857142857144e-06,
+      "loss": 0.3124,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.785714285714286e-06,
+      "loss": 0.3286,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.3195,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.928571428571429e-06,
+      "loss": 0.3204,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 0.2832,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.071428571428572e-06,
+      "loss": 0.3401,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.142857142857143e-06,
+      "loss": 0.2743,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 0.3032,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.285714285714286e-06,
+      "loss": 0.2723,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.357142857142857e-06,
+      "loss": 0.288,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 0.2849,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 0.2688,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.2745,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.642857142857143e-06,
+      "loss": 0.2929,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 0.2916,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.785714285714286e-06,
+      "loss": 0.3044,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.857142857142858e-06,
+      "loss": 0.2995,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 0.2793,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.2757,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.071428571428572e-06,
+      "loss": 0.2613,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1428571428571435e-06,
+      "loss": 0.2793,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2142857142857145e-06,
+      "loss": 0.2984,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.2756,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.357142857142857e-06,
+      "loss": 0.2526,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.428571428571429e-06,
+      "loss": 0.2586,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-06,
+      "loss": 0.2546,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 0.2922,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 0.2592,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.714285714285715e-06,
+      "loss": 0.3144,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.785714285714287e-06,
+      "loss": 0.2447,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 0.2493,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.928571428571429e-06,
+      "loss": 0.2864,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.071428571428571e-06,
+      "loss": 0.2459,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 0.2248,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.214285714285715e-06,
+      "loss": 0.2372,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.285714285714286e-06,
+      "loss": 0.2564,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.2386,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 0.2542,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.2339,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.571428571428572e-06,
+      "loss": 0.2386,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6428571428571435e-06,
+      "loss": 0.2462,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.2359,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.785714285714286e-06,
+      "loss": 0.2509,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.857142857142858e-06,
+      "loss": 0.2248,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.928571428571429e-06,
+      "loss": 0.2542,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6e-06,
+      "loss": 0.2571,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.071428571428571e-06,
+      "loss": 0.2713,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.142857142857144e-06,
+      "loss": 0.2462,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.214285714285715e-06,
+      "loss": 0.2582,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 0.2509,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.357142857142858e-06,
+      "loss": 0.2467,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.2707,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.2297,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.571428571428572e-06,
+      "loss": 0.2437,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.642857142857143e-06,
+      "loss": 0.2325,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.714285714285714e-06,
+      "loss": 0.2501,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.785714285714287e-06,
+      "loss": 0.2327,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.2219,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.928571428571429e-06,
+      "loss": 0.2341,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-06,
+      "loss": 0.236,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.0714285714285726e-06,
+      "loss": 0.2568,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.2524,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2142857142857145e-06,
+      "loss": 0.2298,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.285714285714286e-06,
+      "loss": 0.2221,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.357142857142858e-06,
+      "loss": 0.2371,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.2387,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.2645,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.571428571428572e-06,
+      "loss": 0.2156,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.642857142857143e-06,
+      "loss": 0.2253,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.714285714285716e-06,
+      "loss": 0.2256,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.785714285714287e-06,
+      "loss": 0.2488,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.2682,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.928571428571429e-06,
+      "loss": 0.2575,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.239,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.071428571428572e-06,
+      "loss": 0.2383,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.142857142857143e-06,
+      "loss": 0.2462,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.2534,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.285714285714287e-06,
+      "loss": 0.2395,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.357142857142858e-06,
+      "loss": 0.2285,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.428571428571429e-06,
+      "loss": 0.2476,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.5e-06,
+      "loss": 0.2563,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.2237,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.642857142857144e-06,
+      "loss": 0.2459,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.714285714285715e-06,
+      "loss": 0.2192,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.785714285714286e-06,
+      "loss": 0.2344,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.857142857142858e-06,
+      "loss": 0.251,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.2118,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9e-06,
+      "loss": 0.2413,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.071428571428573e-06,
+      "loss": 0.2397,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.2246,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.214285714285715e-06,
+      "loss": 0.2447,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.2312,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.357142857142859e-06,
+      "loss": 0.2028,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 0.2431,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.5e-06,
+      "loss": 0.236,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.571428571428573e-06,
+      "loss": 0.2298,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 0.2352,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.2507,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.785714285714286e-06,
+      "loss": 0.2299,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.857142857142859e-06,
+      "loss": 0.2256,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.92857142857143e-06,
+      "loss": 0.2297,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.2513,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0071428571428572e-05,
+      "loss": 0.2041,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0142857142857143e-05,
+      "loss": 0.2145,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0214285714285714e-05,
+      "loss": 0.2278,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 0.2136,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0357142857142859e-05,
+      "loss": 0.2212,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.042857142857143e-05,
+      "loss": 0.2095,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.2159,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0571428571428572e-05,
+      "loss": 0.2257,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0642857142857143e-05,
+      "loss": 0.2115,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.214,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0785714285714287e-05,
+      "loss": 0.2428,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.2309,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0928571428571429e-05,
+      "loss": 0.2265,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.2151,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1071428571428572e-05,
+      "loss": 0.2039,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 0.2261,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1214285714285716e-05,
+      "loss": 0.2236,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1285714285714287e-05,
+      "loss": 0.2197,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1357142857142858e-05,
+      "loss": 0.234,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.2417,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.15e-05,
+      "loss": 0.2339,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1571428571428573e-05,
+      "loss": 0.2287,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1642857142857145e-05,
+      "loss": 0.2292,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1714285714285716e-05,
+      "loss": 0.2088,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1785714285714287e-05,
+      "loss": 0.2294,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1857142857142858e-05,
+      "loss": 0.2149,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.192857142857143e-05,
+      "loss": 0.2307,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.2167,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2071428571428571e-05,
+      "loss": 0.2313,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.208,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2214285714285717e-05,
+      "loss": 0.2296,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2285714285714288e-05,
+      "loss": 0.198,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2357142857142859e-05,
+      "loss": 0.2332,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.242857142857143e-05,
+      "loss": 0.2092,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2469,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 0.2251,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2642857142857143e-05,
+      "loss": 0.2194,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2714285714285715e-05,
+      "loss": 0.2167,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2785714285714286e-05,
+      "loss": 0.2108,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.2068,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.292857142857143e-05,
+      "loss": 0.2598,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.2145,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3071428571428572e-05,
+      "loss": 0.2333,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 0.2233,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3214285714285716e-05,
+      "loss": 0.2341,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3285714285714287e-05,
+      "loss": 0.214,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3357142857142858e-05,
+      "loss": 0.2071,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3428571428571429e-05,
+      "loss": 0.2211,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.2081,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.2217,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3642857142857145e-05,
+      "loss": 0.23,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.223,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3785714285714287e-05,
+      "loss": 0.2079,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3857142857142858e-05,
+      "loss": 0.209,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3928571428571429e-05,
+      "loss": 0.2102,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4e-05,
+      "loss": 0.2258,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4071428571428574e-05,
+      "loss": 0.2201,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4142857142857145e-05,
+      "loss": 0.2293,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4214285714285716e-05,
+      "loss": 0.2173,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.2442,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4357142857142858e-05,
+      "loss": 0.2191,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4428571428571429e-05,
+      "loss": 0.2181,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.45e-05,
+      "loss": 0.217,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4571428571428573e-05,
+      "loss": 0.2159,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4642857142857144e-05,
+      "loss": 0.1943,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4714285714285716e-05,
+      "loss": 0.2273,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4785714285714287e-05,
+      "loss": 0.218,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.2173,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.492857142857143e-05,
+      "loss": 0.2357,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2286,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5071428571428573e-05,
+      "loss": 0.1963,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5142857142857144e-05,
+      "loss": 0.2165,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5214285714285715e-05,
+      "loss": 0.2302,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5285714285714286e-05,
+      "loss": 0.2037,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.535714285714286e-05,
+      "loss": 0.1957,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 0.22,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.55e-05,
+      "loss": 0.2032,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5571428571428573e-05,
+      "loss": 0.2211,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5642857142857143e-05,
+      "loss": 0.2052,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.207,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5785714285714288e-05,
+      "loss": 0.2027,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5857142857142857e-05,
+      "loss": 0.2044,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.592857142857143e-05,
+      "loss": 0.2256,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.2089,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.2209,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6142857142857145e-05,
+      "loss": 0.2207,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6214285714285717e-05,
+      "loss": 0.2027,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6285714285714287e-05,
+      "loss": 0.2357,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.635714285714286e-05,
+      "loss": 0.2208,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.2233,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.65e-05,
+      "loss": 0.2117,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 0.1904,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6642857142857147e-05,
+      "loss": 0.2232,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6714285714285716e-05,
+      "loss": 0.2121,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.678571428571429e-05,
+      "loss": 0.2195,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6857142857142858e-05,
+      "loss": 0.1985,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.692857142857143e-05,
+      "loss": 0.2141,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7e-05,
+      "loss": 0.2196,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7071428571428573e-05,
+      "loss": 0.2016,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.2012,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7214285714285718e-05,
+      "loss": 0.2223,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7285714285714287e-05,
+      "loss": 0.2349,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.735714285714286e-05,
+      "loss": 0.2088,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.742857142857143e-05,
+      "loss": 0.2099,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.2029,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.757142857142857e-05,
+      "loss": 0.2091,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7642857142857144e-05,
+      "loss": 0.2114,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 0.2238,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7785714285714286e-05,
+      "loss": 0.2225,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.2381,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.792857142857143e-05,
+      "loss": 0.2222,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8e-05,
+      "loss": 0.2474,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8071428571428573e-05,
+      "loss": 0.2092,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8142857142857146e-05,
+      "loss": 0.2201,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8214285714285715e-05,
+      "loss": 0.2094,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.2312,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8357142857142857e-05,
+      "loss": 0.2157,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842857142857143e-05,
+      "loss": 0.253,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 0.2185,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.211,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8642857142857144e-05,
+      "loss": 0.1989,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8714285714285717e-05,
+      "loss": 0.2054,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8785714285714286e-05,
+      "loss": 0.232,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 0.2102,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.892857142857143e-05,
+      "loss": 0.203,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9e-05,
+      "loss": 0.2074,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9071428571428574e-05,
+      "loss": 0.212,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9142857142857146e-05,
+      "loss": 0.21,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9214285714285716e-05,
+      "loss": 0.2236,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.1964,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9357142857142858e-05,
+      "loss": 0.1871,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.2242,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95e-05,
+      "loss": 0.2289,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9571428571428572e-05,
+      "loss": 0.1956,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.217,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9714285714285718e-05,
+      "loss": 0.2076,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785714285714287e-05,
+      "loss": 0.216,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985714285714286e-05,
+      "loss": 0.2099,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.992857142857143e-05,
+      "loss": 0.2001,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2116,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999396678728e-05,
+      "loss": 0.2415,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997586714986e-05,
+      "loss": 0.206,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999994570108988e-05,
+      "loss": 0.2233,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999990346861104e-05,
+      "loss": 0.202,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999984916971842e-05,
+      "loss": 0.2113,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999978280441853e-05,
+      "loss": 0.2114,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999970437271943e-05,
+      "loss": 0.2311,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999961387463053e-05,
+      "loss": 0.2145,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999951131016282e-05,
+      "loss": 0.2268,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999939667932863e-05,
+      "loss": 0.2306,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999926998214175e-05,
+      "loss": 0.2136,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999991312186176e-05,
+      "loss": 0.1984,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999898038877277e-05,
+      "loss": 0.1812,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999881749262556e-05,
+      "loss": 0.2118,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999986425301956e-05,
+      "loss": 0.216,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999845550150397e-05,
+      "loss": 0.2277,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999982564065733e-05,
+      "loss": 0.212,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999804524542756e-05,
+      "loss": 0.207,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999782201809227e-05,
+      "loss": 0.2048,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999758672459433e-05,
+      "loss": 0.2074,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999733936496213e-05,
+      "loss": 0.2131,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999707993922555e-05,
+      "loss": 0.2578,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999968084474159e-05,
+      "loss": 0.2086,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999652488956585e-05,
+      "loss": 0.2094,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999622926570975e-05,
+      "loss": 0.2217,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999959215758832e-05,
+      "loss": 0.2174,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999560182012328e-05,
+      "loss": 0.2274,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999526999846864e-05,
+      "loss": 0.2113,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999492611095933e-05,
+      "loss": 0.2065,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999945701576368e-05,
+      "loss": 0.2182,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99994202138544e-05,
+      "loss": 0.2177,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999382205372538e-05,
+      "loss": 0.2136,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999342990322677e-05,
+      "loss": 0.2176,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999302568709548e-05,
+      "loss": 0.1984,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999260940538033e-05,
+      "loss": 0.205,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999921810581315e-05,
+      "loss": 0.2117,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999917406454007e-05,
+      "loss": 0.2029,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999912881672411e-05,
+      "loss": 0.2131,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999082362370724e-05,
+      "loss": 0.2069,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999903470148552e-05,
+      "loss": 0.2006,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998985834074246e-05,
+      "loss": 0.2125,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998935760142805e-05,
+      "loss": 0.196,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998884479697237e-05,
+      "loss": 0.1998,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998831992743726e-05,
+      "loss": 0.2068,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998778299288608e-05,
+      "loss": 0.2107,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998723399338364e-05,
+      "loss": 0.2194,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998667292899614e-05,
+      "loss": 0.2411,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999860997997913e-05,
+      "loss": 0.2054,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998551460583826e-05,
+      "loss": 0.2174,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998491734720767e-05,
+      "loss": 0.2005,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998430802397157e-05,
+      "loss": 0.2046,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999836866362035e-05,
+      "loss": 0.2134,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998305318397843e-05,
+      "loss": 0.2078,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999824076673728e-05,
+      "loss": 0.2271,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998175008646447e-05,
+      "loss": 0.1935,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998108044133286e-05,
+      "loss": 0.2115,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998039873205866e-05,
+      "loss": 0.2315,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997970495872425e-05,
+      "loss": 0.2131,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997899912141325e-05,
+      "loss": 0.1914,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999782812202109e-05,
+      "loss": 0.2205,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997755125520376e-05,
+      "loss": 0.2224,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997680922647994e-05,
+      "loss": 0.2474,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99976055134129e-05,
+      "loss": 0.2163,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999752889782419e-05,
+      "loss": 0.1953,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999745107589111e-05,
+      "loss": 0.2049,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997372047623048e-05,
+      "loss": 0.2109,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997291813029546e-05,
+      "loss": 0.2086,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997210372120276e-05,
+      "loss": 0.2173,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997127724905073e-05,
+      "loss": 0.1966,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997043871393906e-05,
+      "loss": 0.1959,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996958811596897e-05,
+      "loss": 0.2109,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996872545524304e-05,
+      "loss": 0.1923,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999678507318654e-05,
+      "loss": 0.213,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996696394594158e-05,
+      "loss": 0.215,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999660650975786e-05,
+      "loss": 0.1899,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996515418688493e-05,
+      "loss": 0.2024,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996423121397043e-05,
+      "loss": 0.1992,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996329617894648e-05,
+      "loss": 0.2106,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99962349081926e-05,
+      "loss": 0.1979,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996138992302314e-05,
+      "loss": 0.1982,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996041870235374e-05,
+      "loss": 0.1802,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999594354200349e-05,
+      "loss": 0.2096,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995844007618536e-05,
+      "loss": 0.2014,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995743267092514e-05,
+      "loss": 0.234,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995641320437586e-05,
+      "loss": 0.2277,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995538167666048e-05,
+      "loss": 0.2069,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995433808790354e-05,
+      "loss": 0.2192,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999532824382309e-05,
+      "loss": 0.1964,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995221472776995e-05,
+      "loss": 0.1963,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995113495664954e-05,
+      "loss": 0.198,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995004312499997e-05,
+      "loss": 0.2106,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994893923295296e-05,
+      "loss": 0.2326,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999478232806417e-05,
+      "loss": 0.1855,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999466952682009e-05,
+      "loss": 0.211,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999455551957666e-05,
+      "loss": 0.2013,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999444030634764e-05,
+      "loss": 0.1995,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994323887146936e-05,
+      "loss": 0.2187,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999420626198859e-05,
+      "loss": 0.1971,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99940874308868e-05,
+      "loss": 0.2036,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99939673938559e-05,
+      "loss": 0.1944,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993846150910372e-05,
+      "loss": 0.185,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993723702064852e-05,
+      "loss": 0.1925,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993600047334113e-05,
+      "loss": 0.1939,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993475186733078e-05,
+      "loss": 0.1998,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999334912027681e-05,
+      "loss": 0.2102,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993221847980517e-05,
+      "loss": 0.1909,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993093369859563e-05,
+      "loss": 0.2202,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999296368592945e-05,
+      "loss": 0.1825,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992832796205824e-05,
+      "loss": 0.1907,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992700700704478e-05,
+      "loss": 0.187,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992567399441353e-05,
+      "loss": 0.1852,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992432892432534e-05,
+      "loss": 0.1962,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992297179694252e-05,
+      "loss": 0.1724,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999216026124288e-05,
+      "loss": 0.1836,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992022137094937e-05,
+      "loss": 0.1827,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991882807267097e-05,
+      "loss": 0.202,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991742271776164e-05,
+      "loss": 0.1966,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99916005306391e-05,
+      "loss": 0.189,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999145758387301e-05,
+      "loss": 0.1995,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999131343149514e-05,
+      "loss": 0.1933,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991168073522885e-05,
+      "loss": 0.179,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991021509973783e-05,
+      "loss": 0.2071,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999087374086552e-05,
+      "loss": 0.1988,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990724766215925e-05,
+      "loss": 0.2076,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999057458604298e-05,
+      "loss": 0.1883,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990423200364794e-05,
+      "loss": 0.1757,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990270609199647e-05,
+      "loss": 0.1972,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990116812565944e-05,
+      "loss": 0.1982,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989961810482246e-05,
+      "loss": 0.1893,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998980560296725e-05,
+      "loss": 0.1827,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998964819003981e-05,
+      "loss": 0.1929,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989489571718924e-05,
+      "loss": 0.1953,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989329748023728e-05,
+      "loss": 0.2008,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99891687189735e-05,
+      "loss": 0.198,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989006484587682e-05,
+      "loss": 0.1837,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998884304488584e-05,
+      "loss": 0.1982,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99886783998877e-05,
+      "loss": 0.2002,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998851254961313e-05,
+      "loss": 0.1965,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998834549408214e-05,
+      "loss": 0.1688,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998817723331489e-05,
+      "loss": 0.1781,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998800776733168e-05,
+      "loss": 0.2042,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998783709615296e-05,
+      "loss": 0.1921,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987665219799323e-05,
+      "loss": 0.195,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987492138291508e-05,
+      "loss": 0.1926,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987317851650402e-05,
+      "loss": 0.1775,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987142359897037e-05,
+      "loss": 0.1965,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998696566305258e-05,
+      "loss": 0.2096,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986787761138363e-05,
+      "loss": 0.1764,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986608654175842e-05,
+      "loss": 0.2042,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986428342186636e-05,
+      "loss": 0.2149,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986246825192502e-05,
+      "loss": 0.1907,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998606410321534e-05,
+      "loss": 0.1858,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985880176277195e-05,
+      "loss": 0.2059,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985695044400265e-05,
+      "loss": 0.2012,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998550870760689e-05,
+      "loss": 0.1944,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985321165919552e-05,
+      "loss": 0.2001,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998513241936088e-05,
+      "loss": 0.2072,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998494246795365e-05,
+      "loss": 0.192,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984751311720777e-05,
+      "loss": 0.2205,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998455895068534e-05,
+      "loss": 0.1982,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984365384870537e-05,
+      "loss": 0.1915,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984170614299727e-05,
+      "loss": 0.1841,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983974638996417e-05,
+      "loss": 0.1859,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998377745898425e-05,
+      "loss": 0.21,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983579074287024e-05,
+      "loss": 0.1948,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983379484928668e-05,
+      "loss": 0.2013,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998317869093327e-05,
+      "loss": 0.2236,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998297669232506e-05,
+      "loss": 0.1958,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982773489128412e-05,
+      "loss": 0.2068,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982569081367844e-05,
+      "loss": 0.1865,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998236346906802e-05,
+      "loss": 0.2056,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998215665225375e-05,
+      "loss": 0.1947,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998194863094999e-05,
+      "loss": 0.1784,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981739405181845e-05,
+      "loss": 0.2001,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981528974974555e-05,
+      "loss": 0.1756,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998131734035351e-05,
+      "loss": 0.1772,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998110450134425e-05,
+      "loss": 0.1924,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980890457972462e-05,
+      "loss": 0.2048,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980675210263967e-05,
+      "loss": 0.1999,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998045875824474e-05,
+      "loss": 0.1919,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980241101940897e-05,
+      "loss": 0.1898,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980022241378703e-05,
+      "loss": 0.1784,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979802176584564e-05,
+      "loss": 0.1801,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979580907585042e-05,
+      "loss": 0.1961,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979358434406826e-05,
+      "loss": 0.189,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979134757076767e-05,
+      "loss": 0.1796,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997890987562185e-05,
+      "loss": 0.2074,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978683790069215e-05,
+      "loss": 0.1881,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978456500446142e-05,
+      "loss": 0.1865,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978228006780056e-05,
+      "loss": 0.1918,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977998309098527e-05,
+      "loss": 0.172,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977767407429268e-05,
+      "loss": 0.1736,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977535301800145e-05,
+      "loss": 0.1897,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977301992239167e-05,
+      "loss": 0.1809,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977067478774483e-05,
+      "loss": 0.2078,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997683176143439e-05,
+      "loss": 0.2124,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997659484024733e-05,
+      "loss": 0.196,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976356715241892e-05,
+      "loss": 0.1863,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976117386446812e-05,
+      "loss": 0.2035,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997587685389096e-05,
+      "loss": 0.202,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997563511760337e-05,
+      "loss": 0.1755,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975392177613207e-05,
+      "loss": 0.2002,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997514803394978e-05,
+      "loss": 0.1945,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997490268664256e-05,
+      "loss": 0.1974,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997465613572114e-05,
+      "loss": 0.1964,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974408381215277e-05,
+      "loss": 0.2335,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997415942315486e-05,
+      "loss": 0.1836,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973909261569934e-05,
+      "loss": 0.1963,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973657896490687e-05,
+      "loss": 0.1856,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973405327947445e-05,
+      "loss": 0.1649,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997315155597068e-05,
+      "loss": 0.1861,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972896580591025e-05,
+      "loss": 0.204,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972640401839235e-05,
+      "loss": 0.1766,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997238301974623e-05,
+      "loss": 0.1924,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997212443434306e-05,
+      "loss": 0.172,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971864645660933e-05,
+      "loss": 0.1876,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971603653731194e-05,
+      "loss": 0.1829,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971341458585334e-05,
+      "loss": 0.1871,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971078060254992e-05,
+      "loss": 0.1701,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997081345877195e-05,
+      "loss": 0.189,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970547654168136e-05,
+      "loss": 0.211,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970280646475623e-05,
+      "loss": 0.1922,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970012435726633e-05,
+      "loss": 0.2048,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996974302195352e-05,
+      "loss": 0.2118,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99694724051888e-05,
+      "loss": 0.1875,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996920058546513e-05,
+      "loss": 0.1984,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968927562815302e-05,
+      "loss": 0.2087,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968653337272262e-05,
+      "loss": 0.2012,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99683779088691e-05,
+      "loss": 0.1933,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996810127763905e-05,
+      "loss": 0.1783,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996782344361549e-05,
+      "loss": 0.1902,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996754440683195e-05,
+      "loss": 0.1873,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996726416732209e-05,
+      "loss": 0.1921,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966982725119737e-05,
+      "loss": 0.1849,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996670008025884e-05,
+      "loss": 0.1965,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966416232773513e-05,
+      "loss": 0.1958,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966131182698e-05,
+      "loss": 0.185,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99658449300667e-05,
+      "loss": 0.2054,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996555747491415e-05,
+      "loss": 0.196,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996526881727504e-05,
+      "loss": 0.1728,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964978957184198e-05,
+      "loss": 0.1826,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99646878946766e-05,
+      "loss": 0.1841,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964395629787364e-05,
+      "loss": 0.1892,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996410216255176e-05,
+      "loss": 0.1991,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99638074930052e-05,
+      "loss": 0.1889,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963511621183236e-05,
+      "loss": 0.201,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963214547121573e-05,
+      "loss": 0.1908,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962916270856053e-05,
+      "loss": 0.1856,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996261679242267e-05,
+      "loss": 0.1799,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996231611185756e-05,
+      "loss": 0.2193,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962014229197003e-05,
+      "loss": 0.1859,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996171114447743e-05,
+      "loss": 0.1789,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961406857735406e-05,
+      "loss": 0.1989,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961101369007654e-05,
+      "loss": 0.1857,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996079467833103e-05,
+      "loss": 0.1954,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960486785742543e-05,
+      "loss": 0.1875,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960177691279346e-05,
+      "loss": 0.1937,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959867394978732e-05,
+      "loss": 0.1965,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959555896878147e-05,
+      "loss": 0.2058,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959243197015172e-05,
+      "loss": 0.1706,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958929295427544e-05,
+      "loss": 0.1929,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958614192153138e-05,
+      "loss": 0.1816,
+      "step": 542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958297887229977e-05,
+      "loss": 0.1863,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9957980380696227e-05,
+      "loss": 0.1751,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99576616725902e-05,
+      "loss": 0.1747,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957341762950346e-05,
+      "loss": 0.18,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957020651815275e-05,
+      "loss": 0.1694,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956698339223735e-05,
+      "loss": 0.1918,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956374825214608e-05,
+      "loss": 0.1955,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995605010982694e-05,
+      "loss": 0.1896,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955724193099905e-05,
+      "loss": 0.1836,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995539707507284e-05,
+      "loss": 0.2087,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955068755785202e-05,
+      "loss": 0.1736,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995473923527662e-05,
+      "loss": 0.1665,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954408513586845e-05,
+      "loss": 0.1831,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954076590755794e-05,
+      "loss": 0.1853,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953743466823513e-05,
+      "loss": 0.1857,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953409141830194e-05,
+      "loss": 0.1879,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953073615816185e-05,
+      "loss": 0.1747,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995273688882197e-05,
+      "loss": 0.2001,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995239896088818e-05,
+      "loss": 0.1965,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952059832055585e-05,
+      "loss": 0.2,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951719502365113e-05,
+      "loss": 0.1814,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995137797185783e-05,
+      "loss": 0.1865,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995103524057494e-05,
+      "loss": 0.2209,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950691308557806e-05,
+      "loss": 0.1823,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995034617584792e-05,
+      "loss": 0.1859,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949999842486933e-05,
+      "loss": 0.1585,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949652308516635e-05,
+      "loss": 0.1714,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949303573978958e-05,
+      "loss": 0.1843,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948953638915986e-05,
+      "loss": 0.1935,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948602503369942e-05,
+      "loss": 0.1942,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994825016738319e-05,
+      "loss": 0.1782,
+      "step": 573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947896630998253e-05,
+      "loss": 0.1957,
+      "step": 574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947541894257782e-05,
+      "loss": 0.1777,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947185957204588e-05,
+      "loss": 0.1871,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946828819881618e-05,
+      "loss": 0.1774,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946470482331962e-05,
+      "loss": 0.1829,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946110944598864e-05,
+      "loss": 0.1911,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945750206725704e-05,
+      "loss": 0.2101,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994538826875601e-05,
+      "loss": 0.1873,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994502513073346e-05,
+      "loss": 0.1896,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994466079270186e-05,
+      "loss": 0.1947,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944295254705187e-05,
+      "loss": 0.1782,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943928516787538e-05,
+      "loss": 0.194,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994356057899317e-05,
+      "loss": 0.1987,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943191441366478e-05,
+      "loss": 0.2076,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942821103952002e-05,
+      "loss": 0.1972,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942449566794436e-05,
+      "loss": 0.1958,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99420768299386e-05,
+      "loss": 0.185,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941702893429483e-05,
+      "loss": 0.1884,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941327757312194e-05,
+      "loss": 0.1841,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940951421632005e-05,
+      "loss": 0.1689,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940573886434324e-05,
+      "loss": 0.1911,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940195151764708e-05,
+      "loss": 0.1692,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939815217668856e-05,
+      "loss": 0.1961,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993943408419261e-05,
+      "loss": 0.1932,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993905175138196e-05,
+      "loss": 0.1889,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938668219283042e-05,
+      "loss": 0.184,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938283487942137e-05,
+      "loss": 0.1677,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993789755740566e-05,
+      "loss": 0.1691,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937510427720187e-05,
+      "loss": 0.1904,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 0.1897,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993673257108924e-05,
+      "loss": 0.1778,
+      "step": 604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936341844237623e-05,
+      "loss": 0.2014,
+      "step": 605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935949918424727e-05,
+      "loss": 0.2116,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993555679369784e-05,
+      "loss": 0.1957,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99351624701044e-05,
+      "loss": 0.1959,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934766947691996e-05,
+      "loss": 0.1951,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934370226508338e-05,
+      "loss": 0.2061,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933972306601308e-05,
+      "loss": 0.1944,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933573188018913e-05,
+      "loss": 0.1976,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993317287080932e-05,
+      "loss": 0.1811,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932771355020826e-05,
+      "loss": 0.1812,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993236864070188e-05,
+      "loss": 0.1966,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993196472790108e-05,
+      "loss": 0.1948,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931559616667164e-05,
+      "loss": 0.1988,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931153307049008e-05,
+      "loss": 0.1737,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930745799095645e-05,
+      "loss": 0.1974,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930337092856243e-05,
+      "loss": 0.1924,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992992718838012e-05,
+      "loss": 0.1745,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929516085716736e-05,
+      "loss": 0.1805,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929103784915698e-05,
+      "loss": 0.1835,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928690286026754e-05,
+      "loss": 0.2055,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928275589099802e-05,
+      "loss": 0.2124,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927859694184873e-05,
+      "loss": 0.174,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927442601332162e-05,
+      "loss": 0.1907,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927024310591992e-05,
+      "loss": 0.1756,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926604822014834e-05,
+      "loss": 0.2038,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926184135651302e-05,
+      "loss": 0.1673,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925762251552164e-05,
+      "loss": 0.2008,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925339169768327e-05,
+      "loss": 0.1848,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992491489035084e-05,
+      "loss": 0.1897,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924489413350894e-05,
+      "loss": 0.1641,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924062738819834e-05,
+      "loss": 0.1734,
+      "step": 635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923634866809144e-05,
+      "loss": 0.1832,
+      "step": 636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992320579737045e-05,
+      "loss": 0.1751,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922775530555533e-05,
+      "loss": 0.1719,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922344066416298e-05,
+      "loss": 0.1795,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921911405004816e-05,
+      "loss": 0.205,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921477546373296e-05,
+      "loss": 0.1679,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992104249057408e-05,
+      "loss": 0.1655,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920606237659674e-05,
+      "loss": 0.1712,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992016878768271e-05,
+      "loss": 0.1747,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919730140695978e-05,
+      "loss": 0.1746,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99192902967524e-05,
+      "loss": 0.1689,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991884925590506e-05,
+      "loss": 0.1758,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918407018207168e-05,
+      "loss": 0.1616,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917963583712085e-05,
+      "loss": 0.1966,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917518952473326e-05,
+      "loss": 0.1931,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917073124544535e-05,
+      "loss": 0.1884,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991662609997951e-05,
+      "loss": 0.1625,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916177878832185e-05,
+      "loss": 0.1861,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991572846115666e-05,
+      "loss": 0.1708,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915277847007143e-05,
+      "loss": 0.1738,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914826036438025e-05,
+      "loss": 0.1735,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914373029503814e-05,
+      "loss": 0.2124,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991391882625917e-05,
+      "loss": 0.1914,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913463426758906e-05,
+      "loss": 0.1958,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913006831057967e-05,
+      "loss": 0.158,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912549039211453e-05,
+      "loss": 0.1928,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99120900512746e-05,
+      "loss": 0.1833,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991162986730279e-05,
+      "loss": 0.1955,
+      "step": 663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911168487351552e-05,
+      "loss": 0.1996,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910705911476558e-05,
+      "loss": 0.1987,
+      "step": 665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910242139733625e-05,
+      "loss": 0.1917,
+      "step": 666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909777172178712e-05,
+      "loss": 0.1733,
+      "step": 667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909311008867926e-05,
+      "loss": 0.1862,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908843649857517e-05,
+      "loss": 0.1824,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908375095203876e-05,
+      "loss": 0.1672,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990790534496354e-05,
+      "loss": 0.2258,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907434399193196e-05,
+      "loss": 0.1873,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906962257949668e-05,
+      "loss": 0.1858,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990648892128992e-05,
+      "loss": 0.1767,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906014389271078e-05,
+      "loss": 0.1845,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905538661950395e-05,
+      "loss": 0.189,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990506173938527e-05,
+      "loss": 0.1779,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990458362163326e-05,
+      "loss": 0.1786,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904104308752053e-05,
+      "loss": 0.1965,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903623800799483e-05,
+      "loss": 0.1728,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990314209783353e-05,
+      "loss": 0.1905,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990265919991232e-05,
+      "loss": 0.1923,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990217510709412e-05,
+      "loss": 0.1768,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901689819437345e-05,
+      "loss": 0.1947,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901203337000556e-05,
+      "loss": 0.1871,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990071565984244e-05,
+      "loss": 0.1754,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900226788021853e-05,
+      "loss": 0.1833,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899736721597787e-05,
+      "loss": 0.2002,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899245460629365e-05,
+      "loss": 0.1618,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898753005175874e-05,
+      "loss": 0.1782,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898259355296728e-05,
+      "loss": 0.1924,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897764511051504e-05,
+      "loss": 0.2008,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98972684724999e-05,
+      "loss": 0.1827,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896771239701772e-05,
+      "loss": 0.1908,
+      "step": 694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896272812717127e-05,
+      "loss": 0.1579,
+      "step": 695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895773191606103e-05,
+      "loss": 0.1814,
+      "step": 696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895272376428983e-05,
+      "loss": 0.1742,
+      "step": 697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894770367246197e-05,
+      "loss": 0.1677,
+      "step": 698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894267164118322e-05,
+      "loss": 0.1654,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989376276710608e-05,
+      "loss": 0.1761,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989325717627033e-05,
+      "loss": 0.1956,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989275039167208e-05,
+      "loss": 0.1938,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989224241337248e-05,
+      "loss": 0.1797,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989173324143282e-05,
+      "loss": 0.1734,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989122287591455e-05,
+      "loss": 0.171,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890711316879243e-05,
+      "loss": 0.1768,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989019856438863e-05,
+      "loss": 0.1896,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988968461850458e-05,
+      "loss": 0.1757,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889169479289112e-05,
+      "loss": 0.1564,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988865314680438e-05,
+      "loss": 0.2034,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988813562111269e-05,
+      "loss": 0.1907,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988761690227649e-05,
+      "loss": 0.1859,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887096990358366e-05,
+      "loss": 0.1615,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886575885421055e-05,
+      "loss": 0.1737,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988605358752744e-05,
+      "loss": 0.1927,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885530096740538e-05,
+      "loss": 0.18,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885005413123515e-05,
+      "loss": 0.1822,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884479536739688e-05,
+      "loss": 0.1769,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883952467652504e-05,
+      "loss": 0.1731,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883424205925567e-05,
+      "loss": 0.1969,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882894751622613e-05,
+      "loss": 0.1666,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882364104807536e-05,
+      "loss": 0.1715,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881832265544366e-05,
+      "loss": 0.1808,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988129923389727e-05,
+      "loss": 0.1813,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988076500993057e-05,
+      "loss": 0.1745,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880229593708726e-05,
+      "loss": 0.181,
+      "step": 726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879692985296345e-05,
+      "loss": 0.1749,
+      "step": 727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879155184758175e-05,
+      "loss": 0.1809,
+      "step": 728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878616192159118e-05,
+      "loss": 0.1686,
+      "step": 729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878076007564197e-05,
+      "loss": 0.1787,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98775346310386e-05,
+      "loss": 0.1682,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876992062647652e-05,
+      "loss": 0.1525,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987644830245682e-05,
+      "loss": 0.1786,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987590335053172e-05,
+      "loss": 0.1813,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875357206938103e-05,
+      "loss": 0.1904,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874809871741877e-05,
+      "loss": 0.1973,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874261345009076e-05,
+      "loss": 0.1661,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987371162680589e-05,
+      "loss": 0.1915,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873160717198655e-05,
+      "loss": 0.1676,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872608616253846e-05,
+      "loss": 0.1751,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872055324038078e-05,
+      "loss": 0.1697,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987150084061811e-05,
+      "loss": 0.1677,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987094516606086e-05,
+      "loss": 0.1787,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870388300433366e-05,
+      "loss": 0.1796,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986983024380283e-05,
+      "loss": 0.1867,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869270996236586e-05,
+      "loss": 0.1742,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868710557802115e-05,
+      "loss": 0.1504,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868148928567046e-05,
+      "loss": 0.1967,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867586108599142e-05,
+      "loss": 0.1625,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867022097966314e-05,
+      "loss": 0.1553,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866456896736627e-05,
+      "loss": 0.1768,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865890504978273e-05,
+      "loss": 0.1746,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865322922759595e-05,
+      "loss": 0.1806,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864754150149086e-05,
+      "loss": 0.1923,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986418418721537e-05,
+      "loss": 0.1828,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 0.1829,
+      "step": 756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863040690653567e-05,
+      "loss": 0.1687,
+      "step": 757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862467157163458e-05,
+      "loss": 0.1754,
+      "step": 758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98618924336261e-05,
+      "loss": 0.1958,
+      "step": 759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986131652011085e-05,
+      "loss": 0.1769,
+      "step": 760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860739416687194e-05,
+      "loss": 0.1833,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860161123424766e-05,
+      "loss": 0.183,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859581640393348e-05,
+      "loss": 0.1683,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859000967662864e-05,
+      "loss": 0.1816,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858419105303378e-05,
+      "loss": 0.1807,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98578360533851e-05,
+      "loss": 0.1897,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857251811978387e-05,
+      "loss": 0.1729,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856666381153733e-05,
+      "loss": 0.1709,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985607976098178e-05,
+      "loss": 0.1855,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855491951533308e-05,
+      "loss": 0.1689,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985490295287925e-05,
+      "loss": 0.1744,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854312765090675e-05,
+      "loss": 0.1832,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98537213882388e-05,
+      "loss": 0.1857,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853128822394976e-05,
+      "loss": 0.1832,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985253506763071e-05,
+      "loss": 0.1799,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985194012401765e-05,
+      "loss": 0.1802,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851343991627575e-05,
+      "loss": 0.1942,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985074667053243e-05,
+      "loss": 0.1728,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850148160804275e-05,
+      "loss": 0.1823,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984954846251534e-05,
+      "loss": 0.173,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848947575737982e-05,
+      "loss": 0.1645,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848345500544712e-05,
+      "loss": 0.1929,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984774223700817e-05,
+      "loss": 0.1618,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847137785201158e-05,
+      "loss": 0.1743,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846532145196607e-05,
+      "loss": 0.1691,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845925317067594e-05,
+      "loss": 0.1871,
+      "step": 786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984531730088734e-05,
+      "loss": 0.179,
+      "step": 787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844708096729223e-05,
+      "loss": 0.1997,
+      "step": 788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984409770466674e-05,
+      "loss": 0.1615,
+      "step": 789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843486124773546e-05,
+      "loss": 0.1636,
+      "step": 790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842873357123438e-05,
+      "loss": 0.1761,
+      "step": 791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842259401790356e-05,
+      "loss": 0.1687,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984164425884838e-05,
+      "loss": 0.1668,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841027928371738e-05,
+      "loss": 0.1789,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840410410434803e-05,
+      "loss": 0.1934,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839791705112077e-05,
+      "loss": 0.177,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839171812478227e-05,
+      "loss": 0.1891,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838550732608042e-05,
+      "loss": 0.1789,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983792846557647e-05,
+      "loss": 0.1689,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837305011458595e-05,
+      "loss": 0.1757,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836680370329643e-05,
+      "loss": 0.1833,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983605454226499e-05,
+      "loss": 0.1541,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835427527340152e-05,
+      "loss": 0.17,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834799325630784e-05,
+      "loss": 0.1749,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834169937212685e-05,
+      "loss": 0.1686,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833539362161804e-05,
+      "loss": 0.1703,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983290760055423e-05,
+      "loss": 0.1873,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983227465246619e-05,
+      "loss": 0.1876,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983164051797406e-05,
+      "loss": 0.1642,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831005197154356e-05,
+      "loss": 0.1787,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983036869008374e-05,
+      "loss": 0.163,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 0.182,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829092117497135e-05,
+      "loss": 0.1916,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982845205213518e-05,
+      "loss": 0.1552,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827810800830384e-05,
+      "loss": 0.1827,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827168363660126e-05,
+      "loss": 0.1783,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826524740701926e-05,
+      "loss": 0.1897,
+      "step": 817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825879932033444e-05,
+      "loss": 0.1652,
+      "step": 818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825233937732485e-05,
+      "loss": 0.1769,
+      "step": 819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824586757876995e-05,
+      "loss": 0.1769,
+      "step": 820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823938392545075e-05,
+      "loss": 0.1675,
+      "step": 821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982328884181495e-05,
+      "loss": 0.1732,
+      "step": 822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822638105765006e-05,
+      "loss": 0.1806,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821986184473757e-05,
+      "loss": 0.172,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821333078019866e-05,
+      "loss": 0.1706,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982067878648214e-05,
+      "loss": 0.1927,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820023309939536e-05,
+      "loss": 0.1874,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981936664847114e-05,
+      "loss": 0.1845,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818708802156185e-05,
+      "loss": 0.188,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818049771074058e-05,
+      "loss": 0.1949,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817389555304274e-05,
+      "loss": 0.1754,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816728154926496e-05,
+      "loss": 0.1902,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816065570020538e-05,
+      "loss": 0.1766,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815401800666346e-05,
+      "loss": 0.1567,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814736846944013e-05,
+      "loss": 0.1907,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814070708933778e-05,
+      "loss": 0.155,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981340338671602e-05,
+      "loss": 0.1574,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812734880371253e-05,
+      "loss": 0.1669,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812065189980153e-05,
+      "loss": 0.1731,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811394315623523e-05,
+      "loss": 0.1715,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981072225738231e-05,
+      "loss": 0.1913,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810049015337614e-05,
+      "loss": 0.1712,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809374589570665e-05,
+      "loss": 0.1817,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808698980162847e-05,
+      "loss": 0.1622,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808022187195678e-05,
+      "loss": 0.1572,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807344210750826e-05,
+      "loss": 0.1907,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806665050910098e-05,
+      "loss": 0.1833,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805984707755444e-05,
+      "loss": 0.1749,
+      "step": 848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805303181368954e-05,
+      "loss": 0.1652,
+      "step": 849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980462047183287e-05,
+      "loss": 0.1914,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803936579229563e-05,
+      "loss": 0.155,
+      "step": 851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803251503641562e-05,
+      "loss": 0.1685,
+      "step": 852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802565245151525e-05,
+      "loss": 0.1794,
+      "step": 853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980187780384226e-05,
+      "loss": 0.1898,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801189179796722e-05,
+      "loss": 0.1907,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800499373097998e-05,
+      "loss": 0.1726,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799808383829326e-05,
+      "loss": 0.1758,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799116212074077e-05,
+      "loss": 0.1673,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798422857915783e-05,
+      "loss": 0.1705,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797728321438096e-05,
+      "loss": 0.1586,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797032602724828e-05,
+      "loss": 0.1877,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796335701859924e-05,
+      "loss": 0.1748,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795637618927478e-05,
+      "loss": 0.1566,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794938354011725e-05,
+      "loss": 0.1696,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794237907197032e-05,
+      "loss": 0.1683,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979353627856793e-05,
+      "loss": 0.1859,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792833468209074e-05,
+      "loss": 0.1885,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979212947620527e-05,
+      "loss": 0.2076,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979142430264146e-05,
+      "loss": 0.1728,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979071794760274e-05,
+      "loss": 0.1748,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979001041117434e-05,
+      "loss": 0.194,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978930169344163e-05,
+      "loss": 0.1818,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788591794490132e-05,
+      "loss": 0.1692,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787880714405503e-05,
+      "loss": 0.1624,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787168453273546e-05,
+      "loss": 0.1719,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786455011180203e-05,
+      "loss": 0.1736,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785740388211567e-05,
+      "loss": 0.1766,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978502458445386e-05,
+      "loss": 0.1709,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784307599993456e-05,
+      "loss": 0.1639,
+      "step": 879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978358943491687e-05,
+      "loss": 0.187,
+      "step": 880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782870089310766e-05,
+      "loss": 0.1794,
+      "step": 881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782149563261933e-05,
+      "loss": 0.1645,
+      "step": 882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781427856857317e-05,
+      "loss": 0.179,
+      "step": 883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780704970184e-05,
+      "loss": 0.1552,
+      "step": 884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977998090332921e-05,
+      "loss": 0.1751,
+      "step": 885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779255656380318e-05,
+      "loss": 0.1853,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778529229424833e-05,
+      "loss": 0.1727,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977780162255041e-05,
+      "loss": 0.1796,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777072835844842e-05,
+      "loss": 0.1766,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776342869396075e-05,
+      "loss": 0.1673,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775611723292182e-05,
+      "loss": 0.1738,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774879397621387e-05,
+      "loss": 0.1687,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977414589247206e-05,
+      "loss": 0.1615,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773411207932707e-05,
+      "loss": 0.1647,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772675344091976e-05,
+      "loss": 0.1704,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771938301038664e-05,
+      "loss": 0.1646,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97712000788617e-05,
+      "loss": 0.1801,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770460677650167e-05,
+      "loss": 0.1768,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769720097493278e-05,
+      "loss": 0.1904,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768978338480396e-05,
+      "loss": 0.1721,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976823540070103e-05,
+      "loss": 0.1692,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976749128424482e-05,
+      "loss": 0.1877,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976674598920156e-05,
+      "loss": 0.1787,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765999515661177e-05,
+      "loss": 0.1677,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976525186371374e-05,
+      "loss": 0.1771,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764503033449475e-05,
+      "loss": 0.1766,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763753024958724e-05,
+      "loss": 0.1738,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763001838331998e-05,
+      "loss": 0.165,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762249473659936e-05,
+      "loss": 0.1724,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976149593103332e-05,
+      "loss": 0.1814,
+      "step": 910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760741210543074e-05,
+      "loss": 0.1635,
+      "step": 911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759985312280264e-05,
+      "loss": 0.1597,
+      "step": 912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759228236336106e-05,
+      "loss": 0.1769,
+      "step": 913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975846998280195e-05,
+      "loss": 0.169,
+      "step": 914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757710551769293e-05,
+      "loss": 0.1647,
+      "step": 915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756949943329763e-05,
+      "loss": 0.153,
+      "step": 916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975618815757514e-05,
+      "loss": 0.1812,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975542519459735e-05,
+      "loss": 0.1741,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754661054488456e-05,
+      "loss": 0.1854,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975389573734066e-05,
+      "loss": 0.187,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753129243246302e-05,
+      "loss": 0.1891,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975236157229788e-05,
+      "loss": 0.1519,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751592724588016e-05,
+      "loss": 0.1727,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975082270020949e-05,
+      "loss": 0.1703,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975005149925521e-05,
+      "loss": 0.1666,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.184,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974850556799177e-05,
+      "loss": 0.1632,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747730837869145e-05,
+      "loss": 0.1754,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746954931543847e-05,
+      "loss": 0.1852,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746177849109497e-05,
+      "loss": 0.1627,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745399590659865e-05,
+      "loss": 0.1857,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974462015628886e-05,
+      "loss": 0.1727,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974383954609053e-05,
+      "loss": 0.1696,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743057760159063e-05,
+      "loss": 0.1643,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742274798588793e-05,
+      "loss": 0.1609,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741490661474202e-05,
+      "loss": 0.1861,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740705348909906e-05,
+      "loss": 0.1706,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739918860990657e-05,
+      "loss": 0.1779,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739131197811365e-05,
+      "loss": 0.1579,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738342359467066e-05,
+      "loss": 0.157,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973755234605295e-05,
+      "loss": 0.1952,
+      "step": 941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736761157664338e-05,
+      "loss": 0.1719,
+      "step": 942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735968794396698e-05,
+      "loss": 0.173,
+      "step": 943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735175256345647e-05,
+      "loss": 0.1726,
+      "step": 944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734380543606932e-05,
+      "loss": 0.1984,
+      "step": 945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733584656276445e-05,
+      "loss": 0.1574,
+      "step": 946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732787594450222e-05,
+      "loss": 0.1564,
+      "step": 947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973198935822444e-05,
+      "loss": 0.165,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973118994769542e-05,
+      "loss": 0.1805,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973038936295962e-05,
+      "loss": 0.1948,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729587604113645e-05,
+      "loss": 0.1778,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728784671254237e-05,
+      "loss": 0.1653,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727980564478274e-05,
+      "loss": 0.1672,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727175283882792e-05,
+      "loss": 0.1714,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972636882956496e-05,
+      "loss": 0.1657,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725561201622085e-05,
+      "loss": 0.1622,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724752400151615e-05,
+      "loss": 0.16,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723942425251147e-05,
+      "loss": 0.1749,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972313127701842e-05,
+      "loss": 0.1625,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722318955551307e-05,
+      "loss": 0.2156,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721505460947826e-05,
+      "loss": 0.1757,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720690793306137e-05,
+      "loss": 0.1621,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971987495272454e-05,
+      "loss": 0.1718,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719057939301477e-05,
+      "loss": 0.1668,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718239753135537e-05,
+      "loss": 0.1607,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717420394325445e-05,
+      "loss": 0.1755,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716599862970063e-05,
+      "loss": 0.1631,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715778159168402e-05,
+      "loss": 0.1891,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971495528301962e-05,
+      "loss": 0.1621,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714131234622996e-05,
+      "loss": 0.1838,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713306014077972e-05,
+      "loss": 0.1575,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712479621484117e-05,
+      "loss": 0.1474,
+      "step": 972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971165205694115e-05,
+      "loss": 0.1576,
+      "step": 973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710823320548937e-05,
+      "loss": 0.1673,
+      "step": 974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970999341240746e-05,
+      "loss": 0.1625,
+      "step": 975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709162332616872e-05,
+      "loss": 0.1691,
+      "step": 976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708330081277447e-05,
+      "loss": 0.1702,
+      "step": 977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707496658489613e-05,
+      "loss": 0.1559,
+      "step": 978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706662064353933e-05,
+      "loss": 0.1827,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705826298971112e-05,
+      "loss": 0.1841,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704989362441998e-05,
+      "loss": 0.1749,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704151254867575e-05,
+      "loss": 0.1623,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970331197634898e-05,
+      "loss": 0.1566,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970247152698748e-05,
+      "loss": 0.1821,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701629906884486e-05,
+      "loss": 0.1554,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700787116141553e-05,
+      "loss": 0.1724,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699943154860376e-05,
+      "loss": 0.1921,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969909802314279e-05,
+      "loss": 0.169,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698251721090776e-05,
+      "loss": 0.1587,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697404248806445e-05,
+      "loss": 0.1691,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696555606392063e-05,
+      "loss": 0.1731,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695705793950025e-05,
+      "loss": 0.1561,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969485481158288e-05,
+      "loss": 0.1828,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694002659393306e-05,
+      "loss": 0.1769,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693149337484127e-05,
+      "loss": 0.1823,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692294845958315e-05,
+      "loss": 0.1731,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969143918491897e-05,
+      "loss": 0.1887,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969058235446934e-05,
+      "loss": 0.1581,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968972435471282e-05,
+      "loss": 0.1599,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968886518575293e-05,
+      "loss": 0.152,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688004847693348e-05,
+      "loss": 0.1687,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687143340637885e-05,
+      "loss": 0.1738,
+      "step": 1002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686280664690495e-05,
+      "loss": 0.2016,
+      "step": 1003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968541681995527e-05,
+      "loss": 0.1695,
+      "step": 1004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684551806536448e-05,
+      "loss": 0.185,
+      "step": 1005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96836856245384e-05,
+      "loss": 0.1767,
+      "step": 1006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682818274065645e-05,
+      "loss": 0.182,
+      "step": 1007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681949755222846e-05,
+      "loss": 0.1712,
+      "step": 1008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681080068114794e-05,
+      "loss": 0.1767,
+      "step": 1009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680209212846437e-05,
+      "loss": 0.189,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679337189522854e-05,
+      "loss": 0.1707,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678463998249264e-05,
+      "loss": 0.1739,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967758963913103e-05,
+      "loss": 0.1826,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676714112273664e-05,
+      "loss": 0.1719,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96758374177828e-05,
+      "loss": 0.1821,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674959555764227e-05,
+      "loss": 0.1571,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967408052632388e-05,
+      "loss": 0.1696,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967320032956781e-05,
+      "loss": 0.1763,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967231896560224e-05,
+      "loss": 0.1674,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671436434533514e-05,
+      "loss": 0.1736,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670552736468117e-05,
+      "loss": 0.1691,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966966787151269e-05,
+      "loss": 0.1667,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668781839773995e-05,
+      "loss": 0.1574,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966789464135895e-05,
+      "loss": 0.1641,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667006276374606e-05,
+      "loss": 0.1789,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966611674492816e-05,
+      "loss": 0.1535,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665226047126942e-05,
+      "loss": 0.1499,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966433418307843e-05,
+      "loss": 0.1549,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663441152890237e-05,
+      "loss": 0.1777,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662546956670123e-05,
+      "loss": 0.1701,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661651594525986e-05,
+      "loss": 0.1609,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660755066565863e-05,
+      "loss": 0.1765,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659857372897934e-05,
+      "loss": 0.1773,
+      "step": 1033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658958513630518e-05,
+      "loss": 0.1761,
+      "step": 1034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965805848887207e-05,
+      "loss": 0.1796,
+      "step": 1035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96571572987312e-05,
+      "loss": 0.1702,
+      "step": 1036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656254943316644e-05,
+      "loss": 0.1496,
+      "step": 1037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655351422737285e-05,
+      "loss": 0.1575,
+      "step": 1038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654446737102147e-05,
+      "loss": 0.1843,
+      "step": 1039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653540886520387e-05,
+      "loss": 0.1771,
+      "step": 1040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965263387110132e-05,
+      "loss": 0.1723,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965172569095438e-05,
+      "loss": 0.1936,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650816346189154e-05,
+      "loss": 0.1718,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649905836915373e-05,
+      "loss": 0.1667,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648994163242898e-05,
+      "loss": 0.1516,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648081325281737e-05,
+      "loss": 0.1795,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647167323142033e-05,
+      "loss": 0.1796,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964625215693408e-05,
+      "loss": 0.1824,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645335826768303e-05,
+      "loss": 0.1634,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964441833275527e-05,
+      "loss": 0.1769,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643499675005692e-05,
+      "loss": 0.1498,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964257985363041e-05,
+      "loss": 0.1715,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641658868740423e-05,
+      "loss": 0.1596,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640736720446857e-05,
+      "loss": 0.1759,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639813408860984e-05,
+      "loss": 0.1682,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638888934094212e-05,
+      "loss": 0.1715,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637963296258094e-05,
+      "loss": 0.1529,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963703649546432e-05,
+      "loss": 0.1563,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636108531824725e-05,
+      "loss": 0.1686,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635179405451277e-05,
+      "loss": 0.1807,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634249116456093e-05,
+      "loss": 0.1594,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633317664951418e-05,
+      "loss": 0.1882,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963238505104965e-05,
+      "loss": 0.1865,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631451274863326e-05,
+      "loss": 0.1796,
+      "step": 1064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963051633650511e-05,
+      "loss": 0.1615,
+      "step": 1065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629580236087823e-05,
+      "loss": 0.1646,
+      "step": 1066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628642973724416e-05,
+      "loss": 0.1839,
+      "step": 1067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627704549527986e-05,
+      "loss": 0.1876,
+      "step": 1068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626764963611764e-05,
+      "loss": 0.1624,
+      "step": 1069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625824216089123e-05,
+      "loss": 0.146,
+      "step": 1070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624882307073584e-05,
+      "loss": 0.1697,
+      "step": 1071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9623939236678793e-05,
+      "loss": 0.1558,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622995005018556e-05,
+      "loss": 0.1929,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622049612206795e-05,
+      "loss": 0.1761,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621103058357594e-05,
+      "loss": 0.1711,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962015534358517e-05,
+      "loss": 0.1718,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619206468003867e-05,
+      "loss": 0.1696,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961825643172819e-05,
+      "loss": 0.1497,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617305234872773e-05,
+      "loss": 0.1971,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616352877552394e-05,
+      "loss": 0.1628,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961539935988196e-05,
+      "loss": 0.1754,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614444681976533e-05,
+      "loss": 0.1739,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961348884395131e-05,
+      "loss": 0.1576,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961253184592162e-05,
+      "loss": 0.2156,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961157368800294e-05,
+      "loss": 0.1575,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610614370310892e-05,
+      "loss": 0.1606,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609653892961226e-05,
+      "loss": 0.1655,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608692256069836e-05,
+      "loss": 0.1512,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960772945975276e-05,
+      "loss": 0.1903,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606765504126174e-05,
+      "loss": 0.1581,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605800389306386e-05,
+      "loss": 0.1871,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960483411540986e-05,
+      "loss": 0.1557,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603866682553187e-05,
+      "loss": 0.1588,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96028980908531e-05,
+      "loss": 0.173,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601928340426478e-05,
+      "loss": 0.1617,
+      "step": 1095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960095743139033e-05,
+      "loss": 0.1832,
+      "step": 1096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959998536386181e-05,
+      "loss": 0.1578,
+      "step": 1097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599012137958214e-05,
+      "loss": 0.1572,
+      "step": 1098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959803775379698e-05,
+      "loss": 0.1759,
+      "step": 1099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959706221149567e-05,
+      "loss": 0.1749,
+      "step": 1100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596085511172006e-05,
+      "loss": 0.1678,
+      "step": 1101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595107652943842e-05,
+      "loss": 0.1694,
+      "step": 1102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594128636929165e-05,
+      "loss": 0.185,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959314846324611e-05,
+      "loss": 0.176,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592167132012946e-05,
+      "loss": 0.1658,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591184643348092e-05,
+      "loss": 0.173,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590200997370092e-05,
+      "loss": 0.1644,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589216194197637e-05,
+      "loss": 0.1529,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958823023394956e-05,
+      "loss": 0.1875,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587243116744832e-05,
+      "loss": 0.1617,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586254842702562e-05,
+      "loss": 0.1756,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585265411941997e-05,
+      "loss": 0.1783,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958427482458253e-05,
+      "loss": 0.1374,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583283080743686e-05,
+      "loss": 0.1893,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582290180545134e-05,
+      "loss": 0.1538,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581296124106682e-05,
+      "loss": 0.1661,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958030091154828e-05,
+      "loss": 0.1835,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579304542990005e-05,
+      "loss": 0.1424,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578307018552096e-05,
+      "loss": 0.1732,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957730833835491e-05,
+      "loss": 0.1735,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957630850251895e-05,
+      "loss": 0.1722,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575307511164873e-05,
+      "loss": 0.1525,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957430536441345e-05,
+      "loss": 0.1794,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957330206238561e-05,
+      "loss": 0.1845,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572297605202412e-05,
+      "loss": 0.1654,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571291992985064e-05,
+      "loss": 0.1619,
+      "step": 1126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95702852258549e-05,
+      "loss": 0.1638,
+      "step": 1127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956927730393341e-05,
+      "loss": 0.1741,
+      "step": 1128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568268227342203e-05,
+      "loss": 0.1703,
+      "step": 1129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956725799620305e-05,
+      "loss": 0.1837,
+      "step": 1130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956624661063784e-05,
+      "loss": 0.1731,
+      "step": 1131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565234070768617e-05,
+      "loss": 0.1927,
+      "step": 1132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564220376717557e-05,
+      "loss": 0.1524,
+      "step": 1133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563205528606977e-05,
+      "loss": 0.1595,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562189526559333e-05,
+      "loss": 0.1481,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561172370697215e-05,
+      "loss": 0.1793,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560154061143365e-05,
+      "loss": 0.1484,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559134598020653e-05,
+      "loss": 0.1905,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558113981452093e-05,
+      "loss": 0.1702,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955709221156083e-05,
+      "loss": 0.1855,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556069288470168e-05,
+      "loss": 0.162,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555045212303528e-05,
+      "loss": 0.1731,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554019983184483e-05,
+      "loss": 0.1632,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552993601236737e-05,
+      "loss": 0.1424,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551966066584144e-05,
+      "loss": 0.1797,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550937379350686e-05,
+      "loss": 0.1549,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954990753966049e-05,
+      "loss": 0.1774,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954887654763782e-05,
+      "loss": 0.1661,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547844403407085e-05,
+      "loss": 0.1919,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546811107092822e-05,
+      "loss": 0.1699,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545776658819715e-05,
+      "loss": 0.1642,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544741058712585e-05,
+      "loss": 0.1658,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543704306896393e-05,
+      "loss": 0.1788,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542666403496232e-05,
+      "loss": 0.1601,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541627348637345e-05,
+      "loss": 0.1909,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540587142445112e-05,
+      "loss": 0.1769,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539545785045045e-05,
+      "loss": 0.1579,
+      "step": 1157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538503276562797e-05,
+      "loss": 0.176,
+      "step": 1158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953745961712416e-05,
+      "loss": 0.1595,
+      "step": 1159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536414806855074e-05,
+      "loss": 0.1438,
+      "step": 1160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535368845881604e-05,
+      "loss": 0.1505,
+      "step": 1161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953432173432996e-05,
+      "loss": 0.1718,
+      "step": 1162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533273472326497e-05,
+      "loss": 0.1596,
+      "step": 1163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532224059997693e-05,
+      "loss": 0.1558,
+      "step": 1164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531173497470185e-05,
+      "loss": 0.1614,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953012178487073e-05,
+      "loss": 0.1508,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529068922326236e-05,
+      "loss": 0.1722,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952801490996375e-05,
+      "loss": 0.1778,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526959747910444e-05,
+      "loss": 0.1897,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525903436293644e-05,
+      "loss": 0.1796,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524845975240806e-05,
+      "loss": 0.1541,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523787364879532e-05,
+      "loss": 0.156,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952272760533756e-05,
+      "loss": 0.1539,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952166669674276e-05,
+      "loss": 0.1767,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520604639223146e-05,
+      "loss": 0.1801,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951954143290687e-05,
+      "loss": 0.154,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951847707792223e-05,
+      "loss": 0.1847,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517411574397647e-05,
+      "loss": 0.1665,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516344922461694e-05,
+      "loss": 0.1691,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515277122243076e-05,
+      "loss": 0.1654,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514208173870638e-05,
+      "loss": 0.1714,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513138077473362e-05,
+      "loss": 0.1676,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951206683318038e-05,
+      "loss": 0.1543,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951099444112094e-05,
+      "loss": 0.1661,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509920901424445e-05,
+      "loss": 0.162,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950884621422044e-05,
+      "loss": 0.1678,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507770379638596e-05,
+      "loss": 0.1642,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506693397808724e-05,
+      "loss": 0.1649,
+      "step": 1188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505615268860786e-05,
+      "loss": 0.176,
+      "step": 1189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950453599292487e-05,
+      "loss": 0.1752,
+      "step": 1190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95034555701312e-05,
+      "loss": 0.1692,
+      "step": 1191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502374000610152e-05,
+      "loss": 0.1611,
+      "step": 1192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950129128449223e-05,
+      "loss": 0.1771,
+      "step": 1193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950020742190808e-05,
+      "loss": 0.1664,
+      "step": 1194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499122412988482e-05,
+      "loss": 0.1703,
+      "step": 1195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498036257864365e-05,
+      "loss": 0.1614,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949694895666678e-05,
+      "loss": 0.1696,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495860509526935e-05,
+      "loss": 0.1514,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494770916576158e-05,
+      "loss": 0.1661,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949368017794593e-05,
+      "loss": 0.1746,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949258829376786e-05,
+      "loss": 0.1715,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94914952641737e-05,
+      "loss": 0.1889,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949040108929534e-05,
+      "loss": 0.1665,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489305769264813e-05,
+      "loss": 0.1742,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488209304214277e-05,
+      "loss": 0.1653,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487111694276043e-05,
+      "loss": 0.1613,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486012939582548e-05,
+      "loss": 0.1681,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484913040266377e-05,
+      "loss": 0.1666,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483811996460243e-05,
+      "loss": 0.1601,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482709808297006e-05,
+      "loss": 0.1599,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948160647590966e-05,
+      "loss": 0.1773,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948050199943134e-05,
+      "loss": 0.1771,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947939637899531e-05,
+      "loss": 0.1562,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947828961473499e-05,
+      "loss": 0.1657,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477181706783917e-05,
+      "loss": 0.1706,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476072655275782e-05,
+      "loss": 0.1473,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474962460344404e-05,
+      "loss": 0.1679,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473851122123743e-05,
+      "loss": 0.1867,
+      "step": 1218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472738640747907e-05,
+      "loss": 0.1613,
+      "step": 1219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471625016351117e-05,
+      "loss": 0.172,
+      "step": 1220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947051024906776e-05,
+      "loss": 0.1792,
+      "step": 1221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469394339032348e-05,
+      "loss": 0.1595,
+      "step": 1222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468277286379527e-05,
+      "loss": 0.159,
+      "step": 1223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467159091244087e-05,
+      "loss": 0.1655,
+      "step": 1224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946603975376095e-05,
+      "loss": 0.1411,
+      "step": 1225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464919274065187e-05,
+      "loss": 0.1561,
+      "step": 1226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463797652291995e-05,
+      "loss": 0.1957,
+      "step": 1227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946267488857672e-05,
+      "loss": 0.1593,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946155098305483e-05,
+      "loss": 0.1656,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946042593586195e-05,
+      "loss": 0.1769,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459299747133823e-05,
+      "loss": 0.1514,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 0.1677,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945704394561555e-05,
+      "loss": 0.1775,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455914333097597e-05,
+      "loss": 0.1589,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454783579588788e-05,
+      "loss": 0.1734,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453651685225567e-05,
+      "loss": 0.1664,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945251865014452e-05,
+      "loss": 0.1678,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451384474482352e-05,
+      "loss": 0.1449,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450249158375926e-05,
+      "loss": 0.1559,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944911270196223e-05,
+      "loss": 0.1498,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447975105378396e-05,
+      "loss": 0.1712,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446836368761692e-05,
+      "loss": 0.1693,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944569649224952e-05,
+      "loss": 0.1571,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444555475979425e-05,
+      "loss": 0.1595,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443413320089082e-05,
+      "loss": 0.152,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442270024716313e-05,
+      "loss": 0.151,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944112558999907e-05,
+      "loss": 0.1644,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439980016075447e-05,
+      "loss": 0.1727,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 0.1592,
+      "step": 1249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437685451162122e-05,
+      "loss": 0.161,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943653646044929e-05,
+      "loss": 0.1569,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435386331083822e-05,
+      "loss": 0.1649,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434235063204497e-05,
+      "loss": 0.1557,
+      "step": 1253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433082656950232e-05,
+      "loss": 0.1668,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943192911246008e-05,
+      "loss": 0.1514,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430774429873235e-05,
+      "loss": 0.1668,
+      "step": 1256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429618609329028e-05,
+      "loss": 0.1766,
+      "step": 1257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942846165096692e-05,
+      "loss": 0.1717,
+      "step": 1258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9427303554926515e-05,
+      "loss": 0.1539,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426144321347557e-05,
+      "loss": 0.1496,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942498395036992e-05,
+      "loss": 0.1723,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942382244213363e-05,
+      "loss": 0.1578,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422659796778825e-05,
+      "loss": 0.1809,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94214960144458e-05,
+      "loss": 0.1799,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420331095274987e-05,
+      "loss": 0.176,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941916503940694e-05,
+      "loss": 0.1696,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417997846982374e-05,
+      "loss": 0.1604,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941682951814212e-05,
+      "loss": 0.1407,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415660053027148e-05,
+      "loss": 0.1453,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414489451778577e-05,
+      "loss": 0.1659,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413317714537657e-05,
+      "loss": 0.1633,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412144841445776e-05,
+      "loss": 0.1703,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941097083264445e-05,
+      "loss": 0.1679,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940979568827535e-05,
+      "loss": 0.161,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408619408480268e-05,
+      "loss": 0.1509,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407441993401137e-05,
+      "loss": 0.158,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406263443180035e-05,
+      "loss": 0.1649,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405083757959168e-05,
+      "loss": 0.1543,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940390293788088e-05,
+      "loss": 0.1771,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402720983087656e-05,
+      "loss": 0.1512,
+      "step": 1280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401537893722117e-05,
+      "loss": 0.1513,
+      "step": 1281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400353669927018e-05,
+      "loss": 0.1607,
+      "step": 1282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939916831184525e-05,
+      "loss": 0.1627,
+      "step": 1283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939798181961985e-05,
+      "loss": 0.1687,
+      "step": 1284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396794193393974e-05,
+      "loss": 0.1521,
+      "step": 1285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395605433310937e-05,
+      "loss": 0.1629,
+      "step": 1286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394415539514176e-05,
+      "loss": 0.2007,
+      "step": 1287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939322451214727e-05,
+      "loss": 0.1629,
+      "step": 1288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939203235135393e-05,
+      "loss": 0.1563,
+      "step": 1289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9390839057278005e-05,
+      "loss": 0.1673,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389644630063495e-05,
+      "loss": 0.168,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938844906985451e-05,
+      "loss": 0.1575,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938725237679532e-05,
+      "loss": 0.1573,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386054551030323e-05,
+      "loss": 0.1591,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938485559270405e-05,
+      "loss": 0.1633,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383655501961173e-05,
+      "loss": 0.1515,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382454278946503e-05,
+      "loss": 0.1646,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938125192380498e-05,
+      "loss": 0.1513,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380048436681695e-05,
+      "loss": 0.1731,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378843817721856e-05,
+      "loss": 0.1572,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937763806707082e-05,
+      "loss": 0.1722,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937643118487408e-05,
+      "loss": 0.1536,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937522317127726e-05,
+      "loss": 0.1672,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374014026426126e-05,
+      "loss": 0.1699,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372803750466577e-05,
+      "loss": 0.1893,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371592343544655e-05,
+      "loss": 0.1553,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370379805806528e-05,
+      "loss": 0.1776,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369166137398513e-05,
+      "loss": 0.1482,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936795133846705e-05,
+      "loss": 0.1704,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936673540915872e-05,
+      "loss": 0.1582,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936551834962025e-05,
+      "loss": 0.1509,
+      "step": 1311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936430015999849e-05,
+      "loss": 0.1526,
+      "step": 1312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363080840440432e-05,
+      "loss": 0.1611,
+      "step": 1313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361860391093207e-05,
+      "loss": 0.1586,
+      "step": 1314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360638812104073e-05,
+      "loss": 0.168,
+      "step": 1315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359416103620444e-05,
+      "loss": 0.1652,
+      "step": 1316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358192265789844e-05,
+      "loss": 0.1738,
+      "step": 1317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356967298759953e-05,
+      "loss": 0.1512,
+      "step": 1318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935574120267858e-05,
+      "loss": 0.1519,
+      "step": 1319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935451397769367e-05,
+      "loss": 0.1691,
+      "step": 1320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9353285623953304e-05,
+      "loss": 0.1753,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352056141605705e-05,
+      "loss": 0.1557,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350825530799223e-05,
+      "loss": 0.1639,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934959379168235e-05,
+      "loss": 0.1435,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934836092440371e-05,
+      "loss": 0.1769,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347126929112077e-05,
+      "loss": 0.1553,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345891805956334e-05,
+      "loss": 0.1473,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934465555508553e-05,
+      "loss": 0.1663,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934341817664883e-05,
+      "loss": 0.1644,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342179670795544e-05,
+      "loss": 0.1702,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934094003767511e-05,
+      "loss": 0.1698,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933969927743711e-05,
+      "loss": 0.1655,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338457390231267e-05,
+      "loss": 0.1732,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337214376207417e-05,
+      "loss": 0.1574,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335970235515563e-05,
+      "loss": 0.16,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334724968305818e-05,
+      "loss": 0.1604,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333478574728447e-05,
+      "loss": 0.1539,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933223105493384e-05,
+      "loss": 0.1653,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330982409072535e-05,
+      "loss": 0.1742,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932973263729519e-05,
+      "loss": 0.1556,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328481739752614e-05,
+      "loss": 0.1575,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327229716595745e-05,
+      "loss": 0.1565,
+      "step": 1342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325976567975658e-05,
+      "loss": 0.1683,
+      "step": 1343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 0.1598,
+      "step": 1344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323466894950796e-05,
+      "loss": 0.1677,
+      "step": 1345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322210370848856e-05,
+      "loss": 0.1643,
+      "step": 1346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932095272188935e-05,
+      "loss": 0.158,
+      "step": 1347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319693948224035e-05,
+      "loss": 0.1562,
+      "step": 1348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318434050004798e-05,
+      "loss": 0.1731,
+      "step": 1349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317173027383662e-05,
+      "loss": 0.1605,
+      "step": 1350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315910880512792e-05,
+      "loss": 0.1753,
+      "step": 1351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931464760954448e-05,
+      "loss": 0.1545,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931338321463116e-05,
+      "loss": 0.1629,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312117695925398e-05,
+      "loss": 0.1437,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310851053579897e-05,
+      "loss": 0.1782,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309583287747496e-05,
+      "loss": 0.1518,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930831439858117e-05,
+      "loss": 0.1679,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307044386234027e-05,
+      "loss": 0.1558,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930577325085931e-05,
+      "loss": 0.1868,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304500992610404e-05,
+      "loss": 0.1575,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303227611640823e-05,
+      "loss": 0.1553,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930195310810422e-05,
+      "loss": 0.173,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930067748215438e-05,
+      "loss": 0.1627,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299400733945226e-05,
+      "loss": 0.1608,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298122863630816e-05,
+      "loss": 0.1667,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296843871365343e-05,
+      "loss": 0.1637,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295563757303136e-05,
+      "loss": 0.164,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929428252159866e-05,
+      "loss": 0.159,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293000164406513e-05,
+      "loss": 0.1496,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929171668588143e-05,
+      "loss": 0.1799,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290432086178282e-05,
+      "loss": 0.1604,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928914636545207e-05,
+      "loss": 0.1541,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287859523857938e-05,
+      "loss": 0.1642,
+      "step": 1373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286571561551165e-05,
+      "loss": 0.1495,
+      "step": 1374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285282478687155e-05,
+      "loss": 0.1773,
+      "step": 1375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283992275421462e-05,
+      "loss": 0.1795,
+      "step": 1376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282700951909762e-05,
+      "loss": 0.1639,
+      "step": 1377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928140850830787e-05,
+      "loss": 0.1621,
+      "step": 1378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928011494477174e-05,
+      "loss": 0.1716,
+      "step": 1379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278820261457465e-05,
+      "loss": 0.1847,
+      "step": 1380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277524458521257e-05,
+      "loss": 0.1658,
+      "step": 1381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927622753611948e-05,
+      "loss": 0.1605,
+      "step": 1382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9274929494408622e-05,
+      "loss": 0.1543,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927363033354531e-05,
+      "loss": 0.1584,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927233005368631e-05,
+      "loss": 0.1554,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927102865498852e-05,
+      "loss": 0.141,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926972613760897e-05,
+      "loss": 0.1629,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268422501704825e-05,
+      "loss": 0.167,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926711774743339e-05,
+      "loss": 0.1933,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92658118749521e-05,
+      "loss": 0.162,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264504884418528e-05,
+      "loss": 0.1907,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263196775990388e-05,
+      "loss": 0.1655,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926188754982551e-05,
+      "loss": 0.1471,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260577206081876e-05,
+      "loss": 0.1531,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92592657449176e-05,
+      "loss": 0.1474,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257953166490928e-05,
+      "loss": 0.1668,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256639470960237e-05,
+      "loss": 0.1627,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255324658484048e-05,
+      "loss": 0.1708,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925400872922101e-05,
+      "loss": 0.1665,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252691683329907e-05,
+      "loss": 0.1708,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925137352096966e-05,
+      "loss": 0.1683,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925005424229933e-05,
+      "loss": 0.1796,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248733847478095e-05,
+      "loss": 0.1833,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924741233666529e-05,
+      "loss": 0.1548,
+      "step": 1404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924608971002037e-05,
+      "loss": 0.1668,
+      "step": 1405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244765967702928e-05,
+      "loss": 0.1715,
+      "step": 1406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243441109872694e-05,
+      "loss": 0.1681,
+      "step": 1407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924211513668953e-05,
+      "loss": 0.1722,
+      "step": 1408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240788048313436e-05,
+      "loss": 0.1681,
+      "step": 1409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923945984490454e-05,
+      "loss": 0.1577,
+      "step": 1410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923813052662311e-05,
+      "loss": 0.1863,
+      "step": 1411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923680009362955e-05,
+      "loss": 0.1821,
+      "step": 1412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9235468546084392e-05,
+      "loss": 0.1754,
+      "step": 1413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923413588414831e-05,
+      "loss": 0.1602,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232802107982103e-05,
+      "loss": 0.1449,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923146721774672e-05,
+      "loss": 0.1615,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923013121360322e-05,
+      "loss": 0.1761,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228794095712824e-05,
+      "loss": 0.154,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922745586423687e-05,
+      "loss": 0.1717,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922611651933683e-05,
+      "loss": 0.1524,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224776061174322e-05,
+      "loss": 0.1647,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922343448991109e-05,
+      "loss": 0.1665,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922209180570901e-05,
+      "loss": 0.1569,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220748008730096e-05,
+      "loss": 0.167,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92194030991365e-05,
+      "loss": 0.1636,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921805707709051e-05,
+      "loss": 0.1754,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921670994275453e-05,
+      "loss": 0.1711,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215361696291114e-05,
+      "loss": 0.1707,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214012337862952e-05,
+      "loss": 0.1694,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212661867632866e-05,
+      "loss": 0.1736,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92113102857638e-05,
+      "loss": 0.1652,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209957592418848e-05,
+      "loss": 0.1649,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920860378776123e-05,
+      "loss": 0.1734,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92072488719543e-05,
+      "loss": 0.1527,
+      "step": 1434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920589284516155e-05,
+      "loss": 0.1691,
+      "step": 1435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204535707546602e-05,
+      "loss": 0.1767,
+      "step": 1436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920317745927322e-05,
+      "loss": 0.1824,
+      "step": 1437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920181810050529e-05,
+      "loss": 0.1778,
+      "step": 1438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200457631406842e-05,
+      "loss": 0.1618,
+      "step": 1439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919909605214203e-05,
+      "loss": 0.1641,
+      "step": 1440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197733362875153e-05,
+      "loss": 0.1596,
+      "step": 1441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919636956377064e-05,
+      "loss": 0.1631,
+      "step": 1442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195004654993047e-05,
+      "loss": 0.1906,
+      "step": 1443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193638636707076e-05,
+      "loss": 0.1642,
+      "step": 1444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192271509077552e-05,
+      "loss": 0.1621,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190903272269438e-05,
+      "loss": 0.153,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189533926447836e-05,
+      "loss": 0.163,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188163471777974e-05,
+      "loss": 0.178,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918679190842522e-05,
+      "loss": 0.1631,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918541923655507e-05,
+      "loss": 0.1586,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918404545633315e-05,
+      "loss": 0.1706,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182670567925237e-05,
+      "loss": 0.1792,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181294571497228e-05,
+      "loss": 0.1785,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179917467215153e-05,
+      "loss": 0.1626,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178539255245182e-05,
+      "loss": 0.1803,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177159935753612e-05,
+      "loss": 0.1647,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175779508906888e-05,
+      "loss": 0.1619,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174397974871563e-05,
+      "loss": 0.1703,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917301533381435e-05,
+      "loss": 0.1537,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171631585902084e-05,
+      "loss": 0.1698,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917024673130173e-05,
+      "loss": 0.1413,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916886077018039e-05,
+      "loss": 0.1598,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91674737027053e-05,
+      "loss": 0.1609,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166085529043834e-05,
+      "loss": 0.1481,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916469624936349e-05,
+      "loss": 0.148,
+      "step": 1465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163305863831908e-05,
+      "loss": 0.1506,
+      "step": 1466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161914372616853e-05,
+      "loss": 0.154,
+      "step": 1467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160521775886237e-05,
+      "loss": 0.1863,
+      "step": 1468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159128073808087e-05,
+      "loss": 0.1594,
+      "step": 1469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157733266550577e-05,
+      "loss": 0.1587,
+      "step": 1470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915633735428201e-05,
+      "loss": 0.1602,
+      "step": 1471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154940337170824e-05,
+      "loss": 0.1483,
+      "step": 1472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153542215385588e-05,
+      "loss": 0.1831,
+      "step": 1473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152142989095007e-05,
+      "loss": 0.145,
+      "step": 1474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150742658467914e-05,
+      "loss": 0.1595,
+      "step": 1475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149341223673282e-05,
+      "loss": 0.1779,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147938684880213e-05,
+      "loss": 0.1585,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914653504225794e-05,
+      "loss": 0.1601,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914513029597584e-05,
+      "loss": 0.1476,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143724446203407e-05,
+      "loss": 0.1739,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142317493110287e-05,
+      "loss": 0.1559,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914090943686624e-05,
+      "loss": 0.1531,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139500277641173e-05,
+      "loss": 0.1598,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138090015605117e-05,
+      "loss": 0.1332,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136678650928244e-05,
+      "loss": 0.155,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135266183780854e-05,
+      "loss": 0.1519,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913385261433338e-05,
+      "loss": 0.1829,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132437942756393e-05,
+      "loss": 0.1694,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131022169220588e-05,
+      "loss": 0.1558,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91296052938968e-05,
+      "loss": 0.1845,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128187316956e-05,
+      "loss": 0.1679,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126768238569283e-05,
+      "loss": 0.1539,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912534805890788e-05,
+      "loss": 0.1645,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912392677814316e-05,
+      "loss": 0.1501,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122504396446615e-05,
+      "loss": 0.1588,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912108091398988e-05,
+      "loss": 0.1571,
+      "step": 1496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119656330944716e-05,
+      "loss": 0.1671,
+      "step": 1497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911823064748302e-05,
+      "loss": 0.1484,
+      "step": 1498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116803863776825e-05,
+      "loss": 0.1526,
+      "step": 1499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115375979998284e-05,
+      "loss": 0.1624,
+      "step": 1500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91139469963197e-05,
+      "loss": 0.1552,
+      "step": 1501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112516912913497e-05,
+      "loss": 0.172,
+      "step": 1502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111085729952235e-05,
+      "loss": 0.1556,
+      "step": 1503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109653447608607e-05,
+      "loss": 0.1577,
+      "step": 1504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910822006605544e-05,
+      "loss": 0.1646,
+      "step": 1505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106785585465687e-05,
+      "loss": 0.1593,
+      "step": 1506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105350006012438e-05,
+      "loss": 0.1968,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103913327868924e-05,
+      "loss": 0.1629,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102475551208493e-05,
+      "loss": 0.164,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910103667620464e-05,
+      "loss": 0.146,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909959670303098e-05,
+      "loss": 0.182,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098155631861272e-05,
+      "loss": 0.1521,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096713462869392e-05,
+      "loss": 0.1731,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909527019622937e-05,
+      "loss": 0.1636,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909382583211535e-05,
+      "loss": 0.1591,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092380370701617e-05,
+      "loss": 0.1712,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090933812162584e-05,
+      "loss": 0.1556,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90894861566728e-05,
+      "loss": 0.1857,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088037404406948e-05,
+      "loss": 0.1444,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908658755553984e-05,
+      "loss": 0.1879,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085136610246415e-05,
+      "loss": 0.1872,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083684568701757e-05,
+      "loss": 0.1532,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082231431081074e-05,
+      "loss": 0.1625,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080777197559707e-05,
+      "loss": 0.1652,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907932186831313e-05,
+      "loss": 0.1807,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077865443516948e-05,
+      "loss": 0.1487,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90764079233469e-05,
+      "loss": 0.1639,
+      "step": 1527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074949307978862e-05,
+      "loss": 0.1834,
+      "step": 1528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907348959758883e-05,
+      "loss": 0.1585,
+      "step": 1529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072028792352938e-05,
+      "loss": 0.1669,
+      "step": 1530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070566892447456e-05,
+      "loss": 0.1452,
+      "step": 1531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069103898048788e-05,
+      "loss": 0.1845,
+      "step": 1532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067639809333456e-05,
+      "loss": 0.1678,
+      "step": 1533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906617462647813e-05,
+      "loss": 0.16,
+      "step": 1534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064708349659598e-05,
+      "loss": 0.1712,
+      "step": 1535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063240979054796e-05,
+      "loss": 0.1591,
+      "step": 1536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061772514840775e-05,
+      "loss": 0.1368,
+      "step": 1537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9060302957194732e-05,
+      "loss": 0.1529,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058832306293987e-05,
+      "loss": 0.1638,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057360562315997e-05,
+      "loss": 0.1467,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055887725438348e-05,
+      "loss": 0.1536,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905441379583876e-05,
+      "loss": 0.1611,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052938773695082e-05,
+      "loss": 0.1864,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051462659185293e-05,
+      "loss": 0.1597,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049985452487516e-05,
+      "loss": 0.1735,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904850715377999e-05,
+      "loss": 0.1411,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047027763241093e-05,
+      "loss": 0.1574,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045547281049338e-05,
+      "loss": 0.1565,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044065707383364e-05,
+      "loss": 0.1678,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042583042421947e-05,
+      "loss": 0.1577,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041099286343988e-05,
+      "loss": 0.1465,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039614439328524e-05,
+      "loss": 0.1831,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038128501554723e-05,
+      "loss": 0.1454,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036641473201887e-05,
+      "loss": 0.157,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035153354449446e-05,
+      "loss": 0.1881,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903366414547696e-05,
+      "loss": 0.163,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032173846464124e-05,
+      "loss": 0.1606,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903068245759077e-05,
+      "loss": 0.1728,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029189979036846e-05,
+      "loss": 0.1659,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902769641098245e-05,
+      "loss": 0.1577,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026201753607792e-05,
+      "loss": 0.1548,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024706007093234e-05,
+      "loss": 0.1614,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023209171619252e-05,
+      "loss": 0.1655,
+      "step": 1563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021711247366463e-05,
+      "loss": 0.158,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9020212234515614e-05,
+      "loss": 0.1629,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018712133247577e-05,
+      "loss": 0.1509,
+      "step": 1566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017210943743373e-05,
+      "loss": 0.1579,
+      "step": 1567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901570866618413e-05,
+      "loss": 0.1607,
+      "step": 1568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014205300751122e-05,
+      "loss": 0.1707,
+      "step": 1569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901270084762575e-05,
+      "loss": 0.16,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011195306989553e-05,
+      "loss": 0.155,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.1538,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008180963911463e-05,
+      "loss": 0.192,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90066721618333e-05,
+      "loss": 0.1811,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900516227297175e-05,
+      "loss": 0.1566,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900365129750901e-05,
+      "loss": 0.1526,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90021392356274e-05,
+      "loss": 0.155,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000626087509376e-05,
+      "loss": 0.1778,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899911185333751e-05,
+      "loss": 0.1663,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8997596533294524e-05,
+      "loss": 0.1643,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996080127563258e-05,
+      "loss": 0.1532,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994562636326694e-05,
+      "loss": 0.1509,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993044059767935e-05,
+      "loss": 0.1544,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899152439807022e-05,
+      "loss": 0.151,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8990003651416916e-05,
+      "loss": 0.1292,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988481819991526e-05,
+      "loss": 0.164,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898695890397768e-05,
+      "loss": 0.1776,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985434903559138e-05,
+      "loss": 0.1728,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898390981891979e-05,
+      "loss": 0.1465,
+      "step": 1589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982383650243666e-05,
+      "loss": 0.1549,
+      "step": 1590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980856397714914e-05,
+      "loss": 0.1579,
+      "step": 1591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979328061517817e-05,
+      "loss": 0.1548,
+      "step": 1592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89777986418368e-05,
+      "loss": 0.1783,
+      "step": 1593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976268138856404e-05,
+      "loss": 0.1667,
+      "step": 1594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974736552761306e-05,
+      "loss": 0.1709,
+      "step": 1595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897320388373631e-05,
+      "loss": 0.1555,
+      "step": 1596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897167013196636e-05,
+      "loss": 0.1605,
+      "step": 1597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897013529763652e-05,
+      "loss": 0.1732,
+      "step": 1598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968599380931994e-05,
+      "loss": 0.1546,
+      "step": 1599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967062382038113e-05,
+      "loss": 0.1695,
+      "step": 1600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8965524301140334e-05,
+      "loss": 0.1337,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896398513842425e-05,
+      "loss": 0.1615,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962444894075582e-05,
+      "loss": 0.1511,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960903568280186e-05,
+      "loss": 0.1543,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959361161224038e-05,
+      "loss": 0.1738,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957817673093258e-05,
+      "loss": 0.16,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956273104074084e-05,
+      "loss": 0.1689,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89547274543529e-05,
+      "loss": 0.1648,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89531807241162e-05,
+      "loss": 0.1697,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951632913550625e-05,
+      "loss": 0.1497,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895008402284294e-05,
+      "loss": 0.1549,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948534052180038e-05,
+      "loss": 0.1597,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946983001748944e-05,
+      "loss": 0.1438,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945430871736818e-05,
+      "loss": 0.1386,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894387766233095e-05,
+      "loss": 0.1688,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894232337371875e-05,
+      "loss": 0.1669,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8940768006087764e-05,
+      "loss": 0.145,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939211559625676e-05,
+      "loss": 0.1649,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937654034520293e-05,
+      "loss": 0.1574,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936095430959545e-05,
+      "loss": 0.204,
+      "step": 1620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934535749131506e-05,
+      "loss": 0.1631,
+      "step": 1621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932974989224374e-05,
+      "loss": 0.1699,
+      "step": 1622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893141315142647e-05,
+      "loss": 0.1623,
+      "step": 1623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929850235926265e-05,
+      "loss": 0.147,
+      "step": 1624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928286242912337e-05,
+      "loss": 0.1674,
+      "step": 1625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926721172573405e-05,
+      "loss": 0.1691,
+      "step": 1626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925155025098318e-05,
+      "loss": 0.1548,
+      "step": 1627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8923587800676054e-05,
+      "loss": 0.1654,
+      "step": 1628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922019499495727e-05,
+      "loss": 0.1536,
+      "step": 1629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920450121746562e-05,
+      "loss": 0.1968,
+      "step": 1630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891887966761794e-05,
+      "loss": 0.1764,
+      "step": 1631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891730813729935e-05,
+      "loss": 0.1512,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891573553098042e-05,
+      "loss": 0.1614,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914161848850913e-05,
+      "loss": 0.1432,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912587091100715e-05,
+      "loss": 0.1627,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911011257919834e-05,
+      "loss": 0.1527,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890943434949843e-05,
+      "loss": 0.1531,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890785636602677e-05,
+      "loss": 0.1568,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890627730769526e-05,
+      "loss": 0.1572,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904697174694447e-05,
+      "loss": 0.1574,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903115967214986e-05,
+      "loss": 0.1645,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901533685447672e-05,
+      "loss": 0.1647,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899950329583435e-05,
+      "loss": 0.1541,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898365899813328e-05,
+      "loss": 0.1634,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896780396328532e-05,
+      "loss": 0.1462,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895193819320366e-05,
+      "loss": 0.1606,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889360616898027e-05,
+      "loss": 0.165,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892017445499812e-05,
+      "loss": 0.1663,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88904276490707e-05,
+      "loss": 0.1676,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888883677988477e-05,
+      "loss": 0.1621,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887244838133973e-05,
+      "loss": 0.1628,
+      "step": 1651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885651824010406e-05,
+      "loss": 0.1723,
+      "step": 1652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884057737706286e-05,
+      "loss": 0.171,
+      "step": 1653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882462579413962e-05,
+      "loss": 0.1546,
+      "step": 1654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880866349325916e-05,
+      "loss": 0.1598,
+      "step": 1655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887926904763475e-05,
+      "loss": 0.153,
+      "step": 1656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8877670674533205e-05,
+      "loss": 0.1799,
+      "step": 1657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887607123021415e-05,
+      "loss": 0.1601,
+      "step": 1658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874470714870578e-05,
+      "loss": 0.1458,
+      "step": 1659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887286912869561e-05,
+      "loss": 0.1608,
+      "step": 1660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887126647188251e-05,
+      "loss": 0.1626,
+      "step": 1661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886966274462465e-05,
+      "loss": 0.1644,
+      "step": 1662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8868057947115554e-05,
+      "loss": 0.1391,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886645207954885e-05,
+      "loss": 0.1443,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886484514211832e-05,
+      "loss": 0.1663,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863237135017865e-05,
+      "loss": 0.1655,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861628058441505e-05,
+      "loss": 0.154,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860017912583406e-05,
+      "loss": 0.1471,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885840669763785e-05,
+      "loss": 0.1449,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856794413799253e-05,
+      "loss": 0.142,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855181061262163e-05,
+      "loss": 0.1397,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853566640221253e-05,
+      "loss": 0.1566,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8851951150871324e-05,
+      "loss": 0.1609,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885033459340731e-05,
+      "loss": 0.1531,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848716968024274e-05,
+      "loss": 0.1688,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847098274917397e-05,
+      "loss": 0.1773,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845478514282008e-05,
+      "loss": 0.15,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843857686313548e-05,
+      "loss": 0.159,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842235791207594e-05,
+      "loss": 0.1556,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884061282915985e-05,
+      "loss": 0.1762,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838988800366152e-05,
+      "loss": 0.1556,
+      "step": 1681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883736370502246e-05,
+      "loss": 0.1666,
+      "step": 1682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8835737543324867e-05,
+      "loss": 0.1462,
+      "step": 1683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883411031546959e-05,
+      "loss": 0.1443,
+      "step": 1684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832482021652975e-05,
+      "loss": 0.1697,
+      "step": 1685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830852662071507e-05,
+      "loss": 0.1557,
+      "step": 1686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829222236921786e-05,
+      "loss": 0.1507,
+      "step": 1687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882759074640055e-05,
+      "loss": 0.1576,
+      "step": 1688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882595819070465e-05,
+      "loss": 0.1506,
+      "step": 1689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824324570031094e-05,
+      "loss": 0.164,
+      "step": 1690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822689884576987e-05,
+      "loss": 0.1452,
+      "step": 1691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882105413453959e-05,
+      "loss": 0.1609,
+      "step": 1692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819417320116266e-05,
+      "loss": 0.1479,
+      "step": 1693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881777944150453e-05,
+      "loss": 0.1672,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816140498902013e-05,
+      "loss": 0.1435,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8814500492506475e-05,
+      "loss": 0.1613,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812859422515804e-05,
+      "loss": 0.1695,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881121728912803e-05,
+      "loss": 0.1537,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809574092541286e-05,
+      "loss": 0.1568,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880792983295385e-05,
+      "loss": 0.1536,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880628451056413e-05,
+      "loss": 0.1518,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880463812557066e-05,
+      "loss": 0.1563,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880299067817209e-05,
+      "loss": 0.1564,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801342168567214e-05,
+      "loss": 0.1446,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799692596954947e-05,
+      "loss": 0.1631,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798041963534337e-05,
+      "loss": 0.1707,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879639026850455e-05,
+      "loss": 0.1672,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 0.1473,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879308369441479e-05,
+      "loss": 0.1398,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791428815753797e-05,
+      "loss": 0.1543,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789772876281606e-05,
+      "loss": 0.162,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788115876198018e-05,
+      "loss": 0.1536,
+      "step": 1712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786457815702987e-05,
+      "loss": 0.1505,
+      "step": 1713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878479869499657e-05,
+      "loss": 0.1644,
+      "step": 1714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783138514278968e-05,
+      "loss": 0.1566,
+      "step": 1715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8781477273750508e-05,
+      "loss": 0.1472,
+      "step": 1716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877981497361164e-05,
+      "loss": 0.1844,
+      "step": 1717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8778151614062944e-05,
+      "loss": 0.1502,
+      "step": 1718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877648719530513e-05,
+      "loss": 0.152,
+      "step": 1719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877482171753903e-05,
+      "loss": 0.145,
+      "step": 1720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773155180965613e-05,
+      "loss": 0.162,
+      "step": 1721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771487585785966e-05,
+      "loss": 0.1561,
+      "step": 1722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769818932201307e-05,
+      "loss": 0.1685,
+      "step": 1723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876814922041299e-05,
+      "loss": 0.1576,
+      "step": 1724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876647845062248e-05,
+      "loss": 0.1498,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764806623031388e-05,
+      "loss": 0.1459,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763133737841436e-05,
+      "loss": 0.1749,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876145979525449e-05,
+      "loss": 0.1379,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759784795472526e-05,
+      "loss": 0.1536,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875810873869766e-05,
+      "loss": 0.1622,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875643162513214e-05,
+      "loss": 0.1566,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875475345497832e-05,
+      "loss": 0.1554,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753074228438705e-05,
+      "loss": 0.1683,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751393945715913e-05,
+      "loss": 0.1316,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87497126070127e-05,
+      "loss": 0.1527,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8748030212531936e-05,
+      "loss": 0.1476,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874634676247663e-05,
+      "loss": 0.1421,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8744662257049908e-05,
+      "loss": 0.1472,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874297669645504e-05,
+      "loss": 0.1603,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8741290080895406e-05,
+      "loss": 0.1554,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739602410574523e-05,
+      "loss": 0.158,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873791368569603e-05,
+      "loss": 0.149,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736223906463698e-05,
+      "loss": 0.1522,
+      "step": 1743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873453307308142e-05,
+      "loss": 0.1415,
+      "step": 1744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732841185753224e-05,
+      "loss": 0.1545,
+      "step": 1745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731148244683253e-05,
+      "loss": 0.1633,
+      "step": 1746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729454250075797e-05,
+      "loss": 0.1603,
+      "step": 1747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727759202135246e-05,
+      "loss": 0.1471,
+      "step": 1748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872606310106614e-05,
+      "loss": 0.15,
+      "step": 1749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8724365947073136e-05,
+      "loss": 0.1544,
+      "step": 1750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8722667740361022e-05,
+      "loss": 0.1492,
+      "step": 1751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720968481134706e-05,
+      "loss": 0.157,
+      "step": 1752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8719268169599233e-05,
+      "loss": 0.1845,
+      "step": 1753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717566805959766e-05,
+      "loss": 0.1573,
+      "step": 1754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87158643904216e-05,
+      "loss": 0.1581,
+      "step": 1755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8714160923190156e-05,
+      "loss": 0.1468,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712456404470982e-05,
+      "loss": 0.163,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871075083446975e-05,
+      "loss": 0.1567,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8709044213392265e-05,
+      "loss": 0.1671,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707336541444454e-05,
+      "loss": 0.1613,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870562781883237e-05,
+      "loss": 0.1463,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703918045762197e-05,
+      "loss": 0.1542,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870220722244024e-05,
+      "loss": 0.1593,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870049534907294e-05,
+      "loss": 0.1566,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698782425866857e-05,
+      "loss": 0.1595,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697068453028677e-05,
+      "loss": 0.1517,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695353430765214e-05,
+      "loss": 0.1563,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869363735928342e-05,
+      "loss": 0.1549,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869192023879035e-05,
+      "loss": 0.1698,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690202069493204e-05,
+      "loss": 0.1739,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868848285159931e-05,
+      "loss": 0.1642,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686762585316102e-05,
+      "loss": 0.1663,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868504127085117e-05,
+      "loss": 0.1448,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8683318908412208e-05,
+      "loss": 0.1426,
+      "step": 1774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681595498207047e-05,
+      "loss": 0.1598,
+      "step": 1775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679871040443632e-05,
+      "loss": 0.1491,
+      "step": 1776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678145535330054e-05,
+      "loss": 0.1408,
+      "step": 1777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676418983074515e-05,
+      "loss": 0.1679,
+      "step": 1778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674691383885345e-05,
+      "loss": 0.1553,
+      "step": 1779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867296273797101e-05,
+      "loss": 0.1569,
+      "step": 1780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867123304554009e-05,
+      "loss": 0.1548,
+      "step": 1781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8669502306801304e-05,
+      "loss": 0.1653,
+      "step": 1782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8667770521963487e-05,
+      "loss": 0.153,
+      "step": 1783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666037691235597e-05,
+      "loss": 0.1514,
+      "step": 1784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664303814826734e-05,
+      "loss": 0.1527,
+      "step": 1785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866256889294611e-05,
+      "loss": 0.1455,
+      "step": 1786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866083292580307e-05,
+      "loss": 0.167,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659095913607083e-05,
+      "loss": 0.1544,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8657357856567744e-05,
+      "loss": 0.1842,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655618754894774e-05,
+      "loss": 0.1448,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865387860879802e-05,
+      "loss": 0.1863,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652137418487454e-05,
+      "loss": 0.1545,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865039518417318e-05,
+      "loss": 0.1426,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648651906065424e-05,
+      "loss": 0.1487,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864690758437453e-05,
+      "loss": 0.1455,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645162219310977e-05,
+      "loss": 0.1551,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643415811085373e-05,
+      "loss": 0.151,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641668359908447e-05,
+      "loss": 0.1527,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863991986599105e-05,
+      "loss": 0.1439,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638170329544164e-05,
+      "loss": 0.1529,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636419750778897e-05,
+      "loss": 0.1498,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8634668129906477e-05,
+      "loss": 0.1591,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8632915467138267e-05,
+      "loss": 0.1686,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863116176268575e-05,
+      "loss": 0.1539,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629407016760533e-05,
+      "loss": 0.1674,
+      "step": 1805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627651229574356e-05,
+      "loss": 0.1734,
+      "step": 1806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862589440133907e-05,
+      "loss": 0.1677,
+      "step": 1807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624136532266673e-05,
+      "loss": 0.1607,
+      "step": 1808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622377622569267e-05,
+      "loss": 0.145,
+      "step": 1809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620617672459097e-05,
+      "loss": 0.1585,
+      "step": 1810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861885668214852e-05,
+      "loss": 0.1469,
+      "step": 1811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617094651850032e-05,
+      "loss": 0.1566,
+      "step": 1812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861533158177624e-05,
+      "loss": 0.1539,
+      "step": 1813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861356747213989e-05,
+      "loss": 0.157,
+      "step": 1814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861180232315384e-05,
+      "loss": 0.1577,
+      "step": 1815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610036135031084e-05,
+      "loss": 0.1778,
+      "step": 1816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608268907984735e-05,
+      "loss": 0.1585,
+      "step": 1817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8606500642228037e-05,
+      "loss": 0.1561,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860473133797436e-05,
+      "loss": 0.1464,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8602960995437187e-05,
+      "loss": 0.1556,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860118961483014e-05,
+      "loss": 0.1504,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599417196366967e-05,
+      "loss": 0.1539,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8597643740261528e-05,
+      "loss": 0.1501,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8595869246727814e-05,
+      "loss": 0.1558,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594093715979948e-05,
+      "loss": 0.132,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859231714823217e-05,
+      "loss": 0.1586,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590539543698852e-05,
+      "loss": 0.1494,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588760902594485e-05,
+      "loss": 0.152,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8586981225133688e-05,
+      "loss": 0.1413,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585200511531203e-05,
+      "loss": 0.1567,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85834187620019e-05,
+      "loss": 0.1673,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858163597676077e-05,
+      "loss": 0.1551,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8579852156022934e-05,
+      "loss": 0.156,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8578067300003634e-05,
+      "loss": 0.1458,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8576281408918242e-05,
+      "loss": 0.1511,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857449448298225e-05,
+      "loss": 0.1547,
+      "step": 1836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857270652241127e-05,
+      "loss": 0.1637,
+      "step": 1837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857091752742105e-05,
+      "loss": 0.1502,
+      "step": 1838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569127498227458e-05,
+      "loss": 0.154,
+      "step": 1839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567336435046482e-05,
+      "loss": 0.1471,
+      "step": 1840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565544338094245e-05,
+      "loss": 0.1584,
+      "step": 1841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856375120758699e-05,
+      "loss": 0.1455,
+      "step": 1842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8561957043741078e-05,
+      "loss": 0.1661,
+      "step": 1843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8560161846773002e-05,
+      "loss": 0.1476,
+      "step": 1844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855836561689938e-05,
+      "loss": 0.146,
+      "step": 1845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556568354336955e-05,
+      "loss": 0.155,
+      "step": 1846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554770059302588e-05,
+      "loss": 0.149,
+      "step": 1847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552970732013267e-05,
+      "loss": 0.159,
+      "step": 1848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8551170372686112e-05,
+      "loss": 0.1487,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854936898153836e-05,
+      "loss": 0.1443,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8547566558787373e-05,
+      "loss": 0.1505,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545763104650643e-05,
+      "loss": 0.1623,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543958619345777e-05,
+      "loss": 0.1447,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542153103090515e-05,
+      "loss": 0.1581,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854034655610272e-05,
+      "loss": 0.1529,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538538978600372e-05,
+      "loss": 0.1579,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8536730370801585e-05,
+      "loss": 0.1516,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534920732924594e-05,
+      "loss": 0.1554,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533110065187755e-05,
+      "loss": 0.1582,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8531298367809557e-05,
+      "loss": 0.1521,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85294856410086e-05,
+      "loss": 0.1473,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527671885003614e-05,
+      "loss": 0.1622,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525857100013463e-05,
+      "loss": 0.1519,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852404128625712e-05,
+      "loss": 0.1788,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852222444395369e-05,
+      "loss": 0.1525,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85204065733224e-05,
+      "loss": 0.1559,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851858767458261e-05,
+      "loss": 0.1553,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516767747953784e-05,
+      "loss": 0.1451,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851494679365553e-05,
+      "loss": 0.1543,
+      "step": 1869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513124811907568e-05,
+      "loss": 0.1809,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511301802929747e-05,
+      "loss": 0.1398,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8509477766942045e-05,
+      "loss": 0.1482,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850765270416455e-05,
+      "loss": 0.1358,
+      "step": 1873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505826614817485e-05,
+      "loss": 0.1619,
+      "step": 1874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503999499121193e-05,
+      "loss": 0.1388,
+      "step": 1875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 0.168,
+      "step": 1876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500342189562926e-05,
+      "loss": 0.1561,
+      "step": 1877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498511996142255e-05,
+      "loss": 0.1544,
+      "step": 1878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849668077725497e-05,
+      "loss": 0.1484,
+      "step": 1879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8494848533122035e-05,
+      "loss": 0.1733,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493015263964535e-05,
+      "loss": 0.1446,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849118097000368e-05,
+      "loss": 0.1509,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489345651460804e-05,
+      "loss": 0.1649,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848750930855737e-05,
+      "loss": 0.1687,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485671941514948e-05,
+      "loss": 0.1696,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483833550555252e-05,
+      "loss": 0.1538,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481994135900104e-05,
+      "loss": 0.1979,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480153697771455e-05,
+      "loss": 0.1395,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478312236391386e-05,
+      "loss": 0.1593,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8476469751982095e-05,
+      "loss": 0.1486,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8474626244765897e-05,
+      "loss": 0.1537,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472781714965242e-05,
+      "loss": 0.1569,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470936162802702e-05,
+      "loss": 0.1413,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469089588500964e-05,
+      "loss": 0.142,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 0.1484,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8465393374371282e-05,
+      "loss": 0.1387,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846354373498934e-05,
+      "loss": 0.1708,
+      "step": 1897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461693074360207e-05,
+      "loss": 0.1588,
+      "step": 1898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459841392707186e-05,
+      "loss": 0.1609,
+      "step": 1899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845798869025371e-05,
+      "loss": 0.1892,
+      "step": 1900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845613496722334e-05,
+      "loss": 0.1441,
+      "step": 1901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454280223839745e-05,
+      "loss": 0.1529,
+      "step": 1902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845242446032673e-05,
+      "loss": 0.1494,
+      "step": 1903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8450567676908222e-05,
+      "loss": 0.1658,
+      "step": 1904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448709873808265e-05,
+      "loss": 0.1515,
+      "step": 1905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446851051251027e-05,
+      "loss": 0.1469,
+      "step": 1906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8444991209460808e-05,
+      "loss": 0.1571,
+      "step": 1907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443130348662022e-05,
+      "loss": 0.1727,
+      "step": 1908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8441268469079204e-05,
+      "loss": 0.1554,
+      "step": 1909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8439405570937027e-05,
+      "loss": 0.1505,
+      "step": 1910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437541654460264e-05,
+      "loss": 0.1454,
+      "step": 1911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8435676719873828e-05,
+      "loss": 0.1576,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843381076740275e-05,
+      "loss": 0.1425,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8431943797272187e-05,
+      "loss": 0.1536,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843007580970741e-05,
+      "loss": 0.1456,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8428206804933825e-05,
+      "loss": 0.1527,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426336783176945e-05,
+      "loss": 0.1503,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842446574466242e-05,
+      "loss": 0.1709,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842259368961602e-05,
+      "loss": 0.1565,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420720618263632e-05,
+      "loss": 0.1613,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841884653083127e-05,
+      "loss": 0.168,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416971427545065e-05,
+      "loss": 0.1565,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841509530863128e-05,
+      "loss": 0.1611,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413218174316295e-05,
+      "loss": 0.1586,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841134002482661e-05,
+      "loss": 0.163,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8409460860388852e-05,
+      "loss": 0.1512,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8407580681229773e-05,
+      "loss": 0.1552,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405699487576237e-05,
+      "loss": 0.1499,
+      "step": 1928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403817279655237e-05,
+      "loss": 0.1446,
+      "step": 1929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8401934057693895e-05,
+      "loss": 0.1417,
+      "step": 1930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400049821919444e-05,
+      "loss": 0.1567,
+      "step": 1931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8398164572559243e-05,
+      "loss": 0.1497,
+      "step": 1932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8396278309840778e-05,
+      "loss": 0.1712,
+      "step": 1933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839439103399165e-05,
+      "loss": 0.1509,
+      "step": 1934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8392502745239586e-05,
+      "loss": 0.1523,
+      "step": 1935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390613443812432e-05,
+      "loss": 0.1433,
+      "step": 1936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838872312993817e-05,
+      "loss": 0.1421,
+      "step": 1937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838683180384488e-05,
+      "loss": 0.1479,
+      "step": 1938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8384939465760786e-05,
+      "loss": 0.153,
+      "step": 1939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383046115914225e-05,
+      "loss": 0.1549,
+      "step": 1940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8381151754533655e-05,
+      "loss": 0.1555,
+      "step": 1941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8379256381847655e-05,
+      "loss": 0.1624,
+      "step": 1942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837735999808493e-05,
+      "loss": 0.1369,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837546260347431e-05,
+      "loss": 0.1437,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837356419824474e-05,
+      "loss": 0.1335,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371664782625287e-05,
+      "loss": 0.1499,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369764356845148e-05,
+      "loss": 0.1392,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836786292113363e-05,
+      "loss": 0.1502,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365960475720172e-05,
+      "loss": 0.1486,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364057020834336e-05,
+      "loss": 0.1608,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362152556705788e-05,
+      "loss": 0.1545,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360247083564343e-05,
+      "loss": 0.1561,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358340601639912e-05,
+      "loss": 0.1637,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356433111162547e-05,
+      "loss": 0.158,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835452461236241e-05,
+      "loss": 0.1463,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835261510546979e-05,
+      "loss": 0.1505,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350704590715097e-05,
+      "loss": 0.1835,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8348793068328864e-05,
+      "loss": 0.1558,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346880538541736e-05,
+      "loss": 0.1381,
+      "step": 1959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834496700158449e-05,
+      "loss": 0.1623,
+      "step": 1960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8343052457688025e-05,
+      "loss": 0.135,
+      "step": 1961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341136907083356e-05,
+      "loss": 0.1563,
+      "step": 1962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339220350001624e-05,
+      "loss": 0.1566,
+      "step": 1963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833730278667408e-05,
+      "loss": 0.1276,
+      "step": 1964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8335384217332114e-05,
+      "loss": 0.1493,
+      "step": 1965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833346464220723e-05,
+      "loss": 0.137,
+      "step": 1966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331544061531044e-05,
+      "loss": 0.1507,
+      "step": 1967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8329622475535306e-05,
+      "loss": 0.1723,
+      "step": 1968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327699884451883e-05,
+      "loss": 0.1538,
+      "step": 1969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325776288512765e-05,
+      "loss": 0.1562,
+      "step": 1970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8323851687950055e-05,
+      "loss": 0.173,
+      "step": 1971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832192608299599e-05,
+      "loss": 0.1808,
+      "step": 1972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.831999947388292e-05,
+      "loss": 0.1471,
+      "step": 1973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8318071860843315e-05,
+      "loss": 0.1558,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8316143244109773e-05,
+      "loss": 0.1334,
+      "step": 1975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314213623915003e-05,
+      "loss": 0.1833,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8312283000491846e-05,
+      "loss": 0.1631,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8310351374073262e-05,
+      "loss": 0.1622,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308418744892322e-05,
+      "loss": 0.1587,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830648511318223e-05,
+      "loss": 0.1607,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304550479176307e-05,
+      "loss": 0.1554,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830261484310799e-05,
+      "loss": 0.1463,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300678205210844e-05,
+      "loss": 0.1455,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829874056571855e-05,
+      "loss": 0.161,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8296801924864914e-05,
+      "loss": 0.158,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294862282883864e-05,
+      "loss": 0.1592,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8292921640009438e-05,
+      "loss": 0.1738,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290979996475804e-05,
+      "loss": 0.185,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289037352517252e-05,
+      "loss": 0.1414,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828709370836819e-05,
+      "loss": 0.1491,
+      "step": 1990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828514906426314e-05,
+      "loss": 0.1508,
+      "step": 1991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828320342043676e-05,
+      "loss": 0.1509,
+      "step": 1992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828125677712381e-05,
+      "loss": 0.1685,
+      "step": 1993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279309134559187e-05,
+      "loss": 0.165,
+      "step": 1994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277360492977908e-05,
+      "loss": 0.1555,
+      "step": 1995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275410852615086e-05,
+      "loss": 0.1489,
+      "step": 1996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827346021370599e-05,
+      "loss": 0.1495,
+      "step": 1997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271508576485987e-05,
+      "loss": 0.143,
+      "step": 1998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269555941190565e-05,
+      "loss": 0.1652,
+      "step": 1999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267602308055342e-05,
+      "loss": 0.1718,
+      "step": 2000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826564767731605e-05,
+      "loss": 0.1521,
+      "step": 2001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826369204920855e-05,
+      "loss": 0.1804,
+      "step": 2002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261735423968804e-05,
+      "loss": 0.1706,
+      "step": 2003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8259777801832916e-05,
+      "loss": 0.148,
+      "step": 2004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8257819183037093e-05,
+      "loss": 0.1369,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825585956781768e-05,
+      "loss": 0.1426,
+      "step": 2006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253898956411124e-05,
+      "loss": 0.1523,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825193734905401e-05,
+      "loss": 0.14,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8249974745983023e-05,
+      "loss": 0.1445,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248011147434987e-05,
+      "loss": 0.1684,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824604655364684e-05,
+      "loss": 0.154,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824408096485563e-05,
+      "loss": 0.172,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8242114381298533e-05,
+      "loss": 0.1513,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240146803212854e-05,
+      "loss": 0.1357,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238178230836005e-05,
+      "loss": 0.1371,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823620866440552e-05,
+      "loss": 0.1575,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823423810415906e-05,
+      "loss": 0.151,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232266550334398e-05,
+      "loss": 0.1449,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823029400316943e-05,
+      "loss": 0.1474,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8228320462902172e-05,
+      "loss": 0.1444,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226345929770758e-05,
+      "loss": 0.1487,
+      "step": 2021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822437040401345e-05,
+      "loss": 0.1436,
+      "step": 2022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822239388586862e-05,
+      "loss": 0.1501,
+      "step": 2023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220416375574757e-05,
+      "loss": 0.141,
+      "step": 2024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8218437873370488e-05,
+      "loss": 0.1842,
+      "step": 2025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216458379494538e-05,
+      "loss": 0.142,
+      "step": 2026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8214477894185764e-05,
+      "loss": 0.1627,
+      "step": 2027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8212496417683135e-05,
+      "loss": 0.1458,
+      "step": 2028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210513950225755e-05,
+      "loss": 0.1524,
+      "step": 2029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820853049205283e-05,
+      "loss": 0.1469,
+      "step": 2030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820654604340369e-05,
+      "loss": 0.1686,
+      "step": 2031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204560604517795e-05,
+      "loss": 0.1405,
+      "step": 2032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202574175634707e-05,
+      "loss": 0.1636,
+      "step": 2033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200586756994126e-05,
+      "loss": 0.1289,
+      "step": 2034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198598348835856e-05,
+      "loss": 0.148,
+      "step": 2035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8196608951399834e-05,
+      "loss": 0.1481,
+      "step": 2036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81946185649261e-05,
+      "loss": 0.1516,
+      "step": 2037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192627189654827e-05,
+      "loss": 0.1505,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190634825826303e-05,
+      "loss": 0.1576,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8188641473680936e-05,
+      "loss": 0.1986,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186647133459248e-05,
+      "loss": 0.1524,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818465180540189e-05,
+      "loss": 0.1473,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818265548974962e-05,
+      "loss": 0.1474,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8180658186743333e-05,
+      "loss": 0.158,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178659896624023e-05,
+      "loss": 0.1508,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817666061963281e-05,
+      "loss": 0.1678,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8174660356010944e-05,
+      "loss": 0.1682,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817265910599978e-05,
+      "loss": 0.1563,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81706568698408e-05,
+      "loss": 0.1496,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81686536477756e-05,
+      "loss": 0.1478,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8166649440045902e-05,
+      "loss": 0.1463,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164644246893532e-05,
+      "loss": 0.1567,
+      "step": 2052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162638068560455e-05,
+      "loss": 0.1527,
+      "step": 2053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816063090528874e-05,
+      "loss": 0.1429,
+      "step": 2054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8158622757320584e-05,
+      "loss": 0.1504,
+      "step": 2055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8156613624898293e-05,
+      "loss": 0.1535,
+      "step": 2056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81546035082643e-05,
+      "loss": 0.1626,
+      "step": 2057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815259240766116e-05,
+      "loss": 0.1671,
+      "step": 2058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815058032333153e-05,
+      "loss": 0.1744,
+      "step": 2059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8148567255518203e-05,
+      "loss": 0.1742,
+      "step": 2060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814655320446409e-05,
+      "loss": 0.1456,
+      "step": 2061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81445381704122e-05,
+      "loss": 0.1492,
+      "step": 2062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814252215360569e-05,
+      "loss": 0.1518,
+      "step": 2063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814050515428781e-05,
+      "loss": 0.1594,
+      "step": 2064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813848717270195e-05,
+      "loss": 0.1426,
+      "step": 2065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81364682090916e-05,
+      "loss": 0.1656,
+      "step": 2066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8134448263700383e-05,
+      "loss": 0.1518,
+      "step": 2067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132427336772033e-05,
+      "loss": 0.1606,
+      "step": 2068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81304054285504e-05,
+      "loss": 0.1568,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8128382539279455e-05,
+      "loss": 0.1588,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8126358669203298e-05,
+      "loss": 0.1388,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124333818566124e-05,
+      "loss": 0.1507,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812230798761227e-05,
+      "loss": 0.1457,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8120281176586178e-05,
+      "loss": 0.145,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118253385732415e-05,
+      "loss": 0.1454,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811622461529566e-05,
+      "loss": 0.1496,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114194865520712e-05,
+      "loss": 0.1574,
+      "step": 2077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112164136652487e-05,
+      "loss": 0.1502,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811013242893603e-05,
+      "loss": 0.1675,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108099742616486e-05,
+      "loss": 0.1488,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8106066077939137e-05,
+      "loss": 0.1423,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104031435149366e-05,
+      "loss": 0.1676,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8101995814492683e-05,
+      "loss": 0.1569,
+      "step": 2083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809995921621472e-05,
+      "loss": 0.1455,
+      "step": 2084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809792164056121e-05,
+      "loss": 0.1444,
+      "step": 2085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809588308777803e-05,
+      "loss": 0.1525,
+      "step": 2086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809384355811115e-05,
+      "loss": 0.1471,
+      "step": 2087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8091803051806676e-05,
+      "loss": 0.1513,
+      "step": 2088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8089761569110816e-05,
+      "loss": 0.147,
+      "step": 2089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808771911026991e-05,
+      "loss": 0.1863,
+      "step": 2090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808567567553041e-05,
+      "loss": 0.1497,
+      "step": 2091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083631265138875e-05,
+      "loss": 0.1565,
+      "step": 2092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081585879342008e-05,
+      "loss": 0.1469,
+      "step": 2093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079539518386605e-05,
+      "loss": 0.1648,
+      "step": 2094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077492182519587e-05,
+      "loss": 0.1506,
+      "step": 2095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075443871988e-05,
+      "loss": 0.168,
+      "step": 2096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8073394587039e-05,
+      "loss": 0.1476,
+      "step": 2097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807134432791986e-05,
+      "loss": 0.1575,
+      "step": 2098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069293094877974e-05,
+      "loss": 0.1414,
+      "step": 2099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067240888160854e-05,
+      "loss": 0.1505,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806518770801613e-05,
+      "loss": 0.151,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806313355469154e-05,
+      "loss": 0.1502,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061078428434956e-05,
+      "loss": 0.1503,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805902232949435e-05,
+      "loss": 0.1351,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056965258117825e-05,
+      "loss": 0.1612,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054907214553593e-05,
+      "loss": 0.1492,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805284819904999e-05,
+      "loss": 0.1435,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050788211855463e-05,
+      "loss": 0.1756,
+      "step": 2108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8048727253218578e-05,
+      "loss": 0.1579,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046665323388023e-05,
+      "loss": 0.1557,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044602422612593e-05,
+      "loss": 0.1447,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042538551141207e-05,
+      "loss": 0.1718,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804047370922291e-05,
+      "loss": 0.15,
+      "step": 2113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8038407897106844e-05,
+      "loss": 0.1448,
+      "step": 2114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036341115042284e-05,
+      "loss": 0.1467,
+      "step": 2115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8034273363278615e-05,
+      "loss": 0.1485,
+      "step": 2116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8032204642065337e-05,
+      "loss": 0.1514,
+      "step": 2117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030134951652082e-05,
+      "loss": 0.1504,
+      "step": 2118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028064292288576e-05,
+      "loss": 0.1518,
+      "step": 2119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025992664224675e-05,
+      "loss": 0.1785,
+      "step": 2120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802392006771036e-05,
+      "loss": 0.1414,
+      "step": 2121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802184650299571e-05,
+      "loss": 0.1592,
+      "step": 2122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801977197033093e-05,
+      "loss": 0.1502,
+      "step": 2123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801769646996635e-05,
+      "loss": 0.1528,
+      "step": 2124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015620002152398e-05,
+      "loss": 0.1541,
+      "step": 2125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8013542567139635e-05,
+      "loss": 0.1551,
+      "step": 2126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8011464165178736e-05,
+      "loss": 0.1503,
+      "step": 2127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009384796520487e-05,
+      "loss": 0.1598,
+      "step": 2128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8007304461415794e-05,
+      "loss": 0.1721,
+      "step": 2129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005223160115676e-05,
+      "loss": 0.1659,
+      "step": 2130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003140892871277e-05,
+      "loss": 0.1595,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001057659933844e-05,
+      "loss": 0.1661,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799897346155476e-05,
+      "loss": 0.1483,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996888297985504e-05,
+      "loss": 0.145,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7994802169477684e-05,
+      "loss": 0.1477,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7992715076283024e-05,
+      "loss": 0.1723,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990627018653353e-05,
+      "loss": 0.1571,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988537996840634e-05,
+      "loss": 0.1711,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7986448011096937e-05,
+      "loss": 0.1451,
+      "step": 2139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984357061674443e-05,
+      "loss": 0.1467,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7982265148825454e-05,
+      "loss": 0.1531,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980172272802398e-05,
+      "loss": 0.1601,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79780784338578e-05,
+      "loss": 0.1502,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7975983632244317e-05,
+      "loss": 0.1415,
+      "step": 2144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797388786821472e-05,
+      "loss": 0.155,
+      "step": 2145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971791142021887e-05,
+      "loss": 0.1346,
+      "step": 2146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969693453918816e-05,
+      "loss": 0.1414,
+      "step": 2147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796759480415863e-05,
+      "loss": 0.1453,
+      "step": 2148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7965495192994555e-05,
+      "loss": 0.1666,
+      "step": 2149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7963394620679945e-05,
+      "loss": 0.138,
+      "step": 2150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7961293087468255e-05,
+      "loss": 0.1438,
+      "step": 2151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7959190593613075e-05,
+      "loss": 0.1381,
+      "step": 2152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7957087139368096e-05,
+      "loss": 0.146,
+      "step": 2153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954982724987132e-05,
+      "loss": 0.1614,
+      "step": 2154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952877350724108e-05,
+      "loss": 0.1427,
+      "step": 2155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950771016833064e-05,
+      "loss": 0.1408,
+      "step": 2156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7948663723568166e-05,
+      "loss": 0.1524,
+      "step": 2157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7946555471183685e-05,
+      "loss": 0.1649,
+      "step": 2158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7944446259934018e-05,
+      "loss": 0.1498,
+      "step": 2159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794233609007366e-05,
+      "loss": 0.1995,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940224961857242e-05,
+      "loss": 0.132,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79381128755395e-05,
+      "loss": 0.135,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793599983137529e-05,
+      "loss": 0.1405,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793388582961957e-05,
+      "loss": 0.1467,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931770870527437e-05,
+      "loss": 0.1484,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929654954354085e-05,
+      "loss": 0.1343,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927538081354826e-05,
+      "loss": 0.144,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79254202517851e-05,
+      "loss": 0.1647,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7923301465900448e-05,
+      "loss": 0.1532,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792118172395653e-05,
+      "loss": 0.1523,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919061026209126e-05,
+      "loss": 0.1296,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791693937291413e-05,
+      "loss": 0.1567,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914816764327546e-05,
+      "loss": 0.1478,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79126932007055e-05,
+      "loss": 0.1413,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910568682304226e-05,
+      "loss": 0.1455,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790844320938008e-05,
+      "loss": 0.1436,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790631678218953e-05,
+      "loss": 0.1697,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7904189400989165e-05,
+      "loss": 0.1532,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7902061066035677e-05,
+      "loss": 0.1335,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789993177758588e-05,
+      "loss": 0.1576,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7897801535896708e-05,
+      "loss": 0.1528,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78956703412252e-05,
+      "loss": 0.1575,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7893538193828523e-05,
+      "loss": 0.1417,
+      "step": 2183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 0.1551,
+      "step": 2184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7889271041888844e-05,
+      "loss": 0.1526,
+      "step": 2185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887136037860742e-05,
+      "loss": 0.1411,
+      "step": 2186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7885000082137248e-05,
+      "loss": 0.1431,
+      "step": 2187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882863174976105e-05,
+      "loss": 0.1597,
+      "step": 2188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7880725316635147e-05,
+      "loss": 0.1535,
+      "step": 2189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787858650737235e-05,
+      "loss": 0.1374,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7876446747445782e-05,
+      "loss": 0.1682,
+      "step": 2191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874306037113642e-05,
+      "loss": 0.1386,
+      "step": 2192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7872164376634236e-05,
+      "loss": 0.1605,
+      "step": 2193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870021766265985e-05,
+      "loss": 0.1497,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7867878206267422e-05,
+      "loss": 0.1714,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7865733696897205e-05,
+      "loss": 0.1505,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78635882384141e-05,
+      "loss": 0.1458,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861441831076975e-05,
+      "loss": 0.1349,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7859294475144837e-05,
+      "loss": 0.1707,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785714617087679e-05,
+      "loss": 0.1366,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785499691853206e-05,
+      "loss": 0.1485,
+      "step": 2201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785284671836998e-05,
+      "loss": 0.1424,
+      "step": 2202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785069557065001e-05,
+      "loss": 0.1506,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784854347563171e-05,
+      "loss": 0.1347,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846390433574768e-05,
+      "loss": 0.1487,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784423644473897e-05,
+      "loss": 0.1693,
+      "step": 2206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784208150938423e-05,
+      "loss": 0.1468,
+      "step": 2207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783992562777057e-05,
+      "loss": 0.1604,
+      "step": 2208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7837768800158133e-05,
+      "loss": 0.1519,
+      "step": 2209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783561102680717e-05,
+      "loss": 0.1587,
+      "step": 2210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783345230797804e-05,
+      "loss": 0.1435,
+      "step": 2211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783129264393123e-05,
+      "loss": 0.1575,
+      "step": 2212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7829132034927332e-05,
+      "loss": 0.1554,
+      "step": 2213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7826970481227053e-05,
+      "loss": 0.1615,
+      "step": 2214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824807983091218e-05,
+      "loss": 0.159,
+      "step": 2215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7822644540780763e-05,
+      "loss": 0.1498,
+      "step": 2216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7820480154556735e-05,
+      "loss": 0.1441,
+      "step": 2217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.1618,
+      "step": 2218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7816148551412738e-05,
+      "loss": 0.1562,
+      "step": 2219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7813981335015436e-05,
+      "loss": 0.1437,
+      "step": 2220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7811813175749904e-05,
+      "loss": 0.1481,
+      "step": 2221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.780964407387776e-05,
+      "loss": 0.1637,
+      "step": 2222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807474029660736e-05,
+      "loss": 0.1416,
+      "step": 2223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780530304336068e-05,
+      "loss": 0.1512,
+      "step": 2224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780313111523955e-05,
+      "loss": 0.1524,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800958245559425e-05,
+      "loss": 0.1634,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798784434582484e-05,
+      "loss": 0.1392,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779660968257104e-05,
+      "loss": 0.1591,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794433989787503e-05,
+      "loss": 0.1458,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792257356494397e-05,
+      "loss": 0.1425,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790079782954366e-05,
+      "loss": 0.1483,
+      "step": 2231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7787901269430168e-05,
+      "loss": 0.1581,
+      "step": 2232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778572181618467e-05,
+      "loss": 0.1607,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7783541423480853e-05,
+      "loss": 0.1438,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7781360091581815e-05,
+      "loss": 0.1524,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779177820750763e-05,
+      "loss": 0.138,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7776994611251016e-05,
+      "loss": 0.1445,
+      "step": 2237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774810463346015e-05,
+      "loss": 0.1683,
+      "step": 2238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777262537729931e-05,
+      "loss": 0.1468,
+      "step": 2239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7770439353374555e-05,
+      "loss": 0.1428,
+      "step": 2240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7768252391835526e-05,
+      "loss": 0.1377,
+      "step": 2241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7766064492946122e-05,
+      "loss": 0.1374,
+      "step": 2242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7763875656970334e-05,
+      "loss": 0.151,
+      "step": 2243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761685884172274e-05,
+      "loss": 0.1558,
+      "step": 2244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775949517481618e-05,
+      "loss": 0.1485,
+      "step": 2245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775730352916638e-05,
+      "loss": 0.1651,
+      "step": 2246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7755110947487338e-05,
+      "loss": 0.1575,
+      "step": 2247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752917430043615e-05,
+      "loss": 0.1483,
+      "step": 2248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775072297709989e-05,
+      "loss": 0.1533,
+      "step": 2249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7748527588920956e-05,
+      "loss": 0.1618,
+      "step": 2250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774633126577172e-05,
+      "loss": 0.1479,
+      "step": 2251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7744134007917195e-05,
+      "loss": 0.1655,
+      "step": 2252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7741935815622512e-05,
+      "loss": 0.1416,
+      "step": 2253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773973668915292e-05,
+      "loss": 0.1356,
+      "step": 2254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737536628773773e-05,
+      "loss": 0.1486,
+      "step": 2255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773533563475053e-05,
+      "loss": 0.1543,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773313370734879e-05,
+      "loss": 0.1471,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773093084683423e-05,
+      "loss": 0.1506,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728727053472664e-05,
+      "loss": 0.1462,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7726522327530013e-05,
+      "loss": 0.1364,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724316669272305e-05,
+      "loss": 0.1356,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722110078965685e-05,
+      "loss": 0.1613,
+      "step": 2262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719902556876412e-05,
+      "loss": 0.1472,
+      "step": 2263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771769410327085e-05,
+      "loss": 0.1584,
+      "step": 2264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715484718415486e-05,
+      "loss": 0.1425,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771327440257691e-05,
+      "loss": 0.1386,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771106315602183e-05,
+      "loss": 0.1551,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770885097901706e-05,
+      "loss": 0.1542,
+      "step": 2268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770663787182954e-05,
+      "loss": 0.1612,
+      "step": 2269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704423834726303e-05,
+      "loss": 0.1487,
+      "step": 2270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702208867974512e-05,
+      "loss": 0.1623,
+      "step": 2271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699992971841427e-05,
+      "loss": 0.1287,
+      "step": 2272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697776146594433e-05,
+      "loss": 0.1376,
+      "step": 2273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769555839250102e-05,
+      "loss": 0.14,
+      "step": 2274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769333970982879e-05,
+      "loss": 0.1579,
+      "step": 2275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769112009884546e-05,
+      "loss": 0.1654,
+      "step": 2276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768889955981886e-05,
+      "loss": 0.1604,
+      "step": 2277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686678093016928e-05,
+      "loss": 0.1301,
+      "step": 2278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684455698707714e-05,
+      "loss": 0.1462,
+      "step": 2279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768223237715938e-05,
+      "loss": 0.1597,
+      "step": 2280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7680008128640204e-05,
+      "loss": 0.1628,
+      "step": 2281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677782953418577e-05,
+      "loss": 0.1369,
+      "step": 2282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7675556851762993e-05,
+      "loss": 0.148,
+      "step": 2283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7673329823942065e-05,
+      "loss": 0.1425,
+      "step": 2284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7671101870224513e-05,
+      "loss": 0.1396,
+      "step": 2285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668872990879175e-05,
+      "loss": 0.1626,
+      "step": 2286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666643186174998e-05,
+      "loss": 0.1637,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7664412456381033e-05,
+      "loss": 0.1561,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7662180801766453e-05,
+      "loss": 0.1583,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765994822260054e-05,
+      "loss": 0.1446,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765771471915269e-05,
+      "loss": 0.1505,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76554802916924e-05,
+      "loss": 0.1446,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765324494048929e-05,
+      "loss": 0.1518,
+      "step": 2293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651008665813083e-05,
+      "loss": 0.1599,
+      "step": 2294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7648771467933625e-05,
+      "loss": 0.1584,
+      "step": 2295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7646533347120852e-05,
+      "loss": 0.1492,
+      "step": 2296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764429430364484e-05,
+      "loss": 0.1463,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7642054337775756e-05,
+      "loss": 0.1461,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763981344978388e-05,
+      "loss": 0.1525,
+      "step": 2299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763757163993961e-05,
+      "loss": 0.1489,
+      "step": 2300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763532890851346e-05,
+      "loss": 0.145,
+      "step": 2301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633085255776033e-05,
+      "loss": 0.1622,
+      "step": 2302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630840681998068e-05,
+      "loss": 0.1372,
+      "step": 2303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76285951874504e-05,
+      "loss": 0.132,
+      "step": 2304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762634877240398e-05,
+      "loss": 0.1582,
+      "step": 2305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7624101437129874e-05,
+      "loss": 0.1533,
+      "step": 2306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621853181899254e-05,
+      "loss": 0.1686,
+      "step": 2307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76196040069834e-05,
+      "loss": 0.1586,
+      "step": 2308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761735391265371e-05,
+      "loss": 0.1523,
+      "step": 2309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761510289918169e-05,
+      "loss": 0.1624,
+      "step": 2310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612850966838956e-05,
+      "loss": 0.1632,
+      "step": 2311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761059811589724e-05,
+      "loss": 0.1495,
+      "step": 2312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760834434662837e-05,
+      "loss": 0.1631,
+      "step": 2313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7606089659304307e-05,
+      "loss": 0.1654,
+      "step": 2314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7603834054197104e-05,
+      "loss": 0.1401,
+      "step": 2315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7601577531578936e-05,
+      "loss": 0.1431,
+      "step": 2316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7599320091722085e-05,
+      "loss": 0.1414,
+      "step": 2317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597061734898936e-05,
+      "loss": 0.1382,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594802461382003e-05,
+      "loss": 0.1496,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7592542271443888e-05,
+      "loss": 0.1362,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590281165357323e-05,
+      "loss": 0.1838,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758801914339514e-05,
+      "loss": 0.1428,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7585756205830287e-05,
+      "loss": 0.1528,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7583492352935817e-05,
+      "loss": 0.1539,
+      "step": 2324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758122758498489e-05,
+      "loss": 0.1566,
+      "step": 2325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75789619022508e-05,
+      "loss": 0.1691,
+      "step": 2326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7576695305006915e-05,
+      "loss": 0.1471,
+      "step": 2327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7574427793526743e-05,
+      "loss": 0.1575,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757215936808389e-05,
+      "loss": 0.1482,
+      "step": 2329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756989002895207e-05,
+      "loss": 0.1445,
+      "step": 2330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7567619776405114e-05,
+      "loss": 0.1441,
+      "step": 2331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7565348610716963e-05,
+      "loss": 0.1632,
+      "step": 2332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756307653216166e-05,
+      "loss": 0.1555,
+      "step": 2333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560803541013365e-05,
+      "loss": 0.1427,
+      "step": 2334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755852963754635e-05,
+      "loss": 0.1567,
+      "step": 2335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755625482203499e-05,
+      "loss": 0.1388,
+      "step": 2336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755397909475378e-05,
+      "loss": 0.1708,
+      "step": 2337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755170245597731e-05,
+      "loss": 0.1547,
+      "step": 2338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75494249059803e-05,
+      "loss": 0.145,
+      "step": 2339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547146445037554e-05,
+      "loss": 0.1413,
+      "step": 2340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544867073424016e-05,
+      "loss": 0.1409,
+      "step": 2341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7542586791414716e-05,
+      "loss": 0.1408,
+      "step": 2342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7540305599284805e-05,
+      "loss": 0.1505,
+      "step": 2343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753802349730954e-05,
+      "loss": 0.1412,
+      "step": 2344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753574048576429e-05,
+      "loss": 0.1556,
+      "step": 2345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7533456564924535e-05,
+      "loss": 0.1525,
+      "step": 2346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7531171735065857e-05,
+      "loss": 0.1562,
+      "step": 2347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752888599646396e-05,
+      "loss": 0.1285,
+      "step": 2348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526599349394646e-05,
+      "loss": 0.1452,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7524311794133833e-05,
+      "loss": 0.1604,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752202333095755e-05,
+      "loss": 0.1586,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519733960141927e-05,
+      "loss": 0.1402,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517443681963214e-05,
+      "loss": 0.1499,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515152496697765e-05,
+      "loss": 0.1436,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7512860404622042e-05,
+      "loss": 0.1359,
+      "step": 2355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751056740601262e-05,
+      "loss": 0.1343,
+      "step": 2356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508273501146182e-05,
+      "loss": 0.1535,
+      "step": 2357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750597869029952e-05,
+      "loss": 0.1681,
+      "step": 2358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503682973749538e-05,
+      "loss": 0.152,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750138635177324e-05,
+      "loss": 0.1424,
+      "step": 2360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499088824647757e-05,
+      "loss": 0.1453,
+      "step": 2361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7496790392650306e-05,
+      "loss": 0.1352,
+      "step": 2362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494491056058235e-05,
+      "loss": 0.1573,
+      "step": 2363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749219081514899e-05,
+      "loss": 0.1693,
+      "step": 2364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7489889670200125e-05,
+      "loss": 0.1497,
+      "step": 2365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748758762148931e-05,
+      "loss": 0.1586,
+      "step": 2366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748528466929431e-05,
+      "loss": 0.137,
+      "step": 2367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748298081389302e-05,
+      "loss": 0.1491,
+      "step": 2368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7480676055563435e-05,
+      "loss": 0.1463,
+      "step": 2369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7478370394583647e-05,
+      "loss": 0.1378,
+      "step": 2370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747606383123187e-05,
+      "loss": 0.1564,
+      "step": 2371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7473756365786426e-05,
+      "loss": 0.1499,
+      "step": 2372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471447998525744e-05,
+      "loss": 0.1561,
+      "step": 2373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7469138729728356e-05,
+      "loss": 0.1455,
+      "step": 2374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7466828559672917e-05,
+      "loss": 0.1448,
+      "step": 2375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464517488638174e-05,
+      "loss": 0.153,
+      "step": 2376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7462205516902995e-05,
+      "loss": 0.1579,
+      "step": 2377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.745989264474635e-05,
+      "loss": 0.155,
+      "step": 2378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457578872447324e-05,
+      "loss": 0.14,
+      "step": 2379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74552642002851e-05,
+      "loss": 0.1628,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745294862853898e-05,
+      "loss": 0.1547,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450632157488378e-05,
+      "loss": 0.1401,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7448314787412794e-05,
+      "loss": 0.1613,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744599651859186e-05,
+      "loss": 0.14,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744367735130531e-05,
+      "loss": 0.1402,
+      "step": 2385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441357285832984e-05,
+      "loss": 0.1408,
+      "step": 2386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7439036322454827e-05,
+      "loss": 0.1566,
+      "step": 2387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74367144614509e-05,
+      "loss": 0.1767,
+      "step": 2388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7434391703101364e-05,
+      "loss": 0.1646,
+      "step": 2389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74320680476865e-05,
+      "loss": 0.1474,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429743495486682e-05,
+      "loss": 0.1648,
+      "step": 2391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742741804678241e-05,
+      "loss": 0.1571,
+      "step": 2392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7425091701854274e-05,
+      "loss": 0.1526,
+      "step": 2393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7422764460982987e-05,
+      "loss": 0.1565,
+      "step": 2394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742043632444936e-05,
+      "loss": 0.1478,
+      "step": 2395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418107292534315e-05,
+      "loss": 0.154,
+      "step": 2396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7415777365518885e-05,
+      "loss": 0.1406,
+      "step": 2397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741344654368421e-05,
+      "loss": 0.1419,
+      "step": 2398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741111482731153e-05,
+      "loss": 0.1433,
+      "step": 2399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7408782216682208e-05,
+      "loss": 0.1512,
+      "step": 2400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406448712077706e-05,
+      "loss": 0.1313,
+      "step": 2401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740411431377959e-05,
+      "loss": 0.1546,
+      "step": 2402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401779022069544e-05,
+      "loss": 0.1409,
+      "step": 2403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7399442837229347e-05,
+      "loss": 0.1375,
+      "step": 2404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.73971057595409e-05,
+      "loss": 0.1532,
+      "step": 2405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.73947677892862e-05,
+      "loss": 0.1519,
+      "step": 2406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739242892674736e-05,
+      "loss": 0.1591,
+      "step": 2407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 0.1629,
+      "step": 2408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7387748525946227e-05,
+      "loss": 0.148,
+      "step": 2409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385406988248696e-05,
+      "loss": 0.1537,
+      "step": 2410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7383064559396535e-05,
+      "loss": 0.1394,
+      "step": 2411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380721239672392e-05,
+      "loss": 0.1559,
+      "step": 2412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378377029359026e-05,
+      "loss": 0.163,
+      "step": 2413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7376031928739296e-05,
+      "loss": 0.1497,
+      "step": 2414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7373685938096174e-05,
+      "loss": 0.1497,
+      "step": 2415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371339057712736e-05,
+      "loss": 0.1482,
+      "step": 2416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368991287872165e-05,
+      "loss": 0.1531,
+      "step": 2417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366642628857757e-05,
+      "loss": 0.1322,
+      "step": 2418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364293080952904e-05,
+      "loss": 0.1629,
+      "step": 2419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361942644441124e-05,
+      "loss": 0.1845,
+      "step": 2420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735959131960602e-05,
+      "loss": 0.1506,
+      "step": 2421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735723910673132e-05,
+      "loss": 0.1579,
+      "step": 2422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7354886006100843e-05,
+      "loss": 0.1404,
+      "step": 2423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7352532017998537e-05,
+      "loss": 0.1426,
+      "step": 2424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350177142708432e-05,
+      "loss": 0.1678,
+      "step": 2425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347821380514682e-05,
+      "loss": 0.1422,
+      "step": 2426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345464731701548e-05,
+      "loss": 0.141,
+      "step": 2427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343107196553385e-05,
+      "loss": 0.148,
+      "step": 2428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340748775354667e-05,
+      "loss": 0.1513,
+      "step": 2429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733838946838997e-05,
+      "loss": 0.1482,
+      "step": 2430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336029275943982e-05,
+      "loss": 0.1489,
+      "step": 2431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333668198301494e-05,
+      "loss": 0.1507,
+      "step": 2432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7331306235747397e-05,
+      "loss": 0.1577,
+      "step": 2433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73289433885667e-05,
+      "loss": 0.1562,
+      "step": 2434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7326579657044515e-05,
+      "loss": 0.1406,
+      "step": 2435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324215041466054e-05,
+      "loss": 0.1449,
+      "step": 2436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732184954211665e-05,
+      "loss": 0.1437,
+      "step": 2437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731948315928173e-05,
+      "loss": 0.1419,
+      "step": 2438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7317115893246833e-05,
+      "loss": 0.1396,
+      "step": 2439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73147477442976e-05,
+      "loss": 0.1698,
+      "step": 2440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7312378712719784e-05,
+      "loss": 0.1336,
+      "step": 2441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7310008798799243e-05,
+      "loss": 0.1394,
+      "step": 2442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7307638002821942e-05,
+      "loss": 0.1457,
+      "step": 2443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7305266325073948e-05,
+      "loss": 0.1427,
+      "step": 2444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730289376584144e-05,
+      "loss": 0.1369,
+      "step": 2445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.73005203254107e-05,
+      "loss": 0.1505,
+      "step": 2446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298146004068122e-05,
+      "loss": 0.1663,
+      "step": 2447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295770802100197e-05,
+      "loss": 0.146,
+      "step": 2448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293394719793524e-05,
+      "loss": 0.1469,
+      "step": 2449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729101775743482e-05,
+      "loss": 0.1489,
+      "step": 2450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7288639915310893e-05,
+      "loss": 0.1536,
+      "step": 2451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7286261193708663e-05,
+      "loss": 0.1701,
+      "step": 2452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7283881592915162e-05,
+      "loss": 0.1417,
+      "step": 2453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7281501113217517e-05,
+      "loss": 0.1261,
+      "step": 2454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727911975490297e-05,
+      "loss": 0.1533,
+      "step": 2455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7276737518258865e-05,
+      "loss": 0.1642,
+      "step": 2456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7274354403572652e-05,
+      "loss": 0.1583,
+      "step": 2457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271970411131888e-05,
+      "loss": 0.1694,
+      "step": 2458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7269585541224238e-05,
+      "loss": 0.1513,
+      "step": 2459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267199794137468e-05,
+      "loss": 0.1372,
+      "step": 2460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264813170159453e-05,
+      "loss": 0.1435,
+      "step": 2461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7262425669578177e-05,
+      "loss": 0.1516,
+      "step": 2462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7260037292681717e-05,
+      "loss": 0.1372,
+      "step": 2463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257648039758273e-05,
+      "loss": 0.15,
+      "step": 2464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7255257911096143e-05,
+      "loss": 0.1549,
+      "step": 2465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725286690698372e-05,
+      "loss": 0.1402,
+      "step": 2466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250475027709523e-05,
+      "loss": 0.129,
+      "step": 2467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7248082273562164e-05,
+      "loss": 0.1325,
+      "step": 2468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724568864483036e-05,
+      "loss": 0.1414,
+      "step": 2469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243294141802936e-05,
+      "loss": 0.1542,
+      "step": 2470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724089876476883e-05,
+      "loss": 0.1528,
+      "step": 2471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723850251401707e-05,
+      "loss": 0.1451,
+      "step": 2472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72361053898368e-05,
+      "loss": 0.1391,
+      "step": 2473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723370739251727e-05,
+      "loss": 0.1604,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723130852234783e-05,
+      "loss": 0.1531,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722890877961794e-05,
+      "loss": 0.1402,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722650816461716e-05,
+      "loss": 0.1567,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224106677635163e-05,
+      "loss": 0.1389,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722170431896172e-05,
+      "loss": 0.1515,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721930108888671e-05,
+      "loss": 0.1532,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7216896987700116e-05,
+      "loss": 0.1414,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7214492015692025e-05,
+      "loss": 0.1527,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721208617315264e-05,
+      "loss": 0.1485,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720967946037225e-05,
+      "loss": 0.181,
+      "step": 2484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207271877641263e-05,
+      "loss": 0.1586,
+      "step": 2485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720486342525019e-05,
+      "loss": 0.1748,
+      "step": 2486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7202454103489646e-05,
+      "loss": 0.1611,
+      "step": 2487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7200043912650346e-05,
+      "loss": 0.1533,
+      "step": 2488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7197632853023114e-05,
+      "loss": 0.1624,
+      "step": 2489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195220924898883e-05,
+      "loss": 0.1452,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7192808128568682e-05,
+      "loss": 0.1469,
+      "step": 2491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719039446432365e-05,
+      "loss": 0.1376,
+      "step": 2492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187979932455032e-05,
+      "loss": 0.1393,
+      "step": 2493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7185564533254178e-05,
+      "loss": 0.1298,
+      "step": 2494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7183148267012533e-05,
+      "loss": 0.1399,
+      "step": 2495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718073113402166e-05,
+      "loss": 0.1438,
+      "step": 2496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717831313457322e-05,
+      "loss": 0.1517,
+      "step": 2497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7175894268958974e-05,
+      "loss": 0.1399,
+      "step": 2498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.71734745374708e-05,
+      "loss": 0.1379,
+      "step": 2499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7171053940400664e-05,
+      "loss": 0.1481,
+      "step": 2500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716863247804066e-05,
+      "loss": 0.138,
+      "step": 2501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7166210150682956e-05,
+      "loss": 0.1479,
+      "step": 2502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716378695861985e-05,
+      "loss": 0.1333,
+      "step": 2503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7161362902143734e-05,
+      "loss": 0.1442,
+      "step": 2504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7158937981547096e-05,
+      "loss": 0.1663,
+      "step": 2505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715651219712255e-05,
+      "loss": 0.1413,
+      "step": 2506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7154085549162794e-05,
+      "loss": 0.1572,
+      "step": 2507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151658037960638e-05,
+      "loss": 0.1506,
+      "step": 2508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7149229663809e-05,
+      "loss": 0.1607,
+      "step": 2509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146800427000893e-05,
+      "loss": 0.1497,
+      "step": 2510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714437032782944e-05,
+      "loss": 0.1493,
+      "step": 2511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7141939366587873e-05,
+      "loss": 0.1347,
+      "step": 2512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7139507543569513e-05,
+      "loss": 0.1535,
+      "step": 2513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71370748590678e-05,
+      "loss": 0.1538,
+      "step": 2514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713464131337627e-05,
+      "loss": 0.1488,
+      "step": 2515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7132206906788565e-05,
+      "loss": 0.1426,
+      "step": 2516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712977163959843e-05,
+      "loss": 0.1375,
+      "step": 2517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712733551209972e-05,
+      "loss": 0.1498,
+      "step": 2518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124898524586383e-05,
+      "loss": 0.1579,
+      "step": 2519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712246067735248e-05,
+      "loss": 0.1459,
+      "step": 2520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7120021970692168e-05,
+      "loss": 0.1462,
+      "step": 2521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117582404899714e-05,
+      "loss": 0.1362,
+      "step": 2522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7115141980269482e-05,
+      "loss": 0.142,
+      "step": 2523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7112700697095955e-05,
+      "loss": 0.1556,
+      "step": 2524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7110258555673694e-05,
+      "loss": 0.1503,
+      "step": 2525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7107815556297392e-05,
+      "loss": 0.1468,
+      "step": 2526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710537169926182e-05,
+      "loss": 0.1436,
+      "step": 2527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7102926984861877e-05,
+      "loss": 0.1389,
+      "step": 2528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100481413392537e-05,
+      "loss": 0.1724,
+      "step": 2529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7098034985148907e-05,
+      "loss": 0.1512,
+      "step": 2530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7095587700426178e-05,
+      "loss": 0.1386,
+      "step": 2531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7093139559519645e-05,
+      "loss": 0.1434,
+      "step": 2532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709069056272472e-05,
+      "loss": 0.1453,
+      "step": 2533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088240710336903e-05,
+      "loss": 0.1545,
+      "step": 2534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085790002651807e-05,
+      "loss": 0.1434,
+      "step": 2535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083338439965142e-05,
+      "loss": 0.1567,
+      "step": 2536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080886022572726e-05,
+      "loss": 0.1433,
+      "step": 2537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707843275077048e-05,
+      "loss": 0.1293,
+      "step": 2538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707597862485442e-05,
+      "loss": 0.1572,
+      "step": 2539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7073523645120677e-05,
+      "loss": 0.1373,
+      "step": 2540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.1521,
+      "step": 2541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706861112538515e-05,
+      "loss": 0.1423,
+      "step": 2542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066153585976137e-05,
+      "loss": 0.1468,
+      "step": 2543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063695193934963e-05,
+      "loss": 0.1448,
+      "step": 2544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706123594955828e-05,
+      "loss": 0.1437,
+      "step": 2545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058775853142825e-05,
+      "loss": 0.1461,
+      "step": 2546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056314904985444e-05,
+      "loss": 0.1473,
+      "step": 2547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705385310538309e-05,
+      "loss": 0.1331,
+      "step": 2548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051390454632805e-05,
+      "loss": 0.1407,
+      "step": 2549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704892695303175e-05,
+      "loss": 0.1232,
+      "step": 2550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704646260087718e-05,
+      "loss": 0.163,
+      "step": 2551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043997398466453e-05,
+      "loss": 0.1587,
+      "step": 2552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7041531346097036e-05,
+      "loss": 0.1603,
+      "step": 2553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039064444066488e-05,
+      "loss": 0.1437,
+      "step": 2554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036596692672476e-05,
+      "loss": 0.1515,
+      "step": 2555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703412809221277e-05,
+      "loss": 0.1447,
+      "step": 2556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703165864298524e-05,
+      "loss": 0.1573,
+      "step": 2557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029188345287868e-05,
+      "loss": 0.1494,
+      "step": 2558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026717199418726e-05,
+      "loss": 0.143,
+      "step": 2559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7024245205675986e-05,
+      "loss": 0.144,
+      "step": 2560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702177236435794e-05,
+      "loss": 0.1533,
+      "step": 2561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019298675762966e-05,
+      "loss": 0.1423,
+      "step": 2562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701682414018955e-05,
+      "loss": 0.1588,
+      "step": 2563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701434875793628e-05,
+      "loss": 0.1403,
+      "step": 2564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011872529301848e-05,
+      "loss": 0.1361,
+      "step": 2565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7009395454585046e-05,
+      "loss": 0.1539,
+      "step": 2566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006917534084768e-05,
+      "loss": 0.1649,
+      "step": 2567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700443876810001e-05,
+      "loss": 0.1466,
+      "step": 2568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700195915692987e-05,
+      "loss": 0.1517,
+      "step": 2569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6999478700873552e-05,
+      "loss": 0.186,
+      "step": 2570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6996997400230347e-05,
+      "loss": 0.14,
+      "step": 2571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6994515255299676e-05,
+      "loss": 0.1519,
+      "step": 2572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6992032266381034e-05,
+      "loss": 0.1497,
+      "step": 2573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698954843377403e-05,
+      "loss": 0.1494,
+      "step": 2574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698706375777838e-05,
+      "loss": 0.1389,
+      "step": 2575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6984578238693888e-05,
+      "loss": 0.1544,
+      "step": 2576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6982091876820474e-05,
+      "loss": 0.1672,
+      "step": 2577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979604672458142e-05,
+      "loss": 0.1612,
+      "step": 2578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977116625907025e-05,
+      "loss": 0.1515,
+      "step": 2579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974627737467327e-05,
+      "loss": 0.1431,
+      "step": 2580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972138007439375e-05,
+      "loss": 0.1456,
+      "step": 2581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696964743612359e-05,
+      "loss": 0.1531,
+      "step": 2582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696715602382049e-05,
+      "loss": 0.1367,
+      "step": 2583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964663770830705e-05,
+      "loss": 0.1418,
+      "step": 2584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962170677454962e-05,
+      "loss": 0.1443,
+      "step": 2585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695967674399408e-05,
+      "loss": 0.1479,
+      "step": 2586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6957181970748996e-05,
+      "loss": 0.1349,
+      "step": 2587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695468635802074e-05,
+      "loss": 0.134,
+      "step": 2588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6952189906110432e-05,
+      "loss": 0.1608,
+      "step": 2589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694969261531932e-05,
+      "loss": 0.1504,
+      "step": 2590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947194485948724e-05,
+      "loss": 0.1413,
+      "step": 2591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944695518300087e-05,
+      "loss": 0.1519,
+      "step": 2592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6942195712674946e-05,
+      "loss": 0.143,
+      "step": 2593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6939695069374932e-05,
+      "loss": 0.1514,
+      "step": 2594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693719358870179e-05,
+      "loss": 0.1448,
+      "step": 2595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934691270957354e-05,
+      "loss": 0.1384,
+      "step": 2596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6932188116443565e-05,
+      "loss": 0.1732,
+      "step": 2597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692968412546247e-05,
+      "loss": 0.1513,
+      "step": 2598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6927179298316204e-05,
+      "loss": 0.1605,
+      "step": 2599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6924673635307013e-05,
+      "loss": 0.1679,
+      "step": 2600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692216713673724e-05,
+      "loss": 0.1614,
+      "step": 2601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6919659802909332e-05,
+      "loss": 0.1492,
+      "step": 2602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6917151634125833e-05,
+      "loss": 0.1545,
+      "step": 2603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691464263068939e-05,
+      "loss": 0.1389,
+      "step": 2604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691213279290275e-05,
+      "loss": 0.1438,
+      "step": 2605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909622121068755e-05,
+      "loss": 0.1364,
+      "step": 2606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6907110615490362e-05,
+      "loss": 0.1667,
+      "step": 2607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904598276470617e-05,
+      "loss": 0.1426,
+      "step": 2608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690208510431267e-05,
+      "loss": 0.1411,
+      "step": 2609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689957109931977e-05,
+      "loss": 0.1395,
+      "step": 2610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897056261795266e-05,
+      "loss": 0.1401,
+      "step": 2611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689454059204261e-05,
+      "loss": 0.1628,
+      "step": 2612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6892024090365357e-05,
+      "loss": 0.1549,
+      "step": 2613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6889506757067157e-05,
+      "loss": 0.1475,
+      "step": 2614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6886988592451757e-05,
+      "loss": 0.1453,
+      "step": 2615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884469596823015e-05,
+      "loss": 0.1188,
+      "step": 2616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688194977048488e-05,
+      "loss": 0.1396,
+      "step": 2617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687942911374141e-05,
+      "loss": 0.1416,
+      "step": 2618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876907626896755e-05,
+      "loss": 0.1419,
+      "step": 2619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687438531025517e-05,
+      "loss": 0.1572,
+      "step": 2620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871862164121002e-05,
+      "loss": 0.1584,
+      "step": 2621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6869338188798713e-05,
+      "loss": 0.1473,
+      "step": 2622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686681338459285e-05,
+      "loss": 0.1434,
+      "step": 2623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6864287751808075e-05,
+      "loss": 0.1509,
+      "step": 2624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861761290749133e-05,
+      "loss": 0.1419,
+      "step": 2625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6859234001720882e-05,
+      "loss": 0.1454,
+      "step": 2626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6856705885028275e-05,
+      "loss": 0.1479,
+      "step": 2627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6854176940976365e-05,
+      "loss": 0.1492,
+      "step": 2628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68516471698703e-05,
+      "loss": 0.1509,
+      "step": 2629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6849116572015347e-05,
+      "loss": 0.1525,
+      "step": 2630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684658514771684e-05,
+      "loss": 0.1617,
+      "step": 2631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684405289728025e-05,
+      "loss": 0.1595,
+      "step": 2632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6841519821011112e-05,
+      "loss": 0.1571,
+      "step": 2633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6838985919215094e-05,
+      "loss": 0.1404,
+      "step": 2634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6836451192197932e-05,
+      "loss": 0.1521,
+      "step": 2635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6833915640265485e-05,
+      "loss": 0.1499,
+      "step": 2636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6831379263723707e-05,
+      "loss": 0.1355,
+      "step": 2637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682884206287864e-05,
+      "loss": 0.1593,
+      "step": 2638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682630403803644e-05,
+      "loss": 0.1431,
+      "step": 2639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823765189503348e-05,
+      "loss": 0.1606,
+      "step": 2640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821225517585723e-05,
+      "loss": 0.129,
+      "step": 2641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6818685022590004e-05,
+      "loss": 0.1677,
+      "step": 2642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681614370482274e-05,
+      "loss": 0.1466,
+      "step": 2643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681360156459058e-05,
+      "loss": 0.1347,
+      "step": 2644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811058602200267e-05,
+      "loss": 0.1513,
+      "step": 2645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808514817958645e-05,
+      "loss": 0.1451,
+      "step": 2646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680597021217266e-05,
+      "loss": 0.1477,
+      "step": 2647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6803424785149352e-05,
+      "loss": 0.1495,
+      "step": 2648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6800878537195864e-05,
+      "loss": 0.1469,
+      "step": 2649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798331468619443e-05,
+      "loss": 0.1522,
+      "step": 2650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679578357972742e-05,
+      "loss": 0.1348,
+      "step": 2651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679323487082724e-05,
+      "loss": 0.1499,
+      "step": 2652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679068534222644e-05,
+      "loss": 0.1469,
+      "step": 2653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788134994232653e-05,
+      "loss": 0.143,
+      "step": 2654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678558382715362e-05,
+      "loss": 0.1433,
+      "step": 2655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6783031841297175e-05,
+      "loss": 0.1404,
+      "step": 2656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6780479036971252e-05,
+      "loss": 0.1665,
+      "step": 2657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6777925414483876e-05,
+      "loss": 0.1265,
+      "step": 2658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6775370974143187e-05,
+      "loss": 0.1455,
+      "step": 2659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 0.1339,
+      "step": 2660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677025964113488e-05,
+      "loss": 0.1568,
+      "step": 2661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676770274908401e-05,
+      "loss": 0.1465,
+      "step": 2662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6765145040413337e-05,
+      "loss": 0.1464,
+      "step": 2663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762586515431484e-05,
+      "loss": 0.1506,
+      "step": 2664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676002717444717e-05,
+      "loss": 0.141,
+      "step": 2665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757467017769216e-05,
+      "loss": 0.1579,
+      "step": 2666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754906045706542e-05,
+      "loss": 0.1581,
+      "step": 2667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6752344258568167e-05,
+      "loss": 0.1689,
+      "step": 2668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674978165666321e-05,
+      "loss": 0.1622,
+      "step": 2669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674721824030088e-05,
+      "loss": 0.1309,
+      "step": 2670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6744654009790488e-05,
+      "loss": 0.1587,
+      "step": 2671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6742088965441456e-05,
+      "loss": 0.1459,
+      "step": 2672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739523107563278e-05,
+      "loss": 0.1415,
+      "step": 2673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6736956436465573e-05,
+      "loss": 0.1474,
+      "step": 2674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734388952458043e-05,
+      "loss": 0.1442,
+      "step": 2675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731820655850492e-05,
+      "loss": 0.1598,
+      "step": 2676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672925154695282e-05,
+      "loss": 0.1513,
+      "step": 2677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6726681626075026e-05,
+      "loss": 0.1563,
+      "step": 2678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672411089352721e-05,
+      "loss": 0.1518,
+      "step": 2679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6721539349619567e-05,
+      "loss": 0.1571,
+      "step": 2680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718966994662388e-05,
+      "loss": 0.1553,
+      "step": 2681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671639382896607e-05,
+      "loss": 0.1523,
+      "step": 2682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6713819852841094e-05,
+      "loss": 0.1681,
+      "step": 2683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711245066598054e-05,
+      "loss": 0.1545,
+      "step": 2684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6708669470547632e-05,
+      "loss": 0.1408,
+      "step": 2685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670609306500061e-05,
+      "loss": 0.1573,
+      "step": 2686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670351585026787e-05,
+      "loss": 0.1609,
+      "step": 2687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6700937826660384e-05,
+      "loss": 0.1536,
+      "step": 2688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6698358994489235e-05,
+      "loss": 0.1345,
+      "step": 2689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669577935406559e-05,
+      "loss": 0.1577,
+      "step": 2690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693198905700722e-05,
+      "loss": 0.1405,
+      "step": 2691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690617649706e-05,
+      "loss": 0.153,
+      "step": 2692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6688035586392885e-05,
+      "loss": 0.1409,
+      "step": 2693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6685452716072946e-05,
+      "loss": 0.1413,
+      "step": 2694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668286903905784e-05,
+      "loss": 0.147,
+      "step": 2695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668028455565932e-05,
+      "loss": 0.1544,
+      "step": 2696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6677699266189247e-05,
+      "loss": 0.1521,
+      "step": 2697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6675113170959568e-05,
+      "loss": 0.1552,
+      "step": 2698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667252627028234e-05,
+      "loss": 0.1463,
+      "step": 2699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669938564469707e-05,
+      "loss": 0.1433,
+      "step": 2700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6667350053833903e-05,
+      "loss": 0.1697,
+      "step": 2701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666476073868728e-05,
+      "loss": 0.1546,
+      "step": 2702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6662170619342267e-05,
+      "loss": 0.1278,
+      "step": 2703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665957969611141e-05,
+      "loss": 0.1608,
+      "step": 2704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665698796930733e-05,
+      "loss": 0.1492,
+      "step": 2705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6654395439242762e-05,
+      "loss": 0.1338,
+      "step": 2706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665180210623053e-05,
+      "loss": 0.1593,
+      "step": 2707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6649207970583557e-05,
+      "loss": 0.1592,
+      "step": 2708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664661303261486e-05,
+      "loss": 0.1692,
+      "step": 2709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664401729263756e-05,
+      "loss": 0.1416,
+      "step": 2710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6641420750964863e-05,
+      "loss": 0.1324,
+      "step": 2711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6638823407910085e-05,
+      "loss": 0.1613,
+      "step": 2712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6636225263786633e-05,
+      "loss": 0.1568,
+      "step": 2713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6633626318908003e-05,
+      "loss": 0.1263,
+      "step": 2714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6631026573587803e-05,
+      "loss": 0.1355,
+      "step": 2715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662842602813972e-05,
+      "loss": 0.1352,
+      "step": 2716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6625824682877557e-05,
+      "loss": 0.157,
+      "step": 2717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66232225381152e-05,
+      "loss": 0.1574,
+      "step": 2718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6620619594166633e-05,
+      "loss": 0.1771,
+      "step": 2719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6618015851345938e-05,
+      "loss": 0.1659,
+      "step": 2720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6615411309967293e-05,
+      "loss": 0.1382,
+      "step": 2721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612805970344978e-05,
+      "loss": 0.1484,
+      "step": 2722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610199832793357e-05,
+      "loss": 0.1414,
+      "step": 2723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6607592897626904e-05,
+      "loss": 0.1386,
+      "step": 2724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604985165160183e-05,
+      "loss": 0.1477,
+      "step": 2725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6602376635707852e-05,
+      "loss": 0.1494,
+      "step": 2726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6599767309584663e-05,
+      "loss": 0.1362,
+      "step": 2727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6597157187105475e-05,
+      "loss": 0.1491,
+      "step": 2728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659454626858523e-05,
+      "loss": 0.1388,
+      "step": 2729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659193455433898e-05,
+      "loss": 0.1715,
+      "step": 2730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6589322044681863e-05,
+      "loss": 0.1466,
+      "step": 2731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658670873992911e-05,
+      "loss": 0.1497,
+      "step": 2732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658409464039606e-05,
+      "loss": 0.1411,
+      "step": 2733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6581479746398145e-05,
+      "loss": 0.1542,
+      "step": 2734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6578864058250876e-05,
+      "loss": 0.1319,
+      "step": 2735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657624757626988e-05,
+      "loss": 0.1508,
+      "step": 2736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657363030077088e-05,
+      "loss": 0.1343,
+      "step": 2737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6571012232069678e-05,
+      "loss": 0.1573,
+      "step": 2738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6568393370482182e-05,
+      "loss": 0.1526,
+      "step": 2739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65657737163244e-05,
+      "loss": 0.1317,
+      "step": 2740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6563153269912428e-05,
+      "loss": 0.1503,
+      "step": 2741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560532031562456e-05,
+      "loss": 0.1635,
+      "step": 2742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6557910001590786e-05,
+      "loss": 0.171,
+      "step": 2743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6555287180313787e-05,
+      "loss": 0.1376,
+      "step": 2744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552663568047952e-05,
+      "loss": 0.1367,
+      "step": 2745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655003916510985e-05,
+      "loss": 0.1571,
+      "step": 2746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6547413971816163e-05,
+      "loss": 0.1525,
+      "step": 2747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544787988483647e-05,
+      "loss": 0.1476,
+      "step": 2748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6542161215429168e-05,
+      "loss": 0.1425,
+      "step": 2749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6539533652969683e-05,
+      "loss": 0.1463,
+      "step": 2750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6536905301422246e-05,
+      "loss": 0.1434,
+      "step": 2751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534276161104003e-05,
+      "loss": 0.1495,
+      "step": 2752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65316462323322e-05,
+      "loss": 0.1503,
+      "step": 2753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529015515424178e-05,
+      "loss": 0.178,
+      "step": 2754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6526384010697363e-05,
+      "loss": 0.1361,
+      "step": 2755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652375171846929e-05,
+      "loss": 0.1413,
+      "step": 2756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652111863905758e-05,
+      "loss": 0.169,
+      "step": 2757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651848477277995e-05,
+      "loss": 0.1294,
+      "step": 2758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515850119954213e-05,
+      "loss": 0.1654,
+      "step": 2759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651321468089828e-05,
+      "loss": 0.146,
+      "step": 2760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651057845593016e-05,
+      "loss": 0.1642,
+      "step": 2761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6507941445367935e-05,
+      "loss": 0.1536,
+      "step": 2762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6505303649529815e-05,
+      "loss": 0.1524,
+      "step": 2763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6502665068734082e-05,
+      "loss": 0.1484,
+      "step": 2764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6500025703299113e-05,
+      "loss": 0.1417,
+      "step": 2765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6497385553543388e-05,
+      "loss": 0.1436,
+      "step": 2766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649474461978548e-05,
+      "loss": 0.1507,
+      "step": 2767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6492102902344056e-05,
+      "loss": 0.1436,
+      "step": 2768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6489460401537875e-05,
+      "loss": 0.1364,
+      "step": 2769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486817117685795e-05,
+      "loss": 0.1463,
+      "step": 2770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484173051106763e-05,
+      "loss": 0.1476,
+      "step": 2771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6481528202119826e-05,
+      "loss": 0.1499,
+      "step": 2772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647888257104412e-05,
+      "loss": 0.1476,
+      "step": 2773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6476236158198878e-05,
+      "loss": 0.1311,
+      "step": 2774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647358896390343e-05,
+      "loss": 0.1333,
+      "step": 2775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6470940988477195e-05,
+      "loss": 0.1429,
+      "step": 2776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646829223223969e-05,
+      "loss": 0.1335,
+      "step": 2777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646564269551053e-05,
+      "loss": 0.1537,
+      "step": 2778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646299237860941e-05,
+      "loss": 0.1452,
+      "step": 2779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6460341281856137e-05,
+      "loss": 0.1312,
+      "step": 2780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64576894055706e-05,
+      "loss": 0.1506,
+      "step": 2781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645503675007278e-05,
+      "loss": 0.1429,
+      "step": 2782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6452383315682768e-05,
+      "loss": 0.1369,
+      "step": 2783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449729102720735e-05,
+      "loss": 0.1278,
+      "step": 2784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6447074111506944e-05,
+      "loss": 0.1447,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6444418342361766e-05,
+      "loss": 0.1444,
+      "step": 2786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441761795605656e-05,
+      "loss": 0.1398,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6439104471559157e-05,
+      "loss": 0.1347,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643644637054292e-05,
+      "loss": 0.1411,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643378749287768e-05,
+      "loss": 0.1471,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643112783888427e-05,
+      "loss": 0.1532,
+      "step": 2791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6428467408883614e-05,
+      "loss": 0.1592,
+      "step": 2792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6425806203196734e-05,
+      "loss": 0.1442,
+      "step": 2793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6423144222144737e-05,
+      "loss": 0.1354,
+      "step": 2794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642048146604883e-05,
+      "loss": 0.1322,
+      "step": 2795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6417817935230318e-05,
+      "loss": 0.1479,
+      "step": 2796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6415153630010587e-05,
+      "loss": 0.1579,
+      "step": 2797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641248855071113e-05,
+      "loss": 0.1348,
+      "step": 2798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409822697653517e-05,
+      "loss": 0.1409,
+      "step": 2799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6407156071159432e-05,
+      "loss": 0.1401,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6404488671550637e-05,
+      "loss": 0.132,
+      "step": 2801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6401820499148988e-05,
+      "loss": 0.1482,
+      "step": 2802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399151554276446e-05,
+      "loss": 0.1494,
+      "step": 2803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639648183725505e-05,
+      "loss": 0.1455,
+      "step": 2804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639381134840694e-05,
+      "loss": 0.1419,
+      "step": 2805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6391140088054353e-05,
+      "loss": 0.1434,
+      "step": 2806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638846805651961e-05,
+      "loss": 0.1263,
+      "step": 2807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6385795254125142e-05,
+      "loss": 0.1558,
+      "step": 2808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383121681193444e-05,
+      "loss": 0.1385,
+      "step": 2809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6380447338047124e-05,
+      "loss": 0.1485,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637777222500889e-05,
+      "loss": 0.1435,
+      "step": 2811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6375096342401523e-05,
+      "loss": 0.1445,
+      "step": 2812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6372419690547908e-05,
+      "loss": 0.1473,
+      "step": 2813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369742269771024e-05,
+      "loss": 0.1358,
+      "step": 2814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367064080393936e-05,
+      "loss": 0.1359,
+      "step": 2815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364385122739812e-05,
+      "loss": 0.1659,
+      "step": 2816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6361705397131902e-05,
+      "loss": 0.169,
+      "step": 2817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6359024903893553e-05,
+      "loss": 0.1434,
+      "step": 2818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6356343643348204e-05,
+      "loss": 0.1456,
+      "step": 2819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353661615819386e-05,
+      "loss": 0.1562,
+      "step": 2820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635097882163073e-05,
+      "loss": 0.1561,
+      "step": 2821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348295261105946e-05,
+      "loss": 0.1613,
+      "step": 2822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634561093456885e-05,
+      "loss": 0.1318,
+      "step": 2823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6342925842343343e-05,
+      "loss": 0.1464,
+      "step": 2824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634023998475342e-05,
+      "loss": 0.1502,
+      "step": 2825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337553362123165e-05,
+      "loss": 0.1486,
+      "step": 2826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6334865974776757e-05,
+      "loss": 0.1425,
+      "step": 2827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6332177823038472e-05,
+      "loss": 0.1569,
+      "step": 2828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632948890723267e-05,
+      "loss": 0.137,
+      "step": 2829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6326799227683806e-05,
+      "loss": 0.1385,
+      "step": 2830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6324108784716432e-05,
+      "loss": 0.1409,
+      "step": 2831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632141757865519e-05,
+      "loss": 0.1597,
+      "step": 2832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6318725609824804e-05,
+      "loss": 0.1465,
+      "step": 2833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6316032878550107e-05,
+      "loss": 0.1447,
+      "step": 2834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631333938515601e-05,
+      "loss": 0.1482,
+      "step": 2835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310645129967525e-05,
+      "loss": 0.1286,
+      "step": 2836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630795011330975e-05,
+      "loss": 0.1404,
+      "step": 2837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630525433550788e-05,
+      "loss": 0.1653,
+      "step": 2838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6302557796887192e-05,
+      "loss": 0.1556,
+      "step": 2839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629986049777307e-05,
+      "loss": 0.1373,
+      "step": 2840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6297162438490976e-05,
+      "loss": 0.1496,
+      "step": 2841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6294463619366473e-05,
+      "loss": 0.1382,
+      "step": 2842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6291764040725212e-05,
+      "loss": 0.1469,
+      "step": 2843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6289063702892932e-05,
+      "loss": 0.1373,
+      "step": 2844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628636260619547e-05,
+      "loss": 0.1501,
+      "step": 2845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283660750958752e-05,
+      "loss": 0.1362,
+      "step": 2846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6280958137508793e-05,
+      "loss": 0.1468,
+      "step": 2847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278254766171703e-05,
+      "loss": 0.1395,
+      "step": 2848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6275550637273685e-05,
+      "loss": 0.1382,
+      "step": 2849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6272845751141026e-05,
+      "loss": 0.1493,
+      "step": 2850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270140108100115e-05,
+      "loss": 0.1381,
+      "step": 2851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626743370847742e-05,
+      "loss": 0.1395,
+      "step": 2852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626472655259951e-05,
+      "loss": 0.1359,
+      "step": 2853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6262018640793042e-05,
+      "loss": 0.1471,
+      "step": 2854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6259309973384764e-05,
+      "loss": 0.1515,
+      "step": 2855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6256600550701513e-05,
+      "loss": 0.1445,
+      "step": 2856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253890373070223e-05,
+      "loss": 0.1597,
+      "step": 2857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6251179440817915e-05,
+      "loss": 0.1485,
+      "step": 2858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62484677542717e-05,
+      "loss": 0.1502,
+      "step": 2859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624575531375878e-05,
+      "loss": 0.1365,
+      "step": 2860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6243042119606455e-05,
+      "loss": 0.1568,
+      "step": 2861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6240328172142104e-05,
+      "loss": 0.1413,
+      "step": 2862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623761347169321e-05,
+      "loss": 0.1517,
+      "step": 2863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.1498,
+      "step": 2864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6232181813152143e-05,
+      "loss": 0.1439,
+      "step": 2865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229464855715376e-05,
+      "loss": 0.1399,
+      "step": 2866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6226747146604876e-05,
+      "loss": 0.1355,
+      "step": 2867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622402868614858e-05,
+      "loss": 0.1424,
+      "step": 2868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6221309474674497e-05,
+      "loss": 0.1543,
+      "step": 2869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621858951251075e-05,
+      "loss": 0.1415,
+      "step": 2870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621586879998553e-05,
+      "loss": 0.1501,
+      "step": 2871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621314733742714e-05,
+      "loss": 0.1336,
+      "step": 2872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6210425125163956e-05,
+      "loss": 0.1354,
+      "step": 2873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620770216352446e-05,
+      "loss": 0.1504,
+      "step": 2874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6204978452837208e-05,
+      "loss": 0.1474,
+      "step": 2875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202253993430853e-05,
+      "loss": 0.1521,
+      "step": 2876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619952878563415e-05,
+      "loss": 0.1629,
+      "step": 2877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6196802829775928e-05,
+      "loss": 0.1433,
+      "step": 2878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619407612618511e-05,
+      "loss": 0.1399,
+      "step": 2879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6191348675190718e-05,
+      "loss": 0.1537,
+      "step": 2880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6188620477121852e-05,
+      "loss": 0.1524,
+      "step": 2881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618589153230771e-05,
+      "loss": 0.1372,
+      "step": 2882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6183161841077582e-05,
+      "loss": 0.1504,
+      "step": 2883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6180431403760836e-05,
+      "loss": 0.1441,
+      "step": 2884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177700220686946e-05,
+      "loss": 0.1523,
+      "step": 2885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6174968292185465e-05,
+      "loss": 0.1481,
+      "step": 2886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6172235618586037e-05,
+      "loss": 0.1444,
+      "step": 2887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169502200218403e-05,
+      "loss": 0.1351,
+      "step": 2888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6166768037412383e-05,
+      "loss": 0.1399,
+      "step": 2889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164033130497894e-05,
+      "loss": 0.139,
+      "step": 2890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161297479804946e-05,
+      "loss": 0.1456,
+      "step": 2891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615856108566363e-05,
+      "loss": 0.1408,
+      "step": 2892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6155823948404138e-05,
+      "loss": 0.1615,
+      "step": 2893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6153086068356732e-05,
+      "loss": 0.1548,
+      "step": 2894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6150347445851786e-05,
+      "loss": 0.1386,
+      "step": 2895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6147608081219753e-05,
+      "loss": 0.1453,
+      "step": 2896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6144867974791176e-05,
+      "loss": 0.1431,
+      "step": 2897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6142127126896682e-05,
+      "loss": 0.1592,
+      "step": 2898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6139385537866998e-05,
+      "loss": 0.1435,
+      "step": 2899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6136643208032937e-05,
+      "loss": 0.1496,
+      "step": 2900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6133900137725397e-05,
+      "loss": 0.1458,
+      "step": 2901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6131156327275372e-05,
+      "loss": 0.1478,
+      "step": 2902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612841177701394e-05,
+      "loss": 0.1574,
+      "step": 2903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6125666487272268e-05,
+      "loss": 0.1413,
+      "step": 2904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612292045838162e-05,
+      "loss": 0.1519,
+      "step": 2905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612017369067334e-05,
+      "loss": 0.1469,
+      "step": 2906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117426184478863e-05,
+      "loss": 0.1318,
+      "step": 2907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114677940129716e-05,
+      "loss": 0.1504,
+      "step": 2908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611192895795752e-05,
+      "loss": 0.1472,
+      "step": 2909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6109179238293967e-05,
+      "loss": 0.1371,
+      "step": 2910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106428781470858e-05,
+      "loss": 0.1473,
+      "step": 2911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6103677587820076e-05,
+      "loss": 0.1516,
+      "step": 2912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6100925657673584e-05,
+      "loss": 0.1419,
+      "step": 2913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6098172991363446e-05,
+      "loss": 0.1341,
+      "step": 2914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6095419589221814e-05,
+      "loss": 0.1451,
+      "step": 2915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092665451580925e-05,
+      "loss": 0.1387,
+      "step": 2916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6089910578773097e-05,
+      "loss": 0.1306,
+      "step": 2917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6087154971130753e-05,
+      "loss": 0.1453,
+      "step": 2918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608439862898639e-05,
+      "loss": 0.1394,
+      "step": 2919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6081641552672604e-05,
+      "loss": 0.148,
+      "step": 2920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6078883742522075e-05,
+      "loss": 0.1409,
+      "step": 2921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076125198867573e-05,
+      "loss": 0.1648,
+      "step": 2922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6073365922041952e-05,
+      "loss": 0.1348,
+      "step": 2923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607060591237816e-05,
+      "loss": 0.1378,
+      "step": 2924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6067845170209233e-05,
+      "loss": 0.1358,
+      "step": 2925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606508369586829e-05,
+      "loss": 0.1452,
+      "step": 2926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606232148968855e-05,
+      "loss": 0.1516,
+      "step": 2927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60595585520033e-05,
+      "loss": 0.1476,
+      "step": 2928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605679488314594e-05,
+      "loss": 0.1307,
+      "step": 2929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6054030483449944e-05,
+      "loss": 0.1562,
+      "step": 2930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605126535324887e-05,
+      "loss": 0.1476,
+      "step": 2931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6048499492876378e-05,
+      "loss": 0.1389,
+      "step": 2932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60457329026662e-05,
+      "loss": 0.1455,
+      "step": 2933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604296558295217e-05,
+      "loss": 0.1527,
+      "step": 2934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6040197534068206e-05,
+      "loss": 0.149,
+      "step": 2935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6037428756348306e-05,
+      "loss": 0.1339,
+      "step": 2936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6034659250126568e-05,
+      "loss": 0.1609,
+      "step": 2937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6031889015737176e-05,
+      "loss": 0.1407,
+      "step": 2938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6029118053514388e-05,
+      "loss": 0.1416,
+      "step": 2939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 0.1381,
+      "step": 2940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602357394690615e-05,
+      "loss": 0.1282,
+      "step": 2941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6020800803189682e-05,
+      "loss": 0.1381,
+      "step": 2942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601802693297777e-05,
+      "loss": 0.1465,
+      "step": 2943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6015252336605124e-05,
+      "loss": 0.133,
+      "step": 2944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012477014406543e-05,
+      "loss": 0.1626,
+      "step": 2945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6009700966716907e-05,
+      "loss": 0.1302,
+      "step": 2946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600692419387118e-05,
+      "loss": 0.1287,
+      "step": 2947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600414669620443e-05,
+      "loss": 0.1469,
+      "step": 2948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600136847405179e-05,
+      "loss": 0.1411,
+      "step": 2949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59985895277485e-05,
+      "loss": 0.134,
+      "step": 2950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599580985762988e-05,
+      "loss": 0.1272,
+      "step": 2951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5993029464031327e-05,
+      "loss": 0.1681,
+      "step": 2952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5990248347288346e-05,
+      "loss": 0.14,
+      "step": 2953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598746650773651e-05,
+      "loss": 0.1395,
+      "step": 2954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5984683945711503e-05,
+      "loss": 0.1487,
+      "step": 2955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981900661549063e-05,
+      "loss": 0.1467,
+      "step": 2956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5979116655585037e-05,
+      "loss": 0.1415,
+      "step": 2957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976331928155366e-05,
+      "loss": 0.1679,
+      "step": 2958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5973546479596053e-05,
+      "loss": 0.1587,
+      "step": 2959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597076031024321e-05,
+      "loss": 0.1524,
+      "step": 2960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5967973420433023e-05,
+      "loss": 0.1243,
+      "step": 2961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965185810501775e-05,
+      "loss": 0.1517,
+      "step": 2962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5962397480785828e-05,
+      "loss": 0.1421,
+      "step": 2963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5959608431621634e-05,
+      "loss": 0.1236,
+      "step": 2964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5956818663345733e-05,
+      "loss": 0.1521,
+      "step": 2965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595402817629475e-05,
+      "loss": 0.1352,
+      "step": 2966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59512369708054e-05,
+      "loss": 0.1422,
+      "step": 2967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594844504721447e-05,
+      "loss": 0.1479,
+      "step": 2968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5945652405858856e-05,
+      "loss": 0.1429,
+      "step": 2969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594285904707553e-05,
+      "loss": 0.1473,
+      "step": 2970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940064971201548e-05,
+      "loss": 0.1458,
+      "step": 2971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5937270178574056e-05,
+      "loss": 0.148,
+      "step": 2972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593447466953028e-05,
+      "loss": 0.1482,
+      "step": 2973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593167844440755e-05,
+      "loss": 0.148,
+      "step": 2974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592888150354326e-05,
+      "loss": 0.1474,
+      "step": 2975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5926083847274904e-05,
+      "loss": 0.1336,
+      "step": 2976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923285475940057e-05,
+      "loss": 0.1565,
+      "step": 2977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5920486389876383e-05,
+      "loss": 0.1423,
+      "step": 2978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5917686589421636e-05,
+      "loss": 0.1371,
+      "step": 2979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5914886074913648e-05,
+      "loss": 0.1436,
+      "step": 2980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5912084846690342e-05,
+      "loss": 0.1406,
+      "step": 2981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590928290508973e-05,
+      "loss": 0.1489,
+      "step": 2982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906480250449895e-05,
+      "loss": 0.1525,
+      "step": 2983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5903676883109032e-05,
+      "loss": 0.1669,
+      "step": 2984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5900872803405393e-05,
+      "loss": 0.1561,
+      "step": 2985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589806801167734e-05,
+      "loss": 0.154,
+      "step": 2986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589526250826331e-05,
+      "loss": 0.1416,
+      "step": 2987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5892456293501823e-05,
+      "loss": 0.1397,
+      "step": 2988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588964936773149e-05,
+      "loss": 0.136,
+      "step": 2989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886841731291008e-05,
+      "loss": 0.1593,
+      "step": 2990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5884033384519154e-05,
+      "loss": 0.1479,
+      "step": 2991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5881224327754807e-05,
+      "loss": 0.1336,
+      "step": 2992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5878414561336905e-05,
+      "loss": 0.1396,
+      "step": 2993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5875604085604496e-05,
+      "loss": 0.1387,
+      "step": 2994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5872792900896697e-05,
+      "loss": 0.1608,
+      "step": 2995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5869981007552725e-05,
+      "loss": 0.1542,
+      "step": 2996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586716840591187e-05,
+      "loss": 0.1511,
+      "step": 2997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5864355096313515e-05,
+      "loss": 0.1312,
+      "step": 2998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5861541079097125e-05,
+      "loss": 0.1446,
+      "step": 2999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5858726354602248e-05,
+      "loss": 0.1436,
+      "step": 3000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5855910923168527e-05,
+      "loss": 0.142,
+      "step": 3001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853094785135682e-05,
+      "loss": 0.1467,
+      "step": 3002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5850277940843513e-05,
+      "loss": 0.1348,
+      "step": 3003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5847460390631925e-05,
+      "loss": 0.1341,
+      "step": 3004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5844642134840885e-05,
+      "loss": 0.1419,
+      "step": 3005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5841823173810457e-05,
+      "loss": 0.1391,
+      "step": 3006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5839003507880796e-05,
+      "loss": 0.1328,
+      "step": 3007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836183137392128e-05,
+      "loss": 0.161,
+      "step": 3008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583336206268477e-05,
+      "loss": 0.15,
+      "step": 3009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5830540284099133e-05,
+      "loss": 0.1378,
+      "step": 3010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827717801975698e-05,
+      "loss": 0.136,
+      "step": 3011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582489461665504e-05,
+      "loss": 0.1599,
+      "step": 3012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582207072847782e-05,
+      "loss": 0.1347,
+      "step": 3013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581924613778477e-05,
+      "loss": 0.1471,
+      "step": 3014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5816420844916728e-05,
+      "loss": 0.1476,
+      "step": 3015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58135948502146e-05,
+      "loss": 0.1397,
+      "step": 3016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5810768154019386e-05,
+      "loss": 0.1487,
+      "step": 3017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580794075667216e-05,
+      "loss": 0.15,
+      "step": 3018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5805112658514098e-05,
+      "loss": 0.1669,
+      "step": 3019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580228385988644e-05,
+      "loss": 0.1435,
+      "step": 3020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5799454361130533e-05,
+      "loss": 0.1354,
+      "step": 3021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5796624162587787e-05,
+      "loss": 0.1384,
+      "step": 3022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5793793264599705e-05,
+      "loss": 0.1368,
+      "step": 3023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5790961667507884e-05,
+      "loss": 0.1607,
+      "step": 3024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578812937165399e-05,
+      "loss": 0.1506,
+      "step": 3025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5785296377379777e-05,
+      "loss": 0.1555,
+      "step": 3026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5782462685027094e-05,
+      "loss": 0.1534,
+      "step": 3027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5779628294937862e-05,
+      "loss": 0.1662,
+      "step": 3028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577679320745409e-05,
+      "loss": 0.1469,
+      "step": 3029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5773957422917873e-05,
+      "loss": 0.1461,
+      "step": 3030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5771120941671386e-05,
+      "loss": 0.1409,
+      "step": 3031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5768283764056898e-05,
+      "loss": 0.1359,
+      "step": 3032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5765445890416748e-05,
+      "loss": 0.1448,
+      "step": 3033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5762607321093368e-05,
+      "loss": 0.1405,
+      "step": 3034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5759768056429274e-05,
+      "loss": 0.1593,
+      "step": 3035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575692809676706e-05,
+      "loss": 0.1473,
+      "step": 3036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575408744244941e-05,
+      "loss": 0.1372,
+      "step": 3037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575124609381909e-05,
+      "loss": 0.1576,
+      "step": 3038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574840405121895e-05,
+      "loss": 0.1555,
+      "step": 3039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574556131499192e-05,
+      "loss": 0.1497,
+      "step": 3040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5742717885481017e-05,
+      "loss": 0.143,
+      "step": 3041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573987376302934e-05,
+      "loss": 0.1426,
+      "step": 3042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5737028947980076e-05,
+      "loss": 0.1424,
+      "step": 3043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5734183440676493e-05,
+      "loss": 0.1476,
+      "step": 3044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573133724146194e-05,
+      "loss": 0.1397,
+      "step": 3045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5728490350679853e-05,
+      "loss": 0.15,
+      "step": 3046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572564276867375e-05,
+      "loss": 0.1476,
+      "step": 3047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722794495787227e-05,
+      "loss": 0.1356,
+      "step": 3048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571994553236398e-05,
+      "loss": 0.1498,
+      "step": 3049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5717095878747764e-05,
+      "loss": 0.135,
+      "step": 3050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5714245535282436e-05,
+      "loss": 0.1579,
+      "step": 3051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5711394502311934e-05,
+      "loss": 0.1413,
+      "step": 3052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570854278018027e-05,
+      "loss": 0.1458,
+      "step": 3053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5705690369231552e-05,
+      "loss": 0.1367,
+      "step": 3054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5702837269809958e-05,
+      "loss": 0.1403,
+      "step": 3055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5699983482259756e-05,
+      "loss": 0.1481,
+      "step": 3056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5697129006925296e-05,
+      "loss": 0.1423,
+      "step": 3057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5694273844151015e-05,
+      "loss": 0.1528,
+      "step": 3058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691417994281426e-05,
+      "loss": 0.1534,
+      "step": 3059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5688561457661125e-05,
+      "loss": 0.1326,
+      "step": 3060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5685704234634803e-05,
+      "loss": 0.1521,
+      "step": 3061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568284632554721e-05,
+      "loss": 0.1499,
+      "step": 3062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5679987730743208e-05,
+      "loss": 0.1618,
+      "step": 3063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567712845056772e-05,
+      "loss": 0.1367,
+      "step": 3064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5674268485365765e-05,
+      "loss": 0.1475,
+      "step": 3065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671407835482426e-05,
+      "loss": 0.1398,
+      "step": 3066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5668546501262897e-05,
+      "loss": 0.1805,
+      "step": 3067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5665684483052425e-05,
+      "loss": 0.1447,
+      "step": 3068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662821781196362e-05,
+      "loss": 0.1543,
+      "step": 3069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565995839604013e-05,
+      "loss": 0.1342,
+      "step": 3070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565709432792924e-05,
+      "loss": 0.1412,
+      "step": 3071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565422957720928e-05,
+      "loss": 0.1385,
+      "step": 3072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565136414422592e-05,
+      "loss": 0.1451,
+      "step": 3073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5648498029324923e-05,
+      "loss": 0.1404,
+      "step": 3074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564563123285212e-05,
+      "loss": 0.1574,
+      "step": 3075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5642763755153438e-05,
+      "loss": 0.1366,
+      "step": 3076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563989559657487e-05,
+      "loss": 0.1392,
+      "step": 3077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563702675746251e-05,
+      "loss": 0.1493,
+      "step": 3078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5634157238162516e-05,
+      "loss": 0.143,
+      "step": 3079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5631287039021143e-05,
+      "loss": 0.1597,
+      "step": 3080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5628416160384714e-05,
+      "loss": 0.1501,
+      "step": 3081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5625544602599647e-05,
+      "loss": 0.1363,
+      "step": 3082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5622672366012436e-05,
+      "loss": 0.1357,
+      "step": 3083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5619799450969655e-05,
+      "loss": 0.1427,
+      "step": 3084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5616925857817965e-05,
+      "loss": 0.1399,
+      "step": 3085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56140515869041e-05,
+      "loss": 0.1297,
+      "step": 3086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561117663857489e-05,
+      "loss": 0.1383,
+      "step": 3087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5608301013177234e-05,
+      "loss": 0.1563,
+      "step": 3088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560542471105812e-05,
+      "loss": 0.1417,
+      "step": 3089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602547732564607e-05,
+      "loss": 0.1408,
+      "step": 3090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559967007804385e-05,
+      "loss": 0.1397,
+      "step": 3091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5596791747843083e-05,
+      "loss": 0.146,
+      "step": 3092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5593912742309608e-05,
+      "loss": 0.1525,
+      "step": 3093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591033061790827e-05,
+      "loss": 0.1327,
+      "step": 3094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5588152706634207e-05,
+      "loss": 0.1369,
+      "step": 3095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558527167718731e-05,
+      "loss": 0.144,
+      "step": 3096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558238997379777e-05,
+      "loss": 0.1611,
+      "step": 3097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579507596813304e-05,
+      "loss": 0.1408,
+      "step": 3098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5576624546581713e-05,
+      "loss": 0.1443,
+      "step": 3099
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5573740823450885e-05,
+      "loss": 0.1376,
+      "step": 3100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557085642776877e-05,
+      "loss": 0.1472,
+      "step": 3101
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556797135988342e-05,
+      "loss": 0.1235,
+      "step": 3102
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565085620142958e-05,
+      "loss": 0.1367,
+      "step": 3103
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5562199208895586e-05,
+      "loss": 0.1535,
+      "step": 3104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559312126489596e-05,
+      "loss": 0.1671,
+      "step": 3105
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5556424373273357e-05,
+      "loss": 0.1446,
+      "step": 3106
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555353594959531e-05,
+      "loss": 0.1554,
+      "step": 3107
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5550646855803986e-05,
+      "loss": 0.1543,
+      "step": 3108
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547757092248e-05,
+      "loss": 0.1278,
+      "step": 3109
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554486665927604e-05,
+      "loss": 0.1226,
+      "step": 3110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554197555723688e-05,
+      "loss": 0.109,
+      "step": 3111
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553908378647937e-05,
+      "loss": 0.1243,
+      "step": 3112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553619134735245e-05,
+      "loss": 0.1217,
+      "step": 3113
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533298240205124e-05,
+      "loss": 0.1171,
+      "step": 3114
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530404465386492e-05,
+      "loss": 0.112,
+      "step": 3115
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552751002324573e-05,
+      "loss": 0.115,
+      "step": 3116
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552461491413209e-05,
+      "loss": 0.1146,
+      "step": 3117
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5521719138394914e-05,
+      "loss": 0.1155,
+      "step": 3118
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5518822696383612e-05,
+      "loss": 0.1013,
+      "step": 3119
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5515925588447683e-05,
+      "loss": 0.1067,
+      "step": 3120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551302781493671e-05,
+      "loss": 0.1237,
+      "step": 3121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551012937620034e-05,
+      "loss": 0.1136,
+      "step": 3122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5507230272588326e-05,
+      "loss": 0.1146,
+      "step": 3123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5504330504450474e-05,
+      "loss": 0.1177,
+      "step": 3124
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5501430072136682e-05,
+      "loss": 0.1054,
+      "step": 3125
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5498528975996933e-05,
+      "loss": 0.1052,
+      "step": 3126
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5495627216381287e-05,
+      "loss": 0.117,
+      "step": 3127
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549272479363988e-05,
+      "loss": 0.1091,
+      "step": 3128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489821708122935e-05,
+      "loss": 0.1141,
+      "step": 3129
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5486917960180742e-05,
+      "loss": 0.1077,
+      "step": 3130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5484013550163683e-05,
+      "loss": 0.1244,
+      "step": 3131
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548110847842222e-05,
+      "loss": 0.1212,
+      "step": 3132
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547820274530689e-05,
+      "loss": 0.1135,
+      "step": 3133
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547529635116831e-05,
+      "loss": 0.1102,
+      "step": 3134
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5472389296357175e-05,
+      "loss": 0.1082,
+      "step": 3135
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.103,
+      "step": 3136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5466573206120448e-05,
+      "loss": 0.1092,
+      "step": 3137
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5463664171396643e-05,
+      "loss": 0.1037,
+      "step": 3138
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460754477403877e-05,
+      "loss": 0.1156,
+      "step": 3139
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5457844124493243e-05,
+      "loss": 0.1155,
+      "step": 3140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5454933113015917e-05,
+      "loss": 0.0955,
+      "step": 3141
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5452021443323155e-05,
+      "loss": 0.1046,
+      "step": 3142
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544910911576629e-05,
+      "loss": 0.1172,
+      "step": 3143
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544619613069673e-05,
+      "loss": 0.1034,
+      "step": 3144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5443282488465983e-05,
+      "loss": 0.1119,
+      "step": 3145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544036818942561e-05,
+      "loss": 0.1154,
+      "step": 3146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437453233927265e-05,
+      "loss": 0.1118,
+      "step": 3147
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543453762232268e-05,
+      "loss": 0.1122,
+      "step": 3148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543162135496367e-05,
+      "loss": 0.1139,
+      "step": 3149
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542870443220211e-05,
+      "loss": 0.1149,
+      "step": 3150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5425786854389983e-05,
+      "loss": 0.1032,
+      "step": 3151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542286862187933e-05,
+      "loss": 0.1184,
+      "step": 3152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5419949735022278e-05,
+      "loss": 0.1336,
+      "step": 3153
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541703019417103e-05,
+      "loss": 0.1103,
+      "step": 3154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5414109999677876e-05,
+      "loss": 0.12,
+      "step": 3155
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5411189151895175e-05,
+      "loss": 0.1118,
+      "step": 3156
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408267651175368e-05,
+      "loss": 0.1062,
+      "step": 3157
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5405345497870977e-05,
+      "loss": 0.1068,
+      "step": 3158
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.54024226923346e-05,
+      "loss": 0.1085,
+      "step": 3159
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5399499234918924e-05,
+      "loss": 0.1093,
+      "step": 3160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5396575125976695e-05,
+      "loss": 0.1121,
+      "step": 3161
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393650365860748e-05,
+      "loss": 0.1057,
+      "step": 3162
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5390724954924007e-05,
+      "loss": 0.1099,
+      "step": 3163
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5387798893519454e-05,
+      "loss": 0.1093,
+      "step": 3164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538487218200017e-05,
+      "loss": 0.1127,
+      "step": 3165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5381944820719296e-05,
+      "loss": 0.117,
+      "step": 3166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5379016810030062e-05,
+      "loss": 0.1122,
+      "step": 3167
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5376088150285777e-05,
+      "loss": 0.1086,
+      "step": 3168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537315884183982e-05,
+      "loss": 0.1138,
+      "step": 3169
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5370228885045662e-05,
+      "loss": 0.114,
+      "step": 3170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5367298280256835e-05,
+      "loss": 0.1038,
+      "step": 3171
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364367027826968e-05,
+      "loss": 0.1057,
+      "step": 3172
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5361435128109753e-05,
+      "loss": 0.1171,
+      "step": 3173
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358502581458964e-05,
+      "loss": 0.1176,
+      "step": 3174
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5355569388228455e-05,
+      "loss": 0.1033,
+      "step": 3175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535263554877216e-05,
+      "loss": 0.1079,
+      "step": 3176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349701063444088e-05,
+      "loss": 0.125,
+      "step": 3177
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346765932598326e-05,
+      "loss": 0.1027,
+      "step": 3178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534383015658904e-05,
+      "loss": 0.1088,
+      "step": 3179
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534089373577047e-05,
+      "loss": 0.1066,
+      "step": 3180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5337956670496943e-05,
+      "loss": 0.1065,
+      "step": 3181
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533501896112285e-05,
+      "loss": 0.1091,
+      "step": 3182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5332080608002672e-05,
+      "loss": 0.1225,
+      "step": 3183
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329141611490968e-05,
+      "loss": 0.1128,
+      "step": 3184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5326201971942363e-05,
+      "loss": 0.1051,
+      "step": 3185
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5323261689711566e-05,
+      "loss": 0.1003,
+      "step": 3186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.1207,
+      "step": 3187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5317379198622632e-05,
+      "loss": 0.1215,
+      "step": 3188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314436990474296e-05,
+      "loss": 0.1131,
+      "step": 3189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5311494141063387e-05,
+      "loss": 0.1208,
+      "step": 3190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5308550650745e-05,
+      "loss": 0.1176,
+      "step": 3191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53056065198743e-05,
+      "loss": 0.1075,
+      "step": 3192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530266174880655e-05,
+      "loss": 0.1063,
+      "step": 3193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529971633789707e-05,
+      "loss": 0.1083,
+      "step": 3194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529677028750127e-05,
+      "loss": 0.1162,
+      "step": 3195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5293823597974638e-05,
+      "loss": 0.1148,
+      "step": 3196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290876269672726e-05,
+      "loss": 0.1043,
+      "step": 3197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528792830295117e-05,
+      "loss": 0.1156,
+      "step": 3198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5284979698165693e-05,
+      "loss": 0.1107,
+      "step": 3199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5282030455672076e-05,
+      "loss": 0.126,
+      "step": 3200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279080575826197e-05,
+      "loss": 0.1166,
+      "step": 3201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5276130058983993e-05,
+      "loss": 0.113,
+      "step": 3202
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5273178905501495e-05,
+      "loss": 0.1179,
+      "step": 3203
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527022711573479e-05,
+      "loss": 0.1176,
+      "step": 3204
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526727469004006e-05,
+      "loss": 0.1184,
+      "step": 3205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526432162877356e-05,
+      "loss": 0.1123,
+      "step": 3206
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5261367932291613e-05,
+      "loss": 0.1002,
+      "step": 3207
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525841360095063e-05,
+      "loss": 0.1234,
+      "step": 3208
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525545863510709e-05,
+      "loss": 0.1161,
+      "step": 3209
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5252503035117552e-05,
+      "loss": 0.1201,
+      "step": 3210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5249546801338648e-05,
+      "loss": 0.1171,
+      "step": 3211
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5246589934127096e-05,
+      "loss": 0.1123,
+      "step": 3212
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243632433839681e-05,
+      "loss": 0.1079,
+      "step": 3213
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5240674300833269e-05,
+      "loss": 0.1271,
+      "step": 3214
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52377155354648e-05,
+      "loss": 0.1143,
+      "step": 3215
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5234756138091288e-05,
+      "loss": 0.1144,
+      "step": 3216
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5231796109069837e-05,
+      "loss": 0.115,
+      "step": 3217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228835448757605e-05,
+      "loss": 0.1113,
+      "step": 3218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522587415751184e-05,
+      "loss": 0.1097,
+      "step": 3219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5222912235689868e-05,
+      "loss": 0.1126,
+      "step": 3220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5219949683649086e-05,
+      "loss": 0.1007,
+      "step": 3221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5216986501746966e-05,
+      "loss": 0.1249,
+      "step": 3222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521402269034106e-05,
+      "loss": 0.1095,
+      "step": 3223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211058249788996e-05,
+      "loss": 0.1095,
+      "step": 3224
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5208093180448469e-05,
+      "loss": 0.1152,
+      "step": 3225
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205127482677265e-05,
+      "loss": 0.1192,
+      "step": 3226
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202161156833236e-05,
+      "loss": 0.1053,
+      "step": 3227
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199194203274306e-05,
+      "loss": 0.1092,
+      "step": 3228
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5196226622358483e-05,
+      "loss": 0.0945,
+      "step": 3229
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5193258414443852e-05,
+      "loss": 0.1117,
+      "step": 3230
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5190289579888567e-05,
+      "loss": 0.1248,
+      "step": 3231
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187320119050854e-05,
+      "loss": 0.1081,
+      "step": 3232
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5184350032289033e-05,
+      "loss": 0.1233,
+      "step": 3233
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518137931996148e-05,
+      "loss": 0.1055,
+      "step": 3234
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517840798242665e-05,
+      "loss": 0.1028,
+      "step": 3235
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175436020043085e-05,
+      "loss": 0.119,
+      "step": 3236
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5172463433169396e-05,
+      "loss": 0.1107,
+      "step": 3237
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169490222164255e-05,
+      "loss": 0.1208,
+      "step": 3238
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5166516387386436e-05,
+      "loss": 0.107,
+      "step": 3239
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163541929194769e-05,
+      "loss": 0.1061,
+      "step": 3240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160566847948163e-05,
+      "loss": 0.1197,
+      "step": 3241
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157591144005609e-05,
+      "loss": 0.1175,
+      "step": 3242
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5154614817726163e-05,
+      "loss": 0.1179,
+      "step": 3243
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515163786946896e-05,
+      "loss": 0.1084,
+      "step": 3244
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514866029959322e-05,
+      "loss": 0.1021,
+      "step": 3245
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5145682108458224e-05,
+      "loss": 0.1021,
+      "step": 3246
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5142703296423328e-05,
+      "loss": 0.1236,
+      "step": 3247
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5139723863847977e-05,
+      "loss": 0.1092,
+      "step": 3248
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5136743811091679e-05,
+      "loss": 0.1055,
+      "step": 3249
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5133763138514015e-05,
+      "loss": 0.1082,
+      "step": 3250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513078184647465e-05,
+      "loss": 0.117,
+      "step": 3251
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5127799935333323e-05,
+      "loss": 0.1172,
+      "step": 3252
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124817405449836e-05,
+      "loss": 0.1114,
+      "step": 3253
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512183425718408e-05,
+      "loss": 0.1121,
+      "step": 3254
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118850490896012e-05,
+      "loss": 0.1048,
+      "step": 3255
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115866106945664e-05,
+      "loss": 0.1122,
+      "step": 3256
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511288110569315e-05,
+      "loss": 0.1146,
+      "step": 3257
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109895487498648e-05,
+      "loss": 0.1121,
+      "step": 3258
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510690925272242e-05,
+      "loss": 0.1039,
+      "step": 3259
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103922401724789e-05,
+      "loss": 0.1108,
+      "step": 3260
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5100934934866169e-05,
+      "loss": 0.1181,
+      "step": 3261
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097946852507042e-05,
+      "loss": 0.1076,
+      "step": 3262
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5094958155007953e-05,
+      "loss": 0.1177,
+      "step": 3263
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509196884272954e-05,
+      "loss": 0.12,
+      "step": 3264
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5088978916032506e-05,
+      "loss": 0.1035,
+      "step": 3265
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508598837527762e-05,
+      "loss": 0.102,
+      "step": 3266
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082997220825744e-05,
+      "loss": 0.1211,
+      "step": 3267
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5080005453037791e-05,
+      "loss": 0.1218,
+      "step": 3268
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5077013072274774e-05,
+      "loss": 0.0995,
+      "step": 3269
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507402007889776e-05,
+      "loss": 0.1149,
+      "step": 3270
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5071026473267894e-05,
+      "loss": 0.121,
+      "step": 3271
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.50680322557464e-05,
+      "loss": 0.1053,
+      "step": 3272
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5065037426694575e-05,
+      "loss": 0.1165,
+      "step": 3273
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506204198647378e-05,
+      "loss": 0.1154,
+      "step": 3274
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505904593544547e-05,
+      "loss": 0.1186,
+      "step": 3275
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505604927397115e-05,
+      "loss": 0.1021,
+      "step": 3276
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5053052002412414e-05,
+      "loss": 0.1178,
+      "step": 3277
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5050054121130925e-05,
+      "loss": 0.1143,
+      "step": 3278
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5047055630488423e-05,
+      "loss": 0.11,
+      "step": 3279
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5044056530846715e-05,
+      "loss": 0.1243,
+      "step": 3280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041056822567685e-05,
+      "loss": 0.1122,
+      "step": 3281
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038056506013297e-05,
+      "loss": 0.1095,
+      "step": 3282
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035055581545574e-05,
+      "loss": 0.1134,
+      "step": 3283
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5032054049526623e-05,
+      "loss": 0.1061,
+      "step": 3284
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5029051910318622e-05,
+      "loss": 0.1263,
+      "step": 3285
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5026049164283824e-05,
+      "loss": 0.1264,
+      "step": 3286
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023045811784548e-05,
+      "loss": 0.1131,
+      "step": 3287
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5020041853183198e-05,
+      "loss": 0.1236,
+      "step": 3288
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017037288842238e-05,
+      "loss": 0.1042,
+      "step": 3289
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5014032119124215e-05,
+      "loss": 0.1074,
+      "step": 3290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5011026344391747e-05,
+      "loss": 0.1168,
+      "step": 3291
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5008019965007518e-05,
+      "loss": 0.1023,
+      "step": 3292
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005012981334294e-05,
+      "loss": 0.1022,
+      "step": 3293
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5002005393734914e-05,
+      "loss": 0.1256,
+      "step": 3294
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4998997202572278e-05,
+      "loss": 0.1088,
+      "step": 3295
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995988408209374e-05,
+      "loss": 0.1099,
+      "step": 3296
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992979011009254e-05,
+      "loss": 0.1262,
+      "step": 3297
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4989969011335043e-05,
+      "loss": 0.105,
+      "step": 3298
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4986958409549943e-05,
+      "loss": 0.1027,
+      "step": 3299
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4983947206017225e-05,
+      "loss": 0.1133,
+      "step": 3300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980935401100234e-05,
+      "loss": 0.1175,
+      "step": 3301
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4977922995162384e-05,
+      "loss": 0.1044,
+      "step": 3302
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4974909988567168e-05,
+      "loss": 0.1153,
+      "step": 3303
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4971896381678152e-05,
+      "loss": 0.1146,
+      "step": 3304
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968882174858962e-05,
+      "loss": 0.1146,
+      "step": 3305
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965867368473308e-05,
+      "loss": 0.1073,
+      "step": 3306
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4962851962884972e-05,
+      "loss": 0.1154,
+      "step": 3307
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959835958457807e-05,
+      "loss": 0.1135,
+      "step": 3308
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495681935555573e-05,
+      "loss": 0.1161,
+      "step": 3309
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4953802154542743e-05,
+      "loss": 0.1146,
+      "step": 3310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4950784355782912e-05,
+      "loss": 0.1157,
+      "step": 3311
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4947765959640379e-05,
+      "loss": 0.1133,
+      "step": 3312
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944746966479355e-05,
+      "loss": 0.1128,
+      "step": 3313
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4941727376664125e-05,
+      "loss": 0.1104,
+      "step": 3314
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938707190559046e-05,
+      "loss": 0.1113,
+      "step": 3315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4935686408528548e-05,
+      "loss": 0.1133,
+      "step": 3316
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4932665030937126e-05,
+      "loss": 0.1009,
+      "step": 3317
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4929643058149357e-05,
+      "loss": 0.1222,
+      "step": 3318
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4926620490529883e-05,
+      "loss": 0.1206,
+      "step": 3319
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923597328443423e-05,
+      "loss": 0.1,
+      "step": 3320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920573572254761e-05,
+      "loss": 0.1031,
+      "step": 3321
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491754922232876e-05,
+      "loss": 0.115,
+      "step": 3322
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914524279030344e-05,
+      "loss": 0.1082,
+      "step": 3323
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4911498742724523e-05,
+      "loss": 0.1094,
+      "step": 3324
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490847261377637e-05,
+      "loss": 0.1264,
+      "step": 3325
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905445892551027e-05,
+      "loss": 0.1224,
+      "step": 3326
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4902418579413713e-05,
+      "loss": 0.1057,
+      "step": 3327
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899390674729716e-05,
+      "loss": 0.1093,
+      "step": 3328
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4896362178864396e-05,
+      "loss": 0.1333,
+      "step": 3329
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4893333092183184e-05,
+      "loss": 0.1115,
+      "step": 3330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890303415051584e-05,
+      "loss": 0.1064,
+      "step": 3331
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488727314783517e-05,
+      "loss": 0.1128,
+      "step": 3332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884242290899585e-05,
+      "loss": 0.1048,
+      "step": 3333
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4881210844610543e-05,
+      "loss": 0.104,
+      "step": 3334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4878178809333834e-05,
+      "loss": 0.1133,
+      "step": 3335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487514618543532e-05,
+      "loss": 0.1077,
+      "step": 3336
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872112973280923e-05,
+      "loss": 0.115,
+      "step": 3337
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869079173236648e-05,
+      "loss": 0.1237,
+      "step": 3338
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4866044785668563e-05,
+      "loss": 0.1028,
+      "step": 3339
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4863009810942814e-05,
+      "loss": 0.1024,
+      "step": 3340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4859974249425615e-05,
+      "loss": 0.1037,
+      "step": 3341
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.485693810148324e-05,
+      "loss": 0.1087,
+      "step": 3342
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853901367482056e-05,
+      "loss": 0.1167,
+      "step": 3343
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850864047788481e-05,
+      "loss": 0.1058,
+      "step": 3344
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484782614276901e-05,
+      "loss": 0.1168,
+      "step": 3345
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4844787652790215e-05,
+      "loss": 0.1051,
+      "step": 3346
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484174857821873e-05,
+      "loss": 0.1033,
+      "step": 3347
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838708919421261e-05,
+      "loss": 0.12,
+      "step": 3348
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4835668676764588e-05,
+      "loss": 0.1071,
+      "step": 3349
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832627850615563e-05,
+      "loss": 0.1112,
+      "step": 3350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829586441341097e-05,
+      "loss": 0.1172,
+      "step": 3351
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4826544449308186e-05,
+      "loss": 0.121,
+      "step": 3352
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482350187488389e-05,
+      "loss": 0.0994,
+      "step": 3353
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820458718435333e-05,
+      "loss": 0.1174,
+      "step": 3354
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481741498032972e-05,
+      "loss": 0.1449,
+      "step": 3355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814370660934323e-05,
+      "loss": 0.1152,
+      "step": 3356
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4811325760616478e-05,
+      "loss": 0.1129,
+      "step": 3357
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808280279743594e-05,
+      "loss": 0.1033,
+      "step": 3358
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4805234218683163e-05,
+      "loss": 0.1122,
+      "step": 3359
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480218757780272e-05,
+      "loss": 0.11,
+      "step": 3360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4799140357469894e-05,
+      "loss": 0.1036,
+      "step": 3361
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4796092558052379e-05,
+      "loss": 0.1097,
+      "step": 3362
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479304417991793e-05,
+      "loss": 0.1263,
+      "step": 3363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4789995223434376e-05,
+      "loss": 0.1087,
+      "step": 3364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4786945688969625e-05,
+      "loss": 0.1127,
+      "step": 3365
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783895576891639e-05,
+      "loss": 0.1118,
+      "step": 3366
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780844887568457e-05,
+      "loss": 0.1077,
+      "step": 3367
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4777793621368197e-05,
+      "loss": 0.1143,
+      "step": 3368
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477474177865903e-05,
+      "loss": 0.1183,
+      "step": 3369
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771689359809202e-05,
+      "loss": 0.1121,
+      "step": 3370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768636365187038e-05,
+      "loss": 0.0956,
+      "step": 3371
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4765582795160926e-05,
+      "loss": 0.1152,
+      "step": 3372
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4762528650099315e-05,
+      "loss": 0.1137,
+      "step": 3373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759473930370738e-05,
+      "loss": 0.129,
+      "step": 3374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756418636343785e-05,
+      "loss": 0.1193,
+      "step": 3375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753362768387125e-05,
+      "loss": 0.1152,
+      "step": 3376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750306326869492e-05,
+      "loss": 0.1133,
+      "step": 3377
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747249312159685e-05,
+      "loss": 0.1033,
+      "step": 3378
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744191724626582e-05,
+      "loss": 0.1129,
+      "step": 3379
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474113356463912e-05,
+      "loss": 0.1138,
+      "step": 3380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738074832566313e-05,
+      "loss": 0.1046,
+      "step": 3381
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4735015528777239e-05,
+      "loss": 0.1103,
+      "step": 3382
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4731955653641045e-05,
+      "loss": 0.1259,
+      "step": 3383
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728895207526952e-05,
+      "loss": 0.1128,
+      "step": 3384
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725834190804244e-05,
+      "loss": 0.1121,
+      "step": 3385
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722772603842277e-05,
+      "loss": 0.1177,
+      "step": 3386
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4719710447010475e-05,
+      "loss": 0.1051,
+      "step": 3387
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471664772067833e-05,
+      "loss": 0.1045,
+      "step": 3388
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713584425215405e-05,
+      "loss": 0.1241,
+      "step": 3389
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4710520560991333e-05,
+      "loss": 0.1126,
+      "step": 3390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4707456128375803e-05,
+      "loss": 0.1113,
+      "step": 3391
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4704391127738596e-05,
+      "loss": 0.117,
+      "step": 3392
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701325559449537e-05,
+      "loss": 0.1143,
+      "step": 3393
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4698259423878534e-05,
+      "loss": 0.1249,
+      "step": 3394
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695192721395563e-05,
+      "loss": 0.1226,
+      "step": 3395
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692125452370664e-05,
+      "loss": 0.1127,
+      "step": 3396
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689057617173944e-05,
+      "loss": 0.1136,
+      "step": 3397
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4685989216175583e-05,
+      "loss": 0.1281,
+      "step": 3398
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468292024974583e-05,
+      "loss": 0.1062,
+      "step": 3399
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467985071825499e-05,
+      "loss": 0.1279,
+      "step": 3400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4676780622073455e-05,
+      "loss": 0.1116,
+      "step": 3401
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673709961571678e-05,
+      "loss": 0.107,
+      "step": 3402
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4670638737120167e-05,
+      "loss": 0.1119,
+      "step": 3403
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4667566949089516e-05,
+      "loss": 0.1024,
+      "step": 3404
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664494597850384e-05,
+      "loss": 0.1152,
+      "step": 3405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4661421683773485e-05,
+      "loss": 0.1144,
+      "step": 3406
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658348207229614e-05,
+      "loss": 0.1157,
+      "step": 3407
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655274168589635e-05,
+      "loss": 0.1047,
+      "step": 3408
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652199568224464e-05,
+      "loss": 0.1163,
+      "step": 3409
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46491244065051e-05,
+      "loss": 0.1138,
+      "step": 3410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464604868380261e-05,
+      "loss": 0.117,
+      "step": 3411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642972400488118e-05,
+      "loss": 0.1107,
+      "step": 3412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639895556932821e-05,
+      "loss": 0.1194,
+      "step": 3413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463681815350799e-05,
+      "loss": 0.1127,
+      "step": 3414
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633740190584954e-05,
+      "loss": 0.102,
+      "step": 3415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630661668535106e-05,
+      "loss": 0.1129,
+      "step": 3416
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627582587729927e-05,
+      "loss": 0.1094,
+      "step": 3417
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4624502948540944e-05,
+      "loss": 0.111,
+      "step": 3418
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462142275133976e-05,
+      "loss": 0.1151,
+      "step": 3419
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618341996498044e-05,
+      "loss": 0.1163,
+      "step": 3420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4615260684387538e-05,
+      "loss": 0.1222,
+      "step": 3421
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4612178815380041e-05,
+      "loss": 0.11,
+      "step": 3422
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4609096389847425e-05,
+      "loss": 0.1114,
+      "step": 3423
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460601340816163e-05,
+      "loss": 0.1191,
+      "step": 3424
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602929870694662e-05,
+      "loss": 0.1268,
+      "step": 3425
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4599845777818594e-05,
+      "loss": 0.1136,
+      "step": 3426
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4596761129905563e-05,
+      "loss": 0.1041,
+      "step": 3427
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459367592732778e-05,
+      "loss": 0.1108,
+      "step": 3428
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459059017045751e-05,
+      "loss": 0.1214,
+      "step": 3429
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587503859667106e-05,
+      "loss": 0.116,
+      "step": 3430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458441699532897e-05,
+      "loss": 0.1141,
+      "step": 3431
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581329577815571e-05,
+      "loss": 0.0966,
+      "step": 3432
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4578241607499458e-05,
+      "loss": 0.1182,
+      "step": 3433
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4575153084753233e-05,
+      "loss": 0.1155,
+      "step": 3434
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4572064009949571e-05,
+      "loss": 0.1053,
+      "step": 3435
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568974383461217e-05,
+      "loss": 0.1264,
+      "step": 3436
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4565884205660975e-05,
+      "loss": 0.1168,
+      "step": 3437
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456279347692172e-05,
+      "loss": 0.1141,
+      "step": 3438
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455970219761639e-05,
+      "loss": 0.1151,
+      "step": 3439
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556610368117996e-05,
+      "loss": 0.1174,
+      "step": 3440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4553517988799609e-05,
+      "loss": 0.1127,
+      "step": 3441
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550425060034367e-05,
+      "loss": 0.1157,
+      "step": 3442
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547331582195479e-05,
+      "loss": 0.1169,
+      "step": 3443
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544237555656217e-05,
+      "loss": 0.1192,
+      "step": 3444
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454114298078992e-05,
+      "loss": 0.1086,
+      "step": 3445
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538047857969988e-05,
+      "loss": 0.1156,
+      "step": 3446
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4534952187569897e-05,
+      "loss": 0.115,
+      "step": 3447
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453185596996318e-05,
+      "loss": 0.108,
+      "step": 3448
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528759205523442e-05,
+      "loss": 0.1081,
+      "step": 3449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4525661894624353e-05,
+      "loss": 0.1085,
+      "step": 3450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4522564037639644e-05,
+      "loss": 0.1079,
+      "step": 3451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4519465634943115e-05,
+      "loss": 0.1098,
+      "step": 3452
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4516366686908639e-05,
+      "loss": 0.1183,
+      "step": 3453
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4513267193910141e-05,
+      "loss": 0.1142,
+      "step": 3454
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510167156321621e-05,
+      "loss": 0.1215,
+      "step": 3455
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4507066574517148e-05,
+      "loss": 0.1087,
+      "step": 3456
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4503965448870848e-05,
+      "loss": 0.1208,
+      "step": 3457
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450086377975691e-05,
+      "loss": 0.1236,
+      "step": 3458
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4497761567549602e-05,
+      "loss": 0.109,
+      "step": 3459
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449465881262325e-05,
+      "loss": 0.113,
+      "step": 3460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4491555515352242e-05,
+      "loss": 0.1208,
+      "step": 3461
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488451676111037e-05,
+      "loss": 0.1107,
+      "step": 3462
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485347295274158e-05,
+      "loss": 0.1179,
+      "step": 3463
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4482242373216191e-05,
+      "loss": 0.1264,
+      "step": 3464
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4479136910311789e-05,
+      "loss": 0.1044,
+      "step": 3465
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4476030906935672e-05,
+      "loss": 0.1261,
+      "step": 3466
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472924363462624e-05,
+      "loss": 0.1177,
+      "step": 3467
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4469817280267491e-05,
+      "loss": 0.1196,
+      "step": 3468
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4466709657725191e-05,
+      "loss": 0.113,
+      "step": 3469
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4463601496210699e-05,
+      "loss": 0.1391,
+      "step": 3470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446049279609906e-05,
+      "loss": 0.1121,
+      "step": 3471
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4457383557765385e-05,
+      "loss": 0.1124,
+      "step": 3472
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4454273781584847e-05,
+      "loss": 0.1068,
+      "step": 3473
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4451163467932681e-05,
+      "loss": 0.1162,
+      "step": 3474
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4448052617184196e-05,
+      "loss": 0.1037,
+      "step": 3475
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444494122971476e-05,
+      "loss": 0.1174,
+      "step": 3476
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44418293058998e-05,
+      "loss": 0.1118,
+      "step": 3477
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443871684611482e-05,
+      "loss": 0.1062,
+      "step": 3478
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4435603850735387e-05,
+      "loss": 0.1228,
+      "step": 3479
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432490320137116e-05,
+      "loss": 0.1256,
+      "step": 3480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429376254695706e-05,
+      "loss": 0.1096,
+      "step": 3481
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426261654786915e-05,
+      "loss": 0.103,
+      "step": 3482
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423146520786561e-05,
+      "loss": 0.1016,
+      "step": 3483
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442003085307053e-05,
+      "loss": 0.1148,
+      "step": 3484
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441691465201477e-05,
+      "loss": 0.1145,
+      "step": 3485
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44137979179953e-05,
+      "loss": 0.1262,
+      "step": 3486
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4410680651388194e-05,
+      "loss": 0.112,
+      "step": 3487
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407562852569594e-05,
+      "loss": 0.1155,
+      "step": 3488
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4404444521915713e-05,
+      "loss": 0.1158,
+      "step": 3489
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401325659802812e-05,
+      "loss": 0.1132,
+      "step": 3490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4398206266607236e-05,
+      "loss": 0.1064,
+      "step": 3491
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439508634270538e-05,
+      "loss": 0.1187,
+      "step": 3492
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4391965888473705e-05,
+      "loss": 0.1062,
+      "step": 3493
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4388844904288743e-05,
+      "loss": 0.1037,
+      "step": 3494
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4385723390527083e-05,
+      "loss": 0.1168,
+      "step": 3495
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4382601347565377e-05,
+      "loss": 0.1071,
+      "step": 3496
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4379478775780351e-05,
+      "loss": 0.1064,
+      "step": 3497
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437635567554878e-05,
+      "loss": 0.1126,
+      "step": 3498
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373232047247515e-05,
+      "loss": 0.1114,
+      "step": 3499
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4370107891253468e-05,
+      "loss": 0.1105,
+      "step": 3500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436698320794361e-05,
+      "loss": 0.1087,
+      "step": 3501
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436385799769498e-05,
+      "loss": 0.1107,
+      "step": 3502
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360732260884678e-05,
+      "loss": 0.1157,
+      "step": 3503
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357605997889867e-05,
+      "loss": 0.1139,
+      "step": 3504
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435447920908778e-05,
+      "loss": 0.1183,
+      "step": 3505
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351351894855705e-05,
+      "loss": 0.133,
+      "step": 3506
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348224055570996e-05,
+      "loss": 0.1086,
+      "step": 3507
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345095691611076e-05,
+      "loss": 0.1103,
+      "step": 3508
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4341966803353423e-05,
+      "loss": 0.1191,
+      "step": 3509
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.115,
+      "step": 3510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335707455455164e-05,
+      "loss": 0.1077,
+      "step": 3511
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332576996569838e-05,
+      "loss": 0.1232,
+      "step": 3512
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329446014897339e-05,
+      "loss": 0.1132,
+      "step": 3513
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4326314510815464e-05,
+      "loss": 0.1142,
+      "step": 3514
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4323182484702076e-05,
+      "loss": 0.1182,
+      "step": 3515
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4320049936935097e-05,
+      "loss": 0.1141,
+      "step": 3516
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.431691686789251e-05,
+      "loss": 0.1173,
+      "step": 3517
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313783277952371e-05,
+      "loss": 0.1153,
+      "step": 3518
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.431064916749279e-05,
+      "loss": 0.1053,
+      "step": 3519
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4307514536891938e-05,
+      "loss": 0.1135,
+      "step": 3520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4304379386528058e-05,
+      "loss": 0.1159,
+      "step": 3521
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301243716779447e-05,
+      "loss": 0.1063,
+      "step": 3522
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429810752802447e-05,
+      "loss": 0.1092,
+      "step": 3523
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4294970820641556e-05,
+      "loss": 0.1169,
+      "step": 3524
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4291833595009188e-05,
+      "loss": 0.1132,
+      "step": 3525
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4288695851505919e-05,
+      "loss": 0.1128,
+      "step": 3526
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.428555759051036e-05,
+      "loss": 0.1201,
+      "step": 3527
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4282418812401197e-05,
+      "loss": 0.1158,
+      "step": 3528
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4279279517557156e-05,
+      "loss": 0.1019,
+      "step": 3529
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276139706357042e-05,
+      "loss": 0.0994,
+      "step": 3530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4272999379179724e-05,
+      "loss": 0.106,
+      "step": 3531
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269858536404116e-05,
+      "loss": 0.1132,
+      "step": 3532
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4266717178409211e-05,
+      "loss": 0.1116,
+      "step": 3533
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4263575305574065e-05,
+      "loss": 0.1108,
+      "step": 3534
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4260432918277779e-05,
+      "loss": 0.108,
+      "step": 3535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425729001689953e-05,
+      "loss": 0.1118,
+      "step": 3536
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425414660181856e-05,
+      "loss": 0.1138,
+      "step": 3537
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4251002673414158e-05,
+      "loss": 0.1126,
+      "step": 3538
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424785823206569e-05,
+      "loss": 0.1123,
+      "step": 3539
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244713278152575e-05,
+      "loss": 0.1155,
+      "step": 3540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4241567812054297e-05,
+      "loss": 0.1088,
+      "step": 3541
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238421834150403e-05,
+      "loss": 0.1183,
+      "step": 3542
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235275344820498e-05,
+      "loss": 0.1145,
+      "step": 3543
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4232128344444251e-05,
+      "loss": 0.1324,
+      "step": 3544
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228980833401391e-05,
+      "loss": 0.1162,
+      "step": 3545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225832812071716e-05,
+      "loss": 0.0998,
+      "step": 3546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4222684280835073e-05,
+      "loss": 0.1207,
+      "step": 3547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219535240071378e-05,
+      "loss": 0.1161,
+      "step": 3548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4216385690160613e-05,
+      "loss": 0.1084,
+      "step": 3549
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213235631482809e-05,
+      "loss": 0.121,
+      "step": 3550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421008506441807e-05,
+      "loss": 0.1044,
+      "step": 3551
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206933989346559e-05,
+      "loss": 0.1121,
+      "step": 3552
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.420378240664849e-05,
+      "loss": 0.1235,
+      "step": 3553
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4200630316704153e-05,
+      "loss": 0.1355,
+      "step": 3554
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4197477719893893e-05,
+      "loss": 0.1195,
+      "step": 3555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419432461659811e-05,
+      "loss": 0.1054,
+      "step": 3556
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4191171007197278e-05,
+      "loss": 0.109,
+      "step": 3557
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418801689207192e-05,
+      "loss": 0.1115,
+      "step": 3558
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4184862271602624e-05,
+      "loss": 0.1218,
+      "step": 3559
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181707146170046e-05,
+      "loss": 0.1157,
+      "step": 3560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4178551516154891e-05,
+      "loss": 0.1085,
+      "step": 3561
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4175395381937933e-05,
+      "loss": 0.1083,
+      "step": 3562
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172238743900005e-05,
+      "loss": 0.1167,
+      "step": 3563
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169081602421996e-05,
+      "loss": 0.1117,
+      "step": 3564
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4165923957884868e-05,
+      "loss": 0.1267,
+      "step": 3565
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416276581066963e-05,
+      "loss": 0.1194,
+      "step": 3566
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4159607161157363e-05,
+      "loss": 0.1114,
+      "step": 3567
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156448009729195e-05,
+      "loss": 0.1097,
+      "step": 3568
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4153288356766328e-05,
+      "loss": 0.1098,
+      "step": 3569
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415012820265002e-05,
+      "loss": 0.1142,
+      "step": 3570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4146967547761585e-05,
+      "loss": 0.1087,
+      "step": 3571
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4143806392482403e-05,
+      "loss": 0.1048,
+      "step": 3572
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140644737193911e-05,
+      "loss": 0.116,
+      "step": 3573
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4137482582277611e-05,
+      "loss": 0.1256,
+      "step": 3574
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134319928115059e-05,
+      "loss": 0.1111,
+      "step": 3575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4131156775087877e-05,
+      "loss": 0.1214,
+      "step": 3576
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4127993123577742e-05,
+      "loss": 0.116,
+      "step": 3577
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124828973966393e-05,
+      "loss": 0.1079,
+      "step": 3578
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4121664326635634e-05,
+      "loss": 0.1198,
+      "step": 3579
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4118499181967322e-05,
+      "loss": 0.1107,
+      "step": 3580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4115333540343375e-05,
+      "loss": 0.1077,
+      "step": 3581
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112167402145773e-05,
+      "loss": 0.1111,
+      "step": 3582
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4109000767756564e-05,
+      "loss": 0.1106,
+      "step": 3583
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105833637557835e-05,
+      "loss": 0.1152,
+      "step": 3584
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102666011931753e-05,
+      "loss": 0.1179,
+      "step": 3585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4099497891260538e-05,
+      "loss": 0.1063,
+      "step": 3586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4096329275926463e-05,
+      "loss": 0.1192,
+      "step": 3587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4093160166311873e-05,
+      "loss": 0.1067,
+      "step": 3588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089990562799163e-05,
+      "loss": 0.113,
+      "step": 3589
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.408682046577079e-05,
+      "loss": 0.1203,
+      "step": 3590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4083649875609272e-05,
+      "loss": 0.114,
+      "step": 3591
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080478792697188e-05,
+      "loss": 0.1075,
+      "step": 3592
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077307217417173e-05,
+      "loss": 0.1105,
+      "step": 3593
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4074135150151922e-05,
+      "loss": 0.1213,
+      "step": 3594
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4070962591284192e-05,
+      "loss": 0.1048,
+      "step": 3595
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067789541196795e-05,
+      "loss": 0.1138,
+      "step": 3596
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4064616000272607e-05,
+      "loss": 0.1124,
+      "step": 3597
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4061441968894559e-05,
+      "loss": 0.1175,
+      "step": 3598
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058267447445644e-05,
+      "loss": 0.1108,
+      "step": 3599
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405509243630891e-05,
+      "loss": 0.1222,
+      "step": 3600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4051916935867476e-05,
+      "loss": 0.1075,
+      "step": 3601
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4048740946504506e-05,
+      "loss": 0.1085,
+      "step": 3602
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4045564468603226e-05,
+      "loss": 0.1139,
+      "step": 3603
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4042387502546925e-05,
+      "loss": 0.1139,
+      "step": 3604
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403921004871895e-05,
+      "loss": 0.1094,
+      "step": 3605
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4036032107502708e-05,
+      "loss": 0.1194,
+      "step": 3606
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032853679281659e-05,
+      "loss": 0.1106,
+      "step": 3607
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4029674764439331e-05,
+      "loss": 0.1056,
+      "step": 3608
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4026495363359301e-05,
+      "loss": 0.1174,
+      "step": 3609
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4023315476425207e-05,
+      "loss": 0.0977,
+      "step": 3610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4020135104020756e-05,
+      "loss": 0.1268,
+      "step": 3611
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 0.1199,
+      "step": 3612
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4013772904335847e-05,
+      "loss": 0.1077,
+      "step": 3613
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010591077823086e-05,
+      "loss": 0.1131,
+      "step": 3614
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400740876737534e-05,
+      "loss": 0.1142,
+      "step": 3615
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004225973376604e-05,
+      "loss": 0.113,
+      "step": 3616
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4001042696210928e-05,
+      "loss": 0.1073,
+      "step": 3617
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3997858936262415e-05,
+      "loss": 0.1119,
+      "step": 3618
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3994674693915237e-05,
+      "loss": 0.1117,
+      "step": 3619
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991489969553615e-05,
+      "loss": 0.1099,
+      "step": 3620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.398830476356183e-05,
+      "loss": 0.1144,
+      "step": 3621
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985119076324228e-05,
+      "loss": 0.1172,
+      "step": 3622
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3981932908225203e-05,
+      "loss": 0.1011,
+      "step": 3623
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397874625964921e-05,
+      "loss": 0.1105,
+      "step": 3624
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3975559130980769e-05,
+      "loss": 0.1115,
+      "step": 3625
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972371522604449e-05,
+      "loss": 0.1154,
+      "step": 3626
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3969183434904878e-05,
+      "loss": 0.1175,
+      "step": 3627
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396599486826675e-05,
+      "loss": 0.1197,
+      "step": 3628
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396280582307481e-05,
+      "loss": 0.1082,
+      "step": 3629
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3959616299713857e-05,
+      "loss": 0.1292,
+      "step": 3630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3956426298568753e-05,
+      "loss": 0.1086,
+      "step": 3631
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953235820024427e-05,
+      "loss": 0.1106,
+      "step": 3632
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3950044864465841e-05,
+      "loss": 0.1065,
+      "step": 3633
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394685343227804e-05,
+      "loss": 0.1193,
+      "step": 3634
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3943661523846115e-05,
+      "loss": 0.1072,
+      "step": 3635
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3940469139555209e-05,
+      "loss": 0.1145,
+      "step": 3636
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937276279790531e-05,
+      "loss": 0.1165,
+      "step": 3637
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393408294493735e-05,
+      "loss": 0.1104,
+      "step": 3638
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3930889135380984e-05,
+      "loss": 0.1227,
+      "step": 3639
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927694851506805e-05,
+      "loss": 0.1094,
+      "step": 3640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3924500093700262e-05,
+      "loss": 0.1126,
+      "step": 3641
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921304862346838e-05,
+      "loss": 0.1199,
+      "step": 3642
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918109157832087e-05,
+      "loss": 0.1057,
+      "step": 3643
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3914912980541617e-05,
+      "loss": 0.1113,
+      "step": 3644
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3911716330861091e-05,
+      "loss": 0.1191,
+      "step": 3645
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3908519209176227e-05,
+      "loss": 0.1334,
+      "step": 3646
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905321615872814e-05,
+      "loss": 0.1126,
+      "step": 3647
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902123551336674e-05,
+      "loss": 0.1185,
+      "step": 3648
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898925015953708e-05,
+      "loss": 0.1135,
+      "step": 3649
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895726010109861e-05,
+      "loss": 0.1199,
+      "step": 3650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389252653419114e-05,
+      "loss": 0.1313,
+      "step": 3651
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3889326588583605e-05,
+      "loss": 0.1078,
+      "step": 3652
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.388612617367338e-05,
+      "loss": 0.1104,
+      "step": 3653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3882925289846637e-05,
+      "loss": 0.1061,
+      "step": 3654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3879723937489606e-05,
+      "loss": 0.1103,
+      "step": 3655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876522116988583e-05,
+      "loss": 0.1153,
+      "step": 3656
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873319828729907e-05,
+      "loss": 0.1142,
+      "step": 3657
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870117073099983e-05,
+      "loss": 0.1101,
+      "step": 3658
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3866913850485264e-05,
+      "loss": 0.0991,
+      "step": 3659
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3863710161272273e-05,
+      "loss": 0.1109,
+      "step": 3660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3860506005847575e-05,
+      "loss": 0.1076,
+      "step": 3661
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3857301384597796e-05,
+      "loss": 0.1161,
+      "step": 3662
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3854096297909625e-05,
+      "loss": 0.1187,
+      "step": 3663
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3850890746169796e-05,
+      "loss": 0.1123,
+      "step": 3664
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3847684729765107e-05,
+      "loss": 0.1041,
+      "step": 3665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844478249082407e-05,
+      "loss": 0.1187,
+      "step": 3666
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841271304508609e-05,
+      "loss": 0.1161,
+      "step": 3667
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383806389643067e-05,
+      "loss": 0.1177,
+      "step": 3668
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3834856025235615e-05,
+      "loss": 0.1022,
+      "step": 3669
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3831647691310514e-05,
+      "loss": 0.1146,
+      "step": 3670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3828438895042502e-05,
+      "loss": 0.1119,
+      "step": 3671
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3825229636818767e-05,
+      "loss": 0.1199,
+      "step": 3672
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3822019917026549e-05,
+      "loss": 0.1035,
+      "step": 3673
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3818809736053147e-05,
+      "loss": 0.1174,
+      "step": 3674
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815599094285917e-05,
+      "loss": 0.1157,
+      "step": 3675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3812387992112264e-05,
+      "loss": 0.106,
+      "step": 3676
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809176429919659e-05,
+      "loss": 0.1071,
+      "step": 3677
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380596440809562e-05,
+      "loss": 0.1087,
+      "step": 3678
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3802751927027722e-05,
+      "loss": 0.1184,
+      "step": 3679
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.37995389871036e-05,
+      "loss": 0.1014,
+      "step": 3680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379632558871094e-05,
+      "loss": 0.1141,
+      "step": 3681
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379311173223748e-05,
+      "loss": 0.116,
+      "step": 3682
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3789897418071022e-05,
+      "loss": 0.1114,
+      "step": 3683
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3786682646599421e-05,
+      "loss": 0.113,
+      "step": 3684
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3783467418210579e-05,
+      "loss": 0.1158,
+      "step": 3685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378025173329246e-05,
+      "loss": 0.1115,
+      "step": 3686
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377703559223309e-05,
+      "loss": 0.1013,
+      "step": 3687
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3773818995420532e-05,
+      "loss": 0.1133,
+      "step": 3688
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3770601943242916e-05,
+      "loss": 0.1089,
+      "step": 3689
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3767384436088434e-05,
+      "loss": 0.1208,
+      "step": 3690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3764166474345315e-05,
+      "loss": 0.1119,
+      "step": 3691
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760948058401855e-05,
+      "loss": 0.1176,
+      "step": 3692
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.37577291886464e-05,
+      "loss": 0.1173,
+      "step": 3693
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754509865467359e-05,
+      "loss": 0.1216,
+      "step": 3694
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375129008925318e-05,
+      "loss": 0.1223,
+      "step": 3695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748069860392381e-05,
+      "loss": 0.1121,
+      "step": 3696
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3744849179273527e-05,
+      "loss": 0.1133,
+      "step": 3697
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741628046285239e-05,
+      "loss": 0.1007,
+      "step": 3698
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3738406461816193e-05,
+      "loss": 0.1091,
+      "step": 3699
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3735184426255117e-05,
+      "loss": 0.1173,
+      "step": 3700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731961939990798e-05,
+      "loss": 0.1123,
+      "step": 3701
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3728739003412075e-05,
+      "loss": 0.1145,
+      "step": 3702
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725515616907841e-05,
+      "loss": 0.1095,
+      "step": 3703
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372229178086704e-05,
+      "loss": 0.1225,
+      "step": 3704
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3719067495678675e-05,
+      "loss": 0.1226,
+      "step": 3705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371584276173181e-05,
+      "loss": 0.1164,
+      "step": 3706
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3712617579415543e-05,
+      "loss": 0.1078,
+      "step": 3707
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3709391949119047e-05,
+      "loss": 0.1128,
+      "step": 3708
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3706165871231539e-05,
+      "loss": 0.1202,
+      "step": 3709
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702939346142286e-05,
+      "loss": 0.1065,
+      "step": 3710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3699712374240615e-05,
+      "loss": 0.1165,
+      "step": 3711
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696484955915917e-05,
+      "loss": 0.1069,
+      "step": 3712
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369325709155761e-05,
+      "loss": 0.1132,
+      "step": 3713
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369002878155519e-05,
+      "loss": 0.1141,
+      "step": 3714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3686800026298204e-05,
+      "loss": 0.1084,
+      "step": 3715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368357082617624e-05,
+      "loss": 0.1198,
+      "step": 3716
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3680341181578946e-05,
+      "loss": 0.1185,
+      "step": 3717
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367711109289603e-05,
+      "loss": 0.1081,
+      "step": 3718
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3673880560517246e-05,
+      "loss": 0.1196,
+      "step": 3719
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36706495848324e-05,
+      "loss": 0.1137,
+      "step": 3720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667418166231361e-05,
+      "loss": 0.1066,
+      "step": 3721
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3664186305104046e-05,
+      "loss": 0.1249,
+      "step": 3722
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3660954001840416e-05,
+      "loss": 0.1112,
+      "step": 3723
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3657721256830507e-05,
+      "loss": 0.1199,
+      "step": 3724
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365448807046439e-05,
+      "loss": 0.1132,
+      "step": 3725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365125444313219e-05,
+      "loss": 0.112,
+      "step": 3726
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3648020375224098e-05,
+      "loss": 0.1098,
+      "step": 3727
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3644785867130343e-05,
+      "loss": 0.1104,
+      "step": 3728
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3641550919241225e-05,
+      "loss": 0.1213,
+      "step": 3729
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363831553194708e-05,
+      "loss": 0.112,
+      "step": 3730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635079705638298e-05,
+      "loss": 0.1001,
+      "step": 3731
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3631843440705338e-05,
+      "loss": 0.1195,
+      "step": 3732
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3628606737538699e-05,
+      "loss": 0.1102,
+      "step": 3733
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362536959652893e-05,
+      "loss": 0.1094,
+      "step": 3734
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622132018066646e-05,
+      "loss": 0.1044,
+      "step": 3735
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36188940025425e-05,
+      "loss": 0.1125,
+      "step": 3736
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3615655550347204e-05,
+      "loss": 0.1149,
+      "step": 3737
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3612416661871532e-05,
+      "loss": 0.1064,
+      "step": 3738
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609177337506294e-05,
+      "loss": 0.1127,
+      "step": 3739
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3605937577642366e-05,
+      "loss": 0.1067,
+      "step": 3740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3602697382670664e-05,
+      "loss": 0.126,
+      "step": 3741
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3599456752982172e-05,
+      "loss": 0.115,
+      "step": 3742
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596215688967917e-05,
+      "loss": 0.1228,
+      "step": 3743
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3592974191018973e-05,
+      "loss": 0.1124,
+      "step": 3744
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3589732259526481e-05,
+      "loss": 0.1074,
+      "step": 3745
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3586489894881618e-05,
+      "loss": 0.1175,
+      "step": 3746
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3583247097475628e-05,
+      "loss": 0.1075,
+      "step": 3747
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3580003867699801e-05,
+      "loss": 0.1121,
+      "step": 3748
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576760205945474e-05,
+      "loss": 0.1117,
+      "step": 3749
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573516112604042e-05,
+      "loss": 0.1163,
+      "step": 3750
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3570271588066957e-05,
+      "loss": 0.1192,
+      "step": 3751
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356702663272571e-05,
+      "loss": 0.1128,
+      "step": 3752
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3563781246971855e-05,
+      "loss": 0.119,
+      "step": 3753
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3560535431196991e-05,
+      "loss": 0.1174,
+      "step": 3754
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3557289185792777e-05,
+      "loss": 0.1189,
+      "step": 3755
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3554042511150913e-05,
+      "loss": 0.1028,
+      "step": 3756
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3550795407663158e-05,
+      "loss": 0.1075,
+      "step": 3757
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3547547875721327e-05,
+      "loss": 0.1144,
+      "step": 3758
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3544299915717276e-05,
+      "loss": 0.1149,
+      "step": 3759
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3541051528042915e-05,
+      "loss": 0.1113,
+      "step": 3760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537802713090216e-05,
+      "loss": 0.1224,
+      "step": 3761
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3534553471251185e-05,
+      "loss": 0.1173,
+      "step": 3762
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35313038029179e-05,
+      "loss": 0.1147,
+      "step": 3763
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3528053708482474e-05,
+      "loss": 0.1037,
+      "step": 3764
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3524803188337074e-05,
+      "loss": 0.1083,
+      "step": 3765
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3521552242873926e-05,
+      "loss": 0.1085,
+      "step": 3766
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3518300872485304e-05,
+      "loss": 0.1094,
+      "step": 3767
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3515049077563531e-05,
+      "loss": 0.1153,
+      "step": 3768
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3511796858500978e-05,
+      "loss": 0.1146,
+      "step": 3769
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350854421569008e-05,
+      "loss": 0.11,
+      "step": 3770
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3505291149523309e-05,
+      "loss": 0.1167,
+      "step": 3771
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502037660393195e-05,
+      "loss": 0.1107,
+      "step": 3772
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3498783748692317e-05,
+      "loss": 0.1167,
+      "step": 3773
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349552941481331e-05,
+      "loss": 0.1211,
+      "step": 3774
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492274659148847e-05,
+      "loss": 0.1054,
+      "step": 3775
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348901948209167e-05,
+      "loss": 0.1252,
+      "step": 3776
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3485763884034558e-05,
+      "loss": 0.1122,
+      "step": 3777
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3482507865370342e-05,
+      "loss": 0.1358,
+      "step": 3778
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3479251426491917e-05,
+      "loss": 0.1098,
+      "step": 3779
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475994567792208e-05,
+      "loss": 0.1118,
+      "step": 3780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347273728966421e-05,
+      "loss": 0.119,
+      "step": 3781
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3469479592500954e-05,
+      "loss": 0.1141,
+      "step": 3782
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466221476695529e-05,
+      "loss": 0.1252,
+      "step": 3783
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3462962942641075e-05,
+      "loss": 0.1172,
+      "step": 3784
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3459703990730781e-05,
+      "loss": 0.111,
+      "step": 3785
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345644462135788e-05,
+      "loss": 0.1215,
+      "step": 3786
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3453184834915671e-05,
+      "loss": 0.1251,
+      "step": 3787
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3449924631797485e-05,
+      "loss": 0.1071,
+      "step": 3788
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3446664012396718e-05,
+      "loss": 0.1095,
+      "step": 3789
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443402977106807e-05,
+      "loss": 0.1181,
+      "step": 3790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3440141526321243e-05,
+      "loss": 0.1266,
+      "step": 3791
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436879660433567e-05,
+      "loss": 0.1135,
+      "step": 3792
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343361737983737e-05,
+      "loss": 0.1131,
+      "step": 3793
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343035468492629e-05,
+      "loss": 0.1218,
+      "step": 3794
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427091576094022e-05,
+      "loss": 0.1043,
+      "step": 3795
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34238280537343e-05,
+      "loss": 0.1215,
+      "step": 3796
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3420564118240922e-05,
+      "loss": 0.1074,
+      "step": 3797
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417299770007728e-05,
+      "loss": 0.123,
+      "step": 3798
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3414035009428598e-05,
+      "loss": 0.1103,
+      "step": 3799
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3410769836897485e-05,
+      "loss": 0.1224,
+      "step": 3800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3407504252808372e-05,
+      "loss": 0.1094,
+      "step": 3801
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3404238257555296e-05,
+      "loss": 0.1208,
+      "step": 3802
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3400971851532352e-05,
+      "loss": 0.103,
+      "step": 3803
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3397705035133672e-05,
+      "loss": 0.1179,
+      "step": 3804
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.339443780875345e-05,
+      "loss": 0.1159,
+      "step": 3805
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3391170172785918e-05,
+      "loss": 0.1176,
+      "step": 3806
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.338790212762537e-05,
+      "loss": 0.1152,
+      "step": 3807
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3384633673666132e-05,
+      "loss": 0.1085,
+      "step": 3808
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381364811302597e-05,
+      "loss": 0.1081,
+      "step": 3809
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3378095540929204e-05,
+      "loss": 0.1018,
+      "step": 3810
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3374825862940425e-05,
+      "loss": 0.1115,
+      "step": 3811
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33715557777308e-05,
+      "loss": 0.1153,
+      "step": 3812
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3368285285694915e-05,
+      "loss": 0.1086,
+      "step": 3813
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3365014387227393e-05,
+      "loss": 0.1038,
+      "step": 3814
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336174308272292e-05,
+      "loss": 0.1241,
+      "step": 3815
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3358471372576229e-05,
+      "loss": 0.1176,
+      "step": 3816
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3355199257182087e-05,
+      "loss": 0.1174,
+      "step": 3817
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3351926736935331e-05,
+      "loss": 0.1131,
+      "step": 3818
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3348653812230836e-05,
+      "loss": 0.1146,
+      "step": 3819
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3345380483463524e-05,
+      "loss": 0.1049,
+      "step": 3820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.334210675102837e-05,
+      "loss": 0.1205,
+      "step": 3821
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3338832615320399e-05,
+      "loss": 0.1158,
+      "step": 3822
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3335558076734678e-05,
+      "loss": 0.1249,
+      "step": 3823
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333228313566633e-05,
+      "loss": 0.1133,
+      "step": 3824
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329007792510517e-05,
+      "loss": 0.1085,
+      "step": 3825
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3325732047662467e-05,
+      "loss": 0.1279,
+      "step": 3826
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322455901517435e-05,
+      "loss": 0.1168,
+      "step": 3827
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331917935447074e-05,
+      "loss": 0.1041,
+      "step": 3828
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315902406917746e-05,
+      "loss": 0.1102,
+      "step": 3829
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312625059253858e-05,
+      "loss": 0.1015,
+      "step": 3830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309347311874536e-05,
+      "loss": 0.1094,
+      "step": 3831
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330606916517529e-05,
+      "loss": 0.1125,
+      "step": 3832
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.1063,
+      "step": 3833
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3299511675399288e-05,
+      "loss": 0.1014,
+      "step": 3834
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3296232333113786e-05,
+      "loss": 0.1179,
+      "step": 3835
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292952593090869e-05,
+      "loss": 0.1209,
+      "step": 3836
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.328967245572628e-05,
+      "loss": 0.1098,
+      "step": 3837
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3286391921415817e-05,
+      "loss": 0.1141,
+      "step": 3838
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3283110990555325e-05,
+      "loss": 0.1066,
+      "step": 3839
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3279829663540693e-05,
+      "loss": 0.1098,
+      "step": 3840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.327654794076786e-05,
+      "loss": 0.1054,
+      "step": 3841
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3273265822632814e-05,
+      "loss": 0.1148,
+      "step": 3842
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3269983309531584e-05,
+      "loss": 0.1156,
+      "step": 3843
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3266700401860262e-05,
+      "loss": 0.0946,
+      "step": 3844
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3263417100014968e-05,
+      "loss": 0.1152,
+      "step": 3845
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3260133404391881e-05,
+      "loss": 0.116,
+      "step": 3846
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.325684931538723e-05,
+      "loss": 0.115,
+      "step": 3847
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3253564833397283e-05,
+      "loss": 0.1045,
+      "step": 3848
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250279958818364e-05,
+      "loss": 0.1154,
+      "step": 3849
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3246994692046837e-05,
+      "loss": 0.099,
+      "step": 3850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3243709033479112e-05,
+      "loss": 0.1198,
+      "step": 3851
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324042298351166e-05,
+      "loss": 0.113,
+      "step": 3852
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3237136542540985e-05,
+      "loss": 0.1159,
+      "step": 3853
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3233849710963638e-05,
+      "loss": 0.1196,
+      "step": 3854
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3230562489176231e-05,
+      "loss": 0.1145,
+      "step": 3855
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3227274877575409e-05,
+      "loss": 0.1203,
+      "step": 3856
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3223986876557869e-05,
+      "loss": 0.111,
+      "step": 3857
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3220698486520358e-05,
+      "loss": 0.1119,
+      "step": 3858
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3217409707859665e-05,
+      "loss": 0.1062,
+      "step": 3859
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3214120540972626e-05,
+      "loss": 0.0989,
+      "step": 3860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3210830986256132e-05,
+      "loss": 0.1102,
+      "step": 3861
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3207541044107109e-05,
+      "loss": 0.127,
+      "step": 3862
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3204250714922538e-05,
+      "loss": 0.1147,
+      "step": 3863
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3200959999099441e-05,
+      "loss": 0.1096,
+      "step": 3864
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3197668897034896e-05,
+      "loss": 0.1124,
+      "step": 3865
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3194377409126016e-05,
+      "loss": 0.1099,
+      "step": 3866
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191085535769968e-05,
+      "loss": 0.1146,
+      "step": 3867
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3187793277363963e-05,
+      "loss": 0.119,
+      "step": 3868
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.318450063430526e-05,
+      "loss": 0.1076,
+      "step": 3869
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.318120760699116e-05,
+      "loss": 0.1166,
+      "step": 3870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3177914195819018e-05,
+      "loss": 0.1072,
+      "step": 3871
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317462040118623e-05,
+      "loss": 0.1111,
+      "step": 3872
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3171326223490234e-05,
+      "loss": 0.1128,
+      "step": 3873
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3168031663128526e-05,
+      "loss": 0.1137,
+      "step": 3874
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3164736720498644e-05,
+      "loss": 0.1087,
+      "step": 3875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3161441395998158e-05,
+      "loss": 0.1267,
+      "step": 3876
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158145690024706e-05,
+      "loss": 0.1163,
+      "step": 3877
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315484960297596e-05,
+      "loss": 0.1069,
+      "step": 3878
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315155313524964e-05,
+      "loss": 0.11,
+      "step": 3879
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3148256287243508e-05,
+      "loss": 0.1047,
+      "step": 3880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3144959059355383e-05,
+      "loss": 0.1228,
+      "step": 3881
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314166145198311e-05,
+      "loss": 0.1047,
+      "step": 3882
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3138363465524607e-05,
+      "loss": 0.1135,
+      "step": 3883
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3135065100377816e-05,
+      "loss": 0.1085,
+      "step": 3884
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3131766356940728e-05,
+      "loss": 0.1084,
+      "step": 3885
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128467235611389e-05,
+      "loss": 0.1134,
+      "step": 3886
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125167736787886e-05,
+      "loss": 0.1262,
+      "step": 3887
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121867860868346e-05,
+      "loss": 0.1148,
+      "step": 3888
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3118567608250947e-05,
+      "loss": 0.1179,
+      "step": 3889
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115266979333917e-05,
+      "loss": 0.1145,
+      "step": 3890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3111965974515517e-05,
+      "loss": 0.1192,
+      "step": 3891
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108664594194063e-05,
+      "loss": 0.1103,
+      "step": 3892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3105362838767914e-05,
+      "loss": 0.1309,
+      "step": 3893
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102060708635475e-05,
+      "loss": 0.1163,
+      "step": 3894
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309875820419519e-05,
+      "loss": 0.1099,
+      "step": 3895
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3095455325845559e-05,
+      "loss": 0.1129,
+      "step": 3896
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3092152073985121e-05,
+      "loss": 0.1133,
+      "step": 3897
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088848449012454e-05,
+      "loss": 0.1204,
+      "step": 3898
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3085544451326196e-05,
+      "loss": 0.1217,
+      "step": 3899
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082240081325015e-05,
+      "loss": 0.1007,
+      "step": 3900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3078935339407632e-05,
+      "loss": 0.1268,
+      "step": 3901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3075630225972813e-05,
+      "loss": 0.1264,
+      "step": 3902
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3072324741419364e-05,
+      "loss": 0.1064,
+      "step": 3903
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069018886146146e-05,
+      "loss": 0.1091,
+      "step": 3904
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306571266055205e-05,
+      "loss": 0.1171,
+      "step": 3905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306240606503602e-05,
+      "loss": 0.1343,
+      "step": 3906
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059099099997046e-05,
+      "loss": 0.11,
+      "step": 3907
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305579176583416e-05,
+      "loss": 0.1099,
+      "step": 3908
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305248406294644e-05,
+      "loss": 0.1048,
+      "step": 3909
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049175991733006e-05,
+      "loss": 0.1189,
+      "step": 3910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045867552593024e-05,
+      "loss": 0.1165,
+      "step": 3911
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3042558745925704e-05,
+      "loss": 0.1178,
+      "step": 3912
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3039249572130304e-05,
+      "loss": 0.1089,
+      "step": 3913
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3035940031606117e-05,
+      "loss": 0.1107,
+      "step": 3914
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3032630124752493e-05,
+      "loss": 0.1082,
+      "step": 3915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3029319851968811e-05,
+      "loss": 0.1154,
+      "step": 3916
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026009213654513e-05,
+      "loss": 0.1123,
+      "step": 3917
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3022698210209069e-05,
+      "loss": 0.1142,
+      "step": 3918
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3019386842031998e-05,
+      "loss": 0.1111,
+      "step": 3919
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3016075109522865e-05,
+      "loss": 0.1137,
+      "step": 3920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3012763013081278e-05,
+      "loss": 0.1229,
+      "step": 3921
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3009450553106886e-05,
+      "loss": 0.1207,
+      "step": 3922
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3006137729999391e-05,
+      "loss": 0.121,
+      "step": 3923
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3002824544158525e-05,
+      "loss": 0.1038,
+      "step": 3924
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2999510995984074e-05,
+      "loss": 0.1125,
+      "step": 3925
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996197085875867e-05,
+      "loss": 0.1114,
+      "step": 3926
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2992882814233773e-05,
+      "loss": 0.1213,
+      "step": 3927
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2989568181457704e-05,
+      "loss": 0.1125,
+      "step": 3928
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2986253187947622e-05,
+      "loss": 0.1183,
+      "step": 3929
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2982937834103523e-05,
+      "loss": 0.0982,
+      "step": 3930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2979622120325455e-05,
+      "loss": 0.115,
+      "step": 3931
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2976306047013508e-05,
+      "loss": 0.1116,
+      "step": 3932
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2972989614567808e-05,
+      "loss": 0.1212,
+      "step": 3933
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2969672823388533e-05,
+      "loss": 0.1045,
+      "step": 3934
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29663556738759e-05,
+      "loss": 0.1144,
+      "step": 3935
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963038166430173e-05,
+      "loss": 0.1082,
+      "step": 3936
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959720301451655e-05,
+      "loss": 0.107,
+      "step": 3937
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295640207934069e-05,
+      "loss": 0.119,
+      "step": 3938
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2953083500497679e-05,
+      "loss": 0.1102,
+      "step": 3939
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2949764565323046e-05,
+      "loss": 0.1104,
+      "step": 3940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2946445274217267e-05,
+      "loss": 0.1141,
+      "step": 3941
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943125627580875e-05,
+      "loss": 0.1078,
+      "step": 3942
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2939805625814416e-05,
+      "loss": 0.1265,
+      "step": 3943
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2936485269318504e-05,
+      "loss": 0.1216,
+      "step": 3944
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293316455849379e-05,
+      "loss": 0.1235,
+      "step": 3945
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2929843493740959e-05,
+      "loss": 0.1095,
+      "step": 3946
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2926522075460746e-05,
+      "loss": 0.1148,
+      "step": 3947
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2923200304053932e-05,
+      "loss": 0.1145,
+      "step": 3948
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291987817992133e-05,
+      "loss": 0.1143,
+      "step": 3949
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2916555703463805e-05,
+      "loss": 0.1176,
+      "step": 3950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2913232875082262e-05,
+      "loss": 0.1139,
+      "step": 3951
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2909909695177647e-05,
+      "loss": 0.1171,
+      "step": 3952
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2906586164150944e-05,
+      "loss": 0.119,
+      "step": 3953
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2903262282403195e-05,
+      "loss": 0.121,
+      "step": 3954
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899938050335464e-05,
+      "loss": 0.1161,
+      "step": 3955
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896613468348869e-05,
+      "loss": 0.111,
+      "step": 3956
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289328853684457e-05,
+      "loss": 0.1207,
+      "step": 3957
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2889963256223767e-05,
+      "loss": 0.1053,
+      "step": 3958
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2886637626887704e-05,
+      "loss": 0.105,
+      "step": 3959
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883311649237665e-05,
+      "loss": 0.1178,
+      "step": 3960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2879985323674973e-05,
+      "loss": 0.1165,
+      "step": 3961
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876658650600999e-05,
+      "loss": 0.1154,
+      "step": 3962
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2873331630417157e-05,
+      "loss": 0.1083,
+      "step": 3963
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.287000426352489e-05,
+      "loss": 0.1123,
+      "step": 3964
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2866676550325704e-05,
+      "loss": 0.1182,
+      "step": 3965
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2863348491221129e-05,
+      "loss": 0.1143,
+      "step": 3966
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2860020086612741e-05,
+      "loss": 0.1162,
+      "step": 3967
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2856691336902163e-05,
+      "loss": 0.1037,
+      "step": 3968
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2853362242491054e-05,
+      "loss": 0.1223,
+      "step": 3969
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2850032803781119e-05,
+      "loss": 0.1055,
+      "step": 3970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28467030211741e-05,
+      "loss": 0.1187,
+      "step": 3971
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2843372895071782e-05,
+      "loss": 0.1115,
+      "step": 3972
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2840042425875996e-05,
+      "loss": 0.1102,
+      "step": 3973
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836711613988607e-05,
+      "loss": 0.1156,
+      "step": 3974
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2833380459811524e-05,
+      "loss": 0.1174,
+      "step": 3975
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830048963746703e-05,
+      "loss": 0.1242,
+      "step": 3976
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2826717126196134e-05,
+      "loss": 0.1204,
+      "step": 3977
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282338494756185e-05,
+      "loss": 0.1043,
+      "step": 3978
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820052428245928e-05,
+      "loss": 0.1054,
+      "step": 3979
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.281671956865048e-05,
+      "loss": 0.1226,
+      "step": 3980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2813386369177667e-05,
+      "loss": 0.1068,
+      "step": 3981
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810052830229684e-05,
+      "loss": 0.1164,
+      "step": 3982
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2806718952208772e-05,
+      "loss": 0.1107,
+      "step": 3983
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2803384735517209e-05,
+      "loss": 0.1093,
+      "step": 3984
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800050180557322e-05,
+      "loss": 0.1156,
+      "step": 3985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2796715287731463e-05,
+      "loss": 0.1044,
+      "step": 3986
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2793380057442038e-05,
+      "loss": 0.1137,
+      "step": 3987
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790044490091496e-05,
+      "loss": 0.114,
+      "step": 3988
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2786708586082312e-05,
+      "loss": 0.1052,
+      "step": 3989
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2783372345817013e-05,
+      "loss": 0.1123,
+      "step": 3990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780035769698172e-05,
+      "loss": 0.1106,
+      "step": 3991
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2776698858128383e-05,
+      "loss": 0.1114,
+      "step": 3992
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2773361611510293e-05,
+      "loss": 0.1126,
+      "step": 3993
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.27700240302466e-05,
+      "loss": 0.1195,
+      "step": 3994
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2766686114740017e-05,
+      "loss": 0.11,
+      "step": 3995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2763347865393317e-05,
+      "loss": 0.1163,
+      "step": 3996
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276000928260931e-05,
+      "loss": 0.1122,
+      "step": 3997
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2756670366790837e-05,
+      "loss": 0.1123,
+      "step": 3998
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2753331118340793e-05,
+      "loss": 0.1151,
+      "step": 3999
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27499915376621e-05,
+      "loss": 0.1016,
+      "step": 4000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274665162515773e-05,
+      "loss": 0.1126,
+      "step": 4001
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274331138123069e-05,
+      "loss": 0.1097,
+      "step": 4002
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2739970806284028e-05,
+      "loss": 0.1083,
+      "step": 4003
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2736629900720832e-05,
+      "loss": 0.1074,
+      "step": 4004
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2733288664944226e-05,
+      "loss": 0.1072,
+      "step": 4005
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2729947099357385e-05,
+      "loss": 0.1165,
+      "step": 4006
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2726605204363512e-05,
+      "loss": 0.1083,
+      "step": 4007
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2723262980365852e-05,
+      "loss": 0.1291,
+      "step": 4008
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719920427767698e-05,
+      "loss": 0.1241,
+      "step": 4009
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2716577546972371e-05,
+      "loss": 0.1355,
+      "step": 4010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271323433838324e-05,
+      "loss": 0.1251,
+      "step": 4011
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2709890802403712e-05,
+      "loss": 0.1301,
+      "step": 4012
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270654693943723e-05,
+      "loss": 0.13,
+      "step": 4013
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2703202749887282e-05,
+      "loss": 0.1471,
+      "step": 4014
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2699858234157385e-05,
+      "loss": 0.1216,
+      "step": 4015
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2696513392651111e-05,
+      "loss": 0.1291,
+      "step": 4016
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269316822577206e-05,
+      "loss": 0.1632,
+      "step": 4017
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268982273392387e-05,
+      "loss": 0.1318,
+      "step": 4018
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2686476917510226e-05,
+      "loss": 0.1293,
+      "step": 4019
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268313077693485e-05,
+      "loss": 0.1324,
+      "step": 4020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2679784312601495e-05,
+      "loss": 0.1459,
+      "step": 4021
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2676437524913967e-05,
+      "loss": 0.1303,
+      "step": 4022
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673090414276101e-05,
+      "loss": 0.1417,
+      "step": 4023
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2669742981091771e-05,
+      "loss": 0.1433,
+      "step": 4024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2666395225764898e-05,
+      "loss": 0.1514,
+      "step": 4025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2663047148699428e-05,
+      "loss": 0.1475,
+      "step": 4026
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2659698750299365e-05,
+      "loss": 0.1531,
+      "step": 4027
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656350030968734e-05,
+      "loss": 0.1382,
+      "step": 4028
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653000991111604e-05,
+      "loss": 0.1486,
+      "step": 4029
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2649651631132089e-05,
+      "loss": 0.1407,
+      "step": 4030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264630195143434e-05,
+      "loss": 0.1208,
+      "step": 4031
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2642951952422533e-05,
+      "loss": 0.1606,
+      "step": 4032
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26396016345009e-05,
+      "loss": 0.1362,
+      "step": 4033
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2636250998073707e-05,
+      "loss": 0.1484,
+      "step": 4034
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2632900043545248e-05,
+      "loss": 0.138,
+      "step": 4035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2629548771319873e-05,
+      "loss": 0.1383,
+      "step": 4036
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2626197181801956e-05,
+      "loss": 0.1397,
+      "step": 4037
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2622845275395912e-05,
+      "loss": 0.1441,
+      "step": 4038
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2619493052506198e-05,
+      "loss": 0.1432,
+      "step": 4039
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616140513537311e-05,
+      "loss": 0.1242,
+      "step": 4040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2612787658893777e-05,
+      "loss": 0.1417,
+      "step": 4041
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2609434488980168e-05,
+      "loss": 0.1288,
+      "step": 4042
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2606081004201096e-05,
+      "loss": 0.1468,
+      "step": 4043
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2602727204961198e-05,
+      "loss": 0.1468,
+      "step": 4044
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2599373091665161e-05,
+      "loss": 0.1412,
+      "step": 4045
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2596018664717713e-05,
+      "loss": 0.1296,
+      "step": 4046
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2592663924523607e-05,
+      "loss": 0.1394,
+      "step": 4047
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258930887148764e-05,
+      "loss": 0.142,
+      "step": 4048
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258595350601465e-05,
+      "loss": 0.1428,
+      "step": 4049
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2582597828509508e-05,
+      "loss": 0.1298,
+      "step": 4050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2579241839377123e-05,
+      "loss": 0.1438,
+      "step": 4051
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2575885539022448e-05,
+      "loss": 0.1472,
+      "step": 4052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2572528927850463e-05,
+      "loss": 0.147,
+      "step": 4053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2569172006266192e-05,
+      "loss": 0.1347,
+      "step": 4054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2565814774674698e-05,
+      "loss": 0.1418,
+      "step": 4055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2562457233481077e-05,
+      "loss": 0.128,
+      "step": 4056
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2559099383090465e-05,
+      "loss": 0.1481,
+      "step": 4057
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2555741223908033e-05,
+      "loss": 0.1335,
+      "step": 4058
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2552382756338993e-05,
+      "loss": 0.1454,
+      "step": 4059
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549023980788588e-05,
+      "loss": 0.1612,
+      "step": 4060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2545664897662108e-05,
+      "loss": 0.1384,
+      "step": 4061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542305507364872e-05,
+      "loss": 0.1485,
+      "step": 4062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538945810302234e-05,
+      "loss": 0.1443,
+      "step": 4063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253558580687959e-05,
+      "loss": 0.1452,
+      "step": 4064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2532225497502381e-05,
+      "loss": 0.1315,
+      "step": 4065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252886488257607e-05,
+      "loss": 0.1467,
+      "step": 4066
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2525503962506163e-05,
+      "loss": 0.1364,
+      "step": 4067
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522142737698202e-05,
+      "loss": 0.1489,
+      "step": 4068
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251878120855777e-05,
+      "loss": 0.138,
+      "step": 4069
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515419375490479e-05,
+      "loss": 0.1495,
+      "step": 4070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512057238901986e-05,
+      "loss": 0.1457,
+      "step": 4071
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2508694799197981e-05,
+      "loss": 0.1356,
+      "step": 4072
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2505332056784184e-05,
+      "loss": 0.1388,
+      "step": 4073
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2501969012066366e-05,
+      "loss": 0.1435,
+      "step": 4074
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2498605665450321e-05,
+      "loss": 0.1401,
+      "step": 4075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2495242017341887e-05,
+      "loss": 0.166,
+      "step": 4076
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2491878068146934e-05,
+      "loss": 0.1349,
+      "step": 4077
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2488513818271372e-05,
+      "loss": 0.1535,
+      "step": 4078
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485149268121148e-05,
+      "loss": 0.1393,
+      "step": 4079
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248178441810224e-05,
+      "loss": 0.1388,
+      "step": 4080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2478419268620662e-05,
+      "loss": 0.1421,
+      "step": 4081
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475053820082476e-05,
+      "loss": 0.1548,
+      "step": 4082
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2471688072893763e-05,
+      "loss": 0.1399,
+      "step": 4083
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2468322027460653e-05,
+      "loss": 0.131,
+      "step": 4084
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2464955684189307e-05,
+      "loss": 0.15,
+      "step": 4085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246158904348592e-05,
+      "loss": 0.1319,
+      "step": 4086
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2458222105756723e-05,
+      "loss": 0.1366,
+      "step": 4087
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.1403,
+      "step": 4088
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.245148734084603e-05,
+      "loss": 0.1443,
+      "step": 4089
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448119514477172e-05,
+      "loss": 0.1321,
+      "step": 4090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.24447513927078e-05,
+      "loss": 0.1338,
+      "step": 4091
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441382975944325e-05,
+      "loss": 0.1291,
+      "step": 4092
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243801426459319e-05,
+      "loss": 0.1328,
+      "step": 4093
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2434645259060884e-05,
+      "loss": 0.1233,
+      "step": 4094
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2431275959753924e-05,
+      "loss": 0.1259,
+      "step": 4095
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2427906367078861e-05,
+      "loss": 0.1407,
+      "step": 4096
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424536481442287e-05,
+      "loss": 0.1354,
+      "step": 4097
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2421166303250827e-05,
+      "loss": 0.1469,
+      "step": 4098
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.241779583291114e-05,
+      "loss": 0.1377,
+      "step": 4099
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414425070829921e-05,
+      "loss": 0.1251,
+      "step": 4100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2411054017413901e-05,
+      "loss": 0.1277,
+      "step": 4101
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2407682673069848e-05,
+      "loss": 0.145,
+      "step": 4102
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2404311038204558e-05,
+      "loss": 0.1427,
+      "step": 4103
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240093911322487e-05,
+      "loss": 0.1464,
+      "step": 4104
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2397566898537655e-05,
+      "loss": 0.135,
+      "step": 4105
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2394194394549815e-05,
+      "loss": 0.141,
+      "step": 4106
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2390821601668295e-05,
+      "loss": 0.1341,
+      "step": 4107
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238744852030007e-05,
+      "loss": 0.133,
+      "step": 4108
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2384075150852147e-05,
+      "loss": 0.1293,
+      "step": 4109
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2380701493731576e-05,
+      "loss": 0.1216,
+      "step": 4110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2377327549345435e-05,
+      "loss": 0.1392,
+      "step": 4111
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373953318100833e-05,
+      "loss": 0.1346,
+      "step": 4112
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2370578800404927e-05,
+      "loss": 0.1282,
+      "step": 4113
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2367203996664901e-05,
+      "loss": 0.1308,
+      "step": 4114
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2363828907287964e-05,
+      "loss": 0.1375,
+      "step": 4115
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2360453532681376e-05,
+      "loss": 0.1343,
+      "step": 4116
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357077873252425e-05,
+      "loss": 0.1431,
+      "step": 4117
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2353701929408425e-05,
+      "loss": 0.137,
+      "step": 4118
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.235032570155674e-05,
+      "loss": 0.1507,
+      "step": 4119
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2346949190104757e-05,
+      "loss": 0.1431,
+      "step": 4120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.23435723954599e-05,
+      "loss": 0.1235,
+      "step": 4121
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2340195318029623e-05,
+      "loss": 0.1441,
+      "step": 4122
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2336817958221427e-05,
+      "loss": 0.1347,
+      "step": 4123
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2333440316442834e-05,
+      "loss": 0.1373,
+      "step": 4124
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2330062393101404e-05,
+      "loss": 0.1351,
+      "step": 4125
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2326684188604733e-05,
+      "loss": 0.1415,
+      "step": 4126
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2323305703360453e-05,
+      "loss": 0.1446,
+      "step": 4127
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319926937776217e-05,
+      "loss": 0.126,
+      "step": 4128
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.231654789225973e-05,
+      "loss": 0.1519,
+      "step": 4129
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2313168567218718e-05,
+      "loss": 0.1281,
+      "step": 4130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2309788963060942e-05,
+      "loss": 0.1379,
+      "step": 4131
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306409080194209e-05,
+      "loss": 0.1309,
+      "step": 4132
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2303028919026339e-05,
+      "loss": 0.1394,
+      "step": 4133
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299648479965205e-05,
+      "loss": 0.1219,
+      "step": 4134
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.22962677634187e-05,
+      "loss": 0.1383,
+      "step": 4135
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2292886769794756e-05,
+      "loss": 0.1352,
+      "step": 4136
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2289505499501341e-05,
+      "loss": 0.1383,
+      "step": 4137
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2286123952946454e-05,
+      "loss": 0.1319,
+      "step": 4138
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2282742130538121e-05,
+      "loss": 0.1222,
+      "step": 4139
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2279360032684415e-05,
+      "loss": 0.1302,
+      "step": 4140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2275977659793427e-05,
+      "loss": 0.148,
+      "step": 4141
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2272595012273292e-05,
+      "loss": 0.1389,
+      "step": 4142
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269212090532176e-05,
+      "loss": 0.1309,
+      "step": 4143
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2265828894978273e-05,
+      "loss": 0.1267,
+      "step": 4144
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2262445426019818e-05,
+      "loss": 0.1396,
+      "step": 4145
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225906168406507e-05,
+      "loss": 0.1262,
+      "step": 4146
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225567766952233e-05,
+      "loss": 0.1252,
+      "step": 4147
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2252293382799927e-05,
+      "loss": 0.1394,
+      "step": 4148
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2248908824306219e-05,
+      "loss": 0.1407,
+      "step": 4149
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2245523994449607e-05,
+      "loss": 0.1346,
+      "step": 4150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2242138893638516e-05,
+      "loss": 0.1335,
+      "step": 4151
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2238753522281407e-05,
+      "loss": 0.1279,
+      "step": 4152
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2235367880786773e-05,
+      "loss": 0.1314,
+      "step": 4153
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223198196956314e-05,
+      "loss": 0.1226,
+      "step": 4154
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2228595789019066e-05,
+      "loss": 0.1394,
+      "step": 4155
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.1343,
+      "step": 4156
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221822621603997e-05,
+      "loss": 0.1124,
+      "step": 4157
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2218435635550278e-05,
+      "loss": 0.1269,
+      "step": 4158
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221504838181068e-05,
+      "loss": 0.1405,
+      "step": 4159
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2211660860793922e-05,
+      "loss": 0.133,
+      "step": 4160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220827307290875e-05,
+      "loss": 0.1427,
+      "step": 4161
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2204885018563955e-05,
+      "loss": 0.1286,
+      "step": 4162
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2201496698168356e-05,
+      "loss": 0.1285,
+      "step": 4163
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198108112130797e-05,
+      "loss": 0.1312,
+      "step": 4164
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2194719260860161e-05,
+      "loss": 0.1197,
+      "step": 4165
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191330144765368e-05,
+      "loss": 0.1325,
+      "step": 4166
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2187940764255348e-05,
+      "loss": 0.1285,
+      "step": 4167
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218455111973909e-05,
+      "loss": 0.1296,
+      "step": 4168
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2181161211625605e-05,
+      "loss": 0.1283,
+      "step": 4169
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177771040323923e-05,
+      "loss": 0.1101,
+      "step": 4170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2174380606243124e-05,
+      "loss": 0.1356,
+      "step": 4171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2170989909792313e-05,
+      "loss": 0.1273,
+      "step": 4172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216759895138062e-05,
+      "loss": 0.1229,
+      "step": 4173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164207731417213e-05,
+      "loss": 0.1175,
+      "step": 4174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2160816250311298e-05,
+      "loss": 0.1481,
+      "step": 4175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.21574245084721e-05,
+      "loss": 0.1458,
+      "step": 4176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215403250630888e-05,
+      "loss": 0.1337,
+      "step": 4177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2150640244230939e-05,
+      "loss": 0.1432,
+      "step": 4178
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2147247722647595e-05,
+      "loss": 0.1327,
+      "step": 4179
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2143854941968203e-05,
+      "loss": 0.133,
+      "step": 4180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2140461902602154e-05,
+      "loss": 0.1229,
+      "step": 4181
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2137068604958866e-05,
+      "loss": 0.1252,
+      "step": 4182
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2133675049447786e-05,
+      "loss": 0.1367,
+      "step": 4183
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21302812364784e-05,
+      "loss": 0.1203,
+      "step": 4184
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2126887166460218e-05,
+      "loss": 0.144,
+      "step": 4185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2123492839802777e-05,
+      "loss": 0.1269,
+      "step": 4186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2120098256915658e-05,
+      "loss": 0.1228,
+      "step": 4187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2116703418208462e-05,
+      "loss": 0.1358,
+      "step": 4188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211330832409083e-05,
+      "loss": 0.1317,
+      "step": 4189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2109912974972424e-05,
+      "loss": 0.1383,
+      "step": 4190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2106517371262938e-05,
+      "loss": 0.1423,
+      "step": 4191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103121513372109e-05,
+      "loss": 0.1465,
+      "step": 4192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2099725401709685e-05,
+      "loss": 0.1242,
+      "step": 4193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2096329036685469e-05,
+      "loss": 0.1381,
+      "step": 4194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2092932418709267e-05,
+      "loss": 0.12,
+      "step": 4195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2089535548190935e-05,
+      "loss": 0.1219,
+      "step": 4196
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2086138425540357e-05,
+      "loss": 0.1324,
+      "step": 4197
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082741051167438e-05,
+      "loss": 0.125,
+      "step": 4198
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079343425482126e-05,
+      "loss": 0.1318,
+      "step": 4199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2075945548894388e-05,
+      "loss": 0.1205,
+      "step": 4200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2072547421814229e-05,
+      "loss": 0.1298,
+      "step": 4201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206914904465168e-05,
+      "loss": 0.1239,
+      "step": 4202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2065750417816806e-05,
+      "loss": 0.124,
+      "step": 4203
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2062351541719697e-05,
+      "loss": 0.1198,
+      "step": 4204
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2058952416770478e-05,
+      "loss": 0.1249,
+      "step": 4205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20555530433793e-05,
+      "loss": 0.1301,
+      "step": 4206
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2052153421956343e-05,
+      "loss": 0.1406,
+      "step": 4207
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2048753552911827e-05,
+      "loss": 0.1319,
+      "step": 4208
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2045353436655994e-05,
+      "loss": 0.1201,
+      "step": 4209
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2041953073599109e-05,
+      "loss": 0.1292,
+      "step": 4210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2038552464151479e-05,
+      "loss": 0.1287,
+      "step": 4211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2035151608723438e-05,
+      "loss": 0.1156,
+      "step": 4212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2031750507725344e-05,
+      "loss": 0.132,
+      "step": 4213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2028349161567588e-05,
+      "loss": 0.1399,
+      "step": 4214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2024947570660599e-05,
+      "loss": 0.1392,
+      "step": 4215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2021545735414816e-05,
+      "loss": 0.1334,
+      "step": 4216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018143656240724e-05,
+      "loss": 0.1211,
+      "step": 4217
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2014741333548836e-05,
+      "loss": 0.129,
+      "step": 4218
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2011338767749685e-05,
+      "loss": 0.1315,
+      "step": 4219
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200793595925384e-05,
+      "loss": 0.1252,
+      "step": 4220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004532908471904e-05,
+      "loss": 0.1274,
+      "step": 4221
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001129615814495e-05,
+      "loss": 0.1273,
+      "step": 4222
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997726081692272e-05,
+      "loss": 0.1283,
+      "step": 4223
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994322306515926e-05,
+      "loss": 0.1274,
+      "step": 4224
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1990918290696161e-05,
+      "loss": 0.1153,
+      "step": 4225
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987514034643725e-05,
+      "loss": 0.1286,
+      "step": 4226
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1984109538769393e-05,
+      "loss": 0.1154,
+      "step": 4227
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980704803483963e-05,
+      "loss": 0.1274,
+      "step": 4228
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197729982919826e-05,
+      "loss": 0.1299,
+      "step": 4229
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1973894616323152e-05,
+      "loss": 0.151,
+      "step": 4230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1970489165269521e-05,
+      "loss": 0.1351,
+      "step": 4231
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967083476448282e-05,
+      "loss": 0.1306,
+      "step": 4232
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1963677550270384e-05,
+      "loss": 0.1223,
+      "step": 4233
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1960271387146799e-05,
+      "loss": 0.1286,
+      "step": 4234
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956864987488525e-05,
+      "loss": 0.1238,
+      "step": 4235
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1953458351706601e-05,
+      "loss": 0.1291,
+      "step": 4236
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950051480212079e-05,
+      "loss": 0.116,
+      "step": 4237
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1946644373416049e-05,
+      "loss": 0.1255,
+      "step": 4238
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1943237031729628e-05,
+      "loss": 0.1297,
+      "step": 4239
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193982945556396e-05,
+      "loss": 0.119,
+      "step": 4240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1936421645330217e-05,
+      "loss": 0.1297,
+      "step": 4241
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1933013601439596e-05,
+      "loss": 0.1184,
+      "step": 4242
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1929605324303336e-05,
+      "loss": 0.1218,
+      "step": 4243
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1926196814332685e-05,
+      "loss": 0.1297,
+      "step": 4244
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192278807193893e-05,
+      "loss": 0.1284,
+      "step": 4245
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919379097533388e-05,
+      "loss": 0.1246,
+      "step": 4246
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.19159698915274e-05,
+      "loss": 0.1255,
+      "step": 4247
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1912560454332329e-05,
+      "loss": 0.119,
+      "step": 4248
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190915078635958e-05,
+      "loss": 0.1297,
+      "step": 4249
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1905740888020576e-05,
+      "loss": 0.1256,
+      "step": 4250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1902330759726766e-05,
+      "loss": 0.1358,
+      "step": 4251
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898920401889636e-05,
+      "loss": 0.1234,
+      "step": 4252
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189550981492069e-05,
+      "loss": 0.1262,
+      "step": 4253
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1892098999231463e-05,
+      "loss": 0.1218,
+      "step": 4254
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1888687955233525e-05,
+      "loss": 0.1167,
+      "step": 4255
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188527668333846e-05,
+      "loss": 0.1495,
+      "step": 4256
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188186518395789e-05,
+      "loss": 0.1336,
+      "step": 4257
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1878453457503465e-05,
+      "loss": 0.1221,
+      "step": 4258
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1875041504386849e-05,
+      "loss": 0.1343,
+      "step": 4259
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187162932501975e-05,
+      "loss": 0.13,
+      "step": 4260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1868216919813896e-05,
+      "loss": 0.1301,
+      "step": 4261
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864804289181036e-05,
+      "loss": 0.1268,
+      "step": 4262
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1861391433532958e-05,
+      "loss": 0.124,
+      "step": 4263
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1857978353281474e-05,
+      "loss": 0.1339,
+      "step": 4264
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1854565048838413e-05,
+      "loss": 0.1191,
+      "step": 4265
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1851151520615645e-05,
+      "loss": 0.1355,
+      "step": 4266
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1847737769025061e-05,
+      "loss": 0.1312,
+      "step": 4267
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1844323794478572e-05,
+      "loss": 0.1366,
+      "step": 4268
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1840909597388128e-05,
+      "loss": 0.129,
+      "step": 4269
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837495178165706e-05,
+      "loss": 0.1385,
+      "step": 4270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1834080537223294e-05,
+      "loss": 0.1186,
+      "step": 4271
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830665674972921e-05,
+      "loss": 0.1284,
+      "step": 4272
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.182725059182664e-05,
+      "loss": 0.1228,
+      "step": 4273
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.182383528819653e-05,
+      "loss": 0.1292,
+      "step": 4274
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1820419764494692e-05,
+      "loss": 0.1243,
+      "step": 4275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1817004021133263e-05,
+      "loss": 0.1213,
+      "step": 4276
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1813588058524398e-05,
+      "loss": 0.1303,
+      "step": 4277
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181017187708028e-05,
+      "loss": 0.1138,
+      "step": 4278
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1806755477213125e-05,
+      "loss": 0.1268,
+      "step": 4279
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1803338859335169e-05,
+      "loss": 0.1302,
+      "step": 4280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.179992202385867e-05,
+      "loss": 0.132,
+      "step": 4281
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1796504971195923e-05,
+      "loss": 0.112,
+      "step": 4282
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1793087701759245e-05,
+      "loss": 0.1378,
+      "step": 4283
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789670215960975e-05,
+      "loss": 0.1278,
+      "step": 4284
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1786252514213483e-05,
+      "loss": 0.1202,
+      "step": 4285
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1782834596929164e-05,
+      "loss": 0.1104,
+      "step": 4286
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1779416464520434e-05,
+      "loss": 0.1273,
+      "step": 4287
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1775998117399747e-05,
+      "loss": 0.1309,
+      "step": 4288
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1772579555979573e-05,
+      "loss": 0.1264,
+      "step": 4289
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1769160780672405e-05,
+      "loss": 0.1249,
+      "step": 4290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1765741791890767e-05,
+      "loss": 0.1293,
+      "step": 4291
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176232259004722e-05,
+      "loss": 0.1163,
+      "step": 4292
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1758903175554326e-05,
+      "loss": 0.1178,
+      "step": 4293
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1755483548824693e-05,
+      "loss": 0.1199,
+      "step": 4294
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175206371027095e-05,
+      "loss": 0.144,
+      "step": 4295
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1748643660305743e-05,
+      "loss": 0.1445,
+      "step": 4296
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174522339934175e-05,
+      "loss": 0.1316,
+      "step": 4297
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1741802927791681e-05,
+      "loss": 0.1304,
+      "step": 4298
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1738382246068258e-05,
+      "loss": 0.1092,
+      "step": 4299
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1734961354584238e-05,
+      "loss": 0.1211,
+      "step": 4300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17315402537524e-05,
+      "loss": 0.13,
+      "step": 4301
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172811894398555e-05,
+      "loss": 0.1351,
+      "step": 4302
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1724697425696511e-05,
+      "loss": 0.1214,
+      "step": 4303
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721275699298147e-05,
+      "loss": 0.1303,
+      "step": 4304
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1717853765203335e-05,
+      "loss": 0.1198,
+      "step": 4305
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1714431623824975e-05,
+      "loss": 0.1318,
+      "step": 4306
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1711009275576006e-05,
+      "loss": 0.1233,
+      "step": 4307
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707586720869375e-05,
+      "loss": 0.117,
+      "step": 4308
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1704163960118069e-05,
+      "loss": 0.1341,
+      "step": 4309
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1700740993735087e-05,
+      "loss": 0.1102,
+      "step": 4310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169731782213346e-05,
+      "loss": 0.1296,
+      "step": 4311
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1693894445726245e-05,
+      "loss": 0.1236,
+      "step": 4312
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1690470864926522e-05,
+      "loss": 0.1253,
+      "step": 4313
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168704708014739e-05,
+      "loss": 0.1342,
+      "step": 4314
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168362309180198e-05,
+      "loss": 0.122,
+      "step": 4315
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1680198900303446e-05,
+      "loss": 0.1296,
+      "step": 4316
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1676774506064963e-05,
+      "loss": 0.123,
+      "step": 4317
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1673349909499736e-05,
+      "loss": 0.1191,
+      "step": 4318
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1669925111020988e-05,
+      "loss": 0.1274,
+      "step": 4319
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1666500111041971e-05,
+      "loss": 0.1452,
+      "step": 4320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1663074909975963e-05,
+      "loss": 0.1202,
+      "step": 4321
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165964950823626e-05,
+      "loss": 0.1196,
+      "step": 4322
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1656223906236186e-05,
+      "loss": 0.1205,
+      "step": 4323
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165279810438909e-05,
+      "loss": 0.1289,
+      "step": 4324
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1649372103108342e-05,
+      "loss": 0.1419,
+      "step": 4325
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164594590280734e-05,
+      "loss": 0.1284,
+      "step": 4326
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1642519503899502e-05,
+      "loss": 0.1253,
+      "step": 4327
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1639092906798276e-05,
+      "loss": 0.1259,
+      "step": 4328
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1635666111917123e-05,
+      "loss": 0.1258,
+      "step": 4329
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632239119669537e-05,
+      "loss": 0.1199,
+      "step": 4330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1628811930469038e-05,
+      "loss": 0.1308,
+      "step": 4331
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1625384544729162e-05,
+      "loss": 0.1316,
+      "step": 4332
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621956962863467e-05,
+      "loss": 0.1279,
+      "step": 4333
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161852918528555e-05,
+      "loss": 0.1351,
+      "step": 4334
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615101212409012e-05,
+      "loss": 0.133,
+      "step": 4335
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1611673044647492e-05,
+      "loss": 0.1353,
+      "step": 4336
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1608244682414647e-05,
+      "loss": 0.1202,
+      "step": 4337
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1604816126124154e-05,
+      "loss": 0.1259,
+      "step": 4338
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.160138737618972e-05,
+      "loss": 0.1212,
+      "step": 4339
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597958433025072e-05,
+      "loss": 0.1329,
+      "step": 4340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1594529297043967e-05,
+      "loss": 0.111,
+      "step": 4341
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1591099968660164e-05,
+      "loss": 0.1299,
+      "step": 4342
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1587670448287473e-05,
+      "loss": 0.1248,
+      "step": 4343
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1584240736339716e-05,
+      "loss": 0.1236,
+      "step": 4344
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1580810833230727e-05,
+      "loss": 0.123,
+      "step": 4345
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1577380739374376e-05,
+      "loss": 0.1384,
+      "step": 4346
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.157395045518456e-05,
+      "loss": 0.1169,
+      "step": 4347
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1570519981075178e-05,
+      "loss": 0.1128,
+      "step": 4348
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567089317460178e-05,
+      "loss": 0.1224,
+      "step": 4349
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563658464753516e-05,
+      "loss": 0.1299,
+      "step": 4350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1560227423369168e-05,
+      "loss": 0.1166,
+      "step": 4351
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1556796193721141e-05,
+      "loss": 0.1283,
+      "step": 4352
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553364776223463e-05,
+      "loss": 0.1324,
+      "step": 4353
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549933171290184e-05,
+      "loss": 0.1312,
+      "step": 4354
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1546501379335374e-05,
+      "loss": 0.119,
+      "step": 4355
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1543069400773126e-05,
+      "loss": 0.1236,
+      "step": 4356
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1539637236017562e-05,
+      "loss": 0.1378,
+      "step": 4357
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1536204885482816e-05,
+      "loss": 0.1253,
+      "step": 4358
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1532772349583055e-05,
+      "loss": 0.1178,
+      "step": 4359
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1529339628732462e-05,
+      "loss": 0.1175,
+      "step": 4360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152590672334524e-05,
+      "loss": 0.1359,
+      "step": 4361
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1522473633835623e-05,
+      "loss": 0.1197,
+      "step": 4362
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1519040360617859e-05,
+      "loss": 0.111,
+      "step": 4363
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1515606904106222e-05,
+      "loss": 0.1171,
+      "step": 4364
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1512173264715012e-05,
+      "loss": 0.1209,
+      "step": 4365
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.150873944285854e-05,
+      "loss": 0.1153,
+      "step": 4366
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1505305438951147e-05,
+      "loss": 0.1235,
+      "step": 4367
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1501871253407196e-05,
+      "loss": 0.1207,
+      "step": 4368
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1498436886641068e-05,
+      "loss": 0.1252,
+      "step": 4369
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1495002339067173e-05,
+      "loss": 0.1127,
+      "step": 4370
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1491567611099934e-05,
+      "loss": 0.1197,
+      "step": 4371
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1488132703153805e-05,
+      "loss": 0.113,
+      "step": 4372
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148469761564325e-05,
+      "loss": 0.1195,
+      "step": 4373
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1481262348982762e-05,
+      "loss": 0.1258,
+      "step": 4374
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1477826903586862e-05,
+      "loss": 0.1277,
+      "step": 4375
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1474391279870076e-05,
+      "loss": 0.1231,
+      "step": 4376
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1470955478246968e-05,
+      "loss": 0.1165,
+      "step": 4377
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467519499132113e-05,
+      "loss": 0.1201,
+      "step": 4378
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464083342940113e-05,
+      "loss": 0.1226,
+      "step": 4379
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1460647010085588e-05,
+      "loss": 0.1308,
+      "step": 4380
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145721050098318e-05,
+      "loss": 0.1266,
+      "step": 4381
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1453773816047554e-05,
+      "loss": 0.1196,
+      "step": 4382
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1450336955693393e-05,
+      "loss": 0.1291,
+      "step": 4383
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1446899920335407e-05,
+      "loss": 0.1196,
+      "step": 4384
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1443462710388316e-05,
+      "loss": 0.1306,
+      "step": 4385
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1440025326266877e-05,
+      "loss": 0.1117,
+      "step": 4386
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1436587768385858e-05,
+      "loss": 0.1174,
+      "step": 4387
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1433150037160043e-05,
+      "loss": 0.1271,
+      "step": 4388
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1429712133004249e-05,
+      "loss": 0.1262,
+      "step": 4389
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1426274056333307e-05,
+      "loss": 0.1176,
+      "step": 4390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1422835807562067e-05,
+      "loss": 0.1261,
+      "step": 4391
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1419397387105405e-05,
+      "loss": 0.123,
+      "step": 4392
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141595879537822e-05,
+      "loss": 0.1218,
+      "step": 4393
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141252003279542e-05,
+      "loss": 0.1271,
+      "step": 4394
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1409081099771942e-05,
+      "loss": 0.1213,
+      "step": 4395
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1405641996722748e-05,
+      "loss": 0.1246,
+      "step": 4396
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1402202724062806e-05,
+      "loss": 0.1252,
+      "step": 4397
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1398763282207118e-05,
+      "loss": 0.1196,
+      "step": 4398
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1395323671570706e-05,
+      "loss": 0.1295,
+      "step": 4399
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1391883892568599e-05,
+      "loss": 0.1201,
+      "step": 4400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138844394561586e-05,
+      "loss": 0.1174,
+      "step": 4401
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1385003831127573e-05,
+      "loss": 0.132,
+      "step": 4402
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1381563549518823e-05,
+      "loss": 0.1201,
+      "step": 4403
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1378123101204742e-05,
+      "loss": 0.12,
+      "step": 4404
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374682486600464e-05,
+      "loss": 0.1277,
+      "step": 4405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137124170612115e-05,
+      "loss": 0.122,
+      "step": 4406
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1367800760181975e-05,
+      "loss": 0.1249,
+      "step": 4407
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1364359649198145e-05,
+      "loss": 0.1221,
+      "step": 4408
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1360918373584873e-05,
+      "loss": 0.1186,
+      "step": 4409
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1357476933757398e-05,
+      "loss": 0.1222,
+      "step": 4410
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1354035330130984e-05,
+      "loss": 0.1218,
+      "step": 4411
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1350593563120907e-05,
+      "loss": 0.1162,
+      "step": 4412
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1347151633142462e-05,
+      "loss": 0.1258,
+      "step": 4413
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1343709540610971e-05,
+      "loss": 0.1226,
+      "step": 4414
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134026728594177e-05,
+      "loss": 0.1256,
+      "step": 4415
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1336824869550216e-05,
+      "loss": 0.1219,
+      "step": 4416
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1333382291851687e-05,
+      "loss": 0.1259,
+      "step": 4417
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1329939553261575e-05,
+      "loss": 0.1338,
+      "step": 4418
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1326496654195302e-05,
+      "loss": 0.1152,
+      "step": 4419
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1323053595068296e-05,
+      "loss": 0.1267,
+      "step": 4420
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1319610376296018e-05,
+      "loss": 0.1128,
+      "step": 4421
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1316166998293937e-05,
+      "loss": 0.1265,
+      "step": 4422
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1312723461477543e-05,
+      "loss": 0.1203,
+      "step": 4423
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1309279766262355e-05,
+      "loss": 0.131,
+      "step": 4424
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13058359130639e-05,
+      "loss": 0.1343,
+      "step": 4425
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1302391902297726e-05,
+      "loss": 0.125,
+      "step": 4426
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1298947734379406e-05,
+      "loss": 0.1386,
+      "step": 4427
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1295503409724526e-05,
+      "loss": 0.121,
+      "step": 4428
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1292058928748692e-05,
+      "loss": 0.1169,
+      "step": 4429
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1288614291867532e-05,
+      "loss": 0.1154,
+      "step": 4430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1285169499496686e-05,
+      "loss": 0.1207,
+      "step": 4431
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1281724552051824e-05,
+      "loss": 0.1286,
+      "step": 4432
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1278279449948626e-05,
+      "loss": 0.1325,
+      "step": 4433
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274834193602786e-05,
+      "loss": 0.1175,
+      "step": 4434
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1271388783430033e-05,
+      "loss": 0.12,
+      "step": 4435
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12679432198461e-05,
+      "loss": 0.127,
+      "step": 4436
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.126449750326674e-05,
+      "loss": 0.1139,
+      "step": 4437
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261051634107735e-05,
+      "loss": 0.1226,
+      "step": 4438
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1257605612784873e-05,
+      "loss": 0.1227,
+      "step": 4439
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125415943971397e-05,
+      "loss": 0.124,
+      "step": 4440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250713115310852e-05,
+      "loss": 0.1096,
+      "step": 4441
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124726663999137e-05,
+      "loss": 0.1145,
+      "step": 4442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1243820014171385e-05,
+      "loss": 0.1215,
+      "step": 4443
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1240373238266786e-05,
+      "loss": 0.1224,
+      "step": 4444
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123692631269348e-05,
+      "loss": 0.1221,
+      "step": 4445
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1233479237867378e-05,
+      "loss": 0.1209,
+      "step": 4446
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230032014204422e-05,
+      "loss": 0.1223,
+      "step": 4447
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1226584642120574e-05,
+      "loss": 0.1198,
+      "step": 4448
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1223137122031803e-05,
+      "loss": 0.1183,
+      "step": 4449
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1219689454354104e-05,
+      "loss": 0.1322,
+      "step": 4450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1216241639503487e-05,
+      "loss": 0.1161,
+      "step": 4451
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1212793677895976e-05,
+      "loss": 0.1318,
+      "step": 4452
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1209345569947622e-05,
+      "loss": 0.1248,
+      "step": 4453
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1205897316074487e-05,
+      "loss": 0.1238,
+      "step": 4454
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1202448916692655e-05,
+      "loss": 0.1211,
+      "step": 4455
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1199000372218214e-05,
+      "loss": 0.1276,
+      "step": 4456
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1195551683067292e-05,
+      "loss": 0.1342,
+      "step": 4457
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1192102849656016e-05,
+      "loss": 0.1356,
+      "step": 4458
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1188653872400539e-05,
+      "loss": 0.1127,
+      "step": 4459
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.118520475171703e-05,
+      "loss": 0.1129,
+      "step": 4460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118175548802167e-05,
+      "loss": 0.1355,
+      "step": 4461
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1178306081730666e-05,
+      "loss": 0.133,
+      "step": 4462
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1174856533260239e-05,
+      "loss": 0.1226,
+      "step": 4463
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1171406843026625e-05,
+      "loss": 0.1234,
+      "step": 4464
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167957011446073e-05,
+      "loss": 0.1303,
+      "step": 4465
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1164507038934864e-05,
+      "loss": 0.1155,
+      "step": 4466
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1161056925909279e-05,
+      "loss": 0.1073,
+      "step": 4467
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157606672785627e-05,
+      "loss": 0.1316,
+      "step": 4468
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154156279980226e-05,
+      "loss": 0.12,
+      "step": 4469
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1150705747909423e-05,
+      "loss": 0.1241,
+      "step": 4470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1147255076989565e-05,
+      "loss": 0.1254,
+      "step": 4471
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1143804267637028e-05,
+      "loss": 0.1393,
+      "step": 4472
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1140353320268205e-05,
+      "loss": 0.121,
+      "step": 4473
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1136902235299495e-05,
+      "loss": 0.1142,
+      "step": 4474
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1133451013147327e-05,
+      "loss": 0.1136,
+      "step": 4475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1129999654228139e-05,
+      "loss": 0.1202,
+      "step": 4476
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1126548158958383e-05,
+      "loss": 0.1206,
+      "step": 4477
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1123096527754533e-05,
+      "loss": 0.116,
+      "step": 4478
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.1254,
+      "step": 4479
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1116192859210527e-05,
+      "loss": 0.1239,
+      "step": 4480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1112740822703392e-05,
+      "loss": 0.1246,
+      "step": 4481
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1109288651928216e-05,
+      "loss": 0.1294,
+      "step": 4482
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1105836347301555e-05,
+      "loss": 0.1199,
+      "step": 4483
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102383909239971e-05,
+      "loss": 0.1176,
+      "step": 4484
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1098931338160057e-05,
+      "loss": 0.133,
+      "step": 4485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1095478634478412e-05,
+      "loss": 0.1187,
+      "step": 4486
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1092025798611655e-05,
+      "loss": 0.1288,
+      "step": 4487
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108857283097642e-05,
+      "loss": 0.1189,
+      "step": 4488
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1085119731989353e-05,
+      "loss": 0.1181,
+      "step": 4489
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1081666502067126e-05,
+      "loss": 0.1143,
+      "step": 4490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1078213141626415e-05,
+      "loss": 0.1202,
+      "step": 4491
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.107475965108392e-05,
+      "loss": 0.1331,
+      "step": 4492
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1071306030856353e-05,
+      "loss": 0.1256,
+      "step": 4493
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1067852281360443e-05,
+      "loss": 0.1187,
+      "step": 4494
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106439840301293e-05,
+      "loss": 0.1326,
+      "step": 4495
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1060944396230583e-05,
+      "loss": 0.131,
+      "step": 4496
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1057490261430168e-05,
+      "loss": 0.1156,
+      "step": 4497
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1054035999028478e-05,
+      "loss": 0.1139,
+      "step": 4498
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1050581609442322e-05,
+      "loss": 0.1077,
+      "step": 4499
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1047127093088521e-05,
+      "loss": 0.1217,
+      "step": 4500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1043672450383907e-05,
+      "loss": 0.113,
+      "step": 4501
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1040217681745335e-05,
+      "loss": 0.1242,
+      "step": 4502
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1036762787589674e-05,
+      "loss": 0.13,
+      "step": 4503
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033307768333806e-05,
+      "loss": 0.1151,
+      "step": 4504
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1029852624394621e-05,
+      "loss": 0.1178,
+      "step": 4505
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102639735618904e-05,
+      "loss": 0.1195,
+      "step": 4506
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1022941964133988e-05,
+      "loss": 0.1312,
+      "step": 4507
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1019486448646407e-05,
+      "loss": 0.1046,
+      "step": 4508
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016030810143253e-05,
+      "loss": 0.1296,
+      "step": 4509
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1012575049041499e-05,
+      "loss": 0.1172,
+      "step": 4510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009119165758129e-05,
+      "loss": 0.1242,
+      "step": 4511
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1005663160710152e-05,
+      "loss": 0.1105,
+      "step": 4512
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002207034314576e-05,
+      "loss": 0.1124,
+      "step": 4513
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0998750786988435e-05,
+      "loss": 0.114,
+      "step": 4514
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995294419148779e-05,
+      "loss": 0.1322,
+      "step": 4515
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099183793121266e-05,
+      "loss": 0.1082,
+      "step": 4516
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0988381323597156e-05,
+      "loss": 0.1148,
+      "step": 4517
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098492459671936e-05,
+      "loss": 0.1237,
+      "step": 4518
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0981467750996373e-05,
+      "loss": 0.1226,
+      "step": 4519
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0978010786845306e-05,
+      "loss": 0.1106,
+      "step": 4520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.09745537046833e-05,
+      "loss": 0.1281,
+      "step": 4521
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0971096504927498e-05,
+      "loss": 0.144,
+      "step": 4522
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0967639187995061e-05,
+      "loss": 0.1332,
+      "step": 4523
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964181754303159e-05,
+      "loss": 0.1325,
+      "step": 4524
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0960724204268989e-05,
+      "loss": 0.1232,
+      "step": 4525
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0957266538309748e-05,
+      "loss": 0.1227,
+      "step": 4526
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095380875684265e-05,
+      "loss": 0.1177,
+      "step": 4527
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0950350860284935e-05,
+      "loss": 0.1172,
+      "step": 4528
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.094689284905384e-05,
+      "loss": 0.1318,
+      "step": 4529
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0943434723566624e-05,
+      "loss": 0.1125,
+      "step": 4530
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0939976484240562e-05,
+      "loss": 0.1254,
+      "step": 4531
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093651813149294e-05,
+      "loss": 0.1227,
+      "step": 4532
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0933059665741052e-05,
+      "loss": 0.1224,
+      "step": 4533
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929601087402217e-05,
+      "loss": 0.1251,
+      "step": 4534
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0926142396893762e-05,
+      "loss": 0.1324,
+      "step": 4535
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092268359463302e-05,
+      "loss": 0.1271,
+      "step": 4536
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0919224681037355e-05,
+      "loss": 0.1251,
+      "step": 4537
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0915765656524127e-05,
+      "loss": 0.1192,
+      "step": 4538
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0912306521510718e-05,
+      "loss": 0.1201,
+      "step": 4539
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908847276414527e-05,
+      "loss": 0.1182,
+      "step": 4540
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0905387921652952e-05,
+      "loss": 0.1123,
+      "step": 4541
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090192845764342e-05,
+      "loss": 0.1271,
+      "step": 4542
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0898468884803366e-05,
+      "loss": 0.114,
+      "step": 4543
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0895009203550231e-05,
+      "loss": 0.1226,
+      "step": 4544
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0891549414301478e-05,
+      "loss": 0.1222,
+      "step": 4545
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088808951747458e-05,
+      "loss": 0.1101,
+      "step": 4546
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0884629513487023e-05,
+      "loss": 0.1365,
+      "step": 4547
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0881169402756305e-05,
+      "loss": 0.1247,
+      "step": 4548
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0877709185699938e-05,
+      "loss": 0.1214,
+      "step": 4549
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0874248862735445e-05,
+      "loss": 0.1188,
+      "step": 4550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0870788434280368e-05,
+      "loss": 0.1304,
+      "step": 4551
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0867327900752252e-05,
+      "loss": 0.1312,
+      "step": 4552
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0863867262568661e-05,
+      "loss": 0.1137,
+      "step": 4553
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0860406520147171e-05,
+      "loss": 0.13,
+      "step": 4554
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085694567390537e-05,
+      "loss": 0.1232,
+      "step": 4555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0853484724260861e-05,
+      "loss": 0.1347,
+      "step": 4556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0850023671631249e-05,
+      "loss": 0.1406,
+      "step": 4557
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0846562516434167e-05,
+      "loss": 0.132,
+      "step": 4558
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0843101259087252e-05,
+      "loss": 0.1239,
+      "step": 4559
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.083963990000815e-05,
+      "loss": 0.1128,
+      "step": 4560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0836178439614526e-05,
+      "loss": 0.1278,
+      "step": 4561
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0832716878324055e-05,
+      "loss": 0.1172,
+      "step": 4562
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082925521655442e-05,
+      "loss": 0.1302,
+      "step": 4563
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.1179,
+      "step": 4564
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0822331593248479e-05,
+      "loss": 0.131,
+      "step": 4565
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0818869632547602e-05,
+      "loss": 0.1281,
+      "step": 4566
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081540757303843e-05,
+      "loss": 0.1207,
+      "step": 4567
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0811945415138717e-05,
+      "loss": 0.1189,
+      "step": 4568
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0808483159266213e-05,
+      "loss": 0.1137,
+      "step": 4569
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0805020805838688e-05,
+      "loss": 0.1255,
+      "step": 4570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0801558355273933e-05,
+      "loss": 0.1137,
+      "step": 4571
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0798095807989735e-05,
+      "loss": 0.1179,
+      "step": 4572
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0794633164403898e-05,
+      "loss": 0.1226,
+      "step": 4573
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0791170424934248e-05,
+      "loss": 0.1243,
+      "step": 4574
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0787707589998603e-05,
+      "loss": 0.1209,
+      "step": 4575
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784244660014813e-05,
+      "loss": 0.1172,
+      "step": 4576
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0780781635400728e-05,
+      "loss": 0.132,
+      "step": 4577
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0777318516574205e-05,
+      "loss": 0.1231,
+      "step": 4578
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0773855303953122e-05,
+      "loss": 0.1296,
+      "step": 4579
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0770391997955367e-05,
+      "loss": 0.1129,
+      "step": 4580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0766928598998837e-05,
+      "loss": 0.1204,
+      "step": 4581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0763465107501438e-05,
+      "loss": 0.1253,
+      "step": 4582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0760001523881094e-05,
+      "loss": 0.1282,
+      "step": 4583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.075653784855573e-05,
+      "loss": 0.1153,
+      "step": 4584
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0753074081943291e-05,
+      "loss": 0.1156,
+      "step": 4585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074961022446173e-05,
+      "loss": 0.1217,
+      "step": 4586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0746146276529011e-05,
+      "loss": 0.1229,
+      "step": 4587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0742682238563105e-05,
+      "loss": 0.1119,
+      "step": 4588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0739218110982002e-05,
+      "loss": 0.1206,
+      "step": 4589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735753894203698e-05,
+      "loss": 0.1223,
+      "step": 4590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0732289588646196e-05,
+      "loss": 0.1076,
+      "step": 4591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072882519472752e-05,
+      "loss": 0.1246,
+      "step": 4592
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0725360712865693e-05,
+      "loss": 0.1259,
+      "step": 4593
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0721896143478759e-05,
+      "loss": 0.1287,
+      "step": 4594
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0718431486984765e-05,
+      "loss": 0.1287,
+      "step": 4595
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0714966743801768e-05,
+      "loss": 0.1282,
+      "step": 4596
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0711501914347842e-05,
+      "loss": 0.1187,
+      "step": 4597
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0708036999041072e-05,
+      "loss": 0.1176,
+      "step": 4598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070457199829954e-05,
+      "loss": 0.1273,
+      "step": 4599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0701106912541357e-05,
+      "loss": 0.1133,
+      "step": 4600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069764174218463e-05,
+      "loss": 0.1302,
+      "step": 4601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0694176487647482e-05,
+      "loss": 0.1278,
+      "step": 4602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0690711149348047e-05,
+      "loss": 0.1228,
+      "step": 4603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687245727704465e-05,
+      "loss": 0.1271,
+      "step": 4604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0683780223134892e-05,
+      "loss": 0.1263,
+      "step": 4605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0680314636057486e-05,
+      "loss": 0.1174,
+      "step": 4606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067684896689042e-05,
+      "loss": 0.1279,
+      "step": 4607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0673383216051883e-05,
+      "loss": 0.1215,
+      "step": 4608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0669917383960058e-05,
+      "loss": 0.1189,
+      "step": 4609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0666451471033153e-05,
+      "loss": 0.113,
+      "step": 4610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0662985477689379e-05,
+      "loss": 0.1082,
+      "step": 4611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0659519404346955e-05,
+      "loss": 0.1184,
+      "step": 4612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0656053251424115e-05,
+      "loss": 0.1171,
+      "step": 4613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0652587019339099e-05,
+      "loss": 0.113,
+      "step": 4614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0649120708510154e-05,
+      "loss": 0.1286,
+      "step": 4615
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0645654319355546e-05,
+      "loss": 0.1196,
+      "step": 4616
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064218785229354e-05,
+      "loss": 0.1238,
+      "step": 4617
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638721307742414e-05,
+      "loss": 0.1274,
+      "step": 4618
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.063525468612046e-05,
+      "loss": 0.1274,
+      "step": 4619
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0631787987845971e-05,
+      "loss": 0.1283,
+      "step": 4620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0628321213337256e-05,
+      "loss": 0.1195,
+      "step": 4621
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624854363012629e-05,
+      "loss": 0.1146,
+      "step": 4622
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0621387437290421e-05,
+      "loss": 0.1049,
+      "step": 4623
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0617920436588955e-05,
+      "loss": 0.1217,
+      "step": 4624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0614453361326582e-05,
+      "loss": 0.1117,
+      "step": 4625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0610986211921657e-05,
+      "loss": 0.1227,
+      "step": 4626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607518988792531e-05,
+      "loss": 0.1083,
+      "step": 4627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060405169235758e-05,
+      "loss": 0.1212,
+      "step": 4628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0600584323035186e-05,
+      "loss": 0.118,
+      "step": 4629
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0597116881243728e-05,
+      "loss": 0.1061,
+      "step": 4630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0593649367401607e-05,
+      "loss": 0.1295,
+      "step": 4631
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0590181781927229e-05,
+      "loss": 0.1214,
+      "step": 4632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0586714125239008e-05,
+      "loss": 0.1208,
+      "step": 4633
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058324639775536e-05,
+      "loss": 0.1238,
+      "step": 4634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0579778599894723e-05,
+      "loss": 0.1173,
+      "step": 4635
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0576310732075534e-05,
+      "loss": 0.1228,
+      "step": 4636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057284279471624e-05,
+      "loss": 0.1165,
+      "step": 4637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0569374788235298e-05,
+      "loss": 0.1252,
+      "step": 4638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0565906713051171e-05,
+      "loss": 0.0994,
+      "step": 4639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056243856958233e-05,
+      "loss": 0.1166,
+      "step": 4640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0558970358247262e-05,
+      "loss": 0.1284,
+      "step": 4641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055550207946445e-05,
+      "loss": 0.1206,
+      "step": 4642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0552033733652393e-05,
+      "loss": 0.1208,
+      "step": 4643
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05485653212296e-05,
+      "loss": 0.1195,
+      "step": 4644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545096842614582e-05,
+      "loss": 0.1256,
+      "step": 4645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541628298225856e-05,
+      "loss": 0.1248,
+      "step": 4646
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0538159688481959e-05,
+      "loss": 0.1206,
+      "step": 4647
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053469101380142e-05,
+      "loss": 0.1123,
+      "step": 4648
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0531222274602795e-05,
+      "loss": 0.1353,
+      "step": 4649
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0527753471304625e-05,
+      "loss": 0.117,
+      "step": 4650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0524284604325476e-05,
+      "loss": 0.1076,
+      "step": 4651
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0520815674083919e-05,
+      "loss": 0.1198,
+      "step": 4652
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0517346680998526e-05,
+      "loss": 0.114,
+      "step": 4653
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051387762548788e-05,
+      "loss": 0.1209,
+      "step": 4654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0510408507970577e-05,
+      "loss": 0.1178,
+      "step": 4655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0506939328865207e-05,
+      "loss": 0.1201,
+      "step": 4656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0503470088590385e-05,
+      "loss": 0.1104,
+      "step": 4657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0500000787564718e-05,
+      "loss": 0.1188,
+      "step": 4658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0496531426206828e-05,
+      "loss": 0.1217,
+      "step": 4659
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0493062004935346e-05,
+      "loss": 0.1089,
+      "step": 4660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0489592524168904e-05,
+      "loss": 0.1253,
+      "step": 4661
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0486122984326146e-05,
+      "loss": 0.1102,
+      "step": 4662
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0482653385825718e-05,
+      "loss": 0.123,
+      "step": 4663
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479183729086279e-05,
+      "loss": 0.1162,
+      "step": 4664
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0475714014526497e-05,
+      "loss": 0.1215,
+      "step": 4665
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472244242565035e-05,
+      "loss": 0.121,
+      "step": 4666
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0468774413620573e-05,
+      "loss": 0.121,
+      "step": 4667
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0465304528111797e-05,
+      "loss": 0.1081,
+      "step": 4668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461834586457398e-05,
+      "loss": 0.1245,
+      "step": 4669
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0458364589076071e-05,
+      "loss": 0.1173,
+      "step": 4670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0454894536386523e-05,
+      "loss": 0.1138,
+      "step": 4671
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0451424428807467e-05,
+      "loss": 0.1236,
+      "step": 4672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0447954266757616e-05,
+      "loss": 0.1045,
+      "step": 4673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444484050655697e-05,
+      "loss": 0.113,
+      "step": 4674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0441013780920446e-05,
+      "loss": 0.1067,
+      "step": 4675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0437543457970591e-05,
+      "loss": 0.1155,
+      "step": 4676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0434073082224883e-05,
+      "loss": 0.1129,
+      "step": 4677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0430602654102071e-05,
+      "loss": 0.1296,
+      "step": 4678
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0427132174020908e-05,
+      "loss": 0.1212,
+      "step": 4679
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423661642400163e-05,
+      "loss": 0.1021,
+      "step": 4680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.04201910596586e-05,
+      "loss": 0.1176,
+      "step": 4681
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0416720426214996e-05,
+      "loss": 0.1197,
+      "step": 4682
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0413249742488132e-05,
+      "loss": 0.1205,
+      "step": 4683
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.04097790088968e-05,
+      "loss": 0.123,
+      "step": 4684
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406308225859784e-05,
+      "loss": 0.1231,
+      "step": 4685
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402837393795891e-05,
+      "loss": 0.1182,
+      "step": 4686
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0399366513123926e-05,
+      "loss": 0.119,
+      "step": 4687
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0395895584262696e-05,
+      "loss": 0.1233,
+      "step": 4688
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0392424607631019e-05,
+      "loss": 0.1297,
+      "step": 4689
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388953583647725e-05,
+      "loss": 0.1159,
+      "step": 4690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385482512731633e-05,
+      "loss": 0.1119,
+      "step": 4691
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038201139530158e-05,
+      "loss": 0.1213,
+      "step": 4692
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037854023177641e-05,
+      "loss": 0.1258,
+      "step": 4693
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375069022574965e-05,
+      "loss": 0.113,
+      "step": 4694
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0371597768116093e-05,
+      "loss": 0.1252,
+      "step": 4695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0368126468818656e-05,
+      "loss": 0.1255,
+      "step": 4696
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.036465512510151e-05,
+      "loss": 0.1212,
+      "step": 4697
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0361183737383526e-05,
+      "loss": 0.1108,
+      "step": 4698
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0357712306083575e-05,
+      "loss": 0.1202,
+      "step": 4699
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0354240831620542e-05,
+      "loss": 0.1164,
+      "step": 4700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0350769314413297e-05,
+      "loss": 0.1264,
+      "step": 4701
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347297754880735e-05,
+      "loss": 0.1042,
+      "step": 4702
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.034382615344175e-05,
+      "loss": 0.1396,
+      "step": 4703
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0340354510515235e-05,
+      "loss": 0.1406,
+      "step": 4704
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.03368828265201e-05,
+      "loss": 0.1072,
+      "step": 4705
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033341110187525e-05,
+      "loss": 0.1129,
+      "step": 4706
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0329939336999598e-05,
+      "loss": 0.1185,
+      "step": 4707
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0326467532312058e-05,
+      "loss": 0.1243,
+      "step": 4708
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0322995688231562e-05,
+      "loss": 0.1099,
+      "step": 4709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0319523805177029e-05,
+      "loss": 0.121,
+      "step": 4710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0316051883567395e-05,
+      "loss": 0.1064,
+      "step": 4711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0312579923821598e-05,
+      "loss": 0.1132,
+      "step": 4712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0309107926358574e-05,
+      "loss": 0.117,
+      "step": 4713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305635891597274e-05,
+      "loss": 0.1167,
+      "step": 4714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0302163819956648e-05,
+      "loss": 0.1133,
+      "step": 4715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029869171185565e-05,
+      "loss": 0.115,
+      "step": 4716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0295219567713238e-05,
+      "loss": 0.1153,
+      "step": 4717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0291747387948376e-05,
+      "loss": 0.1177,
+      "step": 4718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0288275172980033e-05,
+      "loss": 0.1047,
+      "step": 4719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0284802923227183e-05,
+      "loss": 0.1164,
+      "step": 4720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0281330639108801e-05,
+      "loss": 0.1358,
+      "step": 4721
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0277858321043868e-05,
+      "loss": 0.1098,
+      "step": 4722
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0274385969451367e-05,
+      "loss": 0.1226,
+      "step": 4723
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270913584750287e-05,
+      "loss": 0.1148,
+      "step": 4724
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0267441167359619e-05,
+      "loss": 0.1038,
+      "step": 4725
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0263968717698365e-05,
+      "loss": 0.1283,
+      "step": 4726
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0260496236185522e-05,
+      "loss": 0.1068,
+      "step": 4727
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0257023723240093e-05,
+      "loss": 0.119,
+      "step": 4728
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.025355117928109e-05,
+      "loss": 0.1166,
+      "step": 4729
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0250078604727523e-05,
+      "loss": 0.1298,
+      "step": 4730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0246605999998407e-05,
+      "loss": 0.1194,
+      "step": 4731
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024313336551276e-05,
+      "loss": 0.1183,
+      "step": 4732
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023966070168961e-05,
+      "loss": 0.1121,
+      "step": 4733
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023618800894798e-05,
+      "loss": 0.1203,
+      "step": 4734
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0232715287706898e-05,
+      "loss": 0.1207,
+      "step": 4735
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.02292425383854e-05,
+      "loss": 0.1202,
+      "step": 4736
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0225769761402524e-05,
+      "loss": 0.1231,
+      "step": 4737
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0222296957177305e-05,
+      "loss": 0.1353,
+      "step": 4738
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0218824126128793e-05,
+      "loss": 0.1025,
+      "step": 4739
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0215351268676028e-05,
+      "loss": 0.1189,
+      "step": 4740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0211878385238063e-05,
+      "loss": 0.1139,
+      "step": 4741
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0208405476233952e-05,
+      "loss": 0.1201,
+      "step": 4742
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0204932542082748e-05,
+      "loss": 0.1172,
+      "step": 4743
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0201459583203512e-05,
+      "loss": 0.1258,
+      "step": 4744
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0197986600015307e-05,
+      "loss": 0.1184,
+      "step": 4745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0194513592937196e-05,
+      "loss": 0.1228,
+      "step": 4746
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191040562388244e-05,
+      "loss": 0.124,
+      "step": 4747
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187567508787526e-05,
+      "loss": 0.1218,
+      "step": 4748
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0184094432554119e-05,
+      "loss": 0.1109,
+      "step": 4749
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.018062133410709e-05,
+      "loss": 0.11,
+      "step": 4750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0177148213865523e-05,
+      "loss": 0.1137,
+      "step": 4751
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0173675072248504e-05,
+      "loss": 0.1173,
+      "step": 4752
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0170201909675105e-05,
+      "loss": 0.1318,
+      "step": 4753
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0166728726564424e-05,
+      "loss": 0.1258,
+      "step": 4754
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0163255523335542e-05,
+      "loss": 0.112,
+      "step": 4755
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0159782300407556e-05,
+      "loss": 0.1209,
+      "step": 4756
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0156309058199559e-05,
+      "loss": 0.108,
+      "step": 4757
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152835797130644e-05,
+      "loss": 0.1207,
+      "step": 4758
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0149362517619914e-05,
+      "loss": 0.1252,
+      "step": 4759
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0145889220086465e-05,
+      "loss": 0.1148,
+      "step": 4760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0142415904949402e-05,
+      "loss": 0.1246,
+      "step": 4761
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013894257262783e-05,
+      "loss": 0.1164,
+      "step": 4762
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135469223540856e-05,
+      "loss": 0.1253,
+      "step": 4763
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013199585810759e-05,
+      "loss": 0.126,
+      "step": 4764
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0128522476747143e-05,
+      "loss": 0.1199,
+      "step": 4765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0125049079878623e-05,
+      "loss": 0.1104,
+      "step": 4766
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0121575667921156e-05,
+      "loss": 0.0974,
+      "step": 4767
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118102241293848e-05,
+      "loss": 0.1132,
+      "step": 4768
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0114628800415818e-05,
+      "loss": 0.119,
+      "step": 4769
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0111155345706193e-05,
+      "loss": 0.1387,
+      "step": 4770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.010768187758409e-05,
+      "loss": 0.1218,
+      "step": 4771
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0104208396468634e-05,
+      "loss": 0.1182,
+      "step": 4772
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.010073490277895e-05,
+      "loss": 0.1147,
+      "step": 4773
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0097261396934161e-05,
+      "loss": 0.1138,
+      "step": 4774
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093787879353403e-05,
+      "loss": 0.1025,
+      "step": 4775
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0090314350455796e-05,
+      "loss": 0.1076,
+      "step": 4776
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0086840810660476e-05,
+      "loss": 0.1134,
+      "step": 4777
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083367260386578e-05,
+      "loss": 0.1543,
+      "step": 4778
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079893700053228e-05,
+      "loss": 0.1219,
+      "step": 4779
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0076420130079565e-05,
+      "loss": 0.1268,
+      "step": 4780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0072946550884725e-05,
+      "loss": 0.1148,
+      "step": 4781
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069472962887843e-05,
+      "loss": 0.1276,
+      "step": 4782
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0065999366508057e-05,
+      "loss": 0.1214,
+      "step": 4783
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0062525762164507e-05,
+      "loss": 0.1196,
+      "step": 4784
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059052150276334e-05,
+      "loss": 0.1271,
+      "step": 4785
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0055578531262677e-05,
+      "loss": 0.1176,
+      "step": 4786
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0052104905542677e-05,
+      "loss": 0.1119,
+      "step": 4787
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0048631273535476e-05,
+      "loss": 0.1251,
+      "step": 4788
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0045157635660223e-05,
+      "loss": 0.1303,
+      "step": 4789
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0041683992336053e-05,
+      "loss": 0.1103,
+      "step": 4790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038210343982118e-05,
+      "loss": 0.1162,
+      "step": 4791
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003473669101756e-05,
+      "loss": 0.1155,
+      "step": 4792
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0031263033861523e-05,
+      "loss": 0.1122,
+      "step": 4793
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002778937293316e-05,
+      "loss": 0.1276,
+      "step": 4794
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0024315708651609e-05,
+      "loss": 0.1127,
+      "step": 4795
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002084204143602e-05,
+      "loss": 0.1079,
+      "step": 4796
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0017368371705543e-05,
+      "loss": 0.1286,
+      "step": 4797
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013894699879327e-05,
+      "loss": 0.1137,
+      "step": 4798
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0010421026376515e-05,
+      "loss": 0.1214,
+      "step": 4799
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0006947351616256e-05,
+      "loss": 0.1164,
+      "step": 4800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003473676017706e-05,
+      "loss": 0.1178,
+      "step": 4801
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.1187,
+      "step": 4802
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.996526323982297e-06,
+      "loss": 0.1176,
+      "step": 4803
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.993052648383744e-06,
+      "loss": 0.1182,
+      "step": 4804
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.989578973623486e-06,
+      "loss": 0.108,
+      "step": 4805
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986105300120678e-06,
+      "loss": 0.1112,
+      "step": 4806
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98263162829446e-06,
+      "loss": 0.1178,
+      "step": 4807
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.979157958563983e-06,
+      "loss": 0.1183,
+      "step": 4808
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.975684291348395e-06,
+      "loss": 0.123,
+      "step": 4809
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972210627066846e-06,
+      "loss": 0.1039,
+      "step": 4810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.968736966138477e-06,
+      "loss": 0.1075,
+      "step": 4811
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965263308982445e-06,
+      "loss": 0.1159,
+      "step": 4812
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961789656017885e-06,
+      "loss": 0.1176,
+      "step": 4813
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95831600766395e-06,
+      "loss": 0.1208,
+      "step": 4814
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95484236433978e-06,
+      "loss": 0.1092,
+      "step": 4815
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951368726464526e-06,
+      "loss": 0.115,
+      "step": 4816
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.947895094457324e-06,
+      "loss": 0.1264,
+      "step": 4817
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.944421468737325e-06,
+      "loss": 0.12,
+      "step": 4818
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94094784972367e-06,
+      "loss": 0.1113,
+      "step": 4819
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937474237835496e-06,
+      "loss": 0.1066,
+      "step": 4820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.934000633491945e-06,
+      "loss": 0.1509,
+      "step": 4821
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93052703711216e-06,
+      "loss": 0.1138,
+      "step": 4822
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.927053449115278e-06,
+      "loss": 0.1239,
+      "step": 4823
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.923579869920435e-06,
+      "loss": 0.1181,
+      "step": 4824
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.920106299946776e-06,
+      "loss": 0.1224,
+      "step": 4825
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.916632739613428e-06,
+      "loss": 0.1088,
+      "step": 4826
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.913159189339526e-06,
+      "loss": 0.1152,
+      "step": 4827
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.909685649544205e-06,
+      "loss": 0.1171,
+      "step": 4828
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9062121206466e-06,
+      "loss": 0.1125,
+      "step": 4829
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.902738603065839e-06,
+      "loss": 0.1346,
+      "step": 4830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899265097221054e-06,
+      "loss": 0.1168,
+      "step": 4831
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.895791603531371e-06,
+      "loss": 0.1166,
+      "step": 4832
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.892318122415913e-06,
+      "loss": 0.1109,
+      "step": 4833
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88884465429381e-06,
+      "loss": 0.1114,
+      "step": 4834
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.885371199584184e-06,
+      "loss": 0.102,
+      "step": 4835
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.881897758706155e-06,
+      "loss": 0.1073,
+      "step": 4836
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87842433207885e-06,
+      "loss": 0.1088,
+      "step": 4837
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87495092012138e-06,
+      "loss": 0.1177,
+      "step": 4838
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87147752325286e-06,
+      "loss": 0.1058,
+      "step": 4839
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.868004141892412e-06,
+      "loss": 0.1219,
+      "step": 4840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864530776459147e-06,
+      "loss": 0.1108,
+      "step": 4841
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.861057427372171e-06,
+      "loss": 0.1177,
+      "step": 4842
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8575840950506e-06,
+      "loss": 0.1217,
+      "step": 4843
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.854110779913537e-06,
+      "loss": 0.1301,
+      "step": 4844
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.850637482380093e-06,
+      "loss": 0.1212,
+      "step": 4845
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847164202869359e-06,
+      "loss": 0.1084,
+      "step": 4846
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.843690941800445e-06,
+      "loss": 0.1304,
+      "step": 4847
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.840217699592446e-06,
+      "loss": 0.1056,
+      "step": 4848
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.836744476664458e-06,
+      "loss": 0.1201,
+      "step": 4849
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.833271273435578e-06,
+      "loss": 0.1109,
+      "step": 4850
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8297980903249e-06,
+      "loss": 0.1146,
+      "step": 4851
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826324927751501e-06,
+      "loss": 0.1048,
+      "step": 4852
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.822851786134478e-06,
+      "loss": 0.1107,
+      "step": 4853
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.819378665892911e-06,
+      "loss": 0.1122,
+      "step": 4854
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.815905567445883e-06,
+      "loss": 0.1181,
+      "step": 4855
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812432491212474e-06,
+      "loss": 0.1228,
+      "step": 4856
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.808959437611756e-06,
+      "loss": 0.1222,
+      "step": 4857
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.805486407062809e-06,
+      "loss": 0.1105,
+      "step": 4858
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.802013399984696e-06,
+      "loss": 0.1239,
+      "step": 4859
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.798540416796491e-06,
+      "loss": 0.1278,
+      "step": 4860
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.795067457917255e-06,
+      "loss": 0.1143,
+      "step": 4861
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.791594523766052e-06,
+      "loss": 0.1079,
+      "step": 4862
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788121614761939e-06,
+      "loss": 0.1214,
+      "step": 4863
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.784648731323977e-06,
+      "loss": 0.1162,
+      "step": 4864
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.781175873871212e-06,
+      "loss": 0.1215,
+      "step": 4865
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.777703042822699e-06,
+      "loss": 0.1105,
+      "step": 4866
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.774230238597478e-06,
+      "loss": 0.1094,
+      "step": 4867
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770757461614601e-06,
+      "loss": 0.1238,
+      "step": 4868
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.767284712293102e-06,
+      "loss": 0.1196,
+      "step": 4869
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.763811991052021e-06,
+      "loss": 0.115,
+      "step": 4870
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760339298310391e-06,
+      "loss": 0.1288,
+      "step": 4871
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.756866634487243e-06,
+      "loss": 0.124,
+      "step": 4872
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.753394000001597e-06,
+      "loss": 0.1081,
+      "step": 4873
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74992139527248e-06,
+      "loss": 0.1018,
+      "step": 4874
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.746448820718912e-06,
+      "loss": 0.1257,
+      "step": 4875
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.742976276759907e-06,
+      "loss": 0.1093,
+      "step": 4876
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.739503763814481e-06,
+      "loss": 0.1095,
+      "step": 4877
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73603128230164e-06,
+      "loss": 0.1218,
+      "step": 4878
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.732558832640383e-06,
+      "loss": 0.1195,
+      "step": 4879
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.729086415249716e-06,
+      "loss": 0.1094,
+      "step": 4880
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.725614030548637e-06,
+      "loss": 0.1127,
+      "step": 4881
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.722141678956134e-06,
+      "loss": 0.1109,
+      "step": 4882
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.718669360891202e-06,
+      "loss": 0.1199,
+      "step": 4883
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71519707677282e-06,
+      "loss": 0.1172,
+      "step": 4884
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.711724827019969e-06,
+      "loss": 0.113,
+      "step": 4885
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.708252612051625e-06,
+      "loss": 0.1198,
+      "step": 4886
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.704780432286766e-06,
+      "loss": 0.1308,
+      "step": 4887
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.701308288144352e-06,
+      "loss": 0.1171,
+      "step": 4888
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697836180043355e-06,
+      "loss": 0.1013,
+      "step": 4889
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69436410840273e-06,
+      "loss": 0.1224,
+      "step": 4890
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69089207364143e-06,
+      "loss": 0.119,
+      "step": 4891
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.687420076178407e-06,
+      "loss": 0.1088,
+      "step": 4892
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.683948116432609e-06,
+      "loss": 0.1154,
+      "step": 4893
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680476194822973e-06,
+      "loss": 0.1148,
+      "step": 4894
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.677004311768438e-06,
+      "loss": 0.1055,
+      "step": 4895
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.673532467687943e-06,
+      "loss": 0.1082,
+      "step": 4896
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.670060663000408e-06,
+      "loss": 0.1029,
+      "step": 4897
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666588898124754e-06,
+      "loss": 0.1164,
+      "step": 4898
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.663117173479904e-06,
+      "loss": 0.1111,
+      "step": 4899
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.659645489484766e-06,
+      "loss": 0.1049,
+      "step": 4900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.656173846558252e-06,
+      "loss": 0.1241,
+      "step": 4901
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652702245119267e-06,
+      "loss": 0.1082,
+      "step": 4902
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.649230685586708e-06,
+      "loss": 0.1143,
+      "step": 4903
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.645759168379463e-06,
+      "loss": 0.1276,
+      "step": 4904
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.642287693916426e-06,
+      "loss": 0.1084,
+      "step": 4905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.638816262616476e-06,
+      "loss": 0.1073,
+      "step": 4906
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.635344874898491e-06,
+      "loss": 0.1143,
+      "step": 4907
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631873531181349e-06,
+      "loss": 0.1174,
+      "step": 4908
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.628402231883914e-06,
+      "loss": 0.1089,
+      "step": 4909
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62493097742504e-06,
+      "loss": 0.1091,
+      "step": 4910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.621459768223593e-06,
+      "loss": 0.1233,
+      "step": 4911
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.617988604698423e-06,
+      "loss": 0.1183,
+      "step": 4912
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61451748726837e-06,
+      "loss": 0.1172,
+      "step": 4913
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611046416352278e-06,
+      "loss": 0.1059,
+      "step": 4914
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60757539236898e-06,
+      "loss": 0.1073,
+      "step": 4915
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.604104415737309e-06,
+      "loss": 0.1077,
+      "step": 4916
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.600633486876077e-06,
+      "loss": 0.1267,
+      "step": 4917
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597162606204112e-06,
+      "loss": 0.1121,
+      "step": 4918
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.593691774140219e-06,
+      "loss": 0.1181,
+      "step": 4919
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.590220991103206e-06,
+      "loss": 0.1215,
+      "step": 4920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.586750257511868e-06,
+      "loss": 0.1217,
+      "step": 4921
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.583279573785009e-06,
+      "loss": 0.115,
+      "step": 4922
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.579808940341403e-06,
+      "loss": 0.123,
+      "step": 4923
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.576338357599842e-06,
+      "loss": 0.1108,
+      "step": 4924
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.572867825979094e-06,
+      "loss": 0.1146,
+      "step": 4925
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.56939734589793e-06,
+      "loss": 0.1103,
+      "step": 4926
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.565926917775118e-06,
+      "loss": 0.1137,
+      "step": 4927
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.562456542029409e-06,
+      "loss": 0.1212,
+      "step": 4928
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.558986219079559e-06,
+      "loss": 0.127,
+      "step": 4929
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555515949344307e-06,
+      "loss": 0.1133,
+      "step": 4930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.552045733242386e-06,
+      "loss": 0.117,
+      "step": 4931
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.548575571192535e-06,
+      "loss": 0.1103,
+      "step": 4932
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545105463613478e-06,
+      "loss": 0.1156,
+      "step": 4933
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.541635410923929e-06,
+      "loss": 0.116,
+      "step": 4934
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538165413542607e-06,
+      "loss": 0.1137,
+      "step": 4935
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534695471888208e-06,
+      "loss": 0.1223,
+      "step": 4936
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.53122558637943e-06,
+      "loss": 0.1143,
+      "step": 4937
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.527755757434968e-06,
+      "loss": 0.1082,
+      "step": 4938
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524285985473507e-06,
+      "loss": 0.1225,
+      "step": 4939
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52081627091372e-06,
+      "loss": 0.1137,
+      "step": 4940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.517346614174283e-06,
+      "loss": 0.1171,
+      "step": 4941
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513877015673858e-06,
+      "loss": 0.1234,
+      "step": 4942
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.510407475831099e-06,
+      "loss": 0.12,
+      "step": 4943
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.506937995064655e-06,
+      "loss": 0.1268,
+      "step": 4944
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.503468573793173e-06,
+      "loss": 0.1172,
+      "step": 4945
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.499999212435283e-06,
+      "loss": 0.1156,
+      "step": 4946
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496529911409615e-06,
+      "loss": 0.1196,
+      "step": 4947
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.493060671134795e-06,
+      "loss": 0.1081,
+      "step": 4948
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.489591492029428e-06,
+      "loss": 0.111,
+      "step": 4949
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.486122374512122e-06,
+      "loss": 0.1199,
+      "step": 4950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.482653319001478e-06,
+      "loss": 0.1094,
+      "step": 4951
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.479184325916083e-06,
+      "loss": 0.1169,
+      "step": 4952
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475715395674523e-06,
+      "loss": 0.1166,
+      "step": 4953
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.472246528695377e-06,
+      "loss": 0.1119,
+      "step": 4954
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46877772539721e-06,
+      "loss": 0.1115,
+      "step": 4955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465308986198581e-06,
+      "loss": 0.1234,
+      "step": 4956
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.461840311518043e-06,
+      "loss": 0.1174,
+      "step": 4957
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.458371701774145e-06,
+      "loss": 0.1121,
+      "step": 4958
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.454903157385421e-06,
+      "loss": 0.1132,
+      "step": 4959
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.451434678770402e-06,
+      "loss": 0.1087,
+      "step": 4960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44796626634761e-06,
+      "loss": 0.1153,
+      "step": 4961
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444497920535554e-06,
+      "loss": 0.1181,
+      "step": 4962
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.441029641752741e-06,
+      "loss": 0.1197,
+      "step": 4963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437561430417672e-06,
+      "loss": 0.1125,
+      "step": 4964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.434093286948832e-06,
+      "loss": 0.1096,
+      "step": 4965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.430625211764706e-06,
+      "loss": 0.1177,
+      "step": 4966
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427157205283762e-06,
+      "loss": 0.1104,
+      "step": 4967
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423689267924469e-06,
+      "loss": 0.1164,
+      "step": 4968
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.420221400105278e-06,
+      "loss": 0.1209,
+      "step": 4969
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.416753602244643e-06,
+      "loss": 0.116,
+      "step": 4970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.413285874760995e-06,
+      "loss": 0.1158,
+      "step": 4971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.409818218072774e-06,
+      "loss": 0.1212,
+      "step": 4972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.406350632598393e-06,
+      "loss": 0.1166,
+      "step": 4973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.402883118756277e-06,
+      "loss": 0.1196,
+      "step": 4974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39941567696482e-06,
+      "loss": 0.1127,
+      "step": 4975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395948307642423e-06,
+      "loss": 0.115,
+      "step": 4976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39248101120747e-06,
+      "loss": 0.103,
+      "step": 4977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.389013788078344e-06,
+      "loss": 0.1182,
+      "step": 4978
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.385546638673418e-06,
+      "loss": 0.1086,
+      "step": 4979
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.382079563411045e-06,
+      "loss": 0.111,
+      "step": 4980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.378612562709584e-06,
+      "loss": 0.1224,
+      "step": 4981
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.375145636987375e-06,
+      "loss": 0.1117,
+      "step": 4982
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.371678786662746e-06,
+      "loss": 0.1151,
+      "step": 4983
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36821201215403e-06,
+      "loss": 0.1174,
+      "step": 4984
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.364745313879543e-06,
+      "loss": 0.1196,
+      "step": 4985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361278692257588e-06,
+      "loss": 0.1161,
+      "step": 4986
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.357812147706466e-06,
+      "loss": 0.1215,
+      "step": 4987
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.354345680644459e-06,
+      "loss": 0.1045,
+      "step": 4988
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.350879291489848e-06,
+      "loss": 0.1126,
+      "step": 4989
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.347412980660903e-06,
+      "loss": 0.1095,
+      "step": 4990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.343946748575889e-06,
+      "loss": 0.1077,
+      "step": 4991
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.340480595653047e-06,
+      "loss": 0.1217,
+      "step": 4992
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.337014522310621e-06,
+      "loss": 0.1094,
+      "step": 4993
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.333548528966849e-06,
+      "loss": 0.12,
+      "step": 4994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.330082616039946e-06,
+      "loss": 0.1198,
+      "step": 4995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32661678394812e-06,
+      "loss": 0.1176,
+      "step": 4996
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.323151033109581e-06,
+      "loss": 0.111,
+      "step": 4997
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.319685363942516e-06,
+      "loss": 0.1135,
+      "step": 4998
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.316219776865108e-06,
+      "loss": 0.1282,
+      "step": 4999
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312754272295538e-06,
+      "loss": 0.1216,
+      "step": 5000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309288850651956e-06,
+      "loss": 0.1174,
+      "step": 5001
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30582351235252e-06,
+      "loss": 0.1082,
+      "step": 5002
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.302358257815373e-06,
+      "loss": 0.1036,
+      "step": 5003
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.298893087458645e-06,
+      "loss": 0.1163,
+      "step": 5004
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29542800170046e-06,
+      "loss": 0.1141,
+      "step": 5005
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.291963000958932e-06,
+      "loss": 0.1247,
+      "step": 5006
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288498085652162e-06,
+      "loss": 0.1131,
+      "step": 5007
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285033256198237e-06,
+      "loss": 0.1059,
+      "step": 5008
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28156851301524e-06,
+      "loss": 0.116,
+      "step": 5009
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.278103856521246e-06,
+      "loss": 0.1077,
+      "step": 5010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.274639287134309e-06,
+      "loss": 0.0996,
+      "step": 5011
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271174805272482e-06,
+      "loss": 0.1224,
+      "step": 5012
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.267710411353809e-06,
+      "loss": 0.1118,
+      "step": 5013
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.264246105796307e-06,
+      "loss": 0.1481,
+      "step": 5014
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.260781889018e-06,
+      "loss": 0.1089,
+      "step": 5015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.257317761436899e-06,
+      "loss": 0.1181,
+      "step": 5016
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253853723470992e-06,
+      "loss": 0.1283,
+      "step": 5017
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.250389775538273e-06,
+      "loss": 0.1068,
+      "step": 5018
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24692591805671e-06,
+      "loss": 0.107,
+      "step": 5019
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.243462151444275e-06,
+      "loss": 0.1253,
+      "step": 5020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.23999847611891e-06,
+      "loss": 0.1193,
+      "step": 5021
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.236534892498564e-06,
+      "loss": 0.1062,
+      "step": 5022
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233071401001166e-06,
+      "loss": 0.1162,
+      "step": 5023
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.229608002044633e-06,
+      "loss": 0.1204,
+      "step": 5024
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22614469604688e-06,
+      "loss": 0.1176,
+      "step": 5025
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.222681483425801e-06,
+      "loss": 0.1181,
+      "step": 5026
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.219218364599277e-06,
+      "loss": 0.1184,
+      "step": 5027
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21575533998519e-06,
+      "loss": 0.1129,
+      "step": 5028
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.212292410001399e-06,
+      "loss": 0.1273,
+      "step": 5029
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208829575065754e-06,
+      "loss": 0.111,
+      "step": 5030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.205366835596102e-06,
+      "loss": 0.102,
+      "step": 5031
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.201904192010272e-06,
+      "loss": 0.1108,
+      "step": 5032
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198441644726072e-06,
+      "loss": 0.1269,
+      "step": 5033
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.194979194161315e-06,
+      "loss": 0.1237,
+      "step": 5034
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191516840733792e-06,
+      "loss": 0.1081,
+      "step": 5035
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188054584861285e-06,
+      "loss": 0.117,
+      "step": 5036
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18459242696157e-06,
+      "loss": 0.1215,
+      "step": 5037
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.1811303674524e-06,
+      "loss": 0.1119,
+      "step": 5038
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.177668406751526e-06,
+      "loss": 0.1166,
+      "step": 5039
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.1169,
+      "step": 5040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170744783445583e-06,
+      "loss": 0.1102,
+      "step": 5041
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.167283121675949e-06,
+      "loss": 0.1122,
+      "step": 5042
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.163821560385478e-06,
+      "loss": 0.1149,
+      "step": 5043
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.160360099991852e-06,
+      "loss": 0.1151,
+      "step": 5044
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.156898740912753e-06,
+      "loss": 0.1143,
+      "step": 5045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.153437483565835e-06,
+      "loss": 0.1195,
+      "step": 5046
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.149976328368754e-06,
+      "loss": 0.1166,
+      "step": 5047
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.146515275739142e-06,
+      "loss": 0.1218,
+      "step": 5048
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.143054326094632e-06,
+      "loss": 0.1129,
+      "step": 5049
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13959347985283e-06,
+      "loss": 0.1009,
+      "step": 5050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.136132737431339e-06,
+      "loss": 0.1263,
+      "step": 5051
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.132672099247753e-06,
+      "loss": 0.116,
+      "step": 5052
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129211565719637e-06,
+      "loss": 0.1187,
+      "step": 5053
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.125751137264556e-06,
+      "loss": 0.113,
+      "step": 5054
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122290814300064e-06,
+      "loss": 0.1145,
+      "step": 5055
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118830597243697e-06,
+      "loss": 0.1187,
+      "step": 5056
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.115370486512979e-06,
+      "loss": 0.1165,
+      "step": 5057
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111910482525423e-06,
+      "loss": 0.1094,
+      "step": 5058
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.108450585698527e-06,
+      "loss": 0.0989,
+      "step": 5059
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104990796449774e-06,
+      "loss": 0.1158,
+      "step": 5060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.101531115196637e-06,
+      "loss": 0.1066,
+      "step": 5061
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.098071542356581e-06,
+      "loss": 0.1117,
+      "step": 5062
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.09461207834705e-06,
+      "loss": 0.1072,
+      "step": 5063
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091152723585477e-06,
+      "loss": 0.1171,
+      "step": 5064
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087693478489285e-06,
+      "loss": 0.1026,
+      "step": 5065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.084234343475877e-06,
+      "loss": 0.1036,
+      "step": 5066
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.080775318962648e-06,
+      "loss": 0.1204,
+      "step": 5067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07731640536698e-06,
+      "loss": 0.1035,
+      "step": 5068
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.073857603106243e-06,
+      "loss": 0.1092,
+      "step": 5069
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.070398912597785e-06,
+      "loss": 0.1118,
+      "step": 5070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.066940334258951e-06,
+      "loss": 0.1037,
+      "step": 5071
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.063481868507066e-06,
+      "loss": 0.1217,
+      "step": 5072
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06002351575944e-06,
+      "loss": 0.1194,
+      "step": 5073
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056565276433378e-06,
+      "loss": 0.1128,
+      "step": 5074
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.053107150946163e-06,
+      "loss": 0.1138,
+      "step": 5075
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049649139715067e-06,
+      "loss": 0.1067,
+      "step": 5076
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.04619124315735e-06,
+      "loss": 0.1108,
+      "step": 5077
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.042733461690259e-06,
+      "loss": 0.1108,
+      "step": 5078
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039275795731014e-06,
+      "loss": 0.1126,
+      "step": 5079
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035818245696843e-06,
+      "loss": 0.1087,
+      "step": 5080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.032360812004944e-06,
+      "loss": 0.1161,
+      "step": 5081
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.028903495072503e-06,
+      "loss": 0.1172,
+      "step": 5082
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.025446295316701e-06,
+      "loss": 0.1175,
+      "step": 5083
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.021989213154699e-06,
+      "loss": 0.1006,
+      "step": 5084
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.018532249003633e-06,
+      "loss": 0.1202,
+      "step": 5085
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.015075403280644e-06,
+      "loss": 0.1165,
+      "step": 5086
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.011618676402845e-06,
+      "loss": 0.1078,
+      "step": 5087
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008162068787341e-06,
+      "loss": 0.1213,
+      "step": 5088
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004705580851225e-06,
+      "loss": 0.1168,
+      "step": 5089
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001249213011565e-06,
+      "loss": 0.1095,
+      "step": 5090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99779296568543e-06,
+      "loss": 0.1065,
+      "step": 5091
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.994336839289853e-06,
+      "loss": 0.1051,
+      "step": 5092
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.990880834241873e-06,
+      "loss": 0.1115,
+      "step": 5093
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987424950958505e-06,
+      "loss": 0.1083,
+      "step": 5094
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98396918985675e-06,
+      "loss": 0.1124,
+      "step": 5095
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.980513551353595e-06,
+      "loss": 0.1124,
+      "step": 5096
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.977058035866015e-06,
+      "loss": 0.1189,
+      "step": 5097
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.973602643810962e-06,
+      "loss": 0.104,
+      "step": 5098
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97014737560538e-06,
+      "loss": 0.1185,
+      "step": 5099
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.9666922316662e-06,
+      "loss": 0.115,
+      "step": 5100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.963237212410329e-06,
+      "loss": 0.1081,
+      "step": 5101
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.959782318254665e-06,
+      "loss": 0.1074,
+      "step": 5102
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.956327549616093e-06,
+      "loss": 0.1251,
+      "step": 5103
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.952872906911484e-06,
+      "loss": 0.1227,
+      "step": 5104
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.949418390557682e-06,
+      "loss": 0.1174,
+      "step": 5105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945964000971525e-06,
+      "loss": 0.1167,
+      "step": 5106
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.942509738569834e-06,
+      "loss": 0.1151,
+      "step": 5107
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93905560376942e-06,
+      "loss": 0.1112,
+      "step": 5108
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935601596987069e-06,
+      "loss": 0.1266,
+      "step": 5109
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.932147718639562e-06,
+      "loss": 0.1128,
+      "step": 5110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.928693969143652e-06,
+      "loss": 0.1333,
+      "step": 5111
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.925240348916085e-06,
+      "loss": 0.1279,
+      "step": 5112
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.921786858373587e-06,
+      "loss": 0.1168,
+      "step": 5113
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.918333497932878e-06,
+      "loss": 0.1123,
+      "step": 5114
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.914880268010647e-06,
+      "loss": 0.1193,
+      "step": 5115
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.911427169023583e-06,
+      "loss": 0.1183,
+      "step": 5116
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90797420138835e-06,
+      "loss": 0.1112,
+      "step": 5117
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.904521365521591e-06,
+      "loss": 0.1067,
+      "step": 5118
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.901068661839945e-06,
+      "loss": 0.1184,
+      "step": 5119
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897616090760032e-06,
+      "loss": 0.1088,
+      "step": 5120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.894163652698448e-06,
+      "loss": 0.111,
+      "step": 5121
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.890711348071784e-06,
+      "loss": 0.1165,
+      "step": 5122
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887259177296612e-06,
+      "loss": 0.1177,
+      "step": 5123
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.883807140789478e-06,
+      "loss": 0.0923,
+      "step": 5124
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.1111,
+      "step": 5125
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.876903472245469e-06,
+      "loss": 0.1171,
+      "step": 5126
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.873451841041619e-06,
+      "loss": 0.1178,
+      "step": 5127
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.870000345771863e-06,
+      "loss": 0.1116,
+      "step": 5128
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.866548986852673e-06,
+      "loss": 0.0997,
+      "step": 5129
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.863097764700508e-06,
+      "loss": 0.1179,
+      "step": 5130
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.859646679731799e-06,
+      "loss": 0.1181,
+      "step": 5131
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.856195732362975e-06,
+      "loss": 0.114,
+      "step": 5132
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.852744923010438e-06,
+      "loss": 0.1071,
+      "step": 5133
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84929425209058e-06,
+      "loss": 0.1147,
+      "step": 5134
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.845843720019776e-06,
+      "loss": 0.1159,
+      "step": 5135
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.842393327214378e-06,
+      "loss": 0.1198,
+      "step": 5136
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838943074090726e-06,
+      "loss": 0.1224,
+      "step": 5137
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.835492961065139e-06,
+      "loss": 0.1255,
+      "step": 5138
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.832042988553929e-06,
+      "loss": 0.108,
+      "step": 5139
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.828593156973379e-06,
+      "loss": 0.1087,
+      "step": 5140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.825143466739764e-06,
+      "loss": 0.1084,
+      "step": 5141
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.821693918269334e-06,
+      "loss": 0.1124,
+      "step": 5142
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.818244511978335e-06,
+      "loss": 0.1181,
+      "step": 5143
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.814795248282974e-06,
+      "loss": 0.1123,
+      "step": 5144
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811346127599465e-06,
+      "loss": 0.123,
+      "step": 5145
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.807897150343985e-06,
+      "loss": 0.1026,
+      "step": 5146
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.804448316932711e-06,
+      "loss": 0.0981,
+      "step": 5147
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.800999627781786e-06,
+      "loss": 0.1082,
+      "step": 5148
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.797551083307352e-06,
+      "loss": 0.1168,
+      "step": 5149
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.794102683925515e-06,
+      "loss": 0.101,
+      "step": 5150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79065443005238e-06,
+      "loss": 0.1046,
+      "step": 5151
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787206322104025e-06,
+      "loss": 0.1066,
+      "step": 5152
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.783758360496515e-06,
+      "loss": 0.1262,
+      "step": 5153
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.780310545645897e-06,
+      "loss": 0.0963,
+      "step": 5154
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.776862877968198e-06,
+      "loss": 0.106,
+      "step": 5155
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.773415357879429e-06,
+      "loss": 0.1318,
+      "step": 5156
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.769967985795581e-06,
+      "loss": 0.1113,
+      "step": 5157
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.766520762132627e-06,
+      "loss": 0.1094,
+      "step": 5158
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.763073687306523e-06,
+      "loss": 0.1065,
+      "step": 5159
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.759626761733215e-06,
+      "loss": 0.1222,
+      "step": 5160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.756179985828617e-06,
+      "loss": 0.1133,
+      "step": 5161
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.752733360008635e-06,
+      "loss": 0.1104,
+      "step": 5162
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.749286884689153e-06,
+      "loss": 0.108,
+      "step": 5163
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.745840560286034e-06,
+      "loss": 0.1271,
+      "step": 5164
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.742394387215128e-06,
+      "loss": 0.1142,
+      "step": 5165
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.738948365892267e-06,
+      "loss": 0.1094,
+      "step": 5166
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73550249673326e-06,
+      "loss": 0.1085,
+      "step": 5167
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.732056780153904e-06,
+      "loss": 0.1006,
+      "step": 5168
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72861121656997e-06,
+      "loss": 0.1093,
+      "step": 5169
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725165806397216e-06,
+      "loss": 0.1133,
+      "step": 5170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.721720550051378e-06,
+      "loss": 0.1194,
+      "step": 5171
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.718275447948178e-06,
+      "loss": 0.1142,
+      "step": 5172
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.714830500503314e-06,
+      "loss": 0.1159,
+      "step": 5173
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.71138570813247e-06,
+      "loss": 0.1118,
+      "step": 5174
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.707941071251311e-06,
+      "loss": 0.1162,
+      "step": 5175
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.704496590275479e-06,
+      "loss": 0.1237,
+      "step": 5176
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.701052265620597e-06,
+      "loss": 0.1165,
+      "step": 5177
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.697608097702277e-06,
+      "loss": 0.1224,
+      "step": 5178
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.694164086936103e-06,
+      "loss": 0.115,
+      "step": 5179
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.690720233737645e-06,
+      "loss": 0.114,
+      "step": 5180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.687276538522458e-06,
+      "loss": 0.1124,
+      "step": 5181
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.683833001706068e-06,
+      "loss": 0.1215,
+      "step": 5182
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.680389623703985e-06,
+      "loss": 0.1055,
+      "step": 5183
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.676946404931706e-06,
+      "loss": 0.1033,
+      "step": 5184
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6735033458047e-06,
+      "loss": 0.1159,
+      "step": 5185
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670060446738425e-06,
+      "loss": 0.1136,
+      "step": 5186
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.666617708148315e-06,
+      "loss": 0.1222,
+      "step": 5187
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66317513044979e-06,
+      "loss": 0.1055,
+      "step": 5188
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.659732714058233e-06,
+      "loss": 0.1104,
+      "step": 5189
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.656290459389032e-06,
+      "loss": 0.1226,
+      "step": 5190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.652848366857541e-06,
+      "loss": 0.1076,
+      "step": 5191
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.649406436879095e-06,
+      "loss": 0.1122,
+      "step": 5192
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.645964669869018e-06,
+      "loss": 0.1175,
+      "step": 5193
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.642523066242607e-06,
+      "loss": 0.0996,
+      "step": 5194
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.639081626415132e-06,
+      "loss": 0.119,
+      "step": 5195
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63564035080186e-06,
+      "loss": 0.1177,
+      "step": 5196
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.632199239818028e-06,
+      "loss": 0.1036,
+      "step": 5197
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.628758293878853e-06,
+      "loss": 0.0977,
+      "step": 5198
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625317513399538e-06,
+      "loss": 0.1176,
+      "step": 5199
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.62187689879526e-06,
+      "loss": 0.1166,
+      "step": 5200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.618436450481182e-06,
+      "loss": 0.1193,
+      "step": 5201
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.614996168872434e-06,
+      "loss": 0.1191,
+      "step": 5202
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.611556054384144e-06,
+      "loss": 0.1168,
+      "step": 5203
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.608116107431405e-06,
+      "loss": 0.1223,
+      "step": 5204
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.604676328429297e-06,
+      "loss": 0.1119,
+      "step": 5205
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.601236717792884e-06,
+      "loss": 0.1031,
+      "step": 5206
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.5977972759372e-06,
+      "loss": 0.1161,
+      "step": 5207
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.594358003277257e-06,
+      "loss": 0.1273,
+      "step": 5208
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.590918900228062e-06,
+      "loss": 0.1144,
+      "step": 5209
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.587479967204584e-06,
+      "loss": 0.1135,
+      "step": 5210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.584041204621783e-06,
+      "loss": 0.1193,
+      "step": 5211
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580602612894595e-06,
+      "loss": 0.1323,
+      "step": 5212
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.577164192437933e-06,
+      "loss": 0.1178,
+      "step": 5213
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.573725943666698e-06,
+      "loss": 0.1163,
+      "step": 5214
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.570287866995756e-06,
+      "loss": 0.1046,
+      "step": 5215
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56684996283996e-06,
+      "loss": 0.1251,
+      "step": 5216
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.563412231614146e-06,
+      "loss": 0.1136,
+      "step": 5217
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.559974673733125e-06,
+      "loss": 0.1304,
+      "step": 5218
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.556537289611684e-06,
+      "loss": 0.1071,
+      "step": 5219
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.553100079664598e-06,
+      "loss": 0.1232,
+      "step": 5220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54966304430661e-06,
+      "loss": 0.1173,
+      "step": 5221
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.546226183952452e-06,
+      "loss": 0.1093,
+      "step": 5222
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542789499016822e-06,
+      "loss": 0.1152,
+      "step": 5223
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.539352989914416e-06,
+      "loss": 0.121,
+      "step": 5224
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535916657059889e-06,
+      "loss": 0.1213,
+      "step": 5225
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532480500867887e-06,
+      "loss": 0.105,
+      "step": 5226
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.529044521753035e-06,
+      "loss": 0.1021,
+      "step": 5227
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.525608720129928e-06,
+      "loss": 0.1128,
+      "step": 5228
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.522173096413143e-06,
+      "loss": 0.121,
+      "step": 5229
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.518737651017241e-06,
+      "loss": 0.1231,
+      "step": 5230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515302384356753e-06,
+      "loss": 0.109,
+      "step": 5231
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511867296846197e-06,
+      "loss": 0.1203,
+      "step": 5232
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.508432388900069e-06,
+      "loss": 0.1163,
+      "step": 5233
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.504997660932832e-06,
+      "loss": 0.1249,
+      "step": 5234
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.501563113358933e-06,
+      "loss": 0.1126,
+      "step": 5235
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.498128746592806e-06,
+      "loss": 0.0991,
+      "step": 5236
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.494694561048858e-06,
+      "loss": 0.1067,
+      "step": 5237
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.491260557141462e-06,
+      "loss": 0.1122,
+      "step": 5238
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.487826735284991e-06,
+      "loss": 0.1062,
+      "step": 5239
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.484393095893781e-06,
+      "loss": 0.0943,
+      "step": 5240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.480959639382144e-06,
+      "loss": 0.1239,
+      "step": 5241
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.477526366164378e-06,
+      "loss": 0.0964,
+      "step": 5242
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.474093276654764e-06,
+      "loss": 0.0983,
+      "step": 5243
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47066037126754e-06,
+      "loss": 0.1111,
+      "step": 5244
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.467227650416947e-06,
+      "loss": 0.1019,
+      "step": 5245
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.463795114517189e-06,
+      "loss": 0.1264,
+      "step": 5246
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.460362763982443e-06,
+      "loss": 0.109,
+      "step": 5247
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456930599226876e-06,
+      "loss": 0.1103,
+      "step": 5248
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453498620664631e-06,
+      "loss": 0.1165,
+      "step": 5249
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450066828709818e-06,
+      "loss": 0.1001,
+      "step": 5250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.446635223776535e-06,
+      "loss": 0.1145,
+      "step": 5251
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.443203806278859e-06,
+      "loss": 0.1162,
+      "step": 5252
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.439772576630837e-06,
+      "loss": 0.112,
+      "step": 5253
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.436341535246489e-06,
+      "loss": 0.1079,
+      "step": 5254
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432910682539824e-06,
+      "loss": 0.1139,
+      "step": 5255
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.429480018924823e-06,
+      "loss": 0.1205,
+      "step": 5256
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426049544815445e-06,
+      "loss": 0.1188,
+      "step": 5257
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.422619260625626e-06,
+      "loss": 0.1126,
+      "step": 5258
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41918916676928e-06,
+      "loss": 0.1046,
+      "step": 5259
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41575926366029e-06,
+      "loss": 0.1121,
+      "step": 5260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41232955171253e-06,
+      "loss": 0.1205,
+      "step": 5261
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408900031339838e-06,
+      "loss": 0.1033,
+      "step": 5262
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.405470702956039e-06,
+      "loss": 0.0933,
+      "step": 5263
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40204156697493e-06,
+      "loss": 0.1122,
+      "step": 5264
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.398612623810281e-06,
+      "loss": 0.1031,
+      "step": 5265
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.395183873875849e-06,
+      "loss": 0.1249,
+      "step": 5266
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.391755317585358e-06,
+      "loss": 0.1022,
+      "step": 5267
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.38832695535251e-06,
+      "loss": 0.1089,
+      "step": 5268
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.38489878759099e-06,
+      "loss": 0.1107,
+      "step": 5269
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.381470814714455e-06,
+      "loss": 0.1098,
+      "step": 5270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378043037136533e-06,
+      "loss": 0.1136,
+      "step": 5271
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.374615455270843e-06,
+      "loss": 0.1112,
+      "step": 5272
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.371188069530965e-06,
+      "loss": 0.0921,
+      "step": 5273
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.367760880330465e-06,
+      "loss": 0.1003,
+      "step": 5274
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36433388808288e-06,
+      "loss": 0.1094,
+      "step": 5275
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36090709320173e-06,
+      "loss": 0.1087,
+      "step": 5276
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.357480496100498e-06,
+      "loss": 0.1184,
+      "step": 5277
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35405409719266e-06,
+      "loss": 0.1147,
+      "step": 5278
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.350627896891661e-06,
+      "loss": 0.1118,
+      "step": 5279
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.347201895610915e-06,
+      "loss": 0.1233,
+      "step": 5280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.343776093763817e-06,
+      "loss": 0.1112,
+      "step": 5281
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.340350491763745e-06,
+      "loss": 0.1056,
+      "step": 5282
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.336925090024039e-06,
+      "loss": 0.1205,
+      "step": 5283
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33349988895803e-06,
+      "loss": 0.0998,
+      "step": 5284
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.330074888979016e-06,
+      "loss": 0.1124,
+      "step": 5285
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32665009050027e-06,
+      "loss": 0.1186,
+      "step": 5286
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32322549393504e-06,
+      "loss": 0.1049,
+      "step": 5287
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.319801099696558e-06,
+      "loss": 0.1051,
+      "step": 5288
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.316376908198023e-06,
+      "loss": 0.1128,
+      "step": 5289
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.312952919852612e-06,
+      "loss": 0.1123,
+      "step": 5290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.309529135073481e-06,
+      "loss": 0.117,
+      "step": 5291
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.306105554273758e-06,
+      "loss": 0.1153,
+      "step": 5292
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302682177866543e-06,
+      "loss": 0.1364,
+      "step": 5293
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299259006264915e-06,
+      "loss": 0.1011,
+      "step": 5294
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.295836039881934e-06,
+      "loss": 0.1174,
+      "step": 5295
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.292413279130625e-06,
+      "loss": 0.1118,
+      "step": 5296
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.288990724423997e-06,
+      "loss": 0.1052,
+      "step": 5297
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285568376175026e-06,
+      "loss": 0.1158,
+      "step": 5298
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.28214623479667e-06,
+      "loss": 0.1081,
+      "step": 5299
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.278724300701855e-06,
+      "loss": 0.1108,
+      "step": 5300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.27530257430349e-06,
+      "loss": 0.1105,
+      "step": 5301
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.271881056014454e-06,
+      "loss": 0.105,
+      "step": 5302
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.268459746247602e-06,
+      "loss": 0.1102,
+      "step": 5303
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.265038645415764e-06,
+      "loss": 0.1152,
+      "step": 5304
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.261617753931747e-06,
+      "loss": 0.1078,
+      "step": 5305
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.258197072208322e-06,
+      "loss": 0.1179,
+      "step": 5306
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.254776600658253e-06,
+      "loss": 0.1139,
+      "step": 5307
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25135633969426e-06,
+      "loss": 0.1197,
+      "step": 5308
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.247936289729052e-06,
+      "loss": 0.1093,
+      "step": 5309
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.244516451175307e-06,
+      "loss": 0.1068,
+      "step": 5310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.241096824445677e-06,
+      "loss": 0.1086,
+      "step": 5311
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237677409952784e-06,
+      "loss": 0.1158,
+      "step": 5312
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.234258208109234e-06,
+      "loss": 0.1057,
+      "step": 5313
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2308392193276e-06,
+      "loss": 0.1162,
+      "step": 5314
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22742044402043e-06,
+      "loss": 0.1149,
+      "step": 5315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224001882600254e-06,
+      "loss": 0.1068,
+      "step": 5316
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.220583535479564e-06,
+      "loss": 0.1045,
+      "step": 5317
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21716540307084e-06,
+      "loss": 0.114,
+      "step": 5318
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21374748578652e-06,
+      "loss": 0.1131,
+      "step": 5319
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210329784039028e-06,
+      "loss": 0.1055,
+      "step": 5320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.206912298240757e-06,
+      "loss": 0.1087,
+      "step": 5321
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.203495028804079e-06,
+      "loss": 0.1112,
+      "step": 5322
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.200077976141331e-06,
+      "loss": 0.1053,
+      "step": 5323
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.196661140664836e-06,
+      "loss": 0.1097,
+      "step": 5324
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.193244522786877e-06,
+      "loss": 0.1144,
+      "step": 5325
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.189828122919721e-06,
+      "loss": 0.1036,
+      "step": 5326
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.186411941475603e-06,
+      "loss": 0.1194,
+      "step": 5327
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.182995978866738e-06,
+      "loss": 0.1098,
+      "step": 5328
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.17958023550531e-06,
+      "loss": 0.1147,
+      "step": 5329
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.176164711803474e-06,
+      "loss": 0.112,
+      "step": 5330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.172749408173362e-06,
+      "loss": 0.1029,
+      "step": 5331
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.169334325027084e-06,
+      "loss": 0.1041,
+      "step": 5332
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16591946277671e-06,
+      "loss": 0.106,
+      "step": 5333
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.162504821834296e-06,
+      "loss": 0.1136,
+      "step": 5334
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15909040261187e-06,
+      "loss": 0.1125,
+      "step": 5335
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.155676205521428e-06,
+      "loss": 0.0974,
+      "step": 5336
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152262230974945e-06,
+      "loss": 0.1065,
+      "step": 5337
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14884847938436e-06,
+      "loss": 0.1008,
+      "step": 5338
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14543495116159e-06,
+      "loss": 0.107,
+      "step": 5339
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14202164671853e-06,
+      "loss": 0.1127,
+      "step": 5340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.138608566467043e-06,
+      "loss": 0.1164,
+      "step": 5341
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.135195710818966e-06,
+      "loss": 0.0981,
+      "step": 5342
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.131783080186111e-06,
+      "loss": 0.1153,
+      "step": 5343
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.128370674980255e-06,
+      "loss": 0.1116,
+      "step": 5344
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.124958495613154e-06,
+      "loss": 0.109,
+      "step": 5345
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12154654249654e-06,
+      "loss": 0.1149,
+      "step": 5346
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.118134816042111e-06,
+      "loss": 0.1052,
+      "step": 5347
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.114723316661541e-06,
+      "loss": 0.1048,
+      "step": 5348
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.111312044766477e-06,
+      "loss": 0.1112,
+      "step": 5349
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10790100076854e-06,
+      "loss": 0.101,
+      "step": 5350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.104490185079316e-06,
+      "loss": 0.1026,
+      "step": 5351
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.101079598110368e-06,
+      "loss": 0.1122,
+      "step": 5352
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.097669240273237e-06,
+      "loss": 0.1156,
+      "step": 5353
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.094259111979427e-06,
+      "loss": 0.1147,
+      "step": 5354
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.09084921364042e-06,
+      "loss": 0.1069,
+      "step": 5355
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.087439545667674e-06,
+      "loss": 0.1273,
+      "step": 5356
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.084030108472606e-06,
+      "loss": 0.1208,
+      "step": 5357
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.080620902466615e-06,
+      "loss": 0.1002,
+      "step": 5358
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.077211928061072e-06,
+      "loss": 0.1119,
+      "step": 5359
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07380318566732e-06,
+      "loss": 0.107,
+      "step": 5360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.070394675696666e-06,
+      "loss": 0.1132,
+      "step": 5361
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.066986398560405e-06,
+      "loss": 0.1096,
+      "step": 5362
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06357835466979e-06,
+      "loss": 0.1177,
+      "step": 5363
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060170544436045e-06,
+      "loss": 0.1131,
+      "step": 5364
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.056762968270374e-06,
+      "loss": 0.1146,
+      "step": 5365
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.053355626583954e-06,
+      "loss": 0.1162,
+      "step": 5366
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.049948519787923e-06,
+      "loss": 0.1173,
+      "step": 5367
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.046541648293402e-06,
+      "loss": 0.1091,
+      "step": 5368
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043135012511478e-06,
+      "loss": 0.1049,
+      "step": 5369
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.039728612853206e-06,
+      "loss": 0.1205,
+      "step": 5370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.036322449729619e-06,
+      "loss": 0.1111,
+      "step": 5371
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.03291652355172e-06,
+      "loss": 0.116,
+      "step": 5372
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.029510834730482e-06,
+      "loss": 0.1063,
+      "step": 5373
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026105383676851e-06,
+      "loss": 0.1071,
+      "step": 5374
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.022700170801741e-06,
+      "loss": 0.1081,
+      "step": 5375
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.019295196516044e-06,
+      "loss": 0.1004,
+      "step": 5376
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01589046123061e-06,
+      "loss": 0.1009,
+      "step": 5377
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012485965356278e-06,
+      "loss": 0.1048,
+      "step": 5378
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.009081709303842e-06,
+      "loss": 0.1076,
+      "step": 5379
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005677693484077e-06,
+      "loss": 0.1028,
+      "step": 5380
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00227391830773e-06,
+      "loss": 0.1105,
+      "step": 5381
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99887038418551e-06,
+      "loss": 0.1081,
+      "step": 5382
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995467091528101e-06,
+      "loss": 0.1117,
+      "step": 5383
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.992064040746163e-06,
+      "loss": 0.1123,
+      "step": 5384
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.988661232250319e-06,
+      "loss": 0.1131,
+      "step": 5385
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.985258666451166e-06,
+      "loss": 0.1126,
+      "step": 5386
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.981856343759277e-06,
+      "loss": 0.1115,
+      "step": 5387
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.978454264585185e-06,
+      "loss": 0.132,
+      "step": 5388
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.975052429339406e-06,
+      "loss": 0.1021,
+      "step": 5389
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.971650838432414e-06,
+      "loss": 0.1119,
+      "step": 5390
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96824949227466e-06,
+      "loss": 0.118,
+      "step": 5391
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.964848391276565e-06,
+      "loss": 0.1184,
+      "step": 5392
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.961447535848523e-06,
+      "loss": 0.1153,
+      "step": 5393
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.958046926400893e-06,
+      "loss": 0.1131,
+      "step": 5394
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.954646563344013e-06,
+      "loss": 0.1091,
+      "step": 5395
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.951246447088174e-06,
+      "loss": 0.1178,
+      "step": 5396
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.947846578043658e-06,
+      "loss": 0.1154,
+      "step": 5397
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.944446956620703e-06,
+      "loss": 0.1072,
+      "step": 5398
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941047583229526e-06,
+      "loss": 0.1155,
+      "step": 5399
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.937648458280305e-06,
+      "loss": 0.1149,
+      "step": 5400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.934249582183194e-06,
+      "loss": 0.1187,
+      "step": 5401
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.930850955348321e-06,
+      "loss": 0.1078,
+      "step": 5402
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.927452578185774e-06,
+      "loss": 0.114,
+      "step": 5403
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.924054451105614e-06,
+      "loss": 0.1045,
+      "step": 5404
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.920656574517877e-06,
+      "loss": 0.1084,
+      "step": 5405
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917258948832563e-06,
+      "loss": 0.1089,
+      "step": 5406
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.913861574459644e-06,
+      "loss": 0.1082,
+      "step": 5407
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.910464451809068e-06,
+      "loss": 0.1129,
+      "step": 5408
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907067581290738e-06,
+      "loss": 0.1268,
+      "step": 5409
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.903670963314536e-06,
+      "loss": 0.1045,
+      "step": 5410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900274598290316e-06,
+      "loss": 0.11,
+      "step": 5411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.896878486627894e-06,
+      "loss": 0.1173,
+      "step": 5412
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.893482628737062e-06,
+      "loss": 0.1172,
+      "step": 5413
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89008702502758e-06,
+      "loss": 0.1059,
+      "step": 5414
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.886691675909175e-06,
+      "loss": 0.1158,
+      "step": 5415
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88329658179154e-06,
+      "loss": 0.1232,
+      "step": 5416
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.879901743084346e-06,
+      "loss": 0.1211,
+      "step": 5417
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.876507160197226e-06,
+      "loss": 0.1091,
+      "step": 5418
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873112833539787e-06,
+      "loss": 0.1123,
+      "step": 5419
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.869718763521604e-06,
+      "loss": 0.104,
+      "step": 5420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.866324950552219e-06,
+      "loss": 0.1107,
+      "step": 5421
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.862931395041139e-06,
+      "loss": 0.1259,
+      "step": 5422
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.85953809739785e-06,
+      "loss": 0.1137,
+      "step": 5423
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.856145058031802e-06,
+      "loss": 0.1143,
+      "step": 5424
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85275227735241e-06,
+      "loss": 0.1161,
+      "step": 5425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.849359755769063e-06,
+      "loss": 0.1156,
+      "step": 5426
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84596749369112e-06,
+      "loss": 0.1067,
+      "step": 5427
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.842575491527903e-06,
+      "loss": 0.1079,
+      "step": 5428
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839183749688705e-06,
+      "loss": 0.103,
+      "step": 5429
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835792268582789e-06,
+      "loss": 0.1149,
+      "step": 5430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.832401048619385e-06,
+      "loss": 0.1063,
+      "step": 5431
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82901009020769e-06,
+      "loss": 0.1062,
+      "step": 5432
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.825619393756878e-06,
+      "loss": 0.1082,
+      "step": 5433
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822228959676082e-06,
+      "loss": 0.1048,
+      "step": 5434
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8188387883744e-06,
+      "loss": 0.111,
+      "step": 5435
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.815448880260912e-06,
+      "loss": 0.1185,
+      "step": 5436
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.812059235744654e-06,
+      "loss": 0.1189,
+      "step": 5437
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808669855234636e-06,
+      "loss": 0.1119,
+      "step": 5438
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.80528073913984e-06,
+      "loss": 0.1097,
+      "step": 5439
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.801891887869205e-06,
+      "loss": 0.1123,
+      "step": 5440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.798503301831647e-06,
+      "loss": 0.1006,
+      "step": 5441
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795114981436048e-06,
+      "loss": 0.1099,
+      "step": 5442
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.791726927091254e-06,
+      "loss": 0.1097,
+      "step": 5443
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.788339139206083e-06,
+      "loss": 0.0996,
+      "step": 5444
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784951618189322e-06,
+      "loss": 0.1139,
+      "step": 5445
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78156436444972e-06,
+      "loss": 0.1039,
+      "step": 5446
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778177378396006e-06,
+      "loss": 0.1049,
+      "step": 5447
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.1122,
+      "step": 5448
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.771404210980936e-06,
+      "loss": 0.1158,
+      "step": 5449
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.768018030436862e-06,
+      "loss": 0.1146,
+      "step": 5450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76463211921323e-06,
+      "loss": 0.1118,
+      "step": 5451
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.761246477718595e-06,
+      "loss": 0.1156,
+      "step": 5452
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.757861106361486e-06,
+      "loss": 0.1124,
+      "step": 5453
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.754476005550396e-06,
+      "loss": 0.1121,
+      "step": 5454
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.751091175693784e-06,
+      "loss": 0.1093,
+      "step": 5455
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.747706617200078e-06,
+      "loss": 0.1097,
+      "step": 5456
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.744322330477673e-06,
+      "loss": 0.1224,
+      "step": 5457
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.740938315934932e-06,
+      "loss": 0.1134,
+      "step": 5458
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.737554573980182e-06,
+      "loss": 0.0905,
+      "step": 5459
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73417110502173e-06,
+      "loss": 0.1057,
+      "step": 5460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.730787909467829e-06,
+      "loss": 0.104,
+      "step": 5461
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72740498772671e-06,
+      "loss": 0.1083,
+      "step": 5462
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724022340206575e-06,
+      "loss": 0.1062,
+      "step": 5463
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.720639967315588e-06,
+      "loss": 0.1162,
+      "step": 5464
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.717257869461879e-06,
+      "loss": 0.1191,
+      "step": 5465
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.713876047053548e-06,
+      "loss": 0.1124,
+      "step": 5466
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.710494500498662e-06,
+      "loss": 0.1073,
+      "step": 5467
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.707113230205247e-06,
+      "loss": 0.1087,
+      "step": 5468
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703732236581303e-06,
+      "loss": 0.1111,
+      "step": 5469
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700351520034798e-06,
+      "loss": 0.1088,
+      "step": 5470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.696971080973661e-06,
+      "loss": 0.1083,
+      "step": 5471
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693590919805796e-06,
+      "loss": 0.1278,
+      "step": 5472
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69021103693906e-06,
+      "loss": 0.1066,
+      "step": 5473
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.686831432781288e-06,
+      "loss": 0.1121,
+      "step": 5474
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.683452107740274e-06,
+      "loss": 0.111,
+      "step": 5475
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680073062223786e-06,
+      "loss": 0.1103,
+      "step": 5476
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676694296639552e-06,
+      "loss": 0.1077,
+      "step": 5477
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673315811395265e-06,
+      "loss": 0.1071,
+      "step": 5478
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.669937606898596e-06,
+      "loss": 0.1205,
+      "step": 5479
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.666559683557169e-06,
+      "loss": 0.1069,
+      "step": 5480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.663182041778574e-06,
+      "loss": 0.1104,
+      "step": 5481
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.659804681970378e-06,
+      "loss": 0.1204,
+      "step": 5482
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.656427604540104e-06,
+      "loss": 0.1077,
+      "step": 5483
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.653050809895245e-06,
+      "loss": 0.1097,
+      "step": 5484
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.649674298443261e-06,
+      "loss": 0.1094,
+      "step": 5485
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.646298070591578e-06,
+      "loss": 0.1043,
+      "step": 5486
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642922126747579e-06,
+      "loss": 0.119,
+      "step": 5487
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.639546467318629e-06,
+      "loss": 0.1001,
+      "step": 5488
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63617109271204e-06,
+      "loss": 0.1113,
+      "step": 5489
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.632796003335104e-06,
+      "loss": 0.1121,
+      "step": 5490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.629421199595073e-06,
+      "loss": 0.099,
+      "step": 5491
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.626046681899171e-06,
+      "loss": 0.112,
+      "step": 5492
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6226724506545715e-06,
+      "loss": 0.1058,
+      "step": 5493
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.619298506268427e-06,
+      "loss": 0.109,
+      "step": 5494
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6159248491478556e-06,
+      "loss": 0.1048,
+      "step": 5495
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.612551479699933e-06,
+      "loss": 0.1144,
+      "step": 5496
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.609178398331707e-06,
+      "loss": 0.1107,
+      "step": 5497
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.605805605450186e-06,
+      "loss": 0.1125,
+      "step": 5498
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.602433101462351e-06,
+      "loss": 0.1016,
+      "step": 5499
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599060886775134e-06,
+      "loss": 0.1045,
+      "step": 5500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.595688961795445e-06,
+      "loss": 0.1083,
+      "step": 5501
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5923173269301545e-06,
+      "loss": 0.1031,
+      "step": 5502
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5889459825861e-06,
+      "loss": 0.1082,
+      "step": 5503
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.58557492917008e-06,
+      "loss": 0.1161,
+      "step": 5504
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.582204167088864e-06,
+      "loss": 0.103,
+      "step": 5505
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.578833696749175e-06,
+      "loss": 0.1198,
+      "step": 5506
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.575463518557717e-06,
+      "loss": 0.1169,
+      "step": 5507
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.572093632921141e-06,
+      "loss": 0.1043,
+      "step": 5508
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56872404024608e-06,
+      "loss": 0.1171,
+      "step": 5509
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5653547409391175e-06,
+      "loss": 0.1101,
+      "step": 5510
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561985735406809e-06,
+      "loss": 0.1078,
+      "step": 5511
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5586170240556796e-06,
+      "loss": 0.1087,
+      "step": 5512
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555248607292205e-06,
+      "loss": 0.1116,
+      "step": 5513
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.551880485522831e-06,
+      "loss": 0.1082,
+      "step": 5514
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.548512659153973e-06,
+      "loss": 0.1119,
+      "step": 5515
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.1053,
+      "step": 5516
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.541777894243276e-06,
+      "loss": 0.1114,
+      "step": 5517
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538410956514086e-06,
+      "loss": 0.1051,
+      "step": 5518
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.535044315810699e-06,
+      "loss": 0.1086,
+      "step": 5519
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5316779725393495e-06,
+      "loss": 0.0996,
+      "step": 5520
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5283119271062376e-06,
+      "loss": 0.1081,
+      "step": 5521
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.524946179917527e-06,
+      "loss": 0.108,
+      "step": 5522
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.521580731379337e-06,
+      "loss": 0.0949,
+      "step": 5523
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.518215581897763e-06,
+      "loss": 0.101,
+      "step": 5524
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514850731878854e-06,
+      "loss": 0.1154,
+      "step": 5525
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.51148618172863e-06,
+      "loss": 0.114,
+      "step": 5526
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.508121931853067e-06,
+      "loss": 0.1175,
+      "step": 5527
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.504757982658117e-06,
+      "loss": 0.1157,
+      "step": 5528
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.501394334549681e-06,
+      "loss": 0.1155,
+      "step": 5529
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.498030987933635e-06,
+      "loss": 0.1016,
+      "step": 5530
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.494667943215819e-06,
+      "loss": 0.1158,
+      "step": 5531
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.491305200802026e-06,
+      "loss": 0.108,
+      "step": 5532
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.487942761098017e-06,
+      "loss": 0.1237,
+      "step": 5533
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4845806245095245e-06,
+      "loss": 0.1009,
+      "step": 5534
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.481218791442233e-06,
+      "loss": 0.1177,
+      "step": 5535
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.477857262301799e-06,
+      "loss": 0.1032,
+      "step": 5536
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.474496037493839e-06,
+      "loss": 0.1128,
+      "step": 5537
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471135117423935e-06,
+      "loss": 0.1068,
+      "step": 5538
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46777450249762e-06,
+      "loss": 0.108,
+      "step": 5539
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.464414193120411e-06,
+      "loss": 0.0996,
+      "step": 5540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46105418969777e-06,
+      "loss": 0.106,
+      "step": 5541
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.457694492635132e-06,
+      "loss": 0.1117,
+      "step": 5542
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.454335102337895e-06,
+      "loss": 0.1089,
+      "step": 5543
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.450976019211416e-06,
+      "loss": 0.1158,
+      "step": 5544
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.447617243661011e-06,
+      "loss": 0.1222,
+      "step": 5545
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4442587760919696e-06,
+      "loss": 0.1161,
+      "step": 5546
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.440900616909539e-06,
+      "loss": 0.111,
+      "step": 5547
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4375427665189244e-06,
+      "loss": 0.1043,
+      "step": 5548
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.434185225325305e-06,
+      "loss": 0.1076,
+      "step": 5549
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.430827993733809e-06,
+      "loss": 0.1116,
+      "step": 5550
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.427471072149542e-06,
+      "loss": 0.1024,
+      "step": 5551
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.424114460977555e-06,
+      "loss": 0.1108,
+      "step": 5552
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.420758160622878e-06,
+      "loss": 0.098,
+      "step": 5553
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.417402171490495e-06,
+      "loss": 0.1159,
+      "step": 5554
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.414046493985353e-06,
+      "loss": 0.1098,
+      "step": 5555
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.410691128512362e-06,
+      "loss": 0.1139,
+      "step": 5556
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4073360754763975e-06,
+      "loss": 0.1036,
+      "step": 5557
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40398133528229e-06,
+      "loss": 0.1174,
+      "step": 5558
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40062690833484e-06,
+      "loss": 0.1076,
+      "step": 5559
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.397272795038806e-06,
+      "loss": 0.1055,
+      "step": 5560
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.393918995798908e-06,
+      "loss": 0.114,
+      "step": 5561
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.390565511019834e-06,
+      "loss": 0.1116,
+      "step": 5562
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.387212341106224e-06,
+      "loss": 0.1137,
+      "step": 5563
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383859486462693e-06,
+      "loss": 0.1033,
+      "step": 5564
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.380506947493805e-06,
+      "loss": 0.1108,
+      "step": 5565
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.377154724604091e-06,
+      "loss": 0.1105,
+      "step": 5566
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.373802818198046e-06,
+      "loss": 0.1169,
+      "step": 5567
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.370451228680129e-06,
+      "loss": 0.1163,
+      "step": 5568
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.36709995645475e-06,
+      "loss": 0.1146,
+      "step": 5569
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.363749001926298e-06,
+      "loss": 0.1115,
+      "step": 5570
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.360398365499104e-06,
+      "loss": 0.1138,
+      "step": 5571
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.357048047577471e-06,
+      "loss": 0.1103,
+      "step": 5572
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.353698048565665e-06,
+      "loss": 0.1109,
+      "step": 5573
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.350348368867912e-06,
+      "loss": 0.1106,
+      "step": 5574
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.346999008888397e-06,
+      "loss": 0.1062,
+      "step": 5575
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343649969031268e-06,
+      "loss": 0.1087,
+      "step": 5576
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.340301249700639e-06,
+      "loss": 0.1164,
+      "step": 5577
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.336952851300574e-06,
+      "loss": 0.101,
+      "step": 5578
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.333604774235106e-06,
+      "loss": 0.1019,
+      "step": 5579
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.330257018908231e-06,
+      "loss": 0.1143,
+      "step": 5580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.326909585723901e-06,
+      "loss": 0.1089,
+      "step": 5581
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.323562475086033e-06,
+      "loss": 0.1068,
+      "step": 5582
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3202156873985075e-06,
+      "loss": 0.1194,
+      "step": 5583
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.316869223065156e-06,
+      "loss": 0.1105,
+      "step": 5584
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.313523082489777e-06,
+      "loss": 0.1134,
+      "step": 5585
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310177266076134e-06,
+      "loss": 0.1096,
+      "step": 5586
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.306831774227944e-06,
+      "loss": 0.1092,
+      "step": 5587
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3034866073488885e-06,
+      "loss": 0.1198,
+      "step": 5588
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.300141765842614e-06,
+      "loss": 0.1227,
+      "step": 5589
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.296797250112724e-06,
+      "loss": 0.1135,
+      "step": 5590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.293453060562772e-06,
+      "loss": 0.1016,
+      "step": 5591
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.290109197596289e-06,
+      "loss": 0.1122,
+      "step": 5592
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.286765661616761e-06,
+      "loss": 0.1176,
+      "step": 5593
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.283422453027631e-06,
+      "loss": 0.1129,
+      "step": 5594
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2800795722323055e-06,
+      "loss": 0.1044,
+      "step": 5595
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.276737019634155e-06,
+      "loss": 0.1263,
+      "step": 5596
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.273394795636495e-06,
+      "loss": 0.1087,
+      "step": 5597
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.270052900642619e-06,
+      "loss": 0.1131,
+      "step": 5598
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2667113350557775e-06,
+      "loss": 0.1124,
+      "step": 5599
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 0.1013,
+      "step": 5600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.260029193715975e-06,
+      "loss": 0.1092,
+      "step": 5601
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.256688618769311e-06,
+      "loss": 0.1034,
+      "step": 5602
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.253348374842273e-06,
+      "loss": 0.1169,
+      "step": 5603
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2500084623379005e-06,
+      "loss": 0.118,
+      "step": 5604
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2466688816592115e-06,
+      "loss": 0.1044,
+      "step": 5605
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.243329633209164e-06,
+      "loss": 0.1138,
+      "step": 5606
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.239990717390695e-06,
+      "loss": 0.1045,
+      "step": 5607
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.236652134606684e-06,
+      "loss": 0.1075,
+      "step": 5608
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2333138852599884e-06,
+      "loss": 0.1168,
+      "step": 5609
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.229975969753406e-06,
+      "loss": 0.1088,
+      "step": 5610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.226638388489708e-06,
+      "loss": 0.1126,
+      "step": 5611
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.223301141871621e-06,
+      "loss": 0.1063,
+      "step": 5612
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.219964230301831e-06,
+      "loss": 0.1133,
+      "step": 5613
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2166276541829864e-06,
+      "loss": 0.11,
+      "step": 5614
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.213291413917689e-06,
+      "loss": 0.1049,
+      "step": 5615
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209955509908508e-06,
+      "loss": 0.1187,
+      "step": 5616
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2066199425579655e-06,
+      "loss": 0.103,
+      "step": 5617
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.203284712268541e-06,
+      "loss": 0.1037,
+      "step": 5618
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199949819442682e-06,
+      "loss": 0.1054,
+      "step": 5619
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.196615264482791e-06,
+      "loss": 0.1104,
+      "step": 5620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.193281047791229e-06,
+      "loss": 0.1051,
+      "step": 5621
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.18994716977032e-06,
+      "loss": 0.1128,
+      "step": 5622
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1866136308223365e-06,
+      "loss": 0.1044,
+      "step": 5623
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183280431349524e-06,
+      "loss": 0.1247,
+      "step": 5624
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.179947571754076e-06,
+      "loss": 0.11,
+      "step": 5625
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1766150524381515e-06,
+      "loss": 0.1198,
+      "step": 5626
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173282873803866e-06,
+      "loss": 0.107,
+      "step": 5627
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.169951036253296e-06,
+      "loss": 0.1255,
+      "step": 5628
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.166619540188478e-06,
+      "loss": 0.1071,
+      "step": 5629
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163288386011397e-06,
+      "loss": 0.1107,
+      "step": 5630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.159957574124007e-06,
+      "loss": 0.1131,
+      "step": 5631
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.15662710492822e-06,
+      "loss": 0.1037,
+      "step": 5632
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153296978825903e-06,
+      "loss": 0.1113,
+      "step": 5633
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.149967196218882e-06,
+      "loss": 0.1055,
+      "step": 5634
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.14663775750895e-06,
+      "loss": 0.1178,
+      "step": 5635
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143308663097841e-06,
+      "loss": 0.114,
+      "step": 5636
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1399799133872615e-06,
+      "loss": 0.1183,
+      "step": 5637
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.136651508778876e-06,
+      "loss": 0.1087,
+      "step": 5638
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133323449674299e-06,
+      "loss": 0.1107,
+      "step": 5639
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1299957364751085e-06,
+      "loss": 0.1037,
+      "step": 5640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.126668369582849e-06,
+      "loss": 0.1164,
+      "step": 5641
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.123341349399006e-06,
+      "loss": 0.1082,
+      "step": 5642
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.120014676325031e-06,
+      "loss": 0.1113,
+      "step": 5643
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116688350762338e-06,
+      "loss": 0.1045,
+      "step": 5644
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.113362373112298e-06,
+      "loss": 0.113,
+      "step": 5645
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.110036743776234e-06,
+      "loss": 0.1336,
+      "step": 5646
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.106711463155431e-06,
+      "loss": 0.1153,
+      "step": 5647
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.103386531651137e-06,
+      "loss": 0.0945,
+      "step": 5648
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100061949664542e-06,
+      "loss": 0.1013,
+      "step": 5649
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.096737717596809e-06,
+      "loss": 0.111,
+      "step": 5650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0934138358490565e-06,
+      "loss": 0.1063,
+      "step": 5651
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.090090304822356e-06,
+      "loss": 0.1066,
+      "step": 5652
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.086767124917739e-06,
+      "loss": 0.1177,
+      "step": 5653
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.083444296536199e-06,
+      "loss": 0.1165,
+      "step": 5654
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.080121820078673e-06,
+      "loss": 0.1012,
+      "step": 5655
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0767996959460714e-06,
+      "loss": 0.1243,
+      "step": 5656
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.073477924539256e-06,
+      "loss": 0.1162,
+      "step": 5657
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.070156506259044e-06,
+      "loss": 0.1029,
+      "step": 5658
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.066835441506212e-06,
+      "loss": 0.1245,
+      "step": 5659
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0635147306814965e-06,
+      "loss": 0.1082,
+      "step": 5660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.060194374185589e-06,
+      "loss": 0.1164,
+      "step": 5661
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056874372419131e-06,
+      "loss": 0.111,
+      "step": 5662
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.053554725782734e-06,
+      "loss": 0.1109,
+      "step": 5663
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0502354346769575e-06,
+      "loss": 0.1117,
+      "step": 5664
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046916499502323e-06,
+      "loss": 0.1188,
+      "step": 5665
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.043597920659308e-06,
+      "loss": 0.1246,
+      "step": 5666
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0402796985483505e-06,
+      "loss": 0.1068,
+      "step": 5667
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036961833569831e-06,
+      "loss": 0.1003,
+      "step": 5668
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.033644326124104e-06,
+      "loss": 0.1269,
+      "step": 5669
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.030327176611471e-06,
+      "loss": 0.1144,
+      "step": 5670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.027010385432196e-06,
+      "loss": 0.1151,
+      "step": 5671
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.023693952986496e-06,
+      "loss": 0.1185,
+      "step": 5672
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.020377879674546e-06,
+      "loss": 0.1115,
+      "step": 5673
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.01706216589648e-06,
+      "loss": 0.1041,
+      "step": 5674
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.013746812052381e-06,
+      "loss": 0.0946,
+      "step": 5675
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.010431818542298e-06,
+      "loss": 0.116,
+      "step": 5676
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.007117185766228e-06,
+      "loss": 0.1052,
+      "step": 5677
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0038029141241335e-06,
+      "loss": 0.1112,
+      "step": 5678
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0004890040159256e-06,
+      "loss": 0.0988,
+      "step": 5679
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.997175455841478e-06,
+      "loss": 0.0978,
+      "step": 5680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.993862270000613e-06,
+      "loss": 0.1,
+      "step": 5681
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.990549446893116e-06,
+      "loss": 0.0989,
+      "step": 5682
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.987236986918725e-06,
+      "loss": 0.1046,
+      "step": 5683
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983924890477138e-06,
+      "loss": 0.1011,
+      "step": 5684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9806131579680035e-06,
+      "loss": 0.1111,
+      "step": 5685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.977301789790931e-06,
+      "loss": 0.1099,
+      "step": 5686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973990786345489e-06,
+      "loss": 0.1024,
+      "step": 5687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.970680148031191e-06,
+      "loss": 0.1098,
+      "step": 5688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.967369875247512e-06,
+      "loss": 0.1138,
+      "step": 5689
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9640599683938835e-06,
+      "loss": 0.1057,
+      "step": 5690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.960750427869701e-06,
+      "loss": 0.1105,
+      "step": 5691
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.957441254074297e-06,
+      "loss": 0.1137,
+      "step": 5692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.95413244740698e-06,
+      "loss": 0.115,
+      "step": 5693
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9508240082669995e-06,
+      "loss": 0.1059,
+      "step": 5694
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9475159370535635e-06,
+      "loss": 0.0983,
+      "step": 5695
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.944208234165843e-06,
+      "loss": 0.1058,
+      "step": 5696
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940900900002957e-06,
+      "loss": 0.1065,
+      "step": 5697
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9375939349639825e-06,
+      "loss": 0.1165,
+      "step": 5698
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.934287339447954e-06,
+      "loss": 0.1271,
+      "step": 5699
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.930981113853861e-06,
+      "loss": 0.1103,
+      "step": 5700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.927675258580637e-06,
+      "loss": 0.1017,
+      "step": 5701
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.924369774027189e-06,
+      "loss": 0.0996,
+      "step": 5702
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9210646605923715e-06,
+      "loss": 0.11,
+      "step": 5703
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.917759918674988e-06,
+      "loss": 0.1074,
+      "step": 5704
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9144555486738085e-06,
+      "loss": 0.1158,
+      "step": 5705
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.91115155098755e-06,
+      "loss": 0.1078,
+      "step": 5706
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9078479260148855e-06,
+      "loss": 0.1057,
+      "step": 5707
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.904544674154443e-06,
+      "loss": 0.1203,
+      "step": 5708
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.901241795804813e-06,
+      "loss": 0.1027,
+      "step": 5709
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897939291364528e-06,
+      "loss": 0.1221,
+      "step": 5710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.894637161232086e-06,
+      "loss": 0.1014,
+      "step": 5711
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8913354058059374e-06,
+      "loss": 0.1089,
+      "step": 5712
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.888034025484487e-06,
+      "loss": 0.1082,
+      "step": 5713
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.884733020666086e-06,
+      "loss": 0.0993,
+      "step": 5714
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.881432391749054e-06,
+      "loss": 0.1022,
+      "step": 5715
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878132139131657e-06,
+      "loss": 0.119,
+      "step": 5716
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874832263212117e-06,
+      "loss": 0.1045,
+      "step": 5717
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871532764388613e-06,
+      "loss": 0.1105,
+      "step": 5718
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.868233643059277e-06,
+      "loss": 0.1043,
+      "step": 5719
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.864934899622191e-06,
+      "loss": 0.1141,
+      "step": 5720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.861636534475396e-06,
+      "loss": 0.1011,
+      "step": 5721
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.858338548016892e-06,
+      "loss": 0.1038,
+      "step": 5722
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.855040940644622e-06,
+      "loss": 0.1134,
+      "step": 5723
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851743712756494e-06,
+      "loss": 0.1024,
+      "step": 5724
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.848446864750362e-06,
+      "loss": 0.1113,
+      "step": 5725
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.845150397024043e-06,
+      "loss": 0.0957,
+      "step": 5726
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.841854309975296e-06,
+      "loss": 0.106,
+      "step": 5727
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.838558604001843e-06,
+      "loss": 0.1014,
+      "step": 5728
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.83526327950136e-06,
+      "loss": 0.105,
+      "step": 5729
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.831968336871475e-06,
+      "loss": 0.1143,
+      "step": 5730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8286737765097666e-06,
+      "loss": 0.1153,
+      "step": 5731
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825379598813776e-06,
+      "loss": 0.1064,
+      "step": 5732
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.822085804180985e-06,
+      "loss": 0.1132,
+      "step": 5733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.818792393008844e-06,
+      "loss": 0.0982,
+      "step": 5734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.815499365694744e-06,
+      "loss": 0.1161,
+      "step": 5735
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8122067226360415e-06,
+      "loss": 0.1093,
+      "step": 5736
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808914464230034e-06,
+      "loss": 0.1155,
+      "step": 5737
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.805622590873985e-06,
+      "loss": 0.1096,
+      "step": 5738
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.802331102965108e-06,
+      "loss": 0.1136,
+      "step": 5739
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.799040000900563e-06,
+      "loss": 0.1091,
+      "step": 5740
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.795749285077466e-06,
+      "loss": 0.1037,
+      "step": 5741
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.792458955892895e-06,
+      "loss": 0.0971,
+      "step": 5742
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.789169013743872e-06,
+      "loss": 0.1151,
+      "step": 5743
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.785879459027376e-06,
+      "loss": 0.1192,
+      "step": 5744
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.782590292140341e-06,
+      "loss": 0.1123,
+      "step": 5745
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.779301513479648e-06,
+      "loss": 0.1016,
+      "step": 5746
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7760131234421345e-06,
+      "loss": 0.1076,
+      "step": 5747
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772725122424594e-06,
+      "loss": 0.1018,
+      "step": 5748
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7694375108237716e-06,
+      "loss": 0.1041,
+      "step": 5749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7661502890363615e-06,
+      "loss": 0.1081,
+      "step": 5750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.762863457459019e-06,
+      "loss": 0.113,
+      "step": 5751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759577016488343e-06,
+      "loss": 0.1202,
+      "step": 5752
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7562909665208885e-06,
+      "loss": 0.1121,
+      "step": 5753
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7530053079531664e-06,
+      "loss": 0.1059,
+      "step": 5754
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749720041181638e-06,
+      "loss": 0.1068,
+      "step": 5755
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746435166602716e-06,
+      "loss": 0.1254,
+      "step": 5756
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.743150684612769e-06,
+      "loss": 0.1153,
+      "step": 5757
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.739866595608121e-06,
+      "loss": 0.1127,
+      "step": 5758
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.736582899985038e-06,
+      "loss": 0.1021,
+      "step": 5759
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.733299598139742e-06,
+      "loss": 0.1193,
+      "step": 5760
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.730016690468417e-06,
+      "loss": 0.1056,
+      "step": 5761
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.726734177367189e-06,
+      "loss": 0.1198,
+      "step": 5762
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.72345205923214e-06,
+      "loss": 0.112,
+      "step": 5763
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.720170336459308e-06,
+      "loss": 0.1039,
+      "step": 5764
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.716889009444678e-06,
+      "loss": 0.1016,
+      "step": 5765
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.713608078584184e-06,
+      "loss": 0.1097,
+      "step": 5766
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.710327544273724e-06,
+      "loss": 0.0952,
+      "step": 5767
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.707047406909135e-06,
+      "loss": 0.1039,
+      "step": 5768
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.703767666886215e-06,
+      "loss": 0.1038,
+      "step": 5769
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.700488324600715e-06,
+      "loss": 0.1133,
+      "step": 5770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.1199,
+      "step": 5771
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693930834824713e-06,
+      "loss": 0.1071,
+      "step": 5772
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6906526881254655e-06,
+      "loss": 0.1082,
+      "step": 5773
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687374940746145e-06,
+      "loss": 0.106,
+      "step": 5774
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.684097593082256e-06,
+      "loss": 0.1089,
+      "step": 5775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.68082064552926e-06,
+      "loss": 0.1106,
+      "step": 5776
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6775440984825645e-06,
+      "loss": 0.1096,
+      "step": 5777
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.674267952337538e-06,
+      "loss": 0.1257,
+      "step": 5778
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.670992207489484e-06,
+      "loss": 0.1096,
+      "step": 5779
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667716864333675e-06,
+      "loss": 0.1138,
+      "step": 5780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.664441923265326e-06,
+      "loss": 0.1088,
+      "step": 5781
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.661167384679605e-06,
+      "loss": 0.1061,
+      "step": 5782
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657893248971631e-06,
+      "loss": 0.1137,
+      "step": 5783
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.65461951653648e-06,
+      "loss": 0.1119,
+      "step": 5784
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.651346187769168e-06,
+      "loss": 0.1053,
+      "step": 5785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.648073263064672e-06,
+      "loss": 0.1019,
+      "step": 5786
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.644800742817914e-06,
+      "loss": 0.1025,
+      "step": 5787
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6415286274237744e-06,
+      "loss": 0.1106,
+      "step": 5788
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6382569172770805e-06,
+      "loss": 0.1155,
+      "step": 5789
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.634985612772611e-06,
+      "loss": 0.1075,
+      "step": 5790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.631714714305089e-06,
+      "loss": 0.1152,
+      "step": 5791
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.628444222269203e-06,
+      "loss": 0.1107,
+      "step": 5792
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.625174137059578e-06,
+      "loss": 0.1107,
+      "step": 5793
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.621904459070798e-06,
+      "loss": 0.1063,
+      "step": 5794
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.618635188697402e-06,
+      "loss": 0.1111,
+      "step": 5795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.615366326333868e-06,
+      "loss": 0.1034,
+      "step": 5796
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6120978723746356e-06,
+      "loss": 0.1137,
+      "step": 5797
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.608829827214085e-06,
+      "loss": 0.1173,
+      "step": 5798
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.605562191246554e-06,
+      "loss": 0.1073,
+      "step": 5799
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6022949648663294e-06,
+      "loss": 0.1007,
+      "step": 5800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.599028148467652e-06,
+      "loss": 0.11,
+      "step": 5801
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.595761742444705e-06,
+      "loss": 0.1074,
+      "step": 5802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.592495747191634e-06,
+      "loss": 0.1033,
+      "step": 5803
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.589230163102518e-06,
+      "loss": 0.1187,
+      "step": 5804
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585964990571403e-06,
+      "loss": 0.1071,
+      "step": 5805
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.582700229992276e-06,
+      "loss": 0.1017,
+      "step": 5806
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.57943588175908e-06,
+      "loss": 0.1098,
+      "step": 5807
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.576171946265699e-06,
+      "loss": 0.1142,
+      "step": 5808
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.57290842390598e-06,
+      "loss": 0.1049,
+      "step": 5809
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.569645315073713e-06,
+      "loss": 0.1122,
+      "step": 5810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.566382620162635e-06,
+      "loss": 0.1039,
+      "step": 5811
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.563120339566436e-06,
+      "loss": 0.1093,
+      "step": 5812
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.55985847367876e-06,
+      "loss": 0.122,
+      "step": 5813
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556597022893194e-06,
+      "loss": 0.1073,
+      "step": 5814
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.553335987603282e-06,
+      "loss": 0.1049,
+      "step": 5815
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.550075368202518e-06,
+      "loss": 0.1119,
+      "step": 5816
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5468151650843336e-06,
+      "loss": 0.1132,
+      "step": 5817
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.543555378642121e-06,
+      "loss": 0.1198,
+      "step": 5818
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.540296009269223e-06,
+      "loss": 0.0989,
+      "step": 5819
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537037057358928e-06,
+      "loss": 0.1166,
+      "step": 5820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533778523304473e-06,
+      "loss": 0.1061,
+      "step": 5821
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.530520407499049e-06,
+      "loss": 0.1048,
+      "step": 5822
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.527262710335797e-06,
+      "loss": 0.1066,
+      "step": 5823
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524005432207795e-06,
+      "loss": 0.1102,
+      "step": 5824
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.520748573508087e-06,
+      "loss": 0.1101,
+      "step": 5825
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.517492134629659e-06,
+      "loss": 0.1074,
+      "step": 5826
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.514236115965446e-06,
+      "loss": 0.1127,
+      "step": 5827
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.510980517908334e-06,
+      "loss": 0.1077,
+      "step": 5828
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507725340851158e-06,
+      "loss": 0.1136,
+      "step": 5829
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.504470585186696e-06,
+      "loss": 0.1149,
+      "step": 5830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.501216251307685e-06,
+      "loss": 0.1108,
+      "step": 5831
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4979623396068095e-06,
+      "loss": 0.11,
+      "step": 5832
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.494708850476692e-06,
+      "loss": 0.1317,
+      "step": 5833
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.491455784309923e-06,
+      "loss": 0.1073,
+      "step": 5834
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.488203141499021e-06,
+      "loss": 0.1074,
+      "step": 5835
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.484950922436475e-06,
+      "loss": 0.1029,
+      "step": 5836
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.481699127514699e-06,
+      "loss": 0.1043,
+      "step": 5837
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4784477571260775e-06,
+      "loss": 0.0992,
+      "step": 5838
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.475196811662929e-06,
+      "loss": 0.0941,
+      "step": 5839
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4719462915175305e-06,
+      "loss": 0.1226,
+      "step": 5840
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468696197082103e-06,
+      "loss": 0.1059,
+      "step": 5841
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.465446528748818e-06,
+      "loss": 0.1038,
+      "step": 5842
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462197286909789e-06,
+      "loss": 0.1126,
+      "step": 5843
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.458948471957088e-06,
+      "loss": 0.1049,
+      "step": 5844
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4557000842827275e-06,
+      "loss": 0.1091,
+      "step": 5845
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.452452124278674e-06,
+      "loss": 0.0989,
+      "step": 5846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.449204592336842e-06,
+      "loss": 0.0995,
+      "step": 5847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445957488849088e-06,
+      "loss": 0.1069,
+      "step": 5848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.442710814207228e-06,
+      "loss": 0.1101,
+      "step": 5849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.439464568803014e-06,
+      "loss": 0.1032,
+      "step": 5850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4362187530281495e-06,
+      "loss": 0.1075,
+      "step": 5851
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.432973367274292e-06,
+      "loss": 0.1018,
+      "step": 5852
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.429728411933047e-06,
+      "loss": 0.1108,
+      "step": 5853
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.426483887395958e-06,
+      "loss": 0.1135,
+      "step": 5854
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.42323979405453e-06,
+      "loss": 0.1074,
+      "step": 5855
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.419996132300203e-06,
+      "loss": 0.1067,
+      "step": 5856
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.416752902524375e-06,
+      "loss": 0.1049,
+      "step": 5857
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.413510105118383e-06,
+      "loss": 0.1166,
+      "step": 5858
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.410267740473523e-06,
+      "loss": 0.1118,
+      "step": 5859
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.407025808981028e-06,
+      "loss": 0.1027,
+      "step": 5860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.403784311032084e-06,
+      "loss": 0.1001,
+      "step": 5861
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.40054324701783e-06,
+      "loss": 0.1249,
+      "step": 5862
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3973026173293385e-06,
+      "loss": 0.112,
+      "step": 5863
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.394062422357638e-06,
+      "loss": 0.1236,
+      "step": 5864
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.390822662493709e-06,
+      "loss": 0.1065,
+      "step": 5865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.387583338128471e-06,
+      "loss": 0.1019,
+      "step": 5866
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384344449652795e-06,
+      "loss": 0.1043,
+      "step": 5867
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3811059974575065e-06,
+      "loss": 0.1128,
+      "step": 5868
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.377867981933361e-06,
+      "loss": 0.1016,
+      "step": 5869
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.374630403471072e-06,
+      "loss": 0.1027,
+      "step": 5870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.371393262461303e-06,
+      "loss": 0.1108,
+      "step": 5871
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.368156559294662e-06,
+      "loss": 0.1132,
+      "step": 5872
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.364920294361701e-06,
+      "loss": 0.1001,
+      "step": 5873
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361684468052925e-06,
+      "loss": 0.1083,
+      "step": 5874
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35844908075878e-06,
+      "loss": 0.1066,
+      "step": 5875
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.355214132869658e-06,
+      "loss": 0.1212,
+      "step": 5876
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.351979624775906e-06,
+      "loss": 0.1186,
+      "step": 5877
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348745556867814e-06,
+      "loss": 0.1202,
+      "step": 5878
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3455119295356145e-06,
+      "loss": 0.107,
+      "step": 5879
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3422787431694954e-06,
+      "loss": 0.0929,
+      "step": 5880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.339045998159588e-06,
+      "loss": 0.1071,
+      "step": 5881
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33581369489596e-06,
+      "loss": 0.101,
+      "step": 5882
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332581833768641e-06,
+      "loss": 0.1099,
+      "step": 5883
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3293504151676026e-06,
+      "loss": 0.1119,
+      "step": 5884
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3261194394827565e-06,
+      "loss": 0.1027,
+      "step": 5885
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3228889071039705e-06,
+      "loss": 0.1017,
+      "step": 5886
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.319658818421055e-06,
+      "loss": 0.1044,
+      "step": 5887
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.316429173823765e-06,
+      "loss": 0.1007,
+      "step": 5888
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.313199973701798e-06,
+      "loss": 0.1097,
+      "step": 5889
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.30997121844481e-06,
+      "loss": 0.1116,
+      "step": 5890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.306742908442393e-06,
+      "loss": 0.1106,
+      "step": 5891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.303515044084087e-06,
+      "loss": 0.1065,
+      "step": 5892
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3002876257593845e-06,
+      "loss": 0.1163,
+      "step": 5893
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2970606538577205e-06,
+      "loss": 0.1139,
+      "step": 5894
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.293834128768467e-06,
+      "loss": 0.109,
+      "step": 5895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.290608050880957e-06,
+      "loss": 0.1077,
+      "step": 5896
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.287382420584458e-06,
+      "loss": 0.1094,
+      "step": 5897
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.284157238268193e-06,
+      "loss": 0.1029,
+      "step": 5898
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.280932504321324e-06,
+      "loss": 0.1025,
+      "step": 5899
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.277708219132961e-06,
+      "loss": 0.1051,
+      "step": 5900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274484383092165e-06,
+      "loss": 0.1041,
+      "step": 5901
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271260996587928e-06,
+      "loss": 0.1084,
+      "step": 5902
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268038060009205e-06,
+      "loss": 0.1181,
+      "step": 5903
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.264815573744884e-06,
+      "loss": 0.108,
+      "step": 5904
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.26159353818381e-06,
+      "loss": 0.1034,
+      "step": 5905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258371953714762e-06,
+      "loss": 0.111,
+      "step": 5906
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.255150820726477e-06,
+      "loss": 0.1136,
+      "step": 5907
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.251930139607621e-06,
+      "loss": 0.1091,
+      "step": 5908
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.248709910746823e-06,
+      "loss": 0.0997,
+      "step": 5909
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245490134532644e-06,
+      "loss": 0.1121,
+      "step": 5910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242270811353601e-06,
+      "loss": 0.0978,
+      "step": 5911
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.239051941598147e-06,
+      "loss": 0.11,
+      "step": 5912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.235833525654686e-06,
+      "loss": 0.1132,
+      "step": 5913
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.23261556391157e-06,
+      "loss": 0.1131,
+      "step": 5914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.229398056757085e-06,
+      "loss": 0.1122,
+      "step": 5915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.226181004579472e-06,
+      "loss": 0.0997,
+      "step": 5916
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2229644077669135e-06,
+      "loss": 0.1119,
+      "step": 5917
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2197482667075405e-06,
+      "loss": 0.1055,
+      "step": 5918
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.216532581789423e-06,
+      "loss": 0.1038,
+      "step": 5919
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213317353400583e-06,
+      "loss": 0.1103,
+      "step": 5920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.210102581928981e-06,
+      "loss": 0.1158,
+      "step": 5921
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.206888267762522e-06,
+      "loss": 0.1064,
+      "step": 5922
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.203674411289063e-06,
+      "loss": 0.1102,
+      "step": 5923
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.200461012896401e-06,
+      "loss": 0.1103,
+      "step": 5924
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.197248072972277e-06,
+      "loss": 0.1131,
+      "step": 5925
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194035591904382e-06,
+      "loss": 0.1068,
+      "step": 5926
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.190823570080344e-06,
+      "loss": 0.1099,
+      "step": 5927
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.187612007887738e-06,
+      "loss": 0.1134,
+      "step": 5928
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.184400905714085e-06,
+      "loss": 0.1048,
+      "step": 5929
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.181190263946856e-06,
+      "loss": 0.103,
+      "step": 5930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.177980082973453e-06,
+      "loss": 0.1078,
+      "step": 5931
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.174770363181235e-06,
+      "loss": 0.1186,
+      "step": 5932
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1715611049574994e-06,
+      "loss": 0.1032,
+      "step": 5933
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.168352308689489e-06,
+      "loss": 0.1156,
+      "step": 5934
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.165143974764389e-06,
+      "loss": 0.1105,
+      "step": 5935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.161936103569333e-06,
+      "loss": 0.1063,
+      "step": 5936
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.158728695491394e-06,
+      "loss": 0.0983,
+      "step": 5937
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155521750917593e-06,
+      "loss": 0.1048,
+      "step": 5938
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.152315270234897e-06,
+      "loss": 0.1051,
+      "step": 5939
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149109253830208e-06,
+      "loss": 0.1026,
+      "step": 5940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1459037020903775e-06,
+      "loss": 0.107,
+      "step": 5941
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.142698615402205e-06,
+      "loss": 0.1195,
+      "step": 5942
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.139493994152428e-06,
+      "loss": 0.102,
+      "step": 5943
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.136289838727727e-06,
+      "loss": 0.1059,
+      "step": 5944
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.133086149514735e-06,
+      "loss": 0.1069,
+      "step": 5945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129882926900024e-06,
+      "loss": 0.114,
+      "step": 5946
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126680171270096e-06,
+      "loss": 0.1129,
+      "step": 5947
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123477883011422e-06,
+      "loss": 0.109,
+      "step": 5948
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.120276062510395e-06,
+      "loss": 0.115,
+      "step": 5949
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.117074710153366e-06,
+      "loss": 0.1036,
+      "step": 5950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.113873826326623e-06,
+      "loss": 0.1169,
+      "step": 5951
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.110673411416399e-06,
+      "loss": 0.0979,
+      "step": 5952
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.107473465808864e-06,
+      "loss": 0.0941,
+      "step": 5953
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1042739898901416e-06,
+      "loss": 0.1026,
+      "step": 5954
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.101074984046296e-06,
+      "loss": 0.1012,
+      "step": 5955
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0978764486633265e-06,
+      "loss": 0.1117,
+      "step": 5956
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.09467838412719e-06,
+      "loss": 0.1056,
+      "step": 5957
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.091480790823772e-06,
+      "loss": 0.1054,
+      "step": 5958
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.088283669138913e-06,
+      "loss": 0.1067,
+      "step": 5959
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0850870194583854e-06,
+      "loss": 0.112,
+      "step": 5960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.081890842167916e-06,
+      "loss": 0.1025,
+      "step": 5961
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.078695137653164e-06,
+      "loss": 0.0977,
+      "step": 5962
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.075499906299742e-06,
+      "loss": 0.1131,
+      "step": 5963
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.072305148493195e-06,
+      "loss": 0.0966,
+      "step": 5964
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.069110864619023e-06,
+      "loss": 0.1113,
+      "step": 5965
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065917055062654e-06,
+      "loss": 0.1123,
+      "step": 5966
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.062723720209472e-06,
+      "loss": 0.1058,
+      "step": 5967
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0595308604447955e-06,
+      "loss": 0.1088,
+      "step": 5968
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.056338476153889e-06,
+      "loss": 0.1127,
+      "step": 5969
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0531465677219615e-06,
+      "loss": 0.1102,
+      "step": 5970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.049955135534159e-06,
+      "loss": 0.1043,
+      "step": 5971
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0467641799755796e-06,
+      "loss": 0.11,
+      "step": 5972
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.043573701431249e-06,
+      "loss": 0.0994,
+      "step": 5973
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.040383700286148e-06,
+      "loss": 0.1136,
+      "step": 5974
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037194176925194e-06,
+      "loss": 0.1083,
+      "step": 5975
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.034005131733253e-06,
+      "loss": 0.1175,
+      "step": 5976
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.030816565095123e-06,
+      "loss": 0.1062,
+      "step": 5977
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027628477395557e-06,
+      "loss": 0.102,
+      "step": 5978
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.024440869019237e-06,
+      "loss": 0.1221,
+      "step": 5979
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.021253740350793e-06,
+      "loss": 0.1038,
+      "step": 5980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0180670917748e-06,
+      "loss": 0.11,
+      "step": 5981
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014880923675774e-06,
+      "loss": 0.1039,
+      "step": 5982
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.01169523643817e-06,
+      "loss": 0.1116,
+      "step": 5983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.008510030446385e-06,
+      "loss": 0.1042,
+      "step": 5984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.005325306084766e-06,
+      "loss": 0.1274,
+      "step": 5985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.002141063737588e-06,
+      "loss": 0.1009,
+      "step": 5986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.998957303789075e-06,
+      "loss": 0.1021,
+      "step": 5987
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9957740266233975e-06,
+      "loss": 0.1179,
+      "step": 5988
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99259123262466e-06,
+      "loss": 0.1164,
+      "step": 5989
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989408922176916e-06,
+      "loss": 0.1157,
+      "step": 5990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.986227095664155e-06,
+      "loss": 0.1101,
+      "step": 5991
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.1032,
+      "step": 5992
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979864895979248e-06,
+      "loss": 0.1105,
+      "step": 5993
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.976684523574795e-06,
+      "loss": 0.0995,
+      "step": 5994
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.973504636640702e-06,
+      "loss": 0.1205,
+      "step": 5995
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.97032523556067e-06,
+      "loss": 0.1008,
+      "step": 5996
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96714632071834e-06,
+      "loss": 0.1032,
+      "step": 5997
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9639678924972956e-06,
+      "loss": 0.108,
+      "step": 5998
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.960789951281052e-06,
+      "loss": 0.1083,
+      "step": 5999
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.957612497453078e-06,
+      "loss": 0.1065,
+      "step": 6000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9544355313967785e-06,
+      "loss": 0.1006,
+      "step": 6001
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.951259053495497e-06,
+      "loss": 0.1135,
+      "step": 6002
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.948083064132526e-06,
+      "loss": 0.1012,
+      "step": 6003
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944907563691093e-06,
+      "loss": 0.1213,
+      "step": 6004
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941732552554362e-06,
+      "loss": 0.0963,
+      "step": 6005
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9385580311054455e-06,
+      "loss": 0.11,
+      "step": 6006
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.935383999727398e-06,
+      "loss": 0.0994,
+      "step": 6007
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932210458803207e-06,
+      "loss": 0.1016,
+      "step": 6008
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.929037408715812e-06,
+      "loss": 0.1146,
+      "step": 6009
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.92586484984808e-06,
+      "loss": 0.1109,
+      "step": 6010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.922692782582831e-06,
+      "loss": 0.1055,
+      "step": 6011
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9195212073028145e-06,
+      "loss": 0.1103,
+      "step": 6012
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916350124390731e-06,
+      "loss": 0.1053,
+      "step": 6013
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.913179534229213e-06,
+      "loss": 0.0938,
+      "step": 6014
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.910009437200839e-06,
+      "loss": 0.1065,
+      "step": 6015
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9068398336881295e-06,
+      "loss": 0.1058,
+      "step": 6016
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.90367072407354e-06,
+      "loss": 0.114,
+      "step": 6017
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.900502108739466e-06,
+      "loss": 0.0981,
+      "step": 6018
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89733398806825e-06,
+      "loss": 0.1129,
+      "step": 6019
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.894166362442167e-06,
+      "loss": 0.1038,
+      "step": 6020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8909992322434396e-06,
+      "loss": 0.1112,
+      "step": 6021
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887832597854226e-06,
+      "loss": 0.1028,
+      "step": 6022
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.884666459656626e-06,
+      "loss": 0.0975,
+      "step": 6023
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8815008180326825e-06,
+      "loss": 0.1053,
+      "step": 6024
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87833567336437e-06,
+      "loss": 0.1079,
+      "step": 6025
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.875171026033609e-06,
+      "loss": 0.1083,
+      "step": 6026
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87200687642226e-06,
+      "loss": 0.1053,
+      "step": 6027
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868843224912126e-06,
+      "loss": 0.1216,
+      "step": 6028
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.865680071884942e-06,
+      "loss": 0.105,
+      "step": 6029
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.862517417722393e-06,
+      "loss": 0.1018,
+      "step": 6030
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.85935526280609e-06,
+      "loss": 0.0916,
+      "step": 6031
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.856193607517602e-06,
+      "loss": 0.0997,
+      "step": 6032
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.853032452238419e-06,
+      "loss": 0.1002,
+      "step": 6033
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8498717973499845e-06,
+      "loss": 0.1201,
+      "step": 6034
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.846711643233674e-06,
+      "loss": 0.1092,
+      "step": 6035
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843551990270805e-06,
+      "loss": 0.0928,
+      "step": 6036
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.840392838842642e-06,
+      "loss": 0.1104,
+      "step": 6037
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.83723418933037e-06,
+      "loss": 0.1028,
+      "step": 6038
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834076042115133e-06,
+      "loss": 0.1161,
+      "step": 6039
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830918397578005e-06,
+      "loss": 0.1014,
+      "step": 6040
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827761256099996e-06,
+      "loss": 0.1052,
+      "step": 6041
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.824604618062069e-06,
+      "loss": 0.1131,
+      "step": 6042
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.821448483845116e-06,
+      "loss": 0.1022,
+      "step": 6043
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.818292853829957e-06,
+      "loss": 0.1023,
+      "step": 6044
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.815137728397378e-06,
+      "loss": 0.1107,
+      "step": 6045
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.811983107928085e-06,
+      "loss": 0.1123,
+      "step": 6046
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.808828992802723e-06,
+      "loss": 0.1086,
+      "step": 6047
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8056753834018895e-06,
+      "loss": 0.0975,
+      "step": 6048
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.802522280106109e-06,
+      "loss": 0.1243,
+      "step": 6049
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.799369683295848e-06,
+      "loss": 0.1088,
+      "step": 6050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796217593351512e-06,
+      "loss": 0.1136,
+      "step": 6051
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.793066010653447e-06,
+      "loss": 0.1098,
+      "step": 6052
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.789914935581929e-06,
+      "loss": 0.1055,
+      "step": 6053
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.786764368517191e-06,
+      "loss": 0.0932,
+      "step": 6054
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7836143098393894e-06,
+      "loss": 0.1016,
+      "step": 6055
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.780464759928623e-06,
+      "loss": 0.1041,
+      "step": 6056
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7773157191649305e-06,
+      "loss": 0.1118,
+      "step": 6057
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.77416718792829e-06,
+      "loss": 0.1052,
+      "step": 6058
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.771019166598608e-06,
+      "loss": 0.1008,
+      "step": 6059
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7678716555557515e-06,
+      "loss": 0.1051,
+      "step": 6060
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.764724655179505e-06,
+      "loss": 0.1098,
+      "step": 6061
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761578165849596e-06,
+      "loss": 0.1162,
+      "step": 6062
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.758432187945706e-06,
+      "loss": 0.1325,
+      "step": 6063
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7552867218474296e-06,
+      "loss": 0.1098,
+      "step": 6064
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.752141767934312e-06,
+      "loss": 0.1024,
+      "step": 6065
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748997326585844e-06,
+      "loss": 0.1159,
+      "step": 6066
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.745853398181445e-06,
+      "loss": 0.1023,
+      "step": 6067
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7427099831004695e-06,
+      "loss": 0.1081,
+      "step": 6068
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.739567081722227e-06,
+      "loss": 0.0957,
+      "step": 6069
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736424694425943e-06,
+      "loss": 0.1044,
+      "step": 6070
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7332828215907895e-06,
+      "loss": 0.1033,
+      "step": 6071
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.730141463595887e-06,
+      "loss": 0.1104,
+      "step": 6072
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7270006208202825e-06,
+      "loss": 0.1035,
+      "step": 6073
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7238602936429575e-06,
+      "loss": 0.1196,
+      "step": 6074
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7207204824428454e-06,
+      "loss": 0.1072,
+      "step": 6075
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.71758118759881e-06,
+      "loss": 0.1048,
+      "step": 6076
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.714442409489639e-06,
+      "loss": 0.1059,
+      "step": 6077
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.711304148494084e-06,
+      "loss": 0.118,
+      "step": 6078
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7081664049908155e-06,
+      "loss": 0.1046,
+      "step": 6079
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.705029179358445e-06,
+      "loss": 0.1083,
+      "step": 6080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7018924719755294e-06,
+      "loss": 0.1032,
+      "step": 6081
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698756283220559e-06,
+      "loss": 0.106,
+      "step": 6082
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695620613471946e-06,
+      "loss": 0.0947,
+      "step": 6083
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692485463108066e-06,
+      "loss": 0.0923,
+      "step": 6084
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6893508325072165e-06,
+      "loss": 0.1071,
+      "step": 6085
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6862167220476305e-06,
+      "loss": 0.1048,
+      "step": 6086
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683083132107489e-06,
+      "loss": 0.1012,
+      "step": 6087
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.679950063064906e-06,
+      "loss": 0.1228,
+      "step": 6088
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6768175152979274e-06,
+      "loss": 0.1138,
+      "step": 6089
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6736854891845394e-06,
+      "loss": 0.1081,
+      "step": 6090
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6705539851026644e-06,
+      "loss": 0.1119,
+      "step": 6091
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667423003430164e-06,
+      "loss": 0.1128,
+      "step": 6092
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.664292544544838e-06,
+      "loss": 0.1135,
+      "step": 6093
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.1134,
+      "step": 6094
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6580331966465805e-06,
+      "loss": 0.1111,
+      "step": 6095
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6549043083889285e-06,
+      "loss": 0.0986,
+      "step": 6096
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6517759444290084e-06,
+      "loss": 0.0995,
+      "step": 6097
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648648105144299e-06,
+      "loss": 0.1057,
+      "step": 6098
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645520790912224e-06,
+      "loss": 0.101,
+      "step": 6099
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642394002110137e-06,
+      "loss": 0.1042,
+      "step": 6100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639267739115327e-06,
+      "loss": 0.101,
+      "step": 6101
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.636142002305025e-06,
+      "loss": 0.1018,
+      "step": 6102
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.633016792056391e-06,
+      "loss": 0.1024,
+      "step": 6103
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.629892108746533e-06,
+      "loss": 0.1034,
+      "step": 6104
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.626767952752485e-06,
+      "loss": 0.1011,
+      "step": 6105
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.623644324451223e-06,
+      "loss": 0.1037,
+      "step": 6106
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.620521224219651e-06,
+      "loss": 0.1036,
+      "step": 6107
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.617398652434628e-06,
+      "loss": 0.1061,
+      "step": 6108
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6142766094729204e-06,
+      "loss": 0.1039,
+      "step": 6109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.61115509571126e-06,
+      "loss": 0.1044,
+      "step": 6110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.608034111526298e-06,
+      "loss": 0.1,
+      "step": 6111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604913657294625e-06,
+      "loss": 0.0914,
+      "step": 6112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6017937333927644e-06,
+      "loss": 0.107,
+      "step": 6113
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598674340197192e-06,
+      "loss": 0.1069,
+      "step": 6114
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.59555547808429e-06,
+      "loss": 0.1054,
+      "step": 6115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592437147430407e-06,
+      "loss": 0.1012,
+      "step": 6116
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5893193486118105e-06,
+      "loss": 0.1043,
+      "step": 6117
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.586202082004703e-06,
+      "loss": 0.1054,
+      "step": 6118
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.583085347985228e-06,
+      "loss": 0.1055,
+      "step": 6119
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579969146929472e-06,
+      "loss": 0.1037,
+      "step": 6120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576853479213441e-06,
+      "loss": 0.1247,
+      "step": 6121
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573738345213087e-06,
+      "loss": 0.1227,
+      "step": 6122
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570623745304298e-06,
+      "loss": 0.0976,
+      "step": 6123
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567509679862886e-06,
+      "loss": 0.1113,
+      "step": 6124
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.564396149264617e-06,
+      "loss": 0.101,
+      "step": 6125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5612831538851806e-06,
+      "loss": 0.1058,
+      "step": 6126
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5581706941002025e-06,
+      "loss": 0.0954,
+      "step": 6127
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5550587702852465e-06,
+      "loss": 0.1084,
+      "step": 6128
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.55194738281581e-06,
+      "loss": 0.1023,
+      "step": 6129
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548836532067321e-06,
+      "loss": 0.1098,
+      "step": 6130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.545726218415157e-06,
+      "loss": 0.0997,
+      "step": 6131
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.542616442234618e-06,
+      "loss": 0.1032,
+      "step": 6132
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.539507203900942e-06,
+      "loss": 0.1022,
+      "step": 6133
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.536398503789305e-06,
+      "loss": 0.1036,
+      "step": 6134
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.533290342274814e-06,
+      "loss": 0.1034,
+      "step": 6135
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.530182719732509e-06,
+      "loss": 0.1052,
+      "step": 6136
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.527075636537378e-06,
+      "loss": 0.0991,
+      "step": 6137
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.52396909306433e-06,
+      "loss": 0.1048,
+      "step": 6138
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.52086308968821e-06,
+      "loss": 0.106,
+      "step": 6139
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.517757626783813e-06,
+      "loss": 0.1054,
+      "step": 6140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5146527047258465e-06,
+      "loss": 0.1102,
+      "step": 6141
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.511548323888964e-06,
+      "loss": 0.1,
+      "step": 6142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.508444484647759e-06,
+      "loss": 0.1004,
+      "step": 6143
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.505341187376753e-06,
+      "loss": 0.1122,
+      "step": 6144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502238432450396e-06,
+      "loss": 0.1105,
+      "step": 6145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49913622024309e-06,
+      "loss": 0.1026,
+      "step": 6146
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.496034551129159e-06,
+      "loss": 0.1038,
+      "step": 6147
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.492933425482854e-06,
+      "loss": 0.0982,
+      "step": 6148
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4898328436783795e-06,
+      "loss": 0.1121,
+      "step": 6149
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.486732806089863e-06,
+      "loss": 0.1088,
+      "step": 6150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.483633313091363e-06,
+      "loss": 0.1035,
+      "step": 6151
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.480534365056886e-06,
+      "loss": 0.096,
+      "step": 6152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.477435962360364e-06,
+      "loss": 0.1068,
+      "step": 6153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.474338105375651e-06,
+      "loss": 0.1058,
+      "step": 6154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4712407944765625e-06,
+      "loss": 0.1016,
+      "step": 6155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.468144030036824e-06,
+      "loss": 0.1055,
+      "step": 6156
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.465047812430104e-06,
+      "loss": 0.1122,
+      "step": 6157
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4619521420300136e-06,
+      "loss": 0.1126,
+      "step": 6158
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.458857019210083e-06,
+      "loss": 0.1073,
+      "step": 6159
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455762444343785e-06,
+      "loss": 0.111,
+      "step": 6160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.452668417804523e-06,
+      "loss": 0.0991,
+      "step": 6161
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.449574939965637e-06,
+      "loss": 0.0976,
+      "step": 6162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.446482011200394e-06,
+      "loss": 0.1046,
+      "step": 6163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4433896318820065e-06,
+      "loss": 0.1094,
+      "step": 6164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4402978023836124e-06,
+      "loss": 0.101,
+      "step": 6165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.437206523078285e-06,
+      "loss": 0.0938,
+      "step": 6166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.43411579433903e-06,
+      "loss": 0.1,
+      "step": 6167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.431025616538787e-06,
+      "loss": 0.102,
+      "step": 6168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4279359900504294e-06,
+      "loss": 0.1123,
+      "step": 6169
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.42484691524677e-06,
+      "loss": 0.1187,
+      "step": 6170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.421758392500546e-06,
+      "loss": 0.1049,
+      "step": 6171
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418670422184428e-06,
+      "loss": 0.1127,
+      "step": 6172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.415583004671036e-06,
+      "loss": 0.1257,
+      "step": 6173
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.412496140332898e-06,
+      "loss": 0.0969,
+      "step": 6174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.409409829542489e-06,
+      "loss": 0.1044,
+      "step": 6175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.406324072672224e-06,
+      "loss": 0.1084,
+      "step": 6176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.403238870094441e-06,
+      "loss": 0.1128,
+      "step": 6177
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400154222181407e-06,
+      "loss": 0.1125,
+      "step": 6178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.397070129305343e-06,
+      "loss": 0.1119,
+      "step": 6179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.393986591838376e-06,
+      "loss": 0.1048,
+      "step": 6180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.390903610152578e-06,
+      "loss": 0.1039,
+      "step": 6181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387821184619964e-06,
+      "loss": 0.1083,
+      "step": 6182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.384739315612466e-06,
+      "loss": 0.1096,
+      "step": 6183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381658003501955e-06,
+      "loss": 0.1026,
+      "step": 6184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.378577248660242e-06,
+      "loss": 0.102,
+      "step": 6185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.375497051459058e-06,
+      "loss": 0.1009,
+      "step": 6186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372417412270075e-06,
+      "loss": 0.1202,
+      "step": 6187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369338331464895e-06,
+      "loss": 0.0983,
+      "step": 6188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.366259809415053e-06,
+      "loss": 0.1006,
+      "step": 6189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.363181846492012e-06,
+      "loss": 0.0999,
+      "step": 6190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.360104443067179e-06,
+      "loss": 0.1144,
+      "step": 6191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3570275995118844e-06,
+      "loss": 0.1043,
+      "step": 6192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.353951316197393e-06,
+      "loss": 0.1073,
+      "step": 6193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.350875593494902e-06,
+      "loss": 0.1011,
+      "step": 6194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34780043177554e-06,
+      "loss": 0.1002,
+      "step": 6195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.344725831410369e-06,
+      "loss": 0.1108,
+      "step": 6196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.341651792770386e-06,
+      "loss": 0.1083,
+      "step": 6197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.338578316226517e-06,
+      "loss": 0.1033,
+      "step": 6198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.33550540214962e-06,
+      "loss": 0.1073,
+      "step": 6199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.332433050910486e-06,
+      "loss": 0.1138,
+      "step": 6200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.329361262879833e-06,
+      "loss": 0.1032,
+      "step": 6201
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.326290038428326e-06,
+      "loss": 0.1014,
+      "step": 6202
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.323219377926545e-06,
+      "loss": 0.109,
+      "step": 6203
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3201492817450125e-06,
+      "loss": 0.1033,
+      "step": 6204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.317079750254177e-06,
+      "loss": 0.0959,
+      "step": 6205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3140107838244215e-06,
+      "loss": 0.1066,
+      "step": 6206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.310942382826058e-06,
+      "loss": 0.1023,
+      "step": 6207
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.307874547629339e-06,
+      "loss": 0.0988,
+      "step": 6208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.304807278604438e-06,
+      "loss": 0.1099,
+      "step": 6209
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.301740576121468e-06,
+      "loss": 0.1068,
+      "step": 6210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298674440550463e-06,
+      "loss": 0.102,
+      "step": 6211
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.295608872261411e-06,
+      "loss": 0.1133,
+      "step": 6212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2925438716241965e-06,
+      "loss": 0.0913,
+      "step": 6213
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2894794390086715e-06,
+      "loss": 0.0982,
+      "step": 6214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286415574784597e-06,
+      "loss": 0.0935,
+      "step": 6215
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283352279321673e-06,
+      "loss": 0.1108,
+      "step": 6216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.280289552989526e-06,
+      "loss": 0.1043,
+      "step": 6217
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2772273961577285e-06,
+      "loss": 0.0798,
+      "step": 6218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274165809195757e-06,
+      "loss": 0.0775,
+      "step": 6219
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.27110479247305e-06,
+      "loss": 0.0711,
+      "step": 6220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.268044346358957e-06,
+      "loss": 0.0721,
+      "step": 6221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.264984471222761e-06,
+      "loss": 0.0895,
+      "step": 6222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.261925167433688e-06,
+      "loss": 0.0806,
+      "step": 6223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.258866435360881e-06,
+      "loss": 0.0712,
+      "step": 6224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25580827537342e-06,
+      "loss": 0.078,
+      "step": 6225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.252750687840317e-06,
+      "loss": 0.0743,
+      "step": 6226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.249693673130512e-06,
+      "loss": 0.073,
+      "step": 6227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246637231612874e-06,
+      "loss": 0.0676,
+      "step": 6228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.243581363656216e-06,
+      "loss": 0.0732,
+      "step": 6229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240526069629265e-06,
+      "loss": 0.0685,
+      "step": 6230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237471349900687e-06,
+      "loss": 0.0713,
+      "step": 6231
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.234417204839079e-06,
+      "loss": 0.0784,
+      "step": 6232
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.231363634812965e-06,
+      "loss": 0.0751,
+      "step": 6233
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.228310640190798e-06,
+      "loss": 0.0769,
+      "step": 6234
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2252582213409745e-06,
+      "loss": 0.071,
+      "step": 6235
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.222206378631807e-06,
+      "loss": 0.0713,
+      "step": 6236
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.219155112431544e-06,
+      "loss": 0.0722,
+      "step": 6237
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2161044231083656e-06,
+      "loss": 0.0762,
+      "step": 6238
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2130543110303815e-06,
+      "loss": 0.0787,
+      "step": 6239
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.210004776565624e-06,
+      "loss": 0.069,
+      "step": 6240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.206955820082074e-06,
+      "loss": 0.0796,
+      "step": 6241
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2039074419476245e-06,
+      "loss": 0.0754,
+      "step": 6242
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.200859642530105e-06,
+      "loss": 0.0691,
+      "step": 6243
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.197812422197286e-06,
+      "loss": 0.0712,
+      "step": 6244
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.194765781316846e-06,
+      "loss": 0.0684,
+      "step": 6245
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.191719720256407e-06,
+      "loss": 0.0739,
+      "step": 6246
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.188674239383527e-06,
+      "loss": 0.0772,
+      "step": 6247
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.185629339065681e-06,
+      "loss": 0.0736,
+      "step": 6248
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1825850196702785e-06,
+      "loss": 0.0655,
+      "step": 6249
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1795412815646705e-06,
+      "loss": 0.0771,
+      "step": 6250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1764981251161166e-06,
+      "loss": 0.0717,
+      "step": 6251
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.173455550691815e-06,
+      "loss": 0.0867,
+      "step": 6252
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170413558658904e-06,
+      "loss": 0.0765,
+      "step": 6253
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.167372149384442e-06,
+      "loss": 0.0685,
+      "step": 6254
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.164331323235411e-06,
+      "loss": 0.0631,
+      "step": 6255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.161291080578739e-06,
+      "loss": 0.0752,
+      "step": 6256
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158251421781276e-06,
+      "loss": 0.0753,
+      "step": 6257
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.155212347209788e-06,
+      "loss": 0.0727,
+      "step": 6258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.152173857230992e-06,
+      "loss": 0.0788,
+      "step": 6259
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.149135952211524e-06,
+      "loss": 0.0727,
+      "step": 6260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.146098632517945e-06,
+      "loss": 0.081,
+      "step": 6261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.143061898516759e-06,
+      "loss": 0.0801,
+      "step": 6262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.140025750574394e-06,
+      "loss": 0.0698,
+      "step": 6263
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.136990189057187e-06,
+      "loss": 0.0679,
+      "step": 6264
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.133955214331439e-06,
+      "loss": 0.0832,
+      "step": 6265
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.130920826763357e-06,
+      "loss": 0.0704,
+      "step": 6266
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.127887026719078e-06,
+      "loss": 0.0785,
+      "step": 6267
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.124853814564683e-06,
+      "loss": 0.0696,
+      "step": 6268
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1218211906661675e-06,
+      "loss": 0.0696,
+      "step": 6269
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.118789155389461e-06,
+      "loss": 0.0812,
+      "step": 6270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115757709100421e-06,
+      "loss": 0.0619,
+      "step": 6271
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.112726852164836e-06,
+      "loss": 0.075,
+      "step": 6272
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.109696584948417e-06,
+      "loss": 0.0744,
+      "step": 6273
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.106666907816818e-06,
+      "loss": 0.0674,
+      "step": 6274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.103637821135608e-06,
+      "loss": 0.0681,
+      "step": 6275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100609325270288e-06,
+      "loss": 0.0755,
+      "step": 6276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.097581420586292e-06,
+      "loss": 0.07,
+      "step": 6277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094554107448979e-06,
+      "loss": 0.0703,
+      "step": 6278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.091527386223632e-06,
+      "loss": 0.0745,
+      "step": 6279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0885012572754774e-06,
+      "loss": 0.0654,
+      "step": 6280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.085475720969657e-06,
+      "loss": 0.0769,
+      "step": 6281
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.082450777671242e-06,
+      "loss": 0.0705,
+      "step": 6282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.079426427745244e-06,
+      "loss": 0.0781,
+      "step": 6283
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.076402671556578e-06,
+      "loss": 0.0648,
+      "step": 6284
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.073379509470117e-06,
+      "loss": 0.0652,
+      "step": 6285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.070356941850646e-06,
+      "loss": 0.0722,
+      "step": 6286
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0673349690628785e-06,
+      "loss": 0.0684,
+      "step": 6287
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.064313591471455e-06,
+      "loss": 0.0724,
+      "step": 6288
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.061292809440958e-06,
+      "loss": 0.0693,
+      "step": 6289
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.058272623335877e-06,
+      "loss": 0.0677,
+      "step": 6290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.055253033520648e-06,
+      "loss": 0.0753,
+      "step": 6291
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.052234040359625e-06,
+      "loss": 0.069,
+      "step": 6292
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0492156442170914e-06,
+      "loss": 0.0738,
+      "step": 6293
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.046197845457258e-06,
+      "loss": 0.0726,
+      "step": 6294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.043180644444272e-06,
+      "loss": 0.0715,
+      "step": 6295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.040164041542197e-06,
+      "loss": 0.0775,
+      "step": 6296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03714803711503e-06,
+      "loss": 0.0714,
+      "step": 6297
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034132631526696e-06,
+      "loss": 0.0704,
+      "step": 6298
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03111782514104e-06,
+      "loss": 0.0745,
+      "step": 6299
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.028103618321851e-06,
+      "loss": 0.0799,
+      "step": 6300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.025090011432832e-06,
+      "loss": 0.0703,
+      "step": 6301
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.022077004837618e-06,
+      "loss": 0.0721,
+      "step": 6302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.01906459889977e-06,
+      "loss": 0.0742,
+      "step": 6303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.01605279398278e-06,
+      "loss": 0.0744,
+      "step": 6304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.013041590450057e-06,
+      "loss": 0.0846,
+      "step": 6305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.010030988664958e-06,
+      "loss": 0.0682,
+      "step": 6306
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.007020988990748e-06,
+      "loss": 0.0772,
+      "step": 6307
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.00401159179063e-06,
+      "loss": 0.0731,
+      "step": 6308
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.001002797427725e-06,
+      "loss": 0.0796,
+      "step": 6309
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.997994606265092e-06,
+      "loss": 0.0758,
+      "step": 6310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994987018665707e-06,
+      "loss": 0.0757,
+      "step": 6311
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.991980034992484e-06,
+      "loss": 0.0762,
+      "step": 6312
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.988973655608257e-06,
+      "loss": 0.0776,
+      "step": 6313
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.985967880875787e-06,
+      "loss": 0.0708,
+      "step": 6314
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982962711157765e-06,
+      "loss": 0.0684,
+      "step": 6315
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.979958146816807e-06,
+      "loss": 0.0717,
+      "step": 6316
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9769541882154515e-06,
+      "loss": 0.0652,
+      "step": 6317
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.973950835716178e-06,
+      "loss": 0.0703,
+      "step": 6318
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.970948089681379e-06,
+      "loss": 0.0747,
+      "step": 6319
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.967945950473377e-06,
+      "loss": 0.0649,
+      "step": 6320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964944418454426e-06,
+      "loss": 0.0721,
+      "step": 6321
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.961943493986709e-06,
+      "loss": 0.0761,
+      "step": 6322
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.958943177432315e-06,
+      "loss": 0.0709,
+      "step": 6323
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.955943469153287e-06,
+      "loss": 0.0733,
+      "step": 6324
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.952944369511581e-06,
+      "loss": 0.0721,
+      "step": 6325
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.949945878869075e-06,
+      "loss": 0.0728,
+      "step": 6326
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.946947997587588e-06,
+      "loss": 0.074,
+      "step": 6327
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9439507260288565e-06,
+      "loss": 0.0771,
+      "step": 6328
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.940954064554534e-06,
+      "loss": 0.076,
+      "step": 6329
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.937958013526222e-06,
+      "loss": 0.0689,
+      "step": 6330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.934962573305431e-06,
+      "loss": 0.0713,
+      "step": 6331
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.931967744253601e-06,
+      "loss": 0.0765,
+      "step": 6332
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.928973526732107e-06,
+      "loss": 0.069,
+      "step": 6333
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.925979921102243e-06,
+      "loss": 0.0713,
+      "step": 6334
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.922986927725229e-06,
+      "loss": 0.0713,
+      "step": 6335
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91999454696221e-06,
+      "loss": 0.0712,
+      "step": 6336
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.917002779174262e-06,
+      "loss": 0.0698,
+      "step": 6337
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9140116247223816e-06,
+      "loss": 0.0709,
+      "step": 6338
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9110210839674976e-06,
+      "loss": 0.0718,
+      "step": 6339
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.908031157270462e-06,
+      "loss": 0.0725,
+      "step": 6340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.90504184499205e-06,
+      "loss": 0.0727,
+      "step": 6341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902053147492964e-06,
+      "loss": 0.0687,
+      "step": 6342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8990650651338355e-06,
+      "loss": 0.0798,
+      "step": 6343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896077598275213e-06,
+      "loss": 0.0664,
+      "step": 6344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.893090747277585e-06,
+      "loss": 0.0755,
+      "step": 6345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.890104512501355e-06,
+      "loss": 0.0732,
+      "step": 6346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.887118894306849e-06,
+      "loss": 0.0737,
+      "step": 6347
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.884133893054339e-06,
+      "loss": 0.0707,
+      "step": 6348
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881149509103993e-06,
+      "loss": 0.0717,
+      "step": 6349
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878165742815922e-06,
+      "loss": 0.0707,
+      "step": 6350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.875182594550166e-06,
+      "loss": 0.0699,
+      "step": 6351
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872200064666681e-06,
+      "loss": 0.0868,
+      "step": 6352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.869218153525349e-06,
+      "loss": 0.0766,
+      "step": 6353
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.86623686148599e-06,
+      "loss": 0.0747,
+      "step": 6354
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.863256188908329e-06,
+      "loss": 0.0714,
+      "step": 6355
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.860276136152026e-06,
+      "loss": 0.0733,
+      "step": 6356
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.857296703576675e-06,
+      "loss": 0.0707,
+      "step": 6357
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.854317891541782e-06,
+      "loss": 0.072,
+      "step": 6358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.851339700406781e-06,
+      "loss": 0.0701,
+      "step": 6359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.848362130531039e-06,
+      "loss": 0.0754,
+      "step": 6360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845385182273844e-06,
+      "loss": 0.0745,
+      "step": 6361
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.842408855994395e-06,
+      "loss": 0.0756,
+      "step": 6362
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.83943315205184e-06,
+      "loss": 0.0717,
+      "step": 6363
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836458070805235e-06,
+      "loss": 0.0689,
+      "step": 6364
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.833483612613564e-06,
+      "loss": 0.0686,
+      "step": 6365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8305097778357445e-06,
+      "loss": 0.0735,
+      "step": 6366
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.827536566830609e-06,
+      "loss": 0.0712,
+      "step": 6367
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.824563979956915e-06,
+      "loss": 0.0701,
+      "step": 6368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.821592017573351e-06,
+      "loss": 0.0741,
+      "step": 6369
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.818620680038525e-06,
+      "loss": 0.0742,
+      "step": 6370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.815649967710967e-06,
+      "loss": 0.0737,
+      "step": 6371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812679880949145e-06,
+      "loss": 0.0723,
+      "step": 6372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8097104201114365e-06,
+      "loss": 0.0813,
+      "step": 6373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.806741585556151e-06,
+      "loss": 0.0703,
+      "step": 6374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.80377337764152e-06,
+      "loss": 0.0713,
+      "step": 6375
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.800805796725699e-06,
+      "loss": 0.0847,
+      "step": 6376
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.797838843166768e-06,
+      "loss": 0.0691,
+      "step": 6377
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794872517322737e-06,
+      "loss": 0.0703,
+      "step": 6378
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.791906819551533e-06,
+      "loss": 0.0684,
+      "step": 6379
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7889417502110095e-06,
+      "loss": 0.0721,
+      "step": 6380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7859773096589435e-06,
+      "loss": 0.073,
+      "step": 6381
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.783013498253035e-06,
+      "loss": 0.0748,
+      "step": 6382
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.780050316350916e-06,
+      "loss": 0.0734,
+      "step": 6383
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.777087764310134e-06,
+      "loss": 0.0742,
+      "step": 6384
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774125842488163e-06,
+      "loss": 0.066,
+      "step": 6385
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.771164551242401e-06,
+      "loss": 0.0713,
+      "step": 6386
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.768203890930169e-06,
+      "loss": 0.0725,
+      "step": 6387
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.765243861908711e-06,
+      "loss": 0.0836,
+      "step": 6388
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.762284464535202e-06,
+      "loss": 0.0677,
+      "step": 6389
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.759325699166734e-06,
+      "loss": 0.0707,
+      "step": 6390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7563675661603215e-06,
+      "loss": 0.072,
+      "step": 6391
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.753410065872904e-06,
+      "loss": 0.0667,
+      "step": 6392
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.750453198661357e-06,
+      "loss": 0.0701,
+      "step": 6393
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.747496964882452e-06,
+      "loss": 0.0758,
+      "step": 6394
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.744541364892914e-06,
+      "loss": 0.0612,
+      "step": 6395
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7415863990493735e-06,
+      "loss": 0.0757,
+      "step": 6396
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7386320677083866e-06,
+      "loss": 0.0747,
+      "step": 6397
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7356783712264405e-06,
+      "loss": 0.0706,
+      "step": 6398
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.732725309959945e-06,
+      "loss": 0.0692,
+      "step": 6399
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.729772884265212e-06,
+      "loss": 0.0782,
+      "step": 6400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.72682109449851e-06,
+      "loss": 0.0737,
+      "step": 6401
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.72386994101601e-06,
+      "loss": 0.0747,
+      "step": 6402
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7209194241738045e-06,
+      "loss": 0.0725,
+      "step": 6403
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.717969544327924e-06,
+      "loss": 0.0721,
+      "step": 6404
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.715020301834311e-06,
+      "loss": 0.0798,
+      "step": 6405
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.712071697048833e-06,
+      "loss": 0.0734,
+      "step": 6406
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.70912373032728e-06,
+      "loss": 0.0736,
+      "step": 6407
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7061764020253675e-06,
+      "loss": 0.0715,
+      "step": 6408
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703229712498728e-06,
+      "loss": 0.0735,
+      "step": 6409
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.700283662102931e-06,
+      "loss": 0.0695,
+      "step": 6410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697338251193454e-06,
+      "loss": 0.0771,
+      "step": 6411
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.694393480125703e-06,
+      "loss": 0.0725,
+      "step": 6412
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.691449349255008e-06,
+      "loss": 0.0707,
+      "step": 6413
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.688505858936618e-06,
+      "loss": 0.0682,
+      "step": 6414
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685563009525705e-06,
+      "loss": 0.0705,
+      "step": 6415
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.682620801377371e-06,
+      "loss": 0.0741,
+      "step": 6416
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.0733,
+      "step": 6417
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.676738310288433e-06,
+      "loss": 0.0789,
+      "step": 6418
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.673798028057642e-06,
+      "loss": 0.0686,
+      "step": 6419
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.670858388509038e-06,
+      "loss": 0.0718,
+      "step": 6420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6679193919973275e-06,
+      "loss": 0.0747,
+      "step": 6421
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.664981038877152e-06,
+      "loss": 0.0755,
+      "step": 6422
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.662043329503062e-06,
+      "loss": 0.073,
+      "step": 6423
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6591062642295305e-06,
+      "loss": 0.0712,
+      "step": 6424
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6561698434109655e-06,
+      "loss": 0.0726,
+      "step": 6425
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65323406740168e-06,
+      "loss": 0.0721,
+      "step": 6426
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.650298936555913e-06,
+      "loss": 0.072,
+      "step": 6427
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.647364451227841e-06,
+      "loss": 0.0722,
+      "step": 6428
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.644430611771548e-06,
+      "loss": 0.0749,
+      "step": 6429
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641497418541038e-06,
+      "loss": 0.0703,
+      "step": 6430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.638564871890249e-06,
+      "loss": 0.0748,
+      "step": 6431
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.635632972173036e-06,
+      "loss": 0.0756,
+      "step": 6432
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.632701719743164e-06,
+      "loss": 0.0724,
+      "step": 6433
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.629771114954341e-06,
+      "loss": 0.0737,
+      "step": 6434
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.626841158160182e-06,
+      "loss": 0.0673,
+      "step": 6435
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.623911849714226e-06,
+      "loss": 0.0691,
+      "step": 6436
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.62098318996994e-06,
+      "loss": 0.0725,
+      "step": 6437
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.618055179280712e-06,
+      "loss": 0.0659,
+      "step": 6438
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.615127817999835e-06,
+      "loss": 0.065,
+      "step": 6439
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.612201106480548e-06,
+      "loss": 0.071,
+      "step": 6440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.609275045075998e-06,
+      "loss": 0.0712,
+      "step": 6441
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.606349634139251e-06,
+      "loss": 0.0719,
+      "step": 6442
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.603424874023308e-06,
+      "loss": 0.0767,
+      "step": 6443
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.600500765081079e-06,
+      "loss": 0.0665,
+      "step": 6444
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597577307665399e-06,
+      "loss": 0.0724,
+      "step": 6445
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.594654502129026e-06,
+      "loss": 0.0744,
+      "step": 6446
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.591732348824637e-06,
+      "loss": 0.0667,
+      "step": 6447
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5888108481048275e-06,
+      "loss": 0.0715,
+      "step": 6448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.585890000322126e-06,
+      "loss": 0.0751,
+      "step": 6449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.582969805828972e-06,
+      "loss": 0.077,
+      "step": 6450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5800502649777255e-06,
+      "loss": 0.0692,
+      "step": 6451
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.577131378120674e-06,
+      "loss": 0.0645,
+      "step": 6452
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.574213145610021e-06,
+      "loss": 0.0752,
+      "step": 6453
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.571295567797891e-06,
+      "loss": 0.0728,
+      "step": 6454
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.568378645036336e-06,
+      "loss": 0.0723,
+      "step": 6455
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.565462377677322e-06,
+      "loss": 0.0667,
+      "step": 6456
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.562546766072734e-06,
+      "loss": 0.0785,
+      "step": 6457
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559631810574395e-06,
+      "loss": 0.0705,
+      "step": 6458
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.556717511534023e-06,
+      "loss": 0.0675,
+      "step": 6459
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.553803869303271e-06,
+      "loss": 0.0746,
+      "step": 6460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.550890884233716e-06,
+      "loss": 0.0749,
+      "step": 6461
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.547978556676852e-06,
+      "loss": 0.0737,
+      "step": 6462
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.545066886984086e-06,
+      "loss": 0.0761,
+      "step": 6463
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.542155875506763e-06,
+      "loss": 0.0691,
+      "step": 6464
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.539245522596125e-06,
+      "loss": 0.0752,
+      "step": 6465
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5363358286033585e-06,
+      "loss": 0.0776,
+      "step": 6466
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.533426793879556e-06,
+      "loss": 0.0659,
+      "step": 6467
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.0817,
+      "step": 6468
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.527610703642824e-06,
+      "loss": 0.0708,
+      "step": 6469
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524703648831692e-06,
+      "loss": 0.0749,
+      "step": 6470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.521797254693112e-06,
+      "loss": 0.0732,
+      "step": 6471
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518891521577783e-06,
+      "loss": 0.0677,
+      "step": 6472
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5159864498363204e-06,
+      "loss": 0.0771,
+      "step": 6473
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5130820398192645e-06,
+      "loss": 0.07,
+      "step": 6474
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5101782918770685e-06,
+      "loss": 0.068,
+      "step": 6475
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.507275206360121e-06,
+      "loss": 0.0706,
+      "step": 6476
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.504372783618715e-06,
+      "loss": 0.0667,
+      "step": 6477
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.501471024003068e-06,
+      "loss": 0.07,
+      "step": 6478
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.498569927863321e-06,
+      "loss": 0.0845,
+      "step": 6479
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.495669495549529e-06,
+      "loss": 0.0657,
+      "step": 6480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.492769727411676e-06,
+      "loss": 0.0654,
+      "step": 6481
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.489870623799658e-06,
+      "loss": 0.068,
+      "step": 6482
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.486972185063293e-06,
+      "loss": 0.075,
+      "step": 6483
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4840744115523184e-06,
+      "loss": 0.0702,
+      "step": 6484
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.481177303616393e-06,
+      "loss": 0.0709,
+      "step": 6485
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.478280861605089e-06,
+      "loss": 0.076,
+      "step": 6486
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.475385085867912e-06,
+      "loss": 0.0734,
+      "step": 6487
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.472489976754274e-06,
+      "loss": 0.0691,
+      "step": 6488
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4695955346135115e-06,
+      "loss": 0.0688,
+      "step": 6489
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46670175979488e-06,
+      "loss": 0.0697,
+      "step": 6490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.463808652647557e-06,
+      "loss": 0.0806,
+      "step": 6491
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46091621352063e-06,
+      "loss": 0.0728,
+      "step": 6492
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4580244427631215e-06,
+      "loss": 0.0636,
+      "step": 6493
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455133340723962e-06,
+      "loss": 0.0716,
+      "step": 6494
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452242907752e-06,
+      "loss": 0.0654,
+      "step": 6495
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.449353144196015e-06,
+      "loss": 0.0711,
+      "step": 6496
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4464640504046975e-06,
+      "loss": 0.0664,
+      "step": 6497
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.443575626726647e-06,
+      "loss": 0.0778,
+      "step": 6498
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440687873510405e-06,
+      "loss": 0.075,
+      "step": 6499
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.437800791104416e-06,
+      "loss": 0.0733,
+      "step": 6500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434914379857043e-06,
+      "loss": 0.0669,
+      "step": 6501
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.432028640116581e-06,
+      "loss": 0.0764,
+      "step": 6502
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429143572231235e-06,
+      "loss": 0.0693,
+      "step": 6503
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.426259176549119e-06,
+      "loss": 0.0848,
+      "step": 6504
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423375453418288e-06,
+      "loss": 0.0748,
+      "step": 6505
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4204924031866995e-06,
+      "loss": 0.0736,
+      "step": 6506
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.417610026202231e-06,
+      "loss": 0.0734,
+      "step": 6507
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.414728322812691e-06,
+      "loss": 0.0698,
+      "step": 6508
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.411847293365793e-06,
+      "loss": 0.0792,
+      "step": 6509
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4089669382091746e-06,
+      "loss": 0.0682,
+      "step": 6510
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.406087257690393e-06,
+      "loss": 0.0651,
+      "step": 6511
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.403208252156921e-06,
+      "loss": 0.071,
+      "step": 6512
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.400329921956148e-06,
+      "loss": 0.0751,
+      "step": 6513
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397452267435394e-06,
+      "loss": 0.0704,
+      "step": 6514
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394575288941885e-06,
+      "loss": 0.0742,
+      "step": 6515
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.391698986822769e-06,
+      "loss": 0.0698,
+      "step": 6516
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.388823361425113e-06,
+      "loss": 0.0745,
+      "step": 6517
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.385948413095903e-06,
+      "loss": 0.0737,
+      "step": 6518
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3830741421820376e-06,
+      "loss": 0.0762,
+      "step": 6519
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380200549030348e-06,
+      "loss": 0.0626,
+      "step": 6520
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.377327633987567e-06,
+      "loss": 0.0712,
+      "step": 6521
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.374455397400352e-06,
+      "loss": 0.0744,
+      "step": 6522
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.37158383961529e-06,
+      "loss": 0.0698,
+      "step": 6523
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.368712960978863e-06,
+      "loss": 0.0721,
+      "step": 6524
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.365842761837485e-06,
+      "loss": 0.0707,
+      "step": 6525
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.362973242537493e-06,
+      "loss": 0.0758,
+      "step": 6526
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.360104403425131e-06,
+      "loss": 0.0763,
+      "step": 6527
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.357236244846562e-06,
+      "loss": 0.0706,
+      "step": 6528
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.354368767147882e-06,
+      "loss": 0.0755,
+      "step": 6529
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.351501970675082e-06,
+      "loss": 0.0697,
+      "step": 6530
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.348635855774082e-06,
+      "loss": 0.0748,
+      "step": 6531
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.345770422790725e-06,
+      "loss": 0.0755,
+      "step": 6532
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342905672070765e-06,
+      "loss": 0.0714,
+      "step": 6533
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.340041603959871e-06,
+      "loss": 0.0754,
+      "step": 6534
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.33717821880364e-06,
+      "loss": 0.0702,
+      "step": 6535
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.33431551694758e-06,
+      "loss": 0.072,
+      "step": 6536
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331453498737107e-06,
+      "loss": 0.0669,
+      "step": 6537
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3285921645175756e-06,
+      "loss": 0.0745,
+      "step": 6538
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.32573151463424e-06,
+      "loss": 0.0698,
+      "step": 6539
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.32287154943228e-06,
+      "loss": 0.0682,
+      "step": 6540
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.320012269256793e-06,
+      "loss": 0.0718,
+      "step": 6541
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.31715367445279e-06,
+      "loss": 0.0779,
+      "step": 6542
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.314295765365203e-06,
+      "loss": 0.0674,
+      "step": 6543
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.311438542338879e-06,
+      "loss": 0.0675,
+      "step": 6544
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308582005718579e-06,
+      "loss": 0.0711,
+      "step": 6545
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.305726155848986e-06,
+      "loss": 0.0692,
+      "step": 6546
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.302870993074705e-06,
+      "loss": 0.0685,
+      "step": 6547
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.300016517740247e-06,
+      "loss": 0.0853,
+      "step": 6548
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297162730190046e-06,
+      "loss": 0.0669,
+      "step": 6549
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.294309630768452e-06,
+      "loss": 0.066,
+      "step": 6550
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2914572198197325e-06,
+      "loss": 0.0727,
+      "step": 6551
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2886054976880676e-06,
+      "loss": 0.0745,
+      "step": 6552
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.285754464717565e-06,
+      "loss": 0.072,
+      "step": 6553
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.282904121252241e-06,
+      "loss": 0.0704,
+      "step": 6554
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.280054467636027e-06,
+      "loss": 0.07,
+      "step": 6555
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2772055042127755e-06,
+      "loss": 0.0752,
+      "step": 6556
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.274357231326256e-06,
+      "loss": 0.0756,
+      "step": 6557
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.271509649320148e-06,
+      "loss": 0.0703,
+      "step": 6558
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.268662758538062e-06,
+      "loss": 0.0724,
+      "step": 6559
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.265816559323509e-06,
+      "loss": 0.0677,
+      "step": 6560
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2629710520199266e-06,
+      "loss": 0.0847,
+      "step": 6561
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.260126236970664e-06,
+      "loss": 0.0755,
+      "step": 6562
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2572821145189856e-06,
+      "loss": 0.0864,
+      "step": 6563
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.254438685008083e-06,
+      "loss": 0.0698,
+      "step": 6564
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2515959487810535e-06,
+      "loss": 0.0676,
+      "step": 6565
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.248753906180913e-06,
+      "loss": 0.0739,
+      "step": 6566
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.245912557550589e-06,
+      "loss": 0.0694,
+      "step": 6567
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243071903232944e-06,
+      "loss": 0.0675,
+      "step": 6568
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2402319435707274e-06,
+      "loss": 0.0687,
+      "step": 6569
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.237392678906633e-06,
+      "loss": 0.082,
+      "step": 6570
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234554109583255e-06,
+      "loss": 0.0692,
+      "step": 6571
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.231716235943106e-06,
+      "loss": 0.0725,
+      "step": 6572
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2288790583286135e-06,
+      "loss": 0.0745,
+      "step": 6573
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.226042577082133e-06,
+      "loss": 0.0725,
+      "step": 6574
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.223206792545914e-06,
+      "loss": 0.0708,
+      "step": 6575
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.220371705062143e-06,
+      "loss": 0.0748,
+      "step": 6576
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.217537314972911e-06,
+      "loss": 0.0674,
+      "step": 6577
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.214703622620223e-06,
+      "loss": 0.066,
+      "step": 6578
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.211870628346013e-06,
+      "loss": 0.0706,
+      "step": 6579
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.209038332492118e-06,
+      "loss": 0.0822,
+      "step": 6580
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.206206735400296e-06,
+      "loss": 0.0725,
+      "step": 6581
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.203375837412217e-06,
+      "loss": 0.0689,
+      "step": 6582
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200545638869471e-06,
+      "loss": 0.0661,
+      "step": 6583
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.197716140113558e-06,
+      "loss": 0.0686,
+      "step": 6584
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.194887341485904e-06,
+      "loss": 0.0693,
+      "step": 6585
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.19205924332784e-06,
+      "loss": 0.0805,
+      "step": 6586
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189231845980618e-06,
+      "loss": 0.0678,
+      "step": 6587
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.186405149785403e-06,
+      "loss": 0.0718,
+      "step": 6588
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183579155083276e-06,
+      "loss": 0.0797,
+      "step": 6589
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.180753862215229e-06,
+      "loss": 0.072,
+      "step": 6590
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.177929271522183e-06,
+      "loss": 0.0673,
+      "step": 6591
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.17510538334496e-06,
+      "loss": 0.076,
+      "step": 6592
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.172282198024299e-06,
+      "loss": 0.0754,
+      "step": 6593
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.169459715900869e-06,
+      "loss": 0.0692,
+      "step": 6594
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.166637937315231e-06,
+      "loss": 0.072,
+      "step": 6595
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.163816862607874e-06,
+      "loss": 0.0697,
+      "step": 6596
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.160996492119208e-06,
+      "loss": 0.0738,
+      "step": 6597
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158176826189545e-06,
+      "loss": 0.08,
+      "step": 6598
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.155357865159118e-06,
+      "loss": 0.0696,
+      "step": 6599
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.152539609368083e-06,
+      "loss": 0.0736,
+      "step": 6600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.149722059156491e-06,
+      "loss": 0.068,
+      "step": 6601
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.146905214864323e-06,
+      "loss": 0.0709,
+      "step": 6602
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144089076831476e-06,
+      "loss": 0.0737,
+      "step": 6603
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1412736453977545e-06,
+      "loss": 0.0707,
+      "step": 6604
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.138458920902876e-06,
+      "loss": 0.0767,
+      "step": 6605
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.135644903686485e-06,
+      "loss": 0.078,
+      "step": 6606
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.132831594088134e-06,
+      "loss": 0.0725,
+      "step": 6607
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.130018992447276e-06,
+      "loss": 0.0701,
+      "step": 6608
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127207099103304e-06,
+      "loss": 0.0767,
+      "step": 6609
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.124395914395509e-06,
+      "loss": 0.0723,
+      "step": 6610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.121585438663096e-06,
+      "loss": 0.074,
+      "step": 6611
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.118775672245197e-06,
+      "loss": 0.0768,
+      "step": 6612
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1159666154808485e-06,
+      "loss": 0.0754,
+      "step": 6613
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113158268708996e-06,
+      "loss": 0.0667,
+      "step": 6614
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110350632268514e-06,
+      "loss": 0.0726,
+      "step": 6615
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1075437064981824e-06,
+      "loss": 0.0749,
+      "step": 6616
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.104737491736692e-06,
+      "loss": 0.0671,
+      "step": 6617
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.10193198832266e-06,
+      "loss": 0.0704,
+      "step": 6618
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.099127196594608e-06,
+      "loss": 0.0792,
+      "step": 6619
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.096323116890972e-06,
+      "loss": 0.0849,
+      "step": 6620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.093519749550107e-06,
+      "loss": 0.069,
+      "step": 6621
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.090717094910276e-06,
+      "loss": 0.0749,
+      "step": 6622
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.087915153309657e-06,
+      "loss": 0.0701,
+      "step": 6623
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.085113925086352e-06,
+      "loss": 0.0741,
+      "step": 6624
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082313410578366e-06,
+      "loss": 0.0789,
+      "step": 6625
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.079513610123619e-06,
+      "loss": 0.0723,
+      "step": 6626
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076714524059948e-06,
+      "loss": 0.0697,
+      "step": 6627
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.073916152725104e-06,
+      "loss": 0.0693,
+      "step": 6628
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071118496456743e-06,
+      "loss": 0.0678,
+      "step": 6629
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068321555592453e-06,
+      "loss": 0.0741,
+      "step": 6630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.065525330469719e-06,
+      "loss": 0.0734,
+      "step": 6631
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.062729821425944e-06,
+      "loss": 0.0746,
+      "step": 6632
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.059935028798455e-06,
+      "loss": 0.0696,
+      "step": 6633
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.057140952924473e-06,
+      "loss": 0.0719,
+      "step": 6634
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.054347594141144e-06,
+      "loss": 0.0716,
+      "step": 6635
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.051554952785531e-06,
+      "loss": 0.0714,
+      "step": 6636
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.048763029194606e-06,
+      "loss": 0.0689,
+      "step": 6637
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.045971823705249e-06,
+      "loss": 0.065,
+      "step": 6638
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043181336654271e-06,
+      "loss": 0.0726,
+      "step": 6639
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0403915683783664e-06,
+      "loss": 0.0696,
+      "step": 6640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0376025192141744e-06,
+      "loss": 0.0715,
+      "step": 6641
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034814189498229e-06,
+      "loss": 0.0707,
+      "step": 6642
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0320265795669815e-06,
+      "loss": 0.0721,
+      "step": 6643
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.029239689756793e-06,
+      "loss": 0.0653,
+      "step": 6644
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026453520403949e-06,
+      "loss": 0.0678,
+      "step": 6645
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0236680718446376e-06,
+      "loss": 0.0688,
+      "step": 6646
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.020883344414963e-06,
+      "loss": 0.0673,
+      "step": 6647
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0180993384509405e-06,
+      "loss": 0.0697,
+      "step": 6648
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0153160542885015e-06,
+      "loss": 0.0728,
+      "step": 6649
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.012533492263485e-06,
+      "loss": 0.0682,
+      "step": 6650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.009751652711654e-06,
+      "loss": 0.0765,
+      "step": 6651
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.006970535968674e-06,
+      "loss": 0.0714,
+      "step": 6652
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.004190142370126e-06,
+      "loss": 0.077,
+      "step": 6653
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.001410472251504e-06,
+      "loss": 0.0716,
+      "step": 6654
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.998631525948214e-06,
+      "loss": 0.0676,
+      "step": 6655
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995853303795573e-06,
+      "loss": 0.0675,
+      "step": 6656
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99307580612882e-06,
+      "loss": 0.0726,
+      "step": 6657
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.990299033283096e-06,
+      "loss": 0.0669,
+      "step": 6658
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987522985593459e-06,
+      "loss": 0.0679,
+      "step": 6659
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9847476633948765e-06,
+      "loss": 0.0707,
+      "step": 6660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.98197306702223e-06,
+      "loss": 0.069,
+      "step": 6661
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.979199196810319e-06,
+      "loss": 0.0701,
+      "step": 6662
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.976426053093849e-06,
+      "loss": 0.0684,
+      "step": 6663
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.973653636207437e-06,
+      "loss": 0.0777,
+      "step": 6664
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.970881946485617e-06,
+      "loss": 0.0741,
+      "step": 6665
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.96811098426283e-06,
+      "loss": 0.0664,
+      "step": 6666
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.965340749873432e-06,
+      "loss": 0.0765,
+      "step": 6667
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962571243651695e-06,
+      "loss": 0.0704,
+      "step": 6668
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959802465931798e-06,
+      "loss": 0.0712,
+      "step": 6669
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.957034417047832e-06,
+      "loss": 0.0646,
+      "step": 6670
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9542670973338e-06,
+      "loss": 0.0683,
+      "step": 6671
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9515005071236274e-06,
+      "loss": 0.0701,
+      "step": 6672
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.94873464675113e-06,
+      "loss": 0.0749,
+      "step": 6673
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945969516550058e-06,
+      "loss": 0.0769,
+      "step": 6674
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.94320511685406e-06,
+      "loss": 0.0644,
+      "step": 6675
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9404414479966966e-06,
+      "loss": 0.0693,
+      "step": 6676
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.937678510311452e-06,
+      "loss": 0.0704,
+      "step": 6677
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934916304131714e-06,
+      "loss": 0.0654,
+      "step": 6678
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932154829790771e-06,
+      "loss": 0.075,
+      "step": 6679
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929394087621844e-06,
+      "loss": 0.0759,
+      "step": 6680
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926634077958053e-06,
+      "loss": 0.0782,
+      "step": 6681
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92387480113243e-06,
+      "loss": 0.0783,
+      "step": 6682
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.921116257477927e-06,
+      "loss": 0.0651,
+      "step": 6683
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.918358447327399e-06,
+      "loss": 0.0778,
+      "step": 6684
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915601371013612e-06,
+      "loss": 0.0714,
+      "step": 6685
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.912845028869251e-06,
+      "loss": 0.0757,
+      "step": 6686
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.910089421226905e-06,
+      "loss": 0.0716,
+      "step": 6687
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907334548419076e-06,
+      "loss": 0.0736,
+      "step": 6688
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.904580410778185e-06,
+      "loss": 0.0755,
+      "step": 6689
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.901827008636553e-06,
+      "loss": 0.0743,
+      "step": 6690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.899074342326418e-06,
+      "loss": 0.0787,
+      "step": 6691
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.896322412179929e-06,
+      "loss": 0.0727,
+      "step": 6692
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.893571218529146e-06,
+      "loss": 0.0754,
+      "step": 6693
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.890820761706034e-06,
+      "loss": 0.0689,
+      "step": 6694
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.888071042042484e-06,
+      "loss": 0.0762,
+      "step": 6695
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.885322059870284e-06,
+      "loss": 0.0747,
+      "step": 6696
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882573815521139e-06,
+      "loss": 0.0824,
+      "step": 6697
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.879826309326664e-06,
+      "loss": 0.0734,
+      "step": 6698
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.877079541618384e-06,
+      "loss": 0.0651,
+      "step": 6699
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.874333512727732e-06,
+      "loss": 0.0839,
+      "step": 6700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8715882229860626e-06,
+      "loss": 0.0742,
+      "step": 6701
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.86884367272463e-06,
+      "loss": 0.071,
+      "step": 6702
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.866099862274603e-06,
+      "loss": 0.0733,
+      "step": 6703
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.863356791967069e-06,
+      "loss": 0.075,
+      "step": 6704
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.860614462133008e-06,
+      "loss": 0.0666,
+      "step": 6705
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.857872873103322e-06,
+      "loss": 0.0708,
+      "step": 6706
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.855132025208829e-06,
+      "loss": 0.0739,
+      "step": 6707
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.852391918780251e-06,
+      "loss": 0.0718,
+      "step": 6708
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.849652554148213e-06,
+      "loss": 0.0673,
+      "step": 6709
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.846913931643271e-06,
+      "loss": 0.0727,
+      "step": 6710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.844176051595869e-06,
+      "loss": 0.0743,
+      "step": 6711
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.841438914336369e-06,
+      "loss": 0.0699,
+      "step": 6712
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838702520195056e-06,
+      "loss": 0.0805,
+      "step": 6713
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.835966869502108e-06,
+      "loss": 0.07,
+      "step": 6714
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8332319625876195e-06,
+      "loss": 0.071,
+      "step": 6715
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.830497799781601e-06,
+      "loss": 0.0717,
+      "step": 6716
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.827764381413969e-06,
+      "loss": 0.0737,
+      "step": 6717
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.82503170781454e-06,
+      "loss": 0.072,
+      "step": 6718
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822299779313058e-06,
+      "loss": 0.0745,
+      "step": 6719
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.819568596239167e-06,
+      "loss": 0.0744,
+      "step": 6720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.81683815892242e-06,
+      "loss": 0.0674,
+      "step": 6721
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.81410846769229e-06,
+      "loss": 0.0671,
+      "step": 6722
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.81137952287815e-06,
+      "loss": 0.0693,
+      "step": 6723
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.808651324809285e-06,
+      "loss": 0.0664,
+      "step": 6724
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.805923873814892e-06,
+      "loss": 0.0712,
+      "step": 6725
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8031971702240766e-06,
+      "loss": 0.0699,
+      "step": 6726
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8004712143658507e-06,
+      "loss": 0.0766,
+      "step": 6727
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7977460065691463e-06,
+      "loss": 0.0703,
+      "step": 6728
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.795021547162796e-06,
+      "loss": 0.0672,
+      "step": 6729
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.792297836475545e-06,
+      "loss": 0.0663,
+      "step": 6730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7895748748360463e-06,
+      "loss": 0.0756,
+      "step": 6731
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.786852662572865e-06,
+      "loss": 0.0709,
+      "step": 6732
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7841312000144703e-06,
+      "loss": 0.0693,
+      "step": 6733
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7814104874892542e-06,
+      "loss": 0.0667,
+      "step": 6734
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7786905253255056e-06,
+      "loss": 0.0639,
+      "step": 6735
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.775971313851425e-06,
+      "loss": 0.0722,
+      "step": 6736
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7732528533951264e-06,
+      "loss": 0.0722,
+      "step": 6737
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.770535144284625e-06,
+      "loss": 0.0646,
+      "step": 6738
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.767818186847859e-06,
+      "loss": 0.074,
+      "step": 6739
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.0783,
+      "step": 6740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.762386528306793e-06,
+      "loss": 0.0762,
+      "step": 6741
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7596718278578946e-06,
+      "loss": 0.0735,
+      "step": 6742
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7569578803935504e-06,
+      "loss": 0.0781,
+      "step": 6743
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7542446862412207e-06,
+      "loss": 0.0672,
+      "step": 6744
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7515322457283043e-06,
+      "loss": 0.0796,
+      "step": 6745
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7488205591820894e-06,
+      "loss": 0.0781,
+      "step": 6746
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7461096269297804e-06,
+      "loss": 0.069,
+      "step": 6747
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.743399449298488e-06,
+      "loss": 0.0729,
+      "step": 6748
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7406900266152424e-06,
+      "loss": 0.0703,
+      "step": 6749
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7379813592069614e-06,
+      "loss": 0.0767,
+      "step": 6750
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7352734474004937e-06,
+      "loss": 0.0697,
+      "step": 6751
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7325662915225836e-06,
+      "loss": 0.0715,
+      "step": 6752
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7298598918998907e-06,
+      "loss": 0.0721,
+      "step": 6753
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.727154248858974e-06,
+      "loss": 0.0748,
+      "step": 6754
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.724449362726318e-06,
+      "loss": 0.0696,
+      "step": 6755
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.721745233828299e-06,
+      "loss": 0.0767,
+      "step": 6756
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7190418624912104e-06,
+      "loss": 0.0633,
+      "step": 6757
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.716339249041253e-06,
+      "loss": 0.0665,
+      "step": 6758
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7136373938045313e-06,
+      "loss": 0.0779,
+      "step": 6759
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7109362971070707e-06,
+      "loss": 0.0721,
+      "step": 6760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7082359592747906e-06,
+      "loss": 0.0701,
+      "step": 6761
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7055363806335287e-06,
+      "loss": 0.0744,
+      "step": 6762
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702837561509027e-06,
+      "loss": 0.0671,
+      "step": 6763
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.700139502226935e-06,
+      "loss": 0.0664,
+      "step": 6764
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.697442203112809e-06,
+      "loss": 0.0755,
+      "step": 6765
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.694745664492124e-06,
+      "loss": 0.0728,
+      "step": 6766
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.692049886690252e-06,
+      "loss": 0.0764,
+      "step": 6767
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6893548700324775e-06,
+      "loss": 0.0742,
+      "step": 6768
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6866606148439932e-06,
+      "loss": 0.0719,
+      "step": 6769
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.683967121449897e-06,
+      "loss": 0.0771,
+      "step": 6770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6812743901751968e-06,
+      "loss": 0.0782,
+      "step": 6771
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.678582421344814e-06,
+      "loss": 0.07,
+      "step": 6772
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6758912152835704e-06,
+      "loss": 0.0745,
+      "step": 6773
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6732007723161933e-06,
+      "loss": 0.0681,
+      "step": 6774
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.670511092767336e-06,
+      "loss": 0.0674,
+      "step": 6775
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6678221769615343e-06,
+      "loss": 0.0624,
+      "step": 6776
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6651340252232448e-06,
+      "loss": 0.0725,
+      "step": 6777
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6624466378768387e-06,
+      "loss": 0.0714,
+      "step": 6778
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.659760015246584e-06,
+      "loss": 0.0738,
+      "step": 6779
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.657074157656656e-06,
+      "loss": 0.0727,
+      "step": 6780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.654389065431149e-06,
+      "loss": 0.0739,
+      "step": 6781
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6517047388940574e-06,
+      "loss": 0.0773,
+      "step": 6782
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.649021178369273e-06,
+      "loss": 0.0752,
+      "step": 6783
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6463383841806167e-06,
+      "loss": 0.0758,
+      "step": 6784
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6436563566518025e-06,
+      "loss": 0.0797,
+      "step": 6785
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6409750961064507e-06,
+      "loss": 0.0692,
+      "step": 6786
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.638294602868101e-06,
+      "loss": 0.0733,
+      "step": 6787
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.635614877260193e-06,
+      "loss": 0.0725,
+      "step": 6788
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6329359196060643e-06,
+      "loss": 0.0659,
+      "step": 6789
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.630257730228979e-06,
+      "loss": 0.0693,
+      "step": 6790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6275803094520944e-06,
+      "loss": 0.0729,
+      "step": 6791
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6249036575984773e-06,
+      "loss": 0.0588,
+      "step": 6792
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6222277749911115e-06,
+      "loss": 0.0705,
+      "step": 6793
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6195526619528754e-06,
+      "loss": 0.0697,
+      "step": 6794
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6168783188065604e-06,
+      "loss": 0.0712,
+      "step": 6795
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.614204745874863e-06,
+      "loss": 0.0753,
+      "step": 6796
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6115319434803897e-06,
+      "loss": 0.0701,
+      "step": 6797
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.608859911945647e-06,
+      "loss": 0.0764,
+      "step": 6798
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.606188651593061e-06,
+      "loss": 0.0719,
+      "step": 6799
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6035181627449544e-06,
+      "loss": 0.0752,
+      "step": 6800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6008484457235603e-06,
+      "loss": 0.0734,
+      "step": 6801
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.598179500851017e-06,
+      "loss": 0.0704,
+      "step": 6802
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.59551132844937e-06,
+      "loss": 0.0712,
+      "step": 6803
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5928439288405703e-06,
+      "loss": 0.0671,
+      "step": 6804
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5901773023464847e-06,
+      "loss": 0.0592,
+      "step": 6805
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5875114492888754e-06,
+      "loss": 0.0788,
+      "step": 6806
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.584846369989413e-06,
+      "loss": 0.0712,
+      "step": 6807
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.582182064769687e-06,
+      "loss": 0.0715,
+      "step": 6808
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5795185339511727e-06,
+      "loss": 0.0683,
+      "step": 6809
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.576855777855265e-06,
+      "loss": 0.0647,
+      "step": 6810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.574193796803268e-06,
+      "loss": 0.0684,
+      "step": 6811
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5715325911163868e-06,
+      "loss": 0.078,
+      "step": 6812
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5688721611157284e-06,
+      "loss": 0.071,
+      "step": 6813
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5662125071223217e-06,
+      "loss": 0.0746,
+      "step": 6814
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.563553629457084e-06,
+      "loss": 0.0688,
+      "step": 6815
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.560895528440844e-06,
+      "loss": 0.0707,
+      "step": 6816
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5582382043943474e-06,
+      "loss": 0.0676,
+      "step": 6817
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5555816576382353e-06,
+      "loss": 0.0672,
+      "step": 6818
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.552925888493054e-06,
+      "loss": 0.0717,
+      "step": 6819
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5502708972792675e-06,
+      "loss": 0.0729,
+      "step": 6820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5476166843172334e-06,
+      "loss": 0.0695,
+      "step": 6821
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5449632499272215e-06,
+      "loss": 0.0695,
+      "step": 6822
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5423105944294055e-06,
+      "loss": 0.0741,
+      "step": 6823
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.539658718143868e-06,
+      "loss": 0.0734,
+      "step": 6824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5370076213905904e-06,
+      "loss": 0.0741,
+      "step": 6825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.534357304489473e-06,
+      "loss": 0.0767,
+      "step": 6826
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5317077677603107e-06,
+      "loss": 0.0747,
+      "step": 6827
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5290590115228072e-06,
+      "loss": 0.0696,
+      "step": 6828
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.526411036096574e-06,
+      "loss": 0.0645,
+      "step": 6829
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5237638418011255e-06,
+      "loss": 0.0649,
+      "step": 6830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5211174289558815e-06,
+      "loss": 0.0686,
+      "step": 6831
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5184717978801764e-06,
+      "loss": 0.0714,
+      "step": 6832
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5158269488932384e-06,
+      "loss": 0.0696,
+      "step": 6833
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5131828823142065e-06,
+      "loss": 0.0799,
+      "step": 6834
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.510539598462127e-06,
+      "loss": 0.08,
+      "step": 6835
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5078970976559434e-06,
+      "loss": 0.0753,
+      "step": 6836
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5052553802145205e-06,
+      "loss": 0.0712,
+      "step": 6837
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.502614446456615e-06,
+      "loss": 0.074,
+      "step": 6838
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.499974296700891e-06,
+      "loss": 0.0752,
+      "step": 6839
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.497334931265923e-06,
+      "loss": 0.0657,
+      "step": 6840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4946963504701883e-06,
+      "loss": 0.0705,
+      "step": 6841
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.492058554632063e-06,
+      "loss": 0.0699,
+      "step": 6842
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4894215440698444e-06,
+      "loss": 0.0779,
+      "step": 6843
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.486785319101721e-06,
+      "loss": 0.0684,
+      "step": 6844
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4841498800457897e-06,
+      "loss": 0.069,
+      "step": 6845
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4815152272200557e-06,
+      "loss": 0.0774,
+      "step": 6846
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4788813609424266e-06,
+      "loss": 0.0755,
+      "step": 6847
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.476248281530712e-06,
+      "loss": 0.0668,
+      "step": 6848
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4736159893026376e-06,
+      "loss": 0.0697,
+      "step": 6849
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4709844845758244e-06,
+      "loss": 0.0663,
+      "step": 6850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4683537676678e-06,
+      "loss": 0.0746,
+      "step": 6851
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4657238388959957e-06,
+      "loss": 0.0638,
+      "step": 6852
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.463094698577759e-06,
+      "loss": 0.0717,
+      "step": 6853
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.460466347030319e-06,
+      "loss": 0.0679,
+      "step": 6854
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.457838784570835e-06,
+      "loss": 0.0708,
+      "step": 6855
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.455212011516357e-06,
+      "loss": 0.0683,
+      "step": 6856
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4525860281838374e-06,
+      "loss": 0.0694,
+      "step": 6857
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.449960834890147e-06,
+      "loss": 0.0654,
+      "step": 6858
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.447336431952052e-06,
+      "loss": 0.0736,
+      "step": 6859
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4447128196862134e-06,
+      "loss": 0.0787,
+      "step": 6860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4420899984092194e-06,
+      "loss": 0.0729,
+      "step": 6861
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4394679684375456e-06,
+      "loss": 0.0744,
+      "step": 6862
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.436846730087574e-06,
+      "loss": 0.0657,
+      "step": 6863
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4342262836756013e-06,
+      "loss": 0.0698,
+      "step": 6864
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4316066295178197e-06,
+      "loss": 0.0721,
+      "step": 6865
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4289877679303263e-06,
+      "loss": 0.0722,
+      "step": 6866
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.426369699229124e-06,
+      "loss": 0.0722,
+      "step": 6867
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4237524237301213e-06,
+      "loss": 0.0699,
+      "step": 6868
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4211359417491265e-06,
+      "loss": 0.071,
+      "step": 6869
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.41852025360186e-06,
+      "loss": 0.0722,
+      "step": 6870
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4159053596039403e-06,
+      "loss": 0.0716,
+      "step": 6871
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4132912600708925e-06,
+      "loss": 0.0686,
+      "step": 6872
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4106779553181426e-06,
+      "loss": 0.0649,
+      "step": 6873
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.408065445661024e-06,
+      "loss": 0.076,
+      "step": 6874
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4054537314147707e-06,
+      "loss": 0.0743,
+      "step": 6875
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.402842812894529e-06,
+      "loss": 0.0697,
+      "step": 6876
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4002326904153403e-06,
+      "loss": 0.0686,
+      "step": 6877
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.397623364292151e-06,
+      "loss": 0.0688,
+      "step": 6878
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3950148348398204e-06,
+      "loss": 0.0698,
+      "step": 6879
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3924071023730986e-06,
+      "loss": 0.066,
+      "step": 6880
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3898001672066426e-06,
+      "loss": 0.0707,
+      "step": 6881
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3871940296550256e-06,
+      "loss": 0.0662,
+      "step": 6882
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3845886900327086e-06,
+      "loss": 0.0705,
+      "step": 6883
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.381984148654063e-06,
+      "loss": 0.0674,
+      "step": 6884
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3793804058333713e-06,
+      "loss": 0.0755,
+      "step": 6885
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.376777461884805e-06,
+      "loss": 0.0738,
+      "step": 6886
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3741753171224432e-06,
+      "loss": 0.0753,
+      "step": 6887
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3715739718602803e-06,
+      "loss": 0.07,
+      "step": 6888
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3689734264122008e-06,
+      "loss": 0.0748,
+      "step": 6889
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.366373681091997e-06,
+      "loss": 0.0761,
+      "step": 6890
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.36377473621337e-06,
+      "loss": 0.0735,
+      "step": 6891
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.361176592089919e-06,
+      "loss": 0.0767,
+      "step": 6892
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3585792490351387e-06,
+      "loss": 0.0718,
+      "step": 6893
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.355982707362444e-06,
+      "loss": 0.0721,
+      "step": 6894
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3533869673851427e-06,
+      "loss": 0.0826,
+      "step": 6895
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.350792029416444e-06,
+      "loss": 0.0713,
+      "step": 6896
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3481978937694704e-06,
+      "loss": 0.0744,
+      "step": 6897
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3456045607572418e-06,
+      "loss": 0.0657,
+      "step": 6898
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3430120306926716e-06,
+      "loss": 0.0678,
+      "step": 6899
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.340420303888594e-06,
+      "loss": 0.0772,
+      "step": 6900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3378293806577345e-06,
+      "loss": 0.07,
+      "step": 6901
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3352392613127227e-06,
+      "loss": 0.0752,
+      "step": 6902
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3326499461660987e-06,
+      "loss": 0.0656,
+      "step": 6903
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3300614355302986e-06,
+      "loss": 0.065,
+      "step": 6904
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.327473729717662e-06,
+      "loss": 0.0701,
+      "step": 6905
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3248868290404336e-06,
+      "loss": 0.0748,
+      "step": 6906
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3223007338107583e-06,
+      "loss": 0.0711,
+      "step": 6907
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.319715444340682e-06,
+      "loss": 0.0673,
+      "step": 6908
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3171309609421654e-06,
+      "loss": 0.0754,
+      "step": 6909
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3145472839270575e-06,
+      "loss": 0.076,
+      "step": 6910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.311964413607117e-06,
+      "loss": 0.0636,
+      "step": 6911
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.309382350294005e-06,
+      "loss": 0.0658,
+      "step": 6912
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3068010942992824e-06,
+      "loss": 0.0694,
+      "step": 6913
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.304220645934412e-06,
+      "loss": 0.0728,
+      "step": 6914
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3016410055107683e-06,
+      "loss": 0.0676,
+      "step": 6915
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.299062173339619e-06,
+      "loss": 0.0716,
+      "step": 6916
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.296484149732132e-06,
+      "loss": 0.0738,
+      "step": 6917
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2939069349993946e-06,
+      "loss": 0.0689,
+      "step": 6918
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2913305294523688e-06,
+      "loss": 0.0718,
+      "step": 6919
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2887549334019477e-06,
+      "loss": 0.0686,
+      "step": 6920
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2861801471589073e-06,
+      "loss": 0.0765,
+      "step": 6921
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.283606171033934e-06,
+      "loss": 0.0711,
+      "step": 6922
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2810330053376114e-06,
+      "loss": 0.0676,
+      "step": 6923
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2784606503804383e-06,
+      "loss": 0.0699,
+      "step": 6924
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2758891064727917e-06,
+      "loss": 0.0707,
+      "step": 6925
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.273318373924976e-06,
+      "loss": 0.07,
+      "step": 6926
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2707484530471846e-06,
+      "loss": 0.0701,
+      "step": 6927
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2681793441495123e-06,
+      "loss": 0.0655,
+      "step": 6928
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2656110475419576e-06,
+      "loss": 0.0764,
+      "step": 6929
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2630435635344283e-06,
+      "loss": 0.0803,
+      "step": 6930
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2604768924367234e-06,
+      "loss": 0.071,
+      "step": 6931
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.25791103455855e-06,
+      "loss": 0.0717,
+      "step": 6932
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.255345990209514e-06,
+      "loss": 0.0704,
+      "step": 6933
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.252781759699123e-06,
+      "loss": 0.0691,
+      "step": 6934
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.250218343336793e-06,
+      "loss": 0.0679,
+      "step": 6935
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.247655741431833e-06,
+      "loss": 0.0679,
+      "step": 6936
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.24509395429346e-06,
+      "loss": 0.0687,
+      "step": 6937
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2425329822307884e-06,
+      "loss": 0.0841,
+      "step": 6938
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.239972825552835e-06,
+      "loss": 0.0754,
+      "step": 6939
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.237413484568518e-06,
+      "loss": 0.0692,
+      "step": 6940
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.234854959586663e-06,
+      "loss": 0.0719,
+      "step": 6941
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.232297250915991e-06,
+      "loss": 0.0741,
+      "step": 6942
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2297403588651254e-06,
+      "loss": 0.0708,
+      "step": 6943
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.0701,
+      "step": 6944
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.224629025856816e-06,
+      "loss": 0.0676,
+      "step": 6945
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2220745855161226e-06,
+      "loss": 0.0687,
+      "step": 6946
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2195209630287514e-06,
+      "loss": 0.0696,
+      "step": 6947
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2169681587028266e-06,
+      "loss": 0.0677,
+      "step": 6948
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2144161728463806e-06,
+      "loss": 0.0648,
+      "step": 6949
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2118650057673493e-06,
+      "loss": 0.068,
+      "step": 6950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2093146577735656e-06,
+      "loss": 0.0733,
+      "step": 6951
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.206765129172762e-06,
+      "loss": 0.0705,
+      "step": 6952
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.204216420272581e-06,
+      "loss": 0.0756,
+      "step": 6953
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2016685313805605e-06,
+      "loss": 0.0713,
+      "step": 6954
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1991214628041347e-06,
+      "loss": 0.0724,
+      "step": 6955
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196575214850649e-06,
+      "loss": 0.0806,
+      "step": 6956
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.194029787827346e-06,
+      "loss": 0.075,
+      "step": 6957
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1914851820413573e-06,
+      "loss": 0.076,
+      "step": 6958
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1889413977997365e-06,
+      "loss": 0.0657,
+      "step": 6959
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1863984354094236e-06,
+      "loss": 0.0745,
+      "step": 6960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.18385629517726e-06,
+      "loss": 0.0732,
+      "step": 6961
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1813149774099973e-06,
+      "loss": 0.0721,
+      "step": 6962
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.178774482414282e-06,
+      "loss": 0.0732,
+      "step": 6963
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.176234810496651e-06,
+      "loss": 0.0713,
+      "step": 6964
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1736959619635634e-06,
+      "loss": 0.0788,
+      "step": 6965
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.171157937121362e-06,
+      "loss": 0.0723,
+      "step": 6966
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.168620736276293e-06,
+      "loss": 0.0731,
+      "step": 6967
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1660843597345137e-06,
+      "loss": 0.0778,
+      "step": 6968
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1635488078020695e-06,
+      "loss": 0.0716,
+      "step": 6969
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.16101408078491e-06,
+      "loss": 0.0678,
+      "step": 6970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1584801789888887e-06,
+      "loss": 0.0762,
+      "step": 6971
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.155947102719754e-06,
+      "loss": 0.073,
+      "step": 6972
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1534148522831574e-06,
+      "loss": 0.0734,
+      "step": 6973
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.150883427984656e-06,
+      "loss": 0.0746,
+      "step": 6974
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1483528301296985e-06,
+      "loss": 0.072,
+      "step": 6975
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.145823059023638e-06,
+      "loss": 0.0741,
+      "step": 6976
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1432941149717277e-06,
+      "loss": 0.0728,
+      "step": 6977
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1407659982791204e-06,
+      "loss": 0.0691,
+      "step": 6978
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.138238709250867e-06,
+      "loss": 0.0738,
+      "step": 6979
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1357122481919265e-06,
+      "loss": 0.0776,
+      "step": 6980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.13318661540715e-06,
+      "loss": 0.0774,
+      "step": 6981
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1306618112012876e-06,
+      "loss": 0.0679,
+      "step": 6982
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1281378358790015e-06,
+      "loss": 0.0654,
+      "step": 6983
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1256146897448367e-06,
+      "loss": 0.0643,
+      "step": 6984
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.123092373103248e-06,
+      "loss": 0.073,
+      "step": 6985
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1205708862585927e-06,
+      "loss": 0.0734,
+      "step": 6986
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1180502295151214e-06,
+      "loss": 0.0762,
+      "step": 6987
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.115530403176986e-06,
+      "loss": 0.0744,
+      "step": 6988
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1130114075482477e-06,
+      "loss": 0.0667,
+      "step": 6989
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.110493242932849e-06,
+      "loss": 0.0728,
+      "step": 6990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.107975909634644e-06,
+      "loss": 0.0785,
+      "step": 6991
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1054594079573907e-06,
+      "loss": 0.0756,
+      "step": 6992
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1029437382047368e-06,
+      "loss": 0.0725,
+      "step": 6993
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1004289006802304e-06,
+      "loss": 0.0675,
+      "step": 6994
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.097914895687333e-06,
+      "loss": 0.0696,
+      "step": 6995
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0954017235293863e-06,
+      "loss": 0.072,
+      "step": 6996
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0928893845096386e-06,
+      "loss": 0.077,
+      "step": 6997
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0903778789312468e-06,
+      "loss": 0.0742,
+      "step": 6998
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0878672070972557e-06,
+      "loss": 0.0717,
+      "step": 6999
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.085357369310612e-06,
+      "loss": 0.0668,
+      "step": 7000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0828483658741693e-06,
+      "loss": 0.0704,
+      "step": 7001
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0803401970906698e-06,
+      "loss": 0.0711,
+      "step": 7002
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0778328632627618e-06,
+      "loss": 0.0776,
+      "step": 7003
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.075326364692991e-06,
+      "loss": 0.068,
+      "step": 7004
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.072820701683801e-06,
+      "loss": 0.0852,
+      "step": 7005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0703158745375316e-06,
+      "loss": 0.0753,
+      "step": 7006
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.067811883556435e-06,
+      "loss": 0.0689,
+      "step": 7007
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0653087290426484e-06,
+      "loss": 0.0699,
+      "step": 7008
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.062806411298214e-06,
+      "loss": 0.0726,
+      "step": 7009
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0603049306250697e-06,
+      "loss": 0.0762,
+      "step": 7010
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0578042873250592e-06,
+      "loss": 0.072,
+      "step": 7011
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0553044816999133e-06,
+      "loss": 0.0683,
+      "step": 7012
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.052805514051278e-06,
+      "loss": 0.0714,
+      "step": 7013
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0503073846806854e-06,
+      "loss": 0.0739,
+      "step": 7014
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0478100938895705e-06,
+      "loss": 0.0755,
+      "step": 7015
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.045313641979267e-06,
+      "loss": 0.0669,
+      "step": 7016
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0428180292510055e-06,
+      "loss": 0.066,
+      "step": 7017
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0403232560059216e-06,
+      "loss": 0.0672,
+      "step": 7018
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0378293225450426e-06,
+      "loss": 0.0737,
+      "step": 7019
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.035336229169297e-06,
+      "loss": 0.0719,
+      "step": 7020
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0328439761795137e-06,
+      "loss": 0.0715,
+      "step": 7021
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0303525638764163e-06,
+      "loss": 0.0799,
+      "step": 7022
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0278619925606266e-06,
+      "loss": 0.0728,
+      "step": 7023
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0253722625326756e-06,
+      "loss": 0.069,
+      "step": 7024
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0228833740929796e-06,
+      "loss": 0.0702,
+      "step": 7025
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0203953275418584e-06,
+      "loss": 0.0675,
+      "step": 7026
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0179081231795293e-06,
+      "loss": 0.0776,
+      "step": 7027
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0154217613061165e-06,
+      "loss": 0.0655,
+      "step": 7028
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0129362422216223e-06,
+      "loss": 0.0671,
+      "step": 7029
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0104515662259703e-06,
+      "loss": 0.0734,
+      "step": 7030
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0079677336189685e-06,
+      "loss": 0.0678,
+      "step": 7031
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0054847447003243e-06,
+      "loss": 0.0673,
+      "step": 7032
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.003002599769651e-06,
+      "loss": 0.0663,
+      "step": 7033
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0005212991264554e-06,
+      "loss": 0.0671,
+      "step": 7034
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9980408430701326e-06,
+      "loss": 0.0691,
+      "step": 7035
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9955612318999927e-06,
+      "loss": 0.0706,
+      "step": 7036
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9930824659152348e-06,
+      "loss": 0.0718,
+      "step": 7037
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9906045454149535e-06,
+      "loss": 0.0732,
+      "step": 7038
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9881274706981522e-06,
+      "loss": 0.0717,
+      "step": 7039
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9856512420637206e-06,
+      "loss": 0.0701,
+      "step": 7040
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9831758598104533e-06,
+      "loss": 0.0758,
+      "step": 7041
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9807013242370386e-06,
+      "loss": 0.07,
+      "step": 7042
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9782276356420646e-06,
+      "loss": 0.0714,
+      "step": 7043
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.975754794324015e-06,
+      "loss": 0.0745,
+      "step": 7044
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9732828005812787e-06,
+      "loss": 0.0649,
+      "step": 7045
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9708116547121333e-06,
+      "loss": 0.071,
+      "step": 7046
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.96834135701476e-06,
+      "loss": 0.0734,
+      "step": 7047
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9658719077872334e-06,
+      "loss": 0.0652,
+      "step": 7048
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9634033073275283e-06,
+      "loss": 0.0682,
+      "step": 7049
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.960935555933514e-06,
+      "loss": 0.0736,
+      "step": 7050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.958468653902965e-06,
+      "loss": 0.0768,
+      "step": 7051
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9560026015335462e-06,
+      "loss": 0.0712,
+      "step": 7052
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9535373991228177e-06,
+      "loss": 0.0685,
+      "step": 7053
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9510730469682525e-06,
+      "loss": 0.0697,
+      "step": 7054
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.948609545367198e-06,
+      "loss": 0.069,
+      "step": 7055
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9461468946169137e-06,
+      "loss": 0.0646,
+      "step": 7056
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.943685095014557e-06,
+      "loss": 0.0727,
+      "step": 7057
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9412241468571766e-06,
+      "loss": 0.0643,
+      "step": 7058
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.938764050441719e-06,
+      "loss": 0.0722,
+      "step": 7059
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.936304806065039e-06,
+      "loss": 0.066,
+      "step": 7060
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9338464140238687e-06,
+      "loss": 0.0754,
+      "step": 7061
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.93138887461485e-06,
+      "loss": 0.0706,
+      "step": 7062
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.0701,
+      "step": 7063
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9264763548793263e-06,
+      "loss": 0.0728,
+      "step": 7064
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9240213751455804e-06,
+      "loss": 0.0717,
+      "step": 7065
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9215672492295233e-06,
+      "loss": 0.077,
+      "step": 7066
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.919113977427278e-06,
+      "loss": 0.0668,
+      "step": 7067
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9166615600348603e-06,
+      "loss": 0.0657,
+      "step": 7068
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9142099973481963e-06,
+      "loss": 0.0719,
+      "step": 7069
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.911759289663101e-06,
+      "loss": 0.0744,
+      "step": 7070
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9093094372752815e-06,
+      "loss": 0.0577,
+      "step": 7071
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.906860440480356e-06,
+      "loss": 0.0771,
+      "step": 7072
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9044122995738287e-06,
+      "loss": 0.0751,
+      "step": 7073
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.901965014851096e-06,
+      "loss": 0.0701,
+      "step": 7074
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8995185866074647e-06,
+      "loss": 0.075,
+      "step": 7075
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.897073015138129e-06,
+      "loss": 0.0659,
+      "step": 7076
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8946283007381794e-06,
+      "loss": 0.0718,
+      "step": 7077
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8921844437026113e-06,
+      "loss": 0.0815,
+      "step": 7078
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.889741444326307e-06,
+      "loss": 0.0714,
+      "step": 7079
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8872993029040506e-06,
+      "loss": 0.0723,
+      "step": 7080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8848580197305197e-06,
+      "loss": 0.0723,
+      "step": 7081
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8824175951002918e-06,
+      "loss": 0.0676,
+      "step": 7082
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.879978029307835e-06,
+      "loss": 0.0754,
+      "step": 7083
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8775393226475225e-06,
+      "loss": 0.0682,
+      "step": 7084
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8751014754136174e-06,
+      "loss": 0.0707,
+      "step": 7085
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8726644879002817e-06,
+      "loss": 0.0726,
+      "step": 7086
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8702283604015703e-06,
+      "loss": 0.0702,
+      "step": 7087
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.867793093211438e-06,
+      "loss": 0.0694,
+      "step": 7088
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8653586866237313e-06,
+      "loss": 0.0755,
+      "step": 7089
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8629251409322024e-06,
+      "loss": 0.0731,
+      "step": 7090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8604924564304894e-06,
+      "loss": 0.0733,
+      "step": 7091
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8580606334121286e-06,
+      "loss": 0.073,
+      "step": 7092
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8556296721705622e-06,
+      "loss": 0.071,
+      "step": 7093
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.853199572999111e-06,
+      "loss": 0.0692,
+      "step": 7094
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8507703361910024e-06,
+      "loss": 0.0706,
+      "step": 7095
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8483419620393636e-06,
+      "loss": 0.0708,
+      "step": 7096
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8459144508372093e-06,
+      "loss": 0.0622,
+      "step": 7097
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8434878028774503e-06,
+      "loss": 0.0725,
+      "step": 7098
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.841062018452907e-06,
+      "loss": 0.0713,
+      "step": 7099
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8386370978562707e-06,
+      "loss": 0.0701,
+      "step": 7100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8362130413801524e-06,
+      "loss": 0.0757,
+      "step": 7101
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.833789849317046e-06,
+      "loss": 0.0677,
+      "step": 7102
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.831367521959345e-06,
+      "loss": 0.0677,
+      "step": 7103
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8289460595993334e-06,
+      "loss": 0.0707,
+      "step": 7104
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8265254625292014e-06,
+      "loss": 0.0691,
+      "step": 7105
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.824105731041027e-06,
+      "loss": 0.0654,
+      "step": 7106
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8216868654267836e-06,
+      "loss": 0.0704,
+      "step": 7107
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.819268865978343e-06,
+      "loss": 0.0824,
+      "step": 7108
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.81685173298747e-06,
+      "loss": 0.0747,
+      "step": 7109
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.814435466745824e-06,
+      "loss": 0.0714,
+      "step": 7110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8120200675449683e-06,
+      "loss": 0.0718,
+      "step": 7111
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.809605535676352e-06,
+      "loss": 0.0747,
+      "step": 7112
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8071918714313217e-06,
+      "loss": 0.0685,
+      "step": 7113
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8047790751011216e-06,
+      "loss": 0.0716,
+      "step": 7114
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8023671469768866e-06,
+      "loss": 0.0714,
+      "step": 7115
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.799956087349657e-06,
+      "loss": 0.0741,
+      "step": 7116
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7975458965103564e-06,
+      "loss": 0.0711,
+      "step": 7117
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7951365747498114e-06,
+      "loss": 0.0702,
+      "step": 7118
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.792728122358739e-06,
+      "loss": 0.0822,
+      "step": 7119
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7903205396277546e-06,
+      "loss": 0.0681,
+      "step": 7120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.787913826847363e-06,
+      "loss": 0.0697,
+      "step": 7121
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.785507984307975e-06,
+      "loss": 0.071,
+      "step": 7122
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.783103012299887e-06,
+      "loss": 0.0708,
+      "step": 7123
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.780698911113293e-06,
+      "loss": 0.0696,
+      "step": 7124
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7782956810382834e-06,
+      "loss": 0.0718,
+      "step": 7125
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.77589332236484e-06,
+      "loss": 0.0732,
+      "step": 7126
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.773491835382839e-06,
+      "loss": 0.0678,
+      "step": 7127
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7710912203820618e-06,
+      "loss": 0.0676,
+      "step": 7128
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.768691477652171e-06,
+      "loss": 0.0756,
+      "step": 7129
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.76629260748273e-06,
+      "loss": 0.0683,
+      "step": 7130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7638946101632004e-06,
+      "loss": 0.0695,
+      "step": 7131
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.761497485982936e-06,
+      "loss": 0.0658,
+      "step": 7132
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.759101235231173e-06,
+      "loss": 0.0733,
+      "step": 7133
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7567058581970653e-06,
+      "loss": 0.069,
+      "step": 7134
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.754311355169643e-06,
+      "loss": 0.0709,
+      "step": 7135
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7519177264378373e-06,
+      "loss": 0.069,
+      "step": 7136
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7495249722904773e-06,
+      "loss": 0.0757,
+      "step": 7137
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7471330930162834e-06,
+      "loss": 0.0705,
+      "step": 7138
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.744742088903861e-06,
+      "loss": 0.0734,
+      "step": 7139
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.742351960241728e-06,
+      "loss": 0.075,
+      "step": 7140
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7399627073182844e-06,
+      "loss": 0.0672,
+      "step": 7141
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.737574330421825e-06,
+      "loss": 0.0739,
+      "step": 7142
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.735186829840547e-06,
+      "loss": 0.0695,
+      "step": 7143
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.732800205862536e-06,
+      "loss": 0.0723,
+      "step": 7144
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7304144587757643e-06,
+      "loss": 0.0733,
+      "step": 7145
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.728029588868114e-06,
+      "loss": 0.0668,
+      "step": 7146
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.725645596427352e-06,
+      "loss": 0.069,
+      "step": 7147
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.723262481741138e-06,
+      "loss": 0.0775,
+      "step": 7148
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.720880245097033e-06,
+      "loss": 0.0691,
+      "step": 7149
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.718498886782486e-06,
+      "loss": 0.0717,
+      "step": 7150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7161184070848423e-06,
+      "loss": 0.0613,
+      "step": 7151
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7137388062913394e-06,
+      "loss": 0.0744,
+      "step": 7152
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7113600846891117e-06,
+      "loss": 0.0755,
+      "step": 7153
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7089822425651814e-06,
+      "loss": 0.077,
+      "step": 7154
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.706605280206476e-06,
+      "loss": 0.069,
+      "step": 7155
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.704229197899806e-06,
+      "loss": 0.0712,
+      "step": 7156
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7018539959318802e-06,
+      "loss": 0.0677,
+      "step": 7157
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6994796745893e-06,
+      "loss": 0.0663,
+      "step": 7158
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.697106234158563e-06,
+      "loss": 0.0727,
+      "step": 7159
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6947336749260533e-06,
+      "loss": 0.0709,
+      "step": 7160
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.692361997178061e-06,
+      "loss": 0.0738,
+      "step": 7161
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.68999120120076e-06,
+      "loss": 0.0658,
+      "step": 7162
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6876212872802164e-06,
+      "loss": 0.0751,
+      "step": 7163
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.685252255702405e-06,
+      "loss": 0.0698,
+      "step": 7164
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.682884106753174e-06,
+      "loss": 0.0734,
+      "step": 7165
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6805168407182726e-06,
+      "loss": 0.0735,
+      "step": 7166
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.678150457883352e-06,
+      "loss": 0.0636,
+      "step": 7167
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.675784958533948e-06,
+      "loss": 0.0723,
+      "step": 7168
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.673420342955487e-06,
+      "loss": 0.0741,
+      "step": 7169
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6710566114333048e-06,
+      "loss": 0.0779,
+      "step": 7170
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.668693764252608e-06,
+      "loss": 0.0767,
+      "step": 7171
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.66633180169851e-06,
+      "loss": 0.0697,
+      "step": 7172
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6639707240560184e-06,
+      "loss": 0.0641,
+      "step": 7173
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6616105316100317e-06,
+      "loss": 0.0705,
+      "step": 7174
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.659251224645334e-06,
+      "loss": 0.0675,
+      "step": 7175
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6568928034466178e-06,
+      "loss": 0.0696,
+      "step": 7176
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.654535268298457e-06,
+      "loss": 0.0718,
+      "step": 7177
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6521786194853205e-06,
+      "loss": 0.0707,
+      "step": 7178
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.649822857291573e-06,
+      "loss": 0.0687,
+      "step": 7179
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6474679820014694e-06,
+      "loss": 0.0709,
+      "step": 7180
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6451139938991565e-06,
+      "loss": 0.0823,
+      "step": 7181
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.642760893268684e-06,
+      "loss": 0.0755,
+      "step": 7182
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.640408680393982e-06,
+      "loss": 0.0712,
+      "step": 7183
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6380573555588797e-06,
+      "loss": 0.0745,
+      "step": 7184
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.635706919047096e-06,
+      "loss": 0.0694,
+      "step": 7185
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.633357371142248e-06,
+      "loss": 0.0706,
+      "step": 7186
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.631008712127836e-06,
+      "loss": 0.0738,
+      "step": 7187
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6286609422872665e-06,
+      "loss": 0.0709,
+      "step": 7188
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6263140619038284e-06,
+      "loss": 0.0715,
+      "step": 7189
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6239680712607075e-06,
+      "loss": 0.0669,
+      "step": 7190
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.621622970640978e-06,
+      "loss": 0.0741,
+      "step": 7191
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6192787603276125e-06,
+      "loss": 0.0665,
+      "step": 7192
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6169354406034687e-06,
+      "loss": 0.0656,
+      "step": 7193
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6145930117513086e-06,
+      "loss": 0.0737,
+      "step": 7194
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.612251474053775e-06,
+      "loss": 0.0787,
+      "step": 7195
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6099108277934105e-06,
+      "loss": 0.0735,
+      "step": 7196
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6075710732526448e-06,
+      "loss": 0.0671,
+      "step": 7197
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6052322107138017e-06,
+      "loss": 0.0716,
+      "step": 7198
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.602894240459103e-06,
+      "loss": 0.0745,
+      "step": 7199
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.600557162770655e-06,
+      "loss": 0.0671,
+      "step": 7200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.598220977930461e-06,
+      "loss": 0.0743,
+      "step": 7201
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5958856862204098e-06,
+      "loss": 0.0684,
+      "step": 7202
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.593551287922298e-06,
+      "loss": 0.0764,
+      "step": 7203
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.591217783317792e-06,
+      "loss": 0.0885,
+      "step": 7204
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5888851726884712e-06,
+      "loss": 0.0717,
+      "step": 7205
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5865534563157956e-06,
+      "loss": 0.0642,
+      "step": 7206
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5842226344811193e-06,
+      "loss": 0.0802,
+      "step": 7207
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5818927074656874e-06,
+      "loss": 0.0695,
+      "step": 7208
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.579563675550646e-06,
+      "loss": 0.0653,
+      "step": 7209
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5772355390170155e-06,
+      "loss": 0.0788,
+      "step": 7210
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5749082981457273e-06,
+      "loss": 0.069,
+      "step": 7211
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.572581953217592e-06,
+      "loss": 0.0721,
+      "step": 7212
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5702565045133167e-06,
+      "loss": 0.0894,
+      "step": 7213
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5679319523135016e-06,
+      "loss": 0.0705,
+      "step": 7214
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5656082968986372e-06,
+      "loss": 0.0745,
+      "step": 7215
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.563285538549104e-06,
+      "loss": 0.0727,
+      "step": 7216
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5609636775451762e-06,
+      "loss": 0.0679,
+      "step": 7217
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.558642714167021e-06,
+      "loss": 0.067,
+      "step": 7218
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5563226486946902e-06,
+      "loss": 0.068,
+      "step": 7219
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5540034814081406e-06,
+      "loss": 0.0657,
+      "step": 7220
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.551685212587209e-06,
+      "loss": 0.0801,
+      "step": 7221
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5493678425116275e-06,
+      "loss": 0.0717,
+      "step": 7222
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.54705137146102e-06,
+      "loss": 0.0783,
+      "step": 7223
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5447357997149025e-06,
+      "loss": 0.0701,
+      "step": 7224
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5424211275526777e-06,
+      "loss": 0.0741,
+      "step": 7225
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5401073552536505e-06,
+      "loss": 0.0682,
+      "step": 7226
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.537794483097007e-06,
+      "loss": 0.071,
+      "step": 7227
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5354825113618253e-06,
+      "loss": 0.076,
+      "step": 7228
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5331714403270867e-06,
+      "loss": 0.0684,
+      "step": 7229
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.530861270271646e-06,
+      "loss": 0.0731,
+      "step": 7230
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.528552001474258e-06,
+      "loss": 0.0655,
+      "step": 7231
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5262436342135753e-06,
+      "loss": 0.082,
+      "step": 7232
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.523936168768132e-06,
+      "loss": 0.0646,
+      "step": 7233
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5216296054163547e-06,
+      "loss": 0.0671,
+      "step": 7234
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.519323944436569e-06,
+      "loss": 0.0721,
+      "step": 7235
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5170191861069804e-06,
+      "loss": 0.0725,
+      "step": 7236
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.514715330705689e-06,
+      "loss": 0.0747,
+      "step": 7237
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5124123785106945e-06,
+      "loss": 0.0693,
+      "step": 7238
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5101103297998774e-06,
+      "loss": 0.0663,
+      "step": 7239
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.50780918485101e-06,
+      "loss": 0.0778,
+      "step": 7240
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5055089439417646e-06,
+      "loss": 0.0669,
+      "step": 7241
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.503209607349697e-06,
+      "loss": 0.0711,
+      "step": 7242
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5009111753522463e-06,
+      "loss": 0.0697,
+      "step": 7243
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.498613648226761e-06,
+      "loss": 0.0732,
+      "step": 7244
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4963170262504667e-06,
+      "loss": 0.0678,
+      "step": 7245
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4940213097004796e-06,
+      "loss": 0.0751,
+      "step": 7246
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4917264988538194e-06,
+      "loss": 0.0735,
+      "step": 7247
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4894325939873844e-06,
+      "loss": 0.0694,
+      "step": 7248
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4871395953779597e-06,
+      "loss": 0.0712,
+      "step": 7249
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.484847503302238e-06,
+      "loss": 0.074,
+      "step": 7250
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4825563180367882e-06,
+      "loss": 0.0633,
+      "step": 7251
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4802660398580735e-06,
+      "loss": 0.077,
+      "step": 7252
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4779766690424522e-06,
+      "loss": 0.0712,
+      "step": 7253
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4756882058661693e-06,
+      "loss": 0.0709,
+      "step": 7254
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4734006506053574e-06,
+      "loss": 0.0743,
+      "step": 7255
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.471114003536045e-06,
+      "loss": 0.0741,
+      "step": 7256
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4688282649341465e-06,
+      "loss": 0.0728,
+      "step": 7257
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.466543435075468e-06,
+      "loss": 0.0718,
+      "step": 7258
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4642595142357116e-06,
+      "loss": 0.0756,
+      "step": 7259
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461976502690463e-06,
+      "loss": 0.0654,
+      "step": 7260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.459694400715199e-06,
+      "loss": 0.0683,
+      "step": 7261
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4574132085852888e-06,
+      "loss": 0.0693,
+      "step": 7262
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4551329265759894e-06,
+      "loss": 0.0728,
+      "step": 7263
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4528535549624464e-06,
+      "loss": 0.077,
+      "step": 7264
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4505750940197058e-06,
+      "loss": 0.07,
+      "step": 7265
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.448297544022692e-06,
+      "loss": 0.0658,
+      "step": 7266
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4460209052462213e-06,
+      "loss": 0.0716,
+      "step": 7267
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4437451779650122e-06,
+      "loss": 0.0686,
+      "step": 7268
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4414703624536553e-06,
+      "loss": 0.0768,
+      "step": 7269
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.439196458986638e-06,
+      "loss": 0.0727,
+      "step": 7270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4369234678383445e-06,
+      "loss": 0.0756,
+      "step": 7271
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4346513892830427e-06,
+      "loss": 0.0734,
+      "step": 7272
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4323802235948866e-06,
+      "loss": 0.0731,
+      "step": 7273
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.430109971047935e-06,
+      "loss": 0.073,
+      "step": 7274
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.427840631916114e-06,
+      "loss": 0.0729,
+      "step": 7275
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.425572206473259e-06,
+      "loss": 0.0704,
+      "step": 7276
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4233046949930873e-06,
+      "loss": 0.0684,
+      "step": 7277
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4210380977492055e-06,
+      "loss": 0.0667,
+      "step": 7278
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.418772415015107e-06,
+      "loss": 0.072,
+      "step": 7279
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4165076470641856e-06,
+      "loss": 0.0663,
+      "step": 7280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.414243794169715e-06,
+      "loss": 0.0713,
+      "step": 7281
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4119808566048607e-06,
+      "loss": 0.0768,
+      "step": 7282
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4097188346426794e-06,
+      "loss": 0.0658,
+      "step": 7283
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.407457728556115e-06,
+      "loss": 0.0712,
+      "step": 7284
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4051975386180004e-06,
+      "loss": 0.0725,
+      "step": 7285
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4029382651010647e-06,
+      "loss": 0.0736,
+      "step": 7286
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.400679908277919e-06,
+      "loss": 0.0719,
+      "step": 7287
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3984224684210655e-06,
+      "loss": 0.0704,
+      "step": 7288
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3961659458028975e-06,
+      "loss": 0.0739,
+      "step": 7289
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3939103406956964e-06,
+      "loss": 0.075,
+      "step": 7290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3916556533716296e-06,
+      "loss": 0.065,
+      "step": 7291
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3894018841027646e-06,
+      "loss": 0.0731,
+      "step": 7292
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.387149033161046e-06,
+      "loss": 0.0703,
+      "step": 7293
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.384897100818313e-06,
+      "loss": 0.0669,
+      "step": 7294
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.382646087346293e-06,
+      "loss": 0.0747,
+      "step": 7295
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.380395993016602e-06,
+      "loss": 0.0719,
+      "step": 7296
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3781468181007494e-06,
+      "loss": 0.0657,
+      "step": 7297
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3758985628701282e-06,
+      "loss": 0.0711,
+      "step": 7298
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3736512275960224e-06,
+      "loss": 0.067,
+      "step": 7299
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.371404812549605e-06,
+      "loss": 0.0626,
+      "step": 7300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.369159318001937e-06,
+      "loss": 0.0695,
+      "step": 7301
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.366914744223968e-06,
+      "loss": 0.0686,
+      "step": 7302
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.364671091486543e-06,
+      "loss": 0.0723,
+      "step": 7303
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.362428360060389e-06,
+      "loss": 0.0731,
+      "step": 7304
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3601865502161213e-06,
+      "loss": 0.0736,
+      "step": 7305
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.357945662224248e-06,
+      "loss": 0.0717,
+      "step": 7306
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3557056963551627e-06,
+      "loss": 0.0771,
+      "step": 7307
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.353466652879147e-06,
+      "loss": 0.0689,
+      "step": 7308
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.351228532066381e-06,
+      "loss": 0.0714,
+      "step": 7309
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.34899133418692e-06,
+      "loss": 0.0703,
+      "step": 7310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.346755059510711e-06,
+      "loss": 0.0627,
+      "step": 7311
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.344519708307601e-06,
+      "loss": 0.0712,
+      "step": 7312
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342285280847314e-06,
+      "loss": 0.0686,
+      "step": 7313
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.340051777399459e-06,
+      "loss": 0.0687,
+      "step": 7314
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3378191982335484e-06,
+      "loss": 0.0681,
+      "step": 7315
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3355875436189714e-06,
+      "loss": 0.0713,
+      "step": 7316
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.333356813825005e-06,
+      "loss": 0.0709,
+      "step": 7317
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331127009120826e-06,
+      "loss": 0.0707,
+      "step": 7318
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.328898129775491e-06,
+      "loss": 0.0749,
+      "step": 7319
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.326670176057938e-06,
+      "loss": 0.0669,
+      "step": 7320
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.32444314823701e-06,
+      "loss": 0.0761,
+      "step": 7321
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3222170465814265e-06,
+      "loss": 0.0738,
+      "step": 7322
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3199918713597947e-06,
+      "loss": 0.0662,
+      "step": 7323
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.317767622840622e-06,
+      "loss": 0.0817,
+      "step": 7324
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3155443012922897e-06,
+      "loss": 0.0784,
+      "step": 7325
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3133219069830758e-06,
+      "loss": 0.0729,
+      "step": 7326
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.311100440181142e-06,
+      "loss": 0.0728,
+      "step": 7327
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3088799011545414e-06,
+      "loss": 0.0685,
+      "step": 7328
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.306660290171211e-06,
+      "loss": 0.0719,
+      "step": 7329
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3044416074989818e-06,
+      "loss": 0.0694,
+      "step": 7330
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3022238534055685e-06,
+      "loss": 0.0687,
+      "step": 7331
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3000070281585752e-06,
+      "loss": 0.0742,
+      "step": 7332
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.297791132025492e-06,
+      "loss": 0.0688,
+      "step": 7333
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2955761652736996e-06,
+      "loss": 0.0752,
+      "step": 7334
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.293362128170462e-06,
+      "loss": 0.0684,
+      "step": 7335
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.29114902098294e-06,
+      "loss": 0.0669,
+      "step": 7336
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2889368439781733e-06,
+      "loss": 0.067,
+      "step": 7337
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2867255974230905e-06,
+      "loss": 0.0714,
+      "step": 7338
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2845152815845173e-06,
+      "loss": 0.0707,
+      "step": 7339
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2823058967291534e-06,
+      "loss": 0.0728,
+      "step": 7340
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.28009744312359e-06,
+      "loss": 0.0675,
+      "step": 7341
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2778899210343163e-06,
+      "loss": 0.066,
+      "step": 7342
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275683330727697e-06,
+      "loss": 0.0761,
+      "step": 7343
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.273477672469987e-06,
+      "loss": 0.0674,
+      "step": 7344
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2712729465273387e-06,
+      "loss": 0.0736,
+      "step": 7345
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2690691531657748e-06,
+      "loss": 0.0717,
+      "step": 7346
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.266866292651214e-06,
+      "loss": 0.0713,
+      "step": 7347
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2646643652494693e-06,
+      "loss": 0.0713,
+      "step": 7348
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.262463371226232e-06,
+      "loss": 0.074,
+      "step": 7349
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2602633108470794e-06,
+      "loss": 0.0729,
+      "step": 7350
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2580641843774863e-06,
+      "loss": 0.0725,
+      "step": 7351
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2558659920828095e-06,
+      "loss": 0.0708,
+      "step": 7352
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2536687342282826e-06,
+      "loss": 0.0724,
+      "step": 7353
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2514724110790455e-06,
+      "loss": 0.0659,
+      "step": 7354
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2492770229001128e-06,
+      "loss": 0.0674,
+      "step": 7355
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.247082569956386e-06,
+      "loss": 0.0736,
+      "step": 7356
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2448890525126633e-06,
+      "loss": 0.072,
+      "step": 7357
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2426964708336196e-06,
+      "loss": 0.0722,
+      "step": 7358
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2405048251838235e-06,
+      "loss": 0.0717,
+      "step": 7359
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2383141158277277e-06,
+      "loss": 0.0699,
+      "step": 7360
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2361243430296707e-06,
+      "loss": 0.0659,
+      "step": 7361
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2339355070538794e-06,
+      "loss": 0.0695,
+      "step": 7362
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2317476081644727e-06,
+      "loss": 0.0792,
+      "step": 7363
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2295606466254483e-06,
+      "loss": 0.0745,
+      "step": 7364
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.227374622700694e-06,
+      "loss": 0.07,
+      "step": 7365
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2251895366539866e-06,
+      "loss": 0.072,
+      "step": 7366
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2230053887489866e-06,
+      "loss": 0.0745,
+      "step": 7367
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.22082217924924e-06,
+      "loss": 0.0671,
+      "step": 7368
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.218639908418189e-06,
+      "loss": 0.0753,
+      "step": 7369
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2164585765191503e-06,
+      "loss": 0.0679,
+      "step": 7370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.214278183815334e-06,
+      "loss": 0.0648,
+      "step": 7371
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.212098730569836e-06,
+      "loss": 0.0675,
+      "step": 7372
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2099202170456358e-06,
+      "loss": 0.0726,
+      "step": 7373
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.207742643505606e-06,
+      "loss": 0.0738,
+      "step": 7374
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.205566010212501e-06,
+      "loss": 0.0701,
+      "step": 7375
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2033903174289615e-06,
+      "loss": 0.0702,
+      "step": 7376
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2012155654175125e-06,
+      "loss": 0.0703,
+      "step": 7377
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1990417544405796e-06,
+      "loss": 0.0681,
+      "step": 7378
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1968688847604512e-06,
+      "loss": 0.07,
+      "step": 7379
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.194696956639323e-06,
+      "loss": 0.069,
+      "step": 7380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1925259703392678e-06,
+      "loss": 0.0664,
+      "step": 7381
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.190355926122244e-06,
+      "loss": 0.0734,
+      "step": 7382
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1881868242500968e-06,
+      "loss": 0.0735,
+      "step": 7383
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.186018664984568e-06,
+      "loss": 0.077,
+      "step": 7384
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1838514485872653e-06,
+      "loss": 0.0709,
+      "step": 7385
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.0767,
+      "step": 7386
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.179519845443269e-06,
+      "loss": 0.0721,
+      "step": 7387
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1773554592192426e-06,
+      "loss": 0.067,
+      "step": 7388
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1751920169087835e-06,
+      "loss": 0.069,
+      "step": 7389
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.173029518772949e-06,
+      "loss": 0.0736,
+      "step": 7390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1708679650726703e-06,
+      "loss": 0.0693,
+      "step": 7391
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1687073560687732e-06,
+      "loss": 0.0731,
+      "step": 7392
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.166547692021963e-06,
+      "loss": 0.0779,
+      "step": 7393
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1643889731928327e-06,
+      "loss": 0.0676,
+      "step": 7394
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1622311998418665e-06,
+      "loss": 0.0672,
+      "step": 7395
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.160074372229429e-06,
+      "loss": 0.0746,
+      "step": 7396
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1579184906157723e-06,
+      "loss": 0.0704,
+      "step": 7397
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1557635552610346e-06,
+      "loss": 0.0697,
+      "step": 7398
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.153609566425238e-06,
+      "loss": 0.0689,
+      "step": 7399
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1514565243682894e-06,
+      "loss": 0.0726,
+      "step": 7400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.149304429349991e-06,
+      "loss": 0.0762,
+      "step": 7401
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.14715328163002e-06,
+      "loss": 0.0716,
+      "step": 7402
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1450030814679436e-06,
+      "loss": 0.0744,
+      "step": 7403
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1428538291232128e-06,
+      "loss": 0.0721,
+      "step": 7404
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.140705524855167e-06,
+      "loss": 0.0668,
+      "step": 7405
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1385581689230262e-06,
+      "loss": 0.0754,
+      "step": 7406
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.136411761585906e-06,
+      "loss": 0.0741,
+      "step": 7407
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1342663031027967e-06,
+      "loss": 0.0697,
+      "step": 7408
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.132121793732577e-06,
+      "loss": 0.0693,
+      "step": 7409
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1299782337340193e-06,
+      "loss": 0.0708,
+      "step": 7410
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1278356233657682e-06,
+      "loss": 0.0692,
+      "step": 7411
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1256939628863583e-06,
+      "loss": 0.0711,
+      "step": 7412
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.123553252554219e-06,
+      "loss": 0.0716,
+      "step": 7413
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1214134926276543e-06,
+      "loss": 0.0621,
+      "step": 7414
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1192746833648526e-06,
+      "loss": 0.0678,
+      "step": 7415
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.117136825023898e-06,
+      "loss": 0.0707,
+      "step": 7416
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1149999178627544e-06,
+      "loss": 0.068,
+      "step": 7417
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.11286396213926e-06,
+      "loss": 0.0668,
+      "step": 7418
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.110728958111159e-06,
+      "loss": 0.069,
+      "step": 7419
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.0684,
+      "step": 7420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1064618061714816e-06,
+      "loss": 0.068,
+      "step": 7421
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1043296587748e-06,
+      "loss": 0.0686,
+      "step": 7422
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.102198464103297e-06,
+      "loss": 0.0693,
+      "step": 7423
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100068222414121e-06,
+      "loss": 0.0695,
+      "step": 7424
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.097938933964326e-06,
+      "loss": 0.0752,
+      "step": 7425
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.095810599010838e-06,
+      "loss": 0.0746,
+      "step": 7426
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.093683217810468e-06,
+      "loss": 0.0675,
+      "step": 7427
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0915567906199207e-06,
+      "loss": 0.0704,
+      "step": 7428
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.089431317695776e-06,
+      "loss": 0.0649,
+      "step": 7429
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.087306799294504e-06,
+      "loss": 0.0699,
+      "step": 7430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0851832356724567e-06,
+      "loss": 0.0749,
+      "step": 7431
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0830606270858735e-06,
+      "loss": 0.0743,
+      "step": 7432
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0809389737908735e-06,
+      "loss": 0.0732,
+      "step": 7433
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0788182760434707e-06,
+      "loss": 0.0713,
+      "step": 7434
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.076698534099555e-06,
+      "loss": 0.07,
+      "step": 7435
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0745797482149022e-06,
+      "loss": 0.0678,
+      "step": 7436
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0724619186451755e-06,
+      "loss": 0.0716,
+      "step": 7437
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0703450456459208e-06,
+      "loss": 0.067,
+      "step": 7438
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0682291294725652e-06,
+      "loss": 0.0703,
+      "step": 7439
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0661141703804312e-06,
+      "loss": 0.0769,
+      "step": 7440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.064000168624716e-06,
+      "loss": 0.0661,
+      "step": 7441
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.061887124460502e-06,
+      "loss": 0.0767,
+      "step": 7442
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0597750381427607e-06,
+      "loss": 0.0664,
+      "step": 7443
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.057663909926343e-06,
+      "loss": 0.0757,
+      "step": 7444
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0555537400659853e-06,
+      "loss": 0.0701,
+      "step": 7445
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.053444528816315e-06,
+      "loss": 0.069,
+      "step": 7446
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.051336276431837e-06,
+      "loss": 0.0703,
+      "step": 7447
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0492289831669366e-06,
+      "loss": 0.0758,
+      "step": 7448
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.047122649275898e-06,
+      "loss": 0.0699,
+      "step": 7449
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0450172750128728e-06,
+      "loss": 0.0638,
+      "step": 7450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.042912860631905e-06,
+      "loss": 0.0686,
+      "step": 7451
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0408094063869255e-06,
+      "loss": 0.0714,
+      "step": 7452
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.038706912531746e-06,
+      "loss": 0.0682,
+      "step": 7453
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0366053793200567e-06,
+      "loss": 0.0706,
+      "step": 7454
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0345048070054485e-06,
+      "loss": 0.0743,
+      "step": 7455
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0324051958413728e-06,
+      "loss": 0.0708,
+      "step": 7456
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0303065460811865e-06,
+      "loss": 0.0733,
+      "step": 7457
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.028208857978118e-06,
+      "loss": 0.0784,
+      "step": 7458
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.026112131785285e-06,
+      "loss": 0.0717,
+      "step": 7459
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0240163677556823e-06,
+      "loss": 0.0678,
+      "step": 7460
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.021921566142201e-06,
+      "loss": 0.0652,
+      "step": 7461
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.019827727197605e-06,
+      "loss": 0.0753,
+      "step": 7462
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0177348511745477e-06,
+      "loss": 0.0742,
+      "step": 7463
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0156429383255617e-06,
+      "loss": 0.0732,
+      "step": 7464
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0135519889030676e-06,
+      "loss": 0.0727,
+      "step": 7465
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0114620031593646e-06,
+      "loss": 0.0655,
+      "step": 7466
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.009372981346647e-06,
+      "loss": 0.0735,
+      "step": 7467
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0072849237169802e-06,
+      "loss": 0.0751,
+      "step": 7468
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0051978305223185e-06,
+      "loss": 0.0724,
+      "step": 7469
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0031117020144997e-06,
+      "loss": 0.0704,
+      "step": 7470
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.001026538445242e-06,
+      "loss": 0.0751,
+      "step": 7471
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9989423400661557e-06,
+      "loss": 0.0662,
+      "step": 7472
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.996859107128727e-06,
+      "loss": 0.0695,
+      "step": 7473
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.994776839884326e-06,
+      "loss": 0.0681,
+      "step": 7474
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9926955385842093e-06,
+      "loss": 0.0649,
+      "step": 7475
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9906152034795158e-06,
+      "loss": 0.0703,
+      "step": 7476
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9885358348212635e-06,
+      "loss": 0.0746,
+      "step": 7477
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.986457432860365e-06,
+      "loss": 0.0723,
+      "step": 7478
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.984379997847604e-06,
+      "loss": 0.0703,
+      "step": 7479
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9823035300336547e-06,
+      "loss": 0.0653,
+      "step": 7480
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9802280296690722e-06,
+      "loss": 0.0703,
+      "step": 7481
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9781534970042946e-06,
+      "loss": 0.0669,
+      "step": 7482
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.976079932289643e-06,
+      "loss": 0.0731,
+      "step": 7483
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974007335775324e-06,
+      "loss": 0.0804,
+      "step": 7484
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.971935707711428e-06,
+      "loss": 0.0626,
+      "step": 7485
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9698650483479233e-06,
+      "loss": 0.0672,
+      "step": 7486
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.967795357934662e-06,
+      "loss": 0.0772,
+      "step": 7487
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.96572663672139e-06,
+      "loss": 0.0691,
+      "step": 7488
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.963658884957719e-06,
+      "loss": 0.0626,
+      "step": 7489
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9615921028931585e-06,
+      "loss": 0.0711,
+      "step": 7490
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.959526290777094e-06,
+      "loss": 0.0764,
+      "step": 7491
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574614488587917e-06,
+      "loss": 0.0696,
+      "step": 7492
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.955397577387409e-06,
+      "loss": 0.0683,
+      "step": 7493
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.953334676611983e-06,
+      "loss": 0.077,
+      "step": 7494
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.951272746781424e-06,
+      "loss": 0.0685,
+      "step": 7495
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9492117881445393e-06,
+      "loss": 0.0678,
+      "step": 7496
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9471518009500125e-06,
+      "loss": 0.0661,
+      "step": 7497
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9450927854464065e-06,
+      "loss": 0.0683,
+      "step": 7498
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.943034741882176e-06,
+      "loss": 0.0707,
+      "step": 7499
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9409776705056514e-06,
+      "loss": 0.0715,
+      "step": 7500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.938921571565048e-06,
+      "loss": 0.0689,
+      "step": 7501
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9368664453084616e-06,
+      "loss": 0.0713,
+      "step": 7502
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.934812291983874e-06,
+      "loss": 0.0717,
+      "step": 7503
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9327591118391466e-06,
+      "loss": 0.0727,
+      "step": 7504
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9307069051220273e-06,
+      "loss": 0.0691,
+      "step": 7505
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.928655672080143e-06,
+      "loss": 0.0656,
+      "step": 7506
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9266054129610034e-06,
+      "loss": 0.069,
+      "step": 7507
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9245561280120027e-06,
+      "loss": 0.0703,
+      "step": 7508
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.922507817480417e-06,
+      "loss": 0.0726,
+      "step": 7509
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9204604816133986e-06,
+      "loss": 0.0747,
+      "step": 7510
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.918414120657995e-06,
+      "loss": 0.0729,
+      "step": 7511
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.916368734861126e-06,
+      "loss": 0.0692,
+      "step": 7512
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.914324324469594e-06,
+      "loss": 0.0723,
+      "step": 7513
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.912280889730095e-06,
+      "loss": 0.0673,
+      "step": 7514
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9102384308891875e-06,
+      "loss": 0.0654,
+      "step": 7515
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.908196948193327e-06,
+      "loss": 0.0716,
+      "step": 7516
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.90615644188885e-06,
+      "loss": 0.0693,
+      "step": 7517
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9041169122219727e-06,
+      "loss": 0.0682,
+      "step": 7518
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9020783594387882e-06,
+      "loss": 0.0696,
+      "step": 7519
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9000407837852851e-06,
+      "loss": 0.0701,
+      "step": 7520
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8980041855073206e-06,
+      "loss": 0.0702,
+      "step": 7521
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8959685648506365e-06,
+      "loss": 0.0712,
+      "step": 7522
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8939339220608655e-06,
+      "loss": 0.0688,
+      "step": 7523
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8919002573835143e-06,
+      "loss": 0.0715,
+      "step": 7524
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8898675710639714e-06,
+      "loss": 0.0703,
+      "step": 7525
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8878358633475125e-06,
+      "loss": 0.0752,
+      "step": 7526
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8858051344792938e-06,
+      "loss": 0.0691,
+      "step": 7527
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8837753847043438e-06,
+      "loss": 0.0697,
+      "step": 7528
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8817466142675877e-06,
+      "loss": 0.0724,
+      "step": 7529
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.879718823413823e-06,
+      "loss": 0.0671,
+      "step": 7530
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8776920123877307e-06,
+      "loss": 0.0778,
+      "step": 7531
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8756661814338773e-06,
+      "loss": 0.0741,
+      "step": 7532
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8736413307967094e-06,
+      "loss": 0.0716,
+      "step": 7533
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8716174607205462e-06,
+      "loss": 0.075,
+      "step": 7534
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8695945714496034e-06,
+      "loss": 0.0605,
+      "step": 7535
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.86757266322797e-06,
+      "loss": 0.065,
+      "step": 7536
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8655517362996157e-06,
+      "loss": 0.0688,
+      "step": 7537
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8635317909083983e-06,
+      "loss": 0.071,
+      "step": 7538
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.861512827298051e-06,
+      "loss": 0.0704,
+      "step": 7539
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8594948457121897e-06,
+      "loss": 0.071,
+      "step": 7540
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8574778463943145e-06,
+      "loss": 0.0707,
+      "step": 7541
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8554618295878023e-06,
+      "loss": 0.0683,
+      "step": 7542
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8534467955359147e-06,
+      "loss": 0.0713,
+      "step": 7543
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.851432744481797e-06,
+      "loss": 0.0751,
+      "step": 7544
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8494196766684725e-06,
+      "loss": 0.0709,
+      "step": 7545
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8474075923388446e-06,
+      "loss": 0.0687,
+      "step": 7546
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8453964917357015e-06,
+      "loss": 0.0721,
+      "step": 7547
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.843386375101711e-06,
+      "loss": 0.0675,
+      "step": 7548
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8413772426794196e-06,
+      "loss": 0.0779,
+      "step": 7549
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8393690947112619e-06,
+      "loss": 0.0746,
+      "step": 7550
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8373619314395485e-06,
+      "loss": 0.069,
+      "step": 7551
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8353557531064682e-06,
+      "loss": 0.0685,
+      "step": 7552
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8333505599541034e-06,
+      "loss": 0.0772,
+      "step": 7553
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8313463522244001e-06,
+      "loss": 0.0774,
+      "step": 7554
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8293431301592013e-06,
+      "loss": 0.0675,
+      "step": 7555
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8273408940002202e-06,
+      "loss": 0.0683,
+      "step": 7556
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.825339643989058e-06,
+      "loss": 0.0713,
+      "step": 7557
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8233393803671895e-06,
+      "loss": 0.066,
+      "step": 7558
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8213401033759837e-06,
+      "loss": 0.07,
+      "step": 7559
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.819341813256671e-06,
+      "loss": 0.0705,
+      "step": 7560
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8173445102503807e-06,
+      "loss": 0.0738,
+      "step": 7561
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8153481945981144e-06,
+      "loss": 0.0674,
+      "step": 7562
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8133528665407562e-06,
+      "loss": 0.0721,
+      "step": 7563
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.811358526319068e-06,
+      "loss": 0.0745,
+      "step": 7564
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8093651741736994e-06,
+      "loss": 0.0695,
+      "step": 7565
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.807372810345176e-06,
+      "loss": 0.0751,
+      "step": 7566
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.805381435073904e-06,
+      "loss": 0.0705,
+      "step": 7567
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8033910486001704e-06,
+      "loss": 0.0623,
+      "step": 7568
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.801401651164143e-06,
+      "loss": 0.0679,
+      "step": 7569
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7994132430058741e-06,
+      "loss": 0.0704,
+      "step": 7570
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7974258243652931e-06,
+      "loss": 0.0704,
+      "step": 7571
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.795439395482208e-06,
+      "loss": 0.0709,
+      "step": 7572
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7934539565963115e-06,
+      "loss": 0.0746,
+      "step": 7573
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7914695079471744e-06,
+      "loss": 0.0667,
+      "step": 7574
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.789486049774246e-06,
+      "loss": 0.0734,
+      "step": 7575
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7875035823168641e-06,
+      "loss": 0.07,
+      "step": 7576
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7855221058142402e-06,
+      "loss": 0.0689,
+      "step": 7577
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7835416205054656e-06,
+      "loss": 0.078,
+      "step": 7578
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7815621266295158e-06,
+      "loss": 0.0721,
+      "step": 7579
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.779583624425244e-06,
+      "loss": 0.0714,
+      "step": 7580
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7776061141313827e-06,
+      "loss": 0.0721,
+      "step": 7581
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.775629595986551e-06,
+      "loss": 0.0683,
+      "step": 7582
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7736540702292426e-06,
+      "loss": 0.069,
+      "step": 7583
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7716795370978311e-06,
+      "loss": 0.0684,
+      "step": 7584
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7697059968305752e-06,
+      "loss": 0.0703,
+      "step": 7585
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7677334496656073e-06,
+      "loss": 0.069,
+      "step": 7586
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7657618958409428e-06,
+      "loss": 0.0725,
+      "step": 7587
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7637913355944814e-06,
+      "loss": 0.0742,
+      "step": 7588
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7618217691639982e-06,
+      "loss": 0.0696,
+      "step": 7589
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7598531967871468e-06,
+      "loss": 0.0737,
+      "step": 7590
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7578856187014691e-06,
+      "loss": 0.0667,
+      "step": 7591
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7559190351443767e-06,
+      "loss": 0.0697,
+      "step": 7592
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7539534463531638e-06,
+      "loss": 0.0737,
+      "step": 7593
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.751988852565013e-06,
+      "loss": 0.0658,
+      "step": 7594
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7500252540169782e-06,
+      "loss": 0.0751,
+      "step": 7595
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7480626509459919e-06,
+      "loss": 0.0754,
+      "step": 7596
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7461010435888749e-06,
+      "loss": 0.0733,
+      "step": 7597
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7441404321823252e-06,
+      "loss": 0.0689,
+      "step": 7598
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7421808169629084e-06,
+      "loss": 0.0769,
+      "step": 7599
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7402221981670897e-06,
+      "loss": 0.0779,
+      "step": 7600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.738264576031201e-06,
+      "loss": 0.073,
+      "step": 7601
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7363079507914537e-06,
+      "loss": 0.0758,
+      "step": 7602
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7343523226839498e-06,
+      "loss": 0.0734,
+      "step": 7603
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7323976919446617e-06,
+      "loss": 0.0754,
+      "step": 7604
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7304440588094373e-06,
+      "loss": 0.0674,
+      "step": 7605
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.728491423514017e-06,
+      "loss": 0.0675,
+      "step": 7606
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7265397862940126e-06,
+      "loss": 0.0757,
+      "step": 7607
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.724589147384913e-06,
+      "loss": 0.0702,
+      "step": 7608
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7226395070220958e-06,
+      "loss": 0.0714,
+      "step": 7609
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7206908654408117e-06,
+      "loss": 0.0672,
+      "step": 7610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7187432228761902e-06,
+      "loss": 0.0705,
+      "step": 7611
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7167965795632447e-06,
+      "loss": 0.0718,
+      "step": 7612
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7148509357368626e-06,
+      "loss": 0.0696,
+      "step": 7613
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.712906291631814e-06,
+      "loss": 0.0689,
+      "step": 7614
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.71096264748275e-06,
+      "loss": 0.0697,
+      "step": 7615
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7090200035241977e-06,
+      "loss": 0.0674,
+      "step": 7616
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.707078359990566e-06,
+      "loss": 0.0685,
+      "step": 7617
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7051377171161398e-06,
+      "loss": 0.0685,
+      "step": 7618
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7031980751350873e-06,
+      "loss": 0.069,
+      "step": 7619
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7012594342814493e-06,
+      "loss": 0.0707,
+      "step": 7620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6993217947891571e-06,
+      "loss": 0.0696,
+      "step": 7621
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6973851568920119e-06,
+      "loss": 0.0721,
+      "step": 7622
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6954495208236944e-06,
+      "loss": 0.0667,
+      "step": 7623
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.693514886817772e-06,
+      "loss": 0.0724,
+      "step": 7624
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6915812551076815e-06,
+      "loss": 0.0708,
+      "step": 7625
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6896486259267408e-06,
+      "loss": 0.0671,
+      "step": 7626
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6877169995081544e-06,
+      "loss": 0.0677,
+      "step": 7627
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.685786376084999e-06,
+      "loss": 0.0717,
+      "step": 7628
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6838567558902297e-06,
+      "loss": 0.0731,
+      "step": 7629
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6819281391566889e-06,
+      "loss": 0.0713,
+      "step": 7630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.680000526117085e-06,
+      "loss": 0.0742,
+      "step": 7631
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6780739170040105e-06,
+      "loss": 0.0705,
+      "step": 7632
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6761483120499456e-06,
+      "loss": 0.0643,
+      "step": 7633
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6742237114872385e-06,
+      "loss": 0.084,
+      "step": 7634
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6723001155481167e-06,
+      "loss": 0.0706,
+      "step": 7635
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6703775244646947e-06,
+      "loss": 0.0699,
+      "step": 7636
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6684559384689581e-06,
+      "loss": 0.0668,
+      "step": 7637
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6665353577927744e-06,
+      "loss": 0.0703,
+      "step": 7638
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6646157826678878e-06,
+      "loss": 0.0706,
+      "step": 7639
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6626972133259223e-06,
+      "loss": 0.0614,
+      "step": 7640
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6607796499983798e-06,
+      "loss": 0.0665,
+      "step": 7641
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.658863092916645e-06,
+      "loss": 0.0756,
+      "step": 7642
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6569475423119763e-06,
+      "loss": 0.0696,
+      "step": 7643
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.655032998415511e-06,
+      "loss": 0.0729,
+      "step": 7644
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.653119461458268e-06,
+      "loss": 0.0672,
+      "step": 7645
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.651206931671141e-06,
+      "loss": 0.0703,
+      "step": 7646
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6492954092849033e-06,
+      "loss": 0.075,
+      "step": 7647
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6473848945302095e-06,
+      "loss": 0.0651,
+      "step": 7648
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6454753876375917e-06,
+      "loss": 0.074,
+      "step": 7649
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6435668888374557e-06,
+      "loss": 0.0724,
+      "step": 7650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6416593983600904e-06,
+      "loss": 0.0641,
+      "step": 7651
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6397529164356606e-06,
+      "loss": 0.0745,
+      "step": 7652
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6378474432942127e-06,
+      "loss": 0.0708,
+      "step": 7653
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6359429791656689e-06,
+      "loss": 0.0717,
+      "step": 7654
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6340395242798302e-06,
+      "loss": 0.0674,
+      "step": 7655
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6321370788663737e-06,
+      "loss": 0.0697,
+      "step": 7656
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6302356431548572e-06,
+      "loss": 0.0683,
+      "step": 7657
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6283352173747148e-06,
+      "loss": 0.0699,
+      "step": 7658
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6264358017552628e-06,
+      "loss": 0.0785,
+      "step": 7659
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6245373965256917e-06,
+      "loss": 0.0672,
+      "step": 7660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6226400019150712e-06,
+      "loss": 0.064,
+      "step": 7661
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6207436181523461e-06,
+      "loss": 0.0744,
+      "step": 7662
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6188482454663501e-06,
+      "loss": 0.0655,
+      "step": 7663
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6169538840857768e-06,
+      "loss": 0.0709,
+      "step": 7664
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6150605342392146e-06,
+      "loss": 0.069,
+      "step": 7665
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6131681961551215e-06,
+      "loss": 0.0664,
+      "step": 7666
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.611276870061832e-06,
+      "loss": 0.0636,
+      "step": 7667
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6093865561875665e-06,
+      "loss": 0.0709,
+      "step": 7668
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6074972547604194e-06,
+      "loss": 0.0699,
+      "step": 7669
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6056089660083529e-06,
+      "loss": 0.0651,
+      "step": 7670
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6037216901592245e-06,
+      "loss": 0.0718,
+      "step": 7671
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6018354274407588e-06,
+      "loss": 0.0712,
+      "step": 7672
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.599950178080556e-06,
+      "loss": 0.0725,
+      "step": 7673
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5980659423061051e-06,
+      "loss": 0.068,
+      "step": 7674
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5961827203447622e-06,
+      "loss": 0.0709,
+      "step": 7675
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5943005124237654e-06,
+      "loss": 0.0693,
+      "step": 7676
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5924193187702308e-06,
+      "loss": 0.0754,
+      "step": 7677
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5905391396111502e-06,
+      "loss": 0.0619,
+      "step": 7678
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5886599751733911e-06,
+      "loss": 0.0677,
+      "step": 7679
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5867818256837074e-06,
+      "loss": 0.0703,
+      "step": 7680
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5849046913687215e-06,
+      "loss": 0.0706,
+      "step": 7681
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5830285724549376e-06,
+      "loss": 0.0716,
+      "step": 7682
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5811534691687348e-06,
+      "loss": 0.0741,
+      "step": 7683
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5792793817363716e-06,
+      "loss": 0.0651,
+      "step": 7684
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.577406310383981e-06,
+      "loss": 0.0653,
+      "step": 7685
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5755342553375797e-06,
+      "loss": 0.0702,
+      "step": 7686
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.573663216823057e-06,
+      "loss": 0.0679,
+      "step": 7687
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.571793195066177e-06,
+      "loss": 0.0685,
+      "step": 7688
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5699241902925921e-06,
+      "loss": 0.0683,
+      "step": 7689
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5680562027278156e-06,
+      "loss": 0.0665,
+      "step": 7690
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5661892325972493e-06,
+      "loss": 0.0639,
+      "step": 7691
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5643232801261731e-06,
+      "loss": 0.0743,
+      "step": 7692
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.562458345539739e-06,
+      "loss": 0.0727,
+      "step": 7693
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.560594429062975e-06,
+      "loss": 0.0657,
+      "step": 7694
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5587315309207973e-06,
+      "loss": 0.0617,
+      "step": 7695
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5568696513379822e-06,
+      "loss": 0.0656,
+      "step": 7696
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.555008790539193e-06,
+      "loss": 0.0734,
+      "step": 7697
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5531489487489748e-06,
+      "loss": 0.0681,
+      "step": 7698
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5512901261917401e-06,
+      "loss": 0.066,
+      "step": 7699
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5494323230917807e-06,
+      "loss": 0.0705,
+      "step": 7700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5475755396732706e-06,
+      "loss": 0.0693,
+      "step": 7701
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5457197761602593e-06,
+      "loss": 0.0668,
+      "step": 7702
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5438650327766635e-06,
+      "loss": 0.0696,
+      "step": 7703
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5420113097462897e-06,
+      "loss": 0.0711,
+      "step": 7704
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5401586072928165e-06,
+      "loss": 0.0756,
+      "step": 7705
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5383069256397932e-06,
+      "loss": 0.0669,
+      "step": 7706
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5364562650106596e-06,
+      "loss": 0.0653,
+      "step": 7707
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5346066256287217e-06,
+      "loss": 0.0748,
+      "step": 7708
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.0739,
+      "step": 7709
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5309104114990403e-06,
+      "loss": 0.0679,
+      "step": 7710
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5290638371973033e-06,
+      "loss": 0.0672,
+      "step": 7711
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5272182850347583e-06,
+      "loss": 0.0716,
+      "step": 7712
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5253737552341053e-06,
+      "loss": 0.0704,
+      "step": 7713
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5235302480179092e-06,
+      "loss": 0.0698,
+      "step": 7714
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5216877636086158e-06,
+      "loss": 0.0697,
+      "step": 7715
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.519846302228547e-06,
+      "loss": 0.071,
+      "step": 7716
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5180058640999008e-06,
+      "loss": 0.071,
+      "step": 7717
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5161664494447503e-06,
+      "loss": 0.0707,
+      "step": 7718
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5143280584850528e-06,
+      "loss": 0.0693,
+      "step": 7719
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5124906914426329e-06,
+      "loss": 0.0678,
+      "step": 7720
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5106543485391955e-06,
+      "loss": 0.0713,
+      "step": 7721
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5088190299963212e-06,
+      "loss": 0.0601,
+      "step": 7722
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5069847360354672e-06,
+      "loss": 0.0768,
+      "step": 7723
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5051514668779665e-06,
+      "loss": 0.0643,
+      "step": 7724
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.503319222745032e-06,
+      "loss": 0.0699,
+      "step": 7725
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5014880038577485e-06,
+      "loss": 0.0739,
+      "step": 7726
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4996578104370752e-06,
+      "loss": 0.0744,
+      "step": 7727
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4978286427038602e-06,
+      "loss": 0.0664,
+      "step": 7728
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4960005008788103e-06,
+      "loss": 0.0705,
+      "step": 7729
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.494173385182517e-06,
+      "loss": 0.0713,
+      "step": 7730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4923472958354523e-06,
+      "loss": 0.0737,
+      "step": 7731
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4905222330579583e-06,
+      "loss": 0.0741,
+      "step": 7732
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4886981970702507e-06,
+      "loss": 0.0772,
+      "step": 7733
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.486875188092435e-06,
+      "loss": 0.0716,
+      "step": 7734
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4850532063444723e-06,
+      "loss": 0.0731,
+      "step": 7735
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4832322520462183e-06,
+      "loss": 0.0672,
+      "step": 7736
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.481412325417395e-06,
+      "loss": 0.0715,
+      "step": 7737
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4795934266776012e-06,
+      "loss": 0.0751,
+      "step": 7738
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4777755560463114e-06,
+      "loss": 0.0703,
+      "step": 7739
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4759587137428822e-06,
+      "loss": 0.0683,
+      "step": 7740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4741428999865404e-06,
+      "loss": 0.0782,
+      "step": 7741
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4723281149963875e-06,
+      "loss": 0.07,
+      "step": 7742
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.470514358991405e-06,
+      "loss": 0.0669,
+      "step": 7743
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.468701632190447e-06,
+      "loss": 0.0703,
+      "step": 7744
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4668899348122434e-06,
+      "loss": 0.0677,
+      "step": 7745
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4650792670754065e-06,
+      "loss": 0.0684,
+      "step": 7746
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.463269629198416e-06,
+      "loss": 0.0756,
+      "step": 7747
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4614610213996306e-06,
+      "loss": 0.077,
+      "step": 7748
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4596534438972853e-06,
+      "loss": 0.0776,
+      "step": 7749
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.457846896909486e-06,
+      "loss": 0.0683,
+      "step": 7750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.456041380654225e-06,
+      "loss": 0.0661,
+      "step": 7751
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4542368953493601e-06,
+      "loss": 0.069,
+      "step": 7752
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4524334412126285e-06,
+      "loss": 0.0745,
+      "step": 7753
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4506310184616435e-06,
+      "loss": 0.0683,
+      "step": 7754
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4488296273138914e-06,
+      "loss": 0.0676,
+      "step": 7755
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4470292679867348e-06,
+      "loss": 0.0697,
+      "step": 7756
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4452299406974157e-06,
+      "loss": 0.0688,
+      "step": 7757
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4434316456630482e-06,
+      "loss": 0.0672,
+      "step": 7758
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4416343831006207e-06,
+      "loss": 0.0687,
+      "step": 7759
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4398381532270001e-06,
+      "loss": 0.0682,
+      "step": 7760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.438042956258926e-06,
+      "loss": 0.0738,
+      "step": 7761
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4362487924130131e-06,
+      "loss": 0.0673,
+      "step": 7762
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4344556619057558e-06,
+      "loss": 0.0735,
+      "step": 7763
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4326635649535203e-06,
+      "loss": 0.0712,
+      "step": 7764
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4308725017725445e-06,
+      "loss": 0.0709,
+      "step": 7765
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4290824725789542e-06,
+      "loss": 0.0696,
+      "step": 7766
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.427293477588736e-06,
+      "loss": 0.0673,
+      "step": 7767
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4255055170177546e-06,
+      "loss": 0.0718,
+      "step": 7768
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4237185910817597e-06,
+      "loss": 0.0691,
+      "step": 7769
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4219326999963668e-06,
+      "loss": 0.0707,
+      "step": 7770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4201478439770656e-06,
+      "loss": 0.0651,
+      "step": 7771
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4183640232392304e-06,
+      "loss": 0.0765,
+      "step": 7772
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4165812379981048e-06,
+      "loss": 0.0777,
+      "step": 7773
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4147994884687987e-06,
+      "loss": 0.0681,
+      "step": 7774
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4130187748663148e-06,
+      "loss": 0.0725,
+      "step": 7775
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4112390974055168e-06,
+      "loss": 0.0736,
+      "step": 7776
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.409460456301147e-06,
+      "loss": 0.0707,
+      "step": 7777
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.40768285176783e-06,
+      "loss": 0.068,
+      "step": 7778
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4059062840200555e-06,
+      "loss": 0.0686,
+      "step": 7779
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4041307532721882e-06,
+      "loss": 0.0698,
+      "step": 7780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4023562597384765e-06,
+      "loss": 0.0757,
+      "step": 7781
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.400582803633036e-06,
+      "loss": 0.0673,
+      "step": 7782
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3988103851698576e-06,
+      "loss": 0.0721,
+      "step": 7783
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3970390045628135e-06,
+      "loss": 0.0736,
+      "step": 7784
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.395268662025643e-06,
+      "loss": 0.0678,
+      "step": 7785
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3934993577719646e-06,
+      "loss": 0.0677,
+      "step": 7786
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3917310920152683e-06,
+      "loss": 0.0698,
+      "step": 7787
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3899638649689219e-06,
+      "loss": 0.0671,
+      "step": 7788
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3881976768461636e-06,
+      "loss": 0.0712,
+      "step": 7789
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3864325278601142e-06,
+      "loss": 0.0665,
+      "step": 7790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3846684182237614e-06,
+      "loss": 0.069,
+      "step": 7791
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.382905348149971e-06,
+      "loss": 0.0683,
+      "step": 7792
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.381143317851481e-06,
+      "loss": 0.0704,
+      "step": 7793
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3793823275409068e-06,
+      "loss": 0.0721,
+      "step": 7794
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.377622377430734e-06,
+      "loss": 0.0688,
+      "step": 7795
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3758634677333304e-06,
+      "loss": 0.0687,
+      "step": 7796
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3741055986609308e-06,
+      "loss": 0.0726,
+      "step": 7797
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3723487704256466e-06,
+      "loss": 0.0672,
+      "step": 7798
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3705929832394693e-06,
+      "loss": 0.0653,
+      "step": 7799
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3688382373142529e-06,
+      "loss": 0.0718,
+      "step": 7800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3670845328617332e-06,
+      "loss": 0.0743,
+      "step": 7801
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3653318700935237e-06,
+      "loss": 0.0677,
+      "step": 7802
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3635802492211059e-06,
+      "loss": 0.0689,
+      "step": 7803
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3618296704558364e-06,
+      "loss": 0.0654,
+      "step": 7804
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3600801340089532e-06,
+      "loss": 0.0678,
+      "step": 7805
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3583316400915568e-06,
+      "loss": 0.0696,
+      "step": 7806
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.356584188914627e-06,
+      "loss": 0.0677,
+      "step": 7807
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3548377806890244e-06,
+      "loss": 0.0773,
+      "step": 7808
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3530924156254742e-06,
+      "loss": 0.07,
+      "step": 7809
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.351348093934579e-06,
+      "loss": 0.0733,
+      "step": 7810
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3496048158268204e-06,
+      "loss": 0.0693,
+      "step": 7811
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3478625815125468e-06,
+      "loss": 0.069,
+      "step": 7812
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3461213912019832e-06,
+      "loss": 0.0667,
+      "step": 7813
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3443812451052297e-06,
+      "loss": 0.0689,
+      "step": 7814
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3426421434322589e-06,
+      "loss": 0.0721,
+      "step": 7815
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.340904086392918e-06,
+      "loss": 0.0716,
+      "step": 7816
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.339167074196931e-06,
+      "loss": 0.0702,
+      "step": 7817
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3374311070538914e-06,
+      "loss": 0.0654,
+      "step": 7818
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3356961851732686e-06,
+      "loss": 0.0717,
+      "step": 7819
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.333962308764405e-06,
+      "loss": 0.0633,
+      "step": 7820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3322294780365187e-06,
+      "loss": 0.0695,
+      "step": 7821
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3304976931986968e-06,
+      "loss": 0.0677,
+      "step": 7822
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3287669544599091e-06,
+      "loss": 0.0681,
+      "step": 7823
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.327037262028993e-06,
+      "loss": 0.0688,
+      "step": 7824
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3253086161146577e-06,
+      "loss": 0.0654,
+      "step": 7825
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3235810169254903e-06,
+      "loss": 0.0721,
+      "step": 7826
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3218544646699505e-06,
+      "loss": 0.0706,
+      "step": 7827
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3201289595563693e-06,
+      "loss": 0.0707,
+      "step": 7828
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.318404501792958e-06,
+      "loss": 0.0698,
+      "step": 7829
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3166810915877936e-06,
+      "loss": 0.0729,
+      "step": 7830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.314958729148832e-06,
+      "loss": 0.0738,
+      "step": 7831
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3132374146838988e-06,
+      "loss": 0.0675,
+      "step": 7832
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.311517148400695e-06,
+      "loss": 0.0711,
+      "step": 7833
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3097979305067977e-06,
+      "loss": 0.0664,
+      "step": 7834
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.308079761209654e-06,
+      "loss": 0.0676,
+      "step": 7835
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3063626407165853e-06,
+      "loss": 0.0781,
+      "step": 7836
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3046465692347842e-06,
+      "loss": 0.0786,
+      "step": 7837
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3029315469713267e-06,
+      "loss": 0.0662,
+      "step": 7838
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3012175741331446e-06,
+      "loss": 0.0707,
+      "step": 7839
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2995046509270593e-06,
+      "loss": 0.0712,
+      "step": 7840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2977927775597599e-06,
+      "loss": 0.067,
+      "step": 7841
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2960819542378055e-06,
+      "loss": 0.0677,
+      "step": 7842
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2943721811676302e-06,
+      "loss": 0.077,
+      "step": 7843
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2926634585555498e-06,
+      "loss": 0.0771,
+      "step": 7844
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2909557866077361e-06,
+      "loss": 0.0715,
+      "step": 7845
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.289249165530252e-06,
+      "loss": 0.0659,
+      "step": 7846
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2875435955290216e-06,
+      "loss": 0.063,
+      "step": 7847
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.285839076809845e-06,
+      "loss": 0.0701,
+      "step": 7848
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.284135609578402e-06,
+      "loss": 0.0729,
+      "step": 7849
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2824331940402367e-06,
+      "loss": 0.0653,
+      "step": 7850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.28073183040077e-06,
+      "loss": 0.0704,
+      "step": 7851
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2790315188652958e-06,
+      "loss": 0.0745,
+      "step": 7852
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2773322596389814e-06,
+      "loss": 0.0717,
+      "step": 7853
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.275634052926864e-06,
+      "loss": 0.0701,
+      "step": 7854
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2739368989338608e-06,
+      "loss": 0.0667,
+      "step": 7855
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2722407978647556e-06,
+      "loss": 0.0671,
+      "step": 7856
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.270545749924207e-06,
+      "loss": 0.0638,
+      "step": 7857
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2688517553167467e-06,
+      "loss": 0.0655,
+      "step": 7858
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2671588142467795e-06,
+      "loss": 0.0731,
+      "step": 7859
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2654669269185804e-06,
+      "loss": 0.0655,
+      "step": 7860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2637760935363053e-06,
+      "loss": 0.0712,
+      "step": 7861
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.262086314303973e-06,
+      "loss": 0.0616,
+      "step": 7862
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2603975894254783e-06,
+      "loss": 0.0707,
+      "step": 7863
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2587099191045971e-06,
+      "loss": 0.069,
+      "step": 7864
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2570233035449642e-06,
+      "loss": 0.0751,
+      "step": 7865
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2553377429500923e-06,
+      "loss": 0.0689,
+      "step": 7866
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2536532375233745e-06,
+      "loss": 0.0669,
+      "step": 7867
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.251969787468068e-06,
+      "loss": 0.0739,
+      "step": 7868
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2502873929873015e-06,
+      "loss": 0.0694,
+      "step": 7869
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2486060542840882e-06,
+      "loss": 0.0711,
+      "step": 7870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.246925771561297e-06,
+      "loss": 0.0775,
+      "step": 7871
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.24524654502168e-06,
+      "loss": 0.0692,
+      "step": 7872
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2435683748678629e-06,
+      "loss": 0.0676,
+      "step": 7873
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2418912613023392e-06,
+      "loss": 0.0744,
+      "step": 7874
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.240215204527474e-06,
+      "loss": 0.0763,
+      "step": 7875
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2385402047455131e-06,
+      "loss": 0.07,
+      "step": 7876
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2368662621585669e-06,
+      "loss": 0.0653,
+      "step": 7877
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2351933769686154e-06,
+      "loss": 0.0698,
+      "step": 7878
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.233521549377522e-06,
+      "loss": 0.0692,
+      "step": 7879
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2318507795870138e-06,
+      "loss": 0.0765,
+      "step": 7880
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2301810677986925e-06,
+      "loss": 0.0742,
+      "step": 7881
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2285124142140358e-06,
+      "loss": 0.0634,
+      "step": 7882
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2268448190343918e-06,
+      "loss": 0.0676,
+      "step": 7883
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2251782824609703e-06,
+      "loss": 0.0724,
+      "step": 7884
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2235128046948718e-06,
+      "loss": 0.067,
+      "step": 7885
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2218483859370577e-06,
+      "loss": 0.063,
+      "step": 7886
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2201850263883609e-06,
+      "loss": 0.0646,
+      "step": 7887
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2185227262494926e-06,
+      "loss": 0.0736,
+      "step": 7888
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2168614857210327e-06,
+      "loss": 0.0671,
+      "step": 7889
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2152013050034328e-06,
+      "loss": 0.0633,
+      "step": 7890
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2135421842970174e-06,
+      "loss": 0.0685,
+      "step": 7891
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.211884123801984e-06,
+      "loss": 0.0768,
+      "step": 7892
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2102271237183971e-06,
+      "loss": 0.0731,
+      "step": 7893
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2085711842462033e-06,
+      "loss": 0.0683,
+      "step": 7894
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.206916305585213e-06,
+      "loss": 0.071,
+      "step": 7895
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.0702,
+      "step": 7896
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.203609731495452e-06,
+      "loss": 0.0703,
+      "step": 7897
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2019580364656669e-06,
+      "loss": 0.0742,
+      "step": 7898
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2003074030450535e-06,
+      "loss": 0.0662,
+      "step": 7899
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.198657831432788e-06,
+      "loss": 0.0674,
+      "step": 7900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1970093218279133e-06,
+      "loss": 0.063,
+      "step": 7901
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.195361874429345e-06,
+      "loss": 0.0692,
+      "step": 7902
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1937154894358716e-06,
+      "loss": 0.0692,
+      "step": 7903
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1920701670461532e-06,
+      "loss": 0.0717,
+      "step": 7904
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1904259074587176e-06,
+      "loss": 0.07,
+      "step": 7905
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1887827108719752e-06,
+      "loss": 0.0725,
+      "step": 7906
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.187140577484196e-06,
+      "loss": 0.0723,
+      "step": 7907
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.185499507493526e-06,
+      "loss": 0.0694,
+      "step": 7908
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1838595010979902e-06,
+      "loss": 0.0656,
+      "step": 7909
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1822205584954716e-06,
+      "loss": 0.0661,
+      "step": 7910
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.180582679883735e-06,
+      "loss": 0.0742,
+      "step": 7911
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1789458654604146e-06,
+      "loss": 0.0681,
+      "step": 7912
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1773101154230139e-06,
+      "loss": 0.0761,
+      "step": 7913
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1756754299689078e-06,
+      "loss": 0.0692,
+      "step": 7914
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.174041809295351e-06,
+      "loss": 0.0667,
+      "step": 7915
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1724092535994547e-06,
+      "loss": 0.0747,
+      "step": 7916
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1707777630782159e-06,
+      "loss": 0.0787,
+      "step": 7917
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1691473379284945e-06,
+      "loss": 0.0693,
+      "step": 7918
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1675179783470258e-06,
+      "loss": 0.069,
+      "step": 7919
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1658896845304124e-06,
+      "loss": 0.0635,
+      "step": 7920
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1642624566751359e-06,
+      "loss": 0.0731,
+      "step": 7921
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1626362949775428e-06,
+      "loss": 0.0707,
+      "step": 7922
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1610111996338513e-06,
+      "loss": 0.0688,
+      "step": 7923
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1593871708401528e-06,
+      "loss": 0.0673,
+      "step": 7924
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.15776420879241e-06,
+      "loss": 0.0698,
+      "step": 7925
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1561423136864547e-06,
+      "loss": 0.0698,
+      "step": 7926
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1545214857179943e-06,
+      "loss": 0.066,
+      "step": 7927
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1529017250826035e-06,
+      "loss": 0.0733,
+      "step": 7928
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1512830319757306e-06,
+      "loss": 0.0685,
+      "step": 7929
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1496654065926927e-06,
+      "loss": 0.0692,
+      "step": 7930
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1480488491286767e-06,
+      "loss": 0.0651,
+      "step": 7931
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1464333597787491e-06,
+      "loss": 0.0695,
+      "step": 7932
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1448189387378395e-06,
+      "loss": 0.0799,
+      "step": 7933
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1432055862007496e-06,
+      "loss": 0.0687,
+      "step": 7934
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1415933023621539e-06,
+      "loss": 0.0656,
+      "step": 7935
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.139982087416598e-06,
+      "loss": 0.0703,
+      "step": 7936
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.138371941558495e-06,
+      "loss": 0.0673,
+      "step": 7937
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.136762864982137e-06,
+      "loss": 0.069,
+      "step": 7938
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1351548578816795e-06,
+      "loss": 0.0716,
+      "step": 7939
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.13354792045115e-06,
+      "loss": 0.0703,
+      "step": 7940
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1319420528844517e-06,
+      "loss": 0.0609,
+      "step": 7941
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1303372553753522e-06,
+      "loss": 0.071,
+      "step": 7942
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1287335281174938e-06,
+      "loss": 0.0717,
+      "step": 7943
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1271308713043904e-06,
+      "loss": 0.0733,
+      "step": 7944
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1255292851294263e-06,
+      "loss": 0.0729,
+      "step": 7945
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1239287697858513e-06,
+      "loss": 0.071,
+      "step": 7946
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1223293254667954e-06,
+      "loss": 0.0676,
+      "step": 7947
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1207309523652543e-06,
+      "loss": 0.0783,
+      "step": 7948
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.119133650674088e-06,
+      "loss": 0.0726,
+      "step": 7949
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1175374205860412e-06,
+      "loss": 0.0677,
+      "step": 7950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1159422622937178e-06,
+      "loss": 0.0682,
+      "step": 7951
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1143481759895958e-06,
+      "loss": 0.0736,
+      "step": 7952
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1127551618660281e-06,
+      "loss": 0.0692,
+      "step": 7953
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1111632201152344e-06,
+      "loss": 0.0674,
+      "step": 7954
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.109572350929299e-06,
+      "loss": 0.0718,
+      "step": 7955
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1079825545001887e-06,
+      "loss": 0.0691,
+      "step": 7956
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1063938310197342e-06,
+      "loss": 0.0661,
+      "step": 7957
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.104806180679635e-06,
+      "loss": 0.0707,
+      "step": 7958
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1032196036714682e-06,
+      "loss": 0.0677,
+      "step": 7959
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1016341001866738e-06,
+      "loss": 0.0676,
+      "step": 7960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1000496704165664e-06,
+      "loss": 0.0678,
+      "step": 7961
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.09846631455233e-06,
+      "loss": 0.0711,
+      "step": 7962
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0968840327850184e-06,
+      "loss": 0.0735,
+      "step": 7963
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0953028253055541e-06,
+      "loss": 0.0782,
+      "step": 7964
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0937226923047385e-06,
+      "loss": 0.067,
+      "step": 7965
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0921436339732327e-06,
+      "loss": 0.0651,
+      "step": 7966
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0905656505015738e-06,
+      "loss": 0.0685,
+      "step": 7967
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0889887420801671e-06,
+      "loss": 0.0687,
+      "step": 7968
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0874129088992913e-06,
+      "loss": 0.0712,
+      "step": 7969
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0858381511490878e-06,
+      "loss": 0.0686,
+      "step": 7970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0842644690195814e-06,
+      "loss": 0.0671,
+      "step": 7971
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0826918627006544e-06,
+      "loss": 0.0626,
+      "step": 7972
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0811203323820629e-06,
+      "loss": 0.0724,
+      "step": 7973
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0795498782534398e-06,
+      "loss": 0.0682,
+      "step": 7974
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0779805005042786e-06,
+      "loss": 0.0773,
+      "step": 7975
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0764121993239463e-06,
+      "loss": 0.0738,
+      "step": 7976
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0748449749016832e-06,
+      "loss": 0.0679,
+      "step": 7977
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.073278827426598e-06,
+      "loss": 0.069,
+      "step": 7978
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0717137570876646e-06,
+      "loss": 0.0654,
+      "step": 7979
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0701497640737378e-06,
+      "loss": 0.0706,
+      "step": 7980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0685868485735296e-06,
+      "loss": 0.0694,
+      "step": 7981
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0670250107756285e-06,
+      "loss": 0.074,
+      "step": 7982
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0654642508684954e-06,
+      "loss": 0.0677,
+      "step": 7983
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.063904569040457e-06,
+      "loss": 0.069,
+      "step": 7984
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0623459654797097e-06,
+      "loss": 0.0693,
+      "step": 7985
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.060788440374324e-06,
+      "loss": 0.0717,
+      "step": 7986
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0592319939122387e-06,
+      "loss": 0.0736,
+      "step": 7987
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0576766262812532e-06,
+      "loss": 0.0659,
+      "step": 7988
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0561223376690533e-06,
+      "loss": 0.0647,
+      "step": 7989
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0545691282631832e-06,
+      "loss": 0.0705,
+      "step": 7990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0530169982510563e-06,
+      "loss": 0.0697,
+      "step": 7991
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0514659478199652e-06,
+      "loss": 0.073,
+      "step": 7992
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0499159771570633e-06,
+      "loss": 0.0712,
+      "step": 7993
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0483670864493777e-06,
+      "loss": 0.064,
+      "step": 7994
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0468192758838025e-06,
+      "loss": 0.0729,
+      "step": 7995
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0452725456471035e-06,
+      "loss": 0.0668,
+      "step": 7996
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0437268959259151e-06,
+      "loss": 0.0637,
+      "step": 7997
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0421823269067443e-06,
+      "loss": 0.07,
+      "step": 7998
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0406388387759636e-06,
+      "loss": 0.0677,
+      "step": 7999
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0390964317198181e-06,
+      "loss": 0.0728,
+      "step": 8000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0375551059244205e-06,
+      "loss": 0.0805,
+      "step": 8001
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0360148615757536e-06,
+      "loss": 0.0661,
+      "step": 8002
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0344756988596672e-06,
+      "loss": 0.0714,
+      "step": 8003
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0329376179618888e-06,
+      "loss": 0.0769,
+      "step": 8004
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.031400619068006e-06,
+      "loss": 0.0661,
+      "step": 8005
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.029864702363481e-06,
+      "loss": 0.0624,
+      "step": 8006
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.028329868033644e-06,
+      "loss": 0.0714,
+      "step": 8007
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0267961162636919e-06,
+      "loss": 0.0712,
+      "step": 8008
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.025263447238698e-06,
+      "loss": 0.0741,
+      "step": 8009
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0237318611435976e-06,
+      "loss": 0.0706,
+      "step": 8010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0222013581632006e-06,
+      "loss": 0.0696,
+      "step": 8011
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0206719384821806e-06,
+      "loss": 0.0692,
+      "step": 8012
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0191436022850909e-06,
+      "loss": 0.0694,
+      "step": 8013
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0176163497563374e-06,
+      "loss": 0.0733,
+      "step": 8014
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0160901810802114e-06,
+      "loss": 0.0672,
+      "step": 8015
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.014565096440866e-06,
+      "loss": 0.0734,
+      "step": 8016
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0130410960223236e-06,
+      "loss": 0.0683,
+      "step": 8017
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.011518180008475e-06,
+      "loss": 0.0691,
+      "step": 8018
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0099963485830866e-06,
+      "loss": 0.0634,
+      "step": 8019
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0084756019297826e-06,
+      "loss": 0.0677,
+      "step": 8020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0069559402320672e-06,
+      "loss": 0.0709,
+      "step": 8021
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0054373636733084e-06,
+      "loss": 0.0743,
+      "step": 8022
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0039198724367439e-06,
+      "loss": 0.0724,
+      "step": 8023
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0024034667054783e-06,
+      "loss": 0.0714,
+      "step": 8024
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.000888146662492e-06,
+      "loss": 0.0715,
+      "step": 8025
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.993739124906277e-07,
+      "loss": 0.0714,
+      "step": 8026
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.978607643726002e-07,
+      "loss": 0.071,
+      "step": 8027
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.963487024909902e-07,
+      "loss": 0.0648,
+      "step": 8028
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.948377270282493e-07,
+      "loss": 0.0699,
+      "step": 8029
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.933278381667023e-07,
+      "loss": 0.0662,
+      "step": 8030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.918190360885361e-07,
+      "loss": 0.0728,
+      "step": 8031
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.0691,
+      "step": 8032
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.88804693010449e-07,
+      "loss": 0.0635,
+      "step": 8033
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.872991523742526e-07,
+      "loss": 0.0721,
+      "step": 8034
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.85794699248881e-07,
+      "loss": 0.0737,
+      "step": 8035
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.842913338158732e-07,
+      "loss": 0.0673,
+      "step": 8036
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.8278905625663e-07,
+      "loss": 0.0665,
+      "step": 8037
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.812878667524218e-07,
+      "loss": 0.0672,
+      "step": 8038
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.79787765484389e-07,
+      "loss": 0.0698,
+      "step": 8039
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.782887526335405e-07,
+      "loss": 0.0645,
+      "step": 8040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.767908283807503e-07,
+      "loss": 0.0707,
+      "step": 8041
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.752939929067695e-07,
+      "loss": 0.0732,
+      "step": 8042
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.737982463922102e-07,
+      "loss": 0.0721,
+      "step": 8043
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.723035890175524e-07,
+      "loss": 0.0668,
+      "step": 8044
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.708100209631565e-07,
+      "loss": 0.0691,
+      "step": 8045
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.69317542409235e-07,
+      "loss": 0.0675,
+      "step": 8046
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.678261535358758e-07,
+      "loss": 0.065,
+      "step": 8047
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.663358545230428e-07,
+      "loss": 0.0714,
+      "step": 8048
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.648466455505578e-07,
+      "loss": 0.0776,
+      "step": 8049
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.63358526798114e-07,
+      "loss": 0.0678,
+      "step": 8050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.618714984452793e-07,
+      "loss": 0.0761,
+      "step": 8051
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.603855606714795e-07,
+      "loss": 0.0658,
+      "step": 8052
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.589007136560147e-07,
+      "loss": 0.0737,
+      "step": 8053
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.574169575780557e-07,
+      "loss": 0.0715,
+      "step": 8054
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.559342926166371e-07,
+      "loss": 0.0636,
+      "step": 8055
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.544527189506625e-07,
+      "loss": 0.0693,
+      "step": 8056
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.529722367589079e-07,
+      "loss": 0.0672,
+      "step": 8057
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.514928462200146e-07,
+      "loss": 0.0685,
+      "step": 8058
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.500145475124866e-07,
+      "loss": 0.0654,
+      "step": 8059
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.485373408147081e-07,
+      "loss": 0.0623,
+      "step": 8060
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.470612263049217e-07,
+      "loss": 0.0702,
+      "step": 8061
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.455862041612418e-07,
+      "loss": 0.0699,
+      "step": 8062
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.441122745616527e-07,
+      "loss": 0.0674,
+      "step": 8063
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.426394376840054e-07,
+      "loss": 0.07,
+      "step": 8064
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.411676937060143e-07,
+      "loss": 0.0676,
+      "step": 8065
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.396970428052698e-07,
+      "loss": 0.0727,
+      "step": 8066
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.382274851592277e-07,
+      "loss": 0.0663,
+      "step": 8067
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.367590209452071e-07,
+      "loss": 0.0692,
+      "step": 8068
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.352916503404042e-07,
+      "loss": 0.0704,
+      "step": 8069
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.33825373521875e-07,
+      "loss": 0.0711,
+      "step": 8070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.323601906665469e-07,
+      "loss": 0.0671,
+      "step": 8071
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.308961019512164e-07,
+      "loss": 0.0699,
+      "step": 8072
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.294331075525453e-07,
+      "loss": 0.0707,
+      "step": 8073
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.279712076470637e-07,
+      "loss": 0.0726,
+      "step": 8074
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.265104024111737e-07,
+      "loss": 0.0735,
+      "step": 8075
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.250506920211422e-07,
+      "loss": 0.0714,
+      "step": 8076
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.235920766531015e-07,
+      "loss": 0.0721,
+      "step": 8077
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.221345564830553e-07,
+      "loss": 0.0665,
+      "step": 8078
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.206781316868741e-07,
+      "loss": 0.0747,
+      "step": 8079
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.19222802440296e-07,
+      "loss": 0.0709,
+      "step": 8080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.177685689189286e-07,
+      "loss": 0.0722,
+      "step": 8081
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.163154312982459e-07,
+      "loss": 0.0655,
+      "step": 8082
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.148633897535864e-07,
+      "loss": 0.0727,
+      "step": 8083
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.134124444601655e-07,
+      "loss": 0.0742,
+      "step": 8084
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.119625955930566e-07,
+      "loss": 0.0698,
+      "step": 8085
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.105138433272021e-07,
+      "loss": 0.0723,
+      "step": 8086
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.090661878374196e-07,
+      "loss": 0.0662,
+      "step": 8087
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.076196292983874e-07,
+      "loss": 0.0673,
+      "step": 8088
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.061741678846514e-07,
+      "loss": 0.0731,
+      "step": 8089
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.047298037706331e-07,
+      "loss": 0.0706,
+      "step": 8090
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.032865371306077e-07,
+      "loss": 0.0698,
+      "step": 8091
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.018443681387312e-07,
+      "loss": 0.0726,
+      "step": 8092
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.004032969690202e-07,
+      "loss": 0.0703,
+      "step": 8093
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.989633237953621e-07,
+      "loss": 0.0793,
+      "step": 8094
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.975244487915057e-07,
+      "loss": 0.0657,
+      "step": 8095
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.960866721310768e-07,
+      "loss": 0.0657,
+      "step": 8096
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.94649993987563e-07,
+      "loss": 0.0722,
+      "step": 8097
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.932144145343169e-07,
+      "loss": 0.0719,
+      "step": 8098
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.917799339445643e-07,
+      "loss": 0.0664,
+      "step": 8099
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.903465523913957e-07,
+      "loss": 0.0684,
+      "step": 8100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.889142700477649e-07,
+      "loss": 0.0739,
+      "step": 8101
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.874830870865037e-07,
+      "loss": 0.0685,
+      "step": 8102
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.860530036803006e-07,
+      "loss": 0.066,
+      "step": 8103
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.846240200017165e-07,
+      "loss": 0.0662,
+      "step": 8104
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.831961362231789e-07,
+      "loss": 0.0698,
+      "step": 8105
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.817693525169802e-07,
+      "loss": 0.0684,
+      "step": 8106
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.803436690552858e-07,
+      "loss": 0.0699,
+      "step": 8107
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.789190860101226e-07,
+      "loss": 0.0612,
+      "step": 8108
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.774956035533877e-07,
+      "loss": 0.0698,
+      "step": 8109
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.760732218568435e-07,
+      "loss": 0.0655,
+      "step": 8110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.746519410921217e-07,
+      "loss": 0.0733,
+      "step": 8111
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.732317614307173e-07,
+      "loss": 0.0689,
+      "step": 8112
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.718126830439999e-07,
+      "loss": 0.071,
+      "step": 8113
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.703947061031981e-07,
+      "loss": 0.064,
+      "step": 8114
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.689778307794128e-07,
+      "loss": 0.0615,
+      "step": 8115
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.675620572436094e-07,
+      "loss": 0.0691,
+      "step": 8116
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.661473856666214e-07,
+      "loss": 0.0685,
+      "step": 8117
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.647338162191465e-07,
+      "loss": 0.0704,
+      "step": 8118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.633213490717573e-07,
+      "loss": 0.0699,
+      "step": 8119
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.619099843948842e-07,
+      "loss": 0.0715,
+      "step": 8120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.604997223588296e-07,
+      "loss": 0.0647,
+      "step": 8121
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.590905631337599e-07,
+      "loss": 0.0676,
+      "step": 8122
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.576825068897154e-07,
+      "loss": 0.0704,
+      "step": 8123
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.562755537965917e-07,
+      "loss": 0.0679,
+      "step": 8124
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.548697040241627e-07,
+      "loss": 0.0685,
+      "step": 8125
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.534649577420617e-07,
+      "loss": 0.0724,
+      "step": 8126
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.520613151197899e-07,
+      "loss": 0.0638,
+      "step": 8127
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.506587763267204e-07,
+      "loss": 0.0651,
+      "step": 8128
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.492573415320903e-07,
+      "loss": 0.0634,
+      "step": 8129
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.478570109049967e-07,
+      "loss": 0.0729,
+      "step": 8130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.46457784614414e-07,
+      "loss": 0.0678,
+      "step": 8131
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.450596628291785e-07,
+      "loss": 0.0664,
+      "step": 8132
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.436626457179909e-07,
+      "loss": 0.0635,
+      "step": 8133
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.42266733449425e-07,
+      "loss": 0.0722,
+      "step": 8134
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.408719261919163e-07,
+      "loss": 0.0788,
+      "step": 8135
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.394782241137666e-07,
+      "loss": 0.0705,
+      "step": 8136
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.380856273831472e-07,
+      "loss": 0.0708,
+      "step": 8137
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.366941361680947e-07,
+      "loss": 0.0721,
+      "step": 8138
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.353037506365102e-07,
+      "loss": 0.0678,
+      "step": 8139
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.339144709561675e-07,
+      "loss": 0.0702,
+      "step": 8140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.325262972947013e-07,
+      "loss": 0.0705,
+      "step": 8141
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.311392298196131e-07,
+      "loss": 0.0686,
+      "step": 8142
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.297532686982734e-07,
+      "loss": 0.0638,
+      "step": 8143
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.283684140979187e-07,
+      "loss": 0.0717,
+      "step": 8144
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.269846661856496e-07,
+      "loss": 0.0659,
+      "step": 8145
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.256020251284381e-07,
+      "loss": 0.0707,
+      "step": 8146
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.242204910931162e-07,
+      "loss": 0.0733,
+      "step": 8147
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.228400642463874e-07,
+      "loss": 0.0631,
+      "step": 8148
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.214607447548218e-07,
+      "loss": 0.0697,
+      "step": 8149
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.200825327848516e-07,
+      "loss": 0.0674,
+      "step": 8150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.187054285027751e-07,
+      "loss": 0.0738,
+      "step": 8151
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.173294320747649e-07,
+      "loss": 0.0743,
+      "step": 8152
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.159545436668514e-07,
+      "loss": 0.0663,
+      "step": 8153
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.145807634449343e-07,
+      "loss": 0.07,
+      "step": 8154
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.132080915747842e-07,
+      "loss": 0.0638,
+      "step": 8155
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.118365282220287e-07,
+      "loss": 0.0724,
+      "step": 8156
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.104660735521652e-07,
+      "loss": 0.0727,
+      "step": 8157
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.090967277305628e-07,
+      "loss": 0.0683,
+      "step": 8158
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.077284909224514e-07,
+      "loss": 0.072,
+      "step": 8159
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.063613632929268e-07,
+      "loss": 0.0673,
+      "step": 8160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.049953450069547e-07,
+      "loss": 0.0662,
+      "step": 8161
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.036304362293646e-07,
+      "loss": 0.07,
+      "step": 8162
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.022666371248489e-07,
+      "loss": 0.0659,
+      "step": 8163
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.009039478579717e-07,
+      "loss": 0.0707,
+      "step": 8164
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.995423685931624e-07,
+      "loss": 0.0697,
+      "step": 8165
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.98181899494711e-07,
+      "loss": 0.0657,
+      "step": 8166
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.968225407267815e-07,
+      "loss": 0.0715,
+      "step": 8167
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.954642924533995e-07,
+      "loss": 0.0687,
+      "step": 8168
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.941071548384527e-07,
+      "loss": 0.071,
+      "step": 8169
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.927511280457034e-07,
+      "loss": 0.0723,
+      "step": 8170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.913962122387753e-07,
+      "loss": 0.0683,
+      "step": 8171
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.900424075811553e-07,
+      "loss": 0.0734,
+      "step": 8172
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.886897142362027e-07,
+      "loss": 0.0699,
+      "step": 8173
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.873381323671381e-07,
+      "loss": 0.0693,
+      "step": 8174
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.859876621370477e-07,
+      "loss": 0.066,
+      "step": 8175
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.846383037088867e-07,
+      "loss": 0.069,
+      "step": 8176
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.832900572454749e-07,
+      "loss": 0.0712,
+      "step": 8177
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.819429229094933e-07,
+      "loss": 0.0697,
+      "step": 8178
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.805969008634984e-07,
+      "loss": 0.0693,
+      "step": 8179
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.792519912699037e-07,
+      "loss": 0.0787,
+      "step": 8180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.779081942909927e-07,
+      "loss": 0.0658,
+      "step": 8181
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.765655100889135e-07,
+      "loss": 0.0712,
+      "step": 8182
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.752239388256811e-07,
+      "loss": 0.0636,
+      "step": 8183
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.738834806631712e-07,
+      "loss": 0.0699,
+      "step": 8184
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.725441357631336e-07,
+      "loss": 0.0707,
+      "step": 8185
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.712059042871788e-07,
+      "loss": 0.0719,
+      "step": 8186
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.698687863967802e-07,
+      "loss": 0.0741,
+      "step": 8187
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.685327822532862e-07,
+      "loss": 0.0754,
+      "step": 8188
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.671978920178979e-07,
+      "loss": 0.0679,
+      "step": 8189
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.658641158516933e-07,
+      "loss": 0.073,
+      "step": 8190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.645314539156101e-07,
+      "loss": 0.0718,
+      "step": 8191
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.631999063704532e-07,
+      "loss": 0.0697,
+      "step": 8192
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.618694733768906e-07,
+      "loss": 0.0682,
+      "step": 8193
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.605401550954638e-07,
+      "loss": 0.0673,
+      "step": 8194
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.59211951686567e-07,
+      "loss": 0.069,
+      "step": 8195
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.578848633104718e-07,
+      "loss": 0.0747,
+      "step": 8196
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.56558890127308e-07,
+      "loss": 0.0684,
+      "step": 8197
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.552340322970753e-07,
+      "loss": 0.0728,
+      "step": 8198
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.539102899796324e-07,
+      "loss": 0.0703,
+      "step": 8199
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.52587663334714e-07,
+      "loss": 0.0666,
+      "step": 8200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.512661525219067e-07,
+      "loss": 0.0658,
+      "step": 8201
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.499457577006753e-07,
+      "loss": 0.0676,
+      "step": 8202
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.48626479030341e-07,
+      "loss": 0.0705,
+      "step": 8203
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.473083166700946e-07,
+      "loss": 0.0702,
+      "step": 8204
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.459912707789929e-07,
+      "loss": 0.0677,
+      "step": 8205
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.446753415159536e-07,
+      "loss": 0.0723,
+      "step": 8206
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.433605290397649e-07,
+      "loss": 0.0716,
+      "step": 8207
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.420468335090758e-07,
+      "loss": 0.0713,
+      "step": 8208
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.407342550824026e-07,
+      "loss": 0.0661,
+      "step": 8209
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.394227939181242e-07,
+      "loss": 0.0707,
+      "step": 8210
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.381124501744929e-07,
+      "loss": 0.072,
+      "step": 8211
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.368032240096157e-07,
+      "loss": 0.069,
+      "step": 8212
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.354951155814716e-07,
+      "loss": 0.0734,
+      "step": 8213
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.341881250479022e-07,
+      "loss": 0.0691,
+      "step": 8214
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.328822525666135e-07,
+      "loss": 0.0732,
+      "step": 8215
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.315774982951762e-07,
+      "loss": 0.0735,
+      "step": 8216
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.302738623910322e-07,
+      "loss": 0.0698,
+      "step": 8217
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.289713450114811e-07,
+      "loss": 0.069,
+      "step": 8218
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.276699463136872e-07,
+      "loss": 0.0707,
+      "step": 8219
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.263696664546904e-07,
+      "loss": 0.0679,
+      "step": 8220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.250705055913809e-07,
+      "loss": 0.0698,
+      "step": 8221
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.237724638805221e-07,
+      "loss": 0.0679,
+      "step": 8222
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.224755414787444e-07,
+      "loss": 0.0651,
+      "step": 8223
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.211797385425379e-07,
+      "loss": 0.0728,
+      "step": 8224
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.198850552282577e-07,
+      "loss": 0.0675,
+      "step": 8225
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.185914916921322e-07,
+      "loss": 0.0791,
+      "step": 8226
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.172990480902431e-07,
+      "loss": 0.0631,
+      "step": 8227
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.160077245785402e-07,
+      "loss": 0.0727,
+      "step": 8228
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.147175213128454e-07,
+      "loss": 0.0662,
+      "step": 8229
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.134284384488377e-07,
+      "loss": 0.067,
+      "step": 8230
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.121404761420614e-07,
+      "loss": 0.0739,
+      "step": 8231
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.108536345479323e-07,
+      "loss": 0.0746,
+      "step": 8232
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.095679138217237e-07,
+      "loss": 0.0688,
+      "step": 8233
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.08283314118573e-07,
+      "loss": 0.0679,
+      "step": 8234
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.069998355934904e-07,
+      "loss": 0.0692,
+      "step": 8235
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.057174784013432e-07,
+      "loss": 0.0671,
+      "step": 8236
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.044362426968643e-07,
+      "loss": 0.0677,
+      "step": 8237
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.031561286346589e-07,
+      "loss": 0.071,
+      "step": 8238
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.018771363691879e-07,
+      "loss": 0.0663,
+      "step": 8239
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.005992660547767e-07,
+      "loss": 0.0679,
+      "step": 8240
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.99322517845622e-07,
+      "loss": 0.0703,
+      "step": 8241
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.980468918957827e-07,
+      "loss": 0.0643,
+      "step": 8242
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.967723883591771e-07,
+      "loss": 0.0703,
+      "step": 8243
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.954990073895962e-07,
+      "loss": 0.0699,
+      "step": 8244
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.942267491406907e-07,
+      "loss": 0.0657,
+      "step": 8245
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.929556137659766e-07,
+      "loss": 0.0749,
+      "step": 8246
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.916856014188334e-07,
+      "loss": 0.0678,
+      "step": 8247
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.904167122525063e-07,
+      "loss": 0.0675,
+      "step": 8248
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.89148946420104e-07,
+      "loss": 0.0674,
+      "step": 8249
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.878823040746041e-07,
+      "loss": 0.0677,
+      "step": 8250
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.866167853688421e-07,
+      "loss": 0.0662,
+      "step": 8251
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.853523904555226e-07,
+      "loss": 0.0703,
+      "step": 8252
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.840891194872112e-07,
+      "loss": 0.072,
+      "step": 8253
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.828269726163405e-07,
+      "loss": 0.0657,
+      "step": 8254
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.815659499952043e-07,
+      "loss": 0.0646,
+      "step": 8255
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.803060517759674e-07,
+      "loss": 0.0677,
+      "step": 8256
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.790472781106517e-07,
+      "loss": 0.0728,
+      "step": 8257
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.777896291511443e-07,
+      "loss": 0.0713,
+      "step": 8258
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.765331050492041e-07,
+      "loss": 0.071,
+      "step": 8259
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.752777059564431e-07,
+      "loss": 0.0675,
+      "step": 8260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.740234320243444e-07,
+      "loss": 0.0699,
+      "step": 8261
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.72770283404256e-07,
+      "loss": 0.0645,
+      "step": 8262
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.715182602473869e-07,
+      "loss": 0.0746,
+      "step": 8263
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.702673627048107e-07,
+      "loss": 0.0701,
+      "step": 8264
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.690175909274699e-07,
+      "loss": 0.0727,
+      "step": 8265
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.677689450661607e-07,
+      "loss": 0.072,
+      "step": 8266
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.665214252715557e-07,
+      "loss": 0.068,
+      "step": 8267
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.652750316941836e-07,
+      "loss": 0.0661,
+      "step": 8268
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.640297644844395e-07,
+      "loss": 0.0751,
+      "step": 8269
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.627856237925812e-07,
+      "loss": 0.0665,
+      "step": 8270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.615426097687361e-07,
+      "loss": 0.0658,
+      "step": 8271
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.603007225628888e-07,
+      "loss": 0.0639,
+      "step": 8272
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.590599623248917e-07,
+      "loss": 0.0688,
+      "step": 8273
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.578203292044594e-07,
+      "loss": 0.0716,
+      "step": 8274
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.565818233511712e-07,
+      "loss": 0.0689,
+      "step": 8275
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.553444449144697e-07,
+      "loss": 0.069,
+      "step": 8276
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.541081940436644e-07,
+      "loss": 0.0689,
+      "step": 8277
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.528730708879261e-07,
+      "loss": 0.073,
+      "step": 8278
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.516390755962887e-07,
+      "loss": 0.0752,
+      "step": 8279
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.504062083176533e-07,
+      "loss": 0.0746,
+      "step": 8280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.491744692007807e-07,
+      "loss": 0.0666,
+      "step": 8281
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.479438583942976e-07,
+      "loss": 0.0645,
+      "step": 8282
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.467143760466976e-07,
+      "loss": 0.0698,
+      "step": 8283
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.454860223063331e-07,
+      "loss": 0.0736,
+      "step": 8284
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.442587973214231e-07,
+      "loss": 0.073,
+      "step": 8285
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.430327012400506e-07,
+      "loss": 0.0687,
+      "step": 8286
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.418077342101581e-07,
+      "loss": 0.0737,
+      "step": 8287
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.405838963795597e-07,
+      "loss": 0.0675,
+      "step": 8288
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.393611878959272e-07,
+      "loss": 0.0689,
+      "step": 8289
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.38139608906797e-07,
+      "loss": 0.0661,
+      "step": 8290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.369191595595714e-07,
+      "loss": 0.0667,
+      "step": 8291
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.356998400015147e-07,
+      "loss": 0.0727,
+      "step": 8292
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.344816503797524e-07,
+      "loss": 0.0688,
+      "step": 8293
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.332645908412805e-07,
+      "loss": 0.0678,
+      "step": 8294
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.320486615329535e-07,
+      "loss": 0.071,
+      "step": 8295
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.308338626014888e-07,
+      "loss": 0.0784,
+      "step": 8296
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.296201941934699e-07,
+      "loss": 0.0692,
+      "step": 8297
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.284076564553465e-07,
+      "loss": 0.0711,
+      "step": 8298
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.271962495334228e-07,
+      "loss": 0.0749,
+      "step": 8299
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.259859735738772e-07,
+      "loss": 0.0697,
+      "step": 8300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.24776828722744e-07,
+      "loss": 0.075,
+      "step": 8301
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.235688151259234e-07,
+      "loss": 0.0724,
+      "step": 8302
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.22361932929182e-07,
+      "loss": 0.0687,
+      "step": 8303
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.211561822781476e-07,
+      "loss": 0.0808,
+      "step": 8304
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.199515633183073e-07,
+      "loss": 0.0716,
+      "step": 8305
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.187480761950182e-07,
+      "loss": 0.0685,
+      "step": 8306
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.175457210534986e-07,
+      "loss": 0.0696,
+      "step": 8307
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.163444980388267e-07,
+      "loss": 0.0688,
+      "step": 8308
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.151444072959522e-07,
+      "loss": 0.067,
+      "step": 8309
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.139454489696795e-07,
+      "loss": 0.0639,
+      "step": 8310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.127476232046814e-07,
+      "loss": 0.0682,
+      "step": 8311
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.115509301454925e-07,
+      "loss": 0.073,
+      "step": 8312
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.103553699365095e-07,
+      "loss": 0.0701,
+      "step": 8313
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.091609427219947e-07,
+      "loss": 0.0672,
+      "step": 8314
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.079676486460728e-07,
+      "loss": 0.0674,
+      "step": 8315
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.067754878527332e-07,
+      "loss": 0.0687,
+      "step": 8316
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.055844604858252e-07,
+      "loss": 0.0667,
+      "step": 8317
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.043945666890638e-07,
+      "loss": 0.0726,
+      "step": 8318
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.032058066060276e-07,
+      "loss": 0.0651,
+      "step": 8319
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.020181803801539e-07,
+      "loss": 0.0733,
+      "step": 8320
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.008316881547515e-07,
+      "loss": 0.0736,
+      "step": 8321
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.996463300729849e-07,
+      "loss": 0.069,
+      "step": 8322
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.984621062778829e-07,
+      "loss": 0.0697,
+      "step": 8323
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.972790169123454e-07,
+      "loss": 0.0783,
+      "step": 8324
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.96097062119122e-07,
+      "loss": 0.0685,
+      "step": 8325
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.949162420408338e-07,
+      "loss": 0.0753,
+      "step": 8326
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.937365568199671e-07,
+      "loss": 0.0678,
+      "step": 8327
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.925580065988645e-07,
+      "loss": 0.0693,
+      "step": 8328
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.913805915197357e-07,
+      "loss": 0.0712,
+      "step": 8329
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.902043117246547e-07,
+      "loss": 0.0716,
+      "step": 8330
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.890291673555526e-07,
+      "loss": 0.0675,
+      "step": 8331
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.87855158554228e-07,
+      "loss": 0.0647,
+      "step": 8332
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.866822854623456e-07,
+      "loss": 0.0708,
+      "step": 8333
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.855105482214252e-07,
+      "loss": 0.0749,
+      "step": 8334
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.843399469728539e-07,
+      "loss": 0.0626,
+      "step": 8335
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.831704818578842e-07,
+      "loss": 0.0708,
+      "step": 8336
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.82002153017629e-07,
+      "loss": 0.0814,
+      "step": 8337
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.808349605930586e-07,
+      "loss": 0.0708,
+      "step": 8338
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.79668904725017e-07,
+      "loss": 0.0655,
+      "step": 8339
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.785039855542018e-07,
+      "loss": 0.0657,
+      "step": 8340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.773402032211783e-07,
+      "loss": 0.069,
+      "step": 8341
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.761775578663742e-07,
+      "loss": 0.0641,
+      "step": 8342
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.750160496300805e-07,
+      "loss": 0.068,
+      "step": 8343
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.738556786524452e-07,
+      "loss": 0.0663,
+      "step": 8344
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.726964450734874e-07,
+      "loss": 0.0678,
+      "step": 8345
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.715383490330839e-07,
+      "loss": 0.0747,
+      "step": 8346
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.703813906709743e-07,
+      "loss": 0.0709,
+      "step": 8347
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.692255701267657e-07,
+      "loss": 0.0719,
+      "step": 8348
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.68070887539921e-07,
+      "loss": 0.0684,
+      "step": 8349
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.66917343049771e-07,
+      "loss": 0.0743,
+      "step": 8350
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.657649367955065e-07,
+      "loss": 0.0656,
+      "step": 8351
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.646136689161808e-07,
+      "loss": 0.0666,
+      "step": 8352
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.634635395507116e-07,
+      "loss": 0.0643,
+      "step": 8353
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.62314548837879e-07,
+      "loss": 0.0633,
+      "step": 8354
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.0731,
+      "step": 8355
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.600199839245535e-07,
+      "loss": 0.0713,
+      "step": 8356
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.588744100009325e-07,
+      "loss": 0.0729,
+      "step": 8357
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.577299752836917e-07,
+      "loss": 0.0714,
+      "step": 8358
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.56586679910921e-07,
+      "loss": 0.0638,
+      "step": 8359
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.554445240205797e-07,
+      "loss": 0.0708,
+      "step": 8360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.543035077504832e-07,
+      "loss": 0.072,
+      "step": 8361
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.531636312383115e-07,
+      "loss": 0.071,
+      "step": 8362
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.520248946216056e-07,
+      "loss": 0.0693,
+      "step": 8363
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.508872980377711e-07,
+      "loss": 0.073,
+      "step": 8364
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.49750841624076e-07,
+      "loss": 0.0737,
+      "step": 8365
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.486155255176495e-07,
+      "loss": 0.0638,
+      "step": 8366
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.474813498554843e-07,
+      "loss": 0.0758,
+      "step": 8367
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.463483147744319e-07,
+      "loss": 0.0742,
+      "step": 8368
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.452164204112154e-07,
+      "loss": 0.0647,
+      "step": 8369
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.440856669024063e-07,
+      "loss": 0.066,
+      "step": 8370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.429560543844514e-07,
+      "loss": 0.0712,
+      "step": 8371
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.0697,
+      "step": 8372
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.407002528661787e-07,
+      "loss": 0.0604,
+      "step": 8373
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.395740641380532e-07,
+      "loss": 0.0631,
+      "step": 8374
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.384490169451717e-07,
+      "loss": 0.0621,
+      "step": 8375
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.373251114232824e-07,
+      "loss": 0.0657,
+      "step": 8376
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.362023477080048e-07,
+      "loss": 0.0681,
+      "step": 8377
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.350807259348145e-07,
+      "loss": 0.0704,
+      "step": 8378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.339602462390525e-07,
+      "loss": 0.0726,
+      "step": 8379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.328409087559161e-07,
+      "loss": 0.069,
+      "step": 8380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.317227136204761e-07,
+      "loss": 0.0686,
+      "step": 8381
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.306056609676546e-07,
+      "loss": 0.071,
+      "step": 8382
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.294897509322405e-07,
+      "loss": 0.0673,
+      "step": 8383
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.283749836488839e-07,
+      "loss": 0.065,
+      "step": 8384
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.272613592520969e-07,
+      "loss": 0.0698,
+      "step": 8385
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.261488778762569e-07,
+      "loss": 0.0722,
+      "step": 8386
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.250375396555985e-07,
+      "loss": 0.0657,
+      "step": 8387
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.239273447242199e-07,
+      "loss": 0.0646,
+      "step": 8388
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.228182932160841e-07,
+      "loss": 0.0704,
+      "step": 8389
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.217103852650129e-07,
+      "loss": 0.0727,
+      "step": 8390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.206036210046883e-07,
+      "loss": 0.0661,
+      "step": 8391
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.194980005686623e-07,
+      "loss": 0.0709,
+      "step": 8392
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.183935240903415e-07,
+      "loss": 0.0739,
+      "step": 8393
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.172901917029971e-07,
+      "loss": 0.0679,
+      "step": 8394
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.161880035397604e-07,
+      "loss": 0.0696,
+      "step": 8395
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.150869597336272e-07,
+      "loss": 0.0723,
+      "step": 8396
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.139870604174535e-07,
+      "loss": 0.0647,
+      "step": 8397
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.128883057239587e-07,
+      "loss": 0.0663,
+      "step": 8398
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.117906957857233e-07,
+      "loss": 0.0671,
+      "step": 8399
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.10694230735188e-07,
+      "loss": 0.0724,
+      "step": 8400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.095989107046606e-07,
+      "loss": 0.0761,
+      "step": 8401
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.08504735826304e-07,
+      "loss": 0.0764,
+      "step": 8402
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.074117062321438e-07,
+      "loss": 0.0684,
+      "step": 8403
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.063198220540744e-07,
+      "loss": 0.0679,
+      "step": 8404
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.052290834238461e-07,
+      "loss": 0.063,
+      "step": 8405
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.04139490473069e-07,
+      "loss": 0.0671,
+      "step": 8406
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.030510433332215e-07,
+      "loss": 0.0741,
+      "step": 8407
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.019637421356405e-07,
+      "loss": 0.073,
+      "step": 8408
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.00877587011519e-07,
+      "loss": 0.0711,
+      "step": 8409
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.997925780919232e-07,
+      "loss": 0.0751,
+      "step": 8410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.98708715507773e-07,
+      "loss": 0.0685,
+      "step": 8411
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.976259993898503e-07,
+      "loss": 0.0616,
+      "step": 8412
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.965444298688015e-07,
+      "loss": 0.0703,
+      "step": 8413
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.95464007075136e-07,
+      "loss": 0.0768,
+      "step": 8414
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.943847311392158e-07,
+      "loss": 0.0668,
+      "step": 8415
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.933066021912758e-07,
+      "loss": 0.0681,
+      "step": 8416
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.922296203614075e-07,
+      "loss": 0.071,
+      "step": 8417
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.911537857795612e-07,
+      "loss": 0.0751,
+      "step": 8418
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.900790985755544e-07,
+      "loss": 0.0705,
+      "step": 8419
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.890055588790632e-07,
+      "loss": 0.074,
+      "step": 8420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.879331668196241e-07,
+      "loss": 0.0651,
+      "step": 8421
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.86861922526638e-07,
+      "loss": 0.0689,
+      "step": 8422
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.857918261293649e-07,
+      "loss": 0.0744,
+      "step": 8423
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.847228777569258e-07,
+      "loss": 0.068,
+      "step": 8424
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.836550775383076e-07,
+      "loss": 0.0678,
+      "step": 8425
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.825884256023538e-07,
+      "loss": 0.0757,
+      "step": 8426
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.815229220777717e-07,
+      "loss": 0.0659,
+      "step": 8427
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.804585670931294e-07,
+      "loss": 0.0747,
+      "step": 8428
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.793953607768553e-07,
+      "loss": 0.067,
+      "step": 8429
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.783333032572413e-07,
+      "loss": 0.0674,
+      "step": 8430
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.772723946624414e-07,
+      "loss": 0.0779,
+      "step": 8431
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7621263512046654e-07,
+      "loss": 0.07,
+      "step": 8432
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7515402475919235e-07,
+      "loss": 0.0722,
+      "step": 8433
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.740965637063588e-07,
+      "loss": 0.0664,
+      "step": 8434
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.730402520895594e-07,
+      "loss": 0.071,
+      "step": 8435
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7198509003625434e-07,
+      "loss": 0.0693,
+      "step": 8436
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7093107767376414e-07,
+      "loss": 0.0677,
+      "step": 8437
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.698782151292713e-07,
+      "loss": 0.0684,
+      "step": 8438
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6882650252981663e-07,
+      "loss": 0.0664,
+      "step": 8439
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6777594000230855e-07,
+      "loss": 0.0752,
+      "step": 8440
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6672652767350777e-07,
+      "loss": 0.07,
+      "step": 8441
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6567826567004094e-07,
+      "loss": 0.0691,
+      "step": 8442
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.64631154118399e-07,
+      "loss": 0.0687,
+      "step": 8443
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.635851931449287e-07,
+      "loss": 0.0659,
+      "step": 8444
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6254038287584013e-07,
+      "loss": 0.0658,
+      "step": 8445
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6149672343720587e-07,
+      "loss": 0.063,
+      "step": 8446
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6045421495495845e-07,
+      "loss": 0.0675,
+      "step": 8447
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5941285755488953e-07,
+      "loss": 0.0742,
+      "step": 8448
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5837265136265517e-07,
+      "loss": 0.0645,
+      "step": 8449
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.573335965037706e-07,
+      "loss": 0.0639,
+      "step": 8450
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.562956931036111e-07,
+      "loss": 0.0693,
+      "step": 8451
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5525894128741753e-07,
+      "loss": 0.0668,
+      "step": 8452
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5422334118028657e-07,
+      "loss": 0.0659,
+      "step": 8453
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.531888929071804e-07,
+      "loss": 0.0643,
+      "step": 8454
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.52155596592917e-07,
+      "loss": 0.0652,
+      "step": 8455
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.511234523621799e-07,
+      "loss": 0.071,
+      "step": 8456
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.500924603395107e-07,
+      "loss": 0.0746,
+      "step": 8457
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.490626206493154e-07,
+      "loss": 0.062,
+      "step": 8458
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4803393341585786e-07,
+      "loss": 0.0686,
+      "step": 8459
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4700639876326333e-07,
+      "loss": 0.0751,
+      "step": 8460
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.459800168155204e-07,
+      "loss": 0.0677,
+      "step": 8461
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.449547876964722e-07,
+      "loss": 0.0711,
+      "step": 8462
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4393071152983305e-07,
+      "loss": 0.0676,
+      "step": 8463
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.429077884391686e-07,
+      "loss": 0.0661,
+      "step": 8464
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4188601854791126e-07,
+      "loss": 0.068,
+      "step": 8465
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4086540197935013e-07,
+      "loss": 0.0742,
+      "step": 8466
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.398459388566378e-07,
+      "loss": 0.07,
+      "step": 8467
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.38827629302786e-07,
+      "loss": 0.0725,
+      "step": 8468
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3781047344067074e-07,
+      "loss": 0.0715,
+      "step": 8469
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.36794471393025e-07,
+      "loss": 0.0726,
+      "step": 8470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3577962328244516e-07,
+      "loss": 0.0686,
+      "step": 8471
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3476592923138327e-07,
+      "loss": 0.0637,
+      "step": 8472
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3375338936216147e-07,
+      "loss": 0.0652,
+      "step": 8473
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.327420037969532e-07,
+      "loss": 0.0615,
+      "step": 8474
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.317317726577974e-07,
+      "loss": 0.0687,
+      "step": 8475
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3072269606659333e-07,
+      "loss": 0.0636,
+      "step": 8476
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.297147741451013e-07,
+      "loss": 0.0738,
+      "step": 8477
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2870800701493854e-07,
+      "loss": 0.0656,
+      "step": 8478
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2770239479759e-07,
+      "loss": 0.0687,
+      "step": 8479
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2669793761439313e-07,
+      "loss": 0.0649,
+      "step": 8480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2569463558655213e-07,
+      "loss": 0.0643,
+      "step": 8481
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2469248883513e-07,
+      "loss": 0.0683,
+      "step": 8482
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.23691497481048e-07,
+      "loss": 0.0689,
+      "step": 8483
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.226916616450916e-07,
+      "loss": 0.0751,
+      "step": 8484
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.216929814479065e-07,
+      "loss": 0.074,
+      "step": 8485
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2069545700999524e-07,
+      "loss": 0.072,
+      "step": 8486
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.196990884517249e-07,
+      "loss": 0.0648,
+      "step": 8487
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.187038758933204e-07,
+      "loss": 0.0659,
+      "step": 8488
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.177098194548679e-07,
+      "loss": 0.0668,
+      "step": 8489
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1671691925631697e-07,
+      "loss": 0.0665,
+      "step": 8490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1572517541747294e-07,
+      "loss": 0.0658,
+      "step": 8491
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.147345880580056e-07,
+      "loss": 0.0732,
+      "step": 8492
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1374515729744157e-07,
+      "loss": 0.0684,
+      "step": 8493
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.127568832551709e-07,
+      "loss": 0.0783,
+      "step": 8494
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.117697660504416e-07,
+      "loss": 0.0713,
+      "step": 8495
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.107838058023661e-07,
+      "loss": 0.0749,
+      "step": 8496
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0979900262991147e-07,
+      "loss": 0.0707,
+      "step": 8497
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.088153566519115e-07,
+      "loss": 0.0766,
+      "step": 8498
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.078328679870547e-07,
+      "loss": 0.0675,
+      "step": 8499
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0685153675389276e-07,
+      "loss": 0.0698,
+      "step": 8500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.058713630708355e-07,
+      "loss": 0.0667,
+      "step": 8501
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.048923470561594e-07,
+      "loss": 0.0734,
+      "step": 8502
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0391448882799334e-07,
+      "loss": 0.0681,
+      "step": 8503
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.029377885043295e-07,
+      "loss": 0.0712,
+      "step": 8504
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.019622462030248e-07,
+      "loss": 0.0686,
+      "step": 8505
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.009878620417873e-07,
+      "loss": 0.0655,
+      "step": 8506
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.000146361381918e-07,
+      "loss": 0.0672,
+      "step": 8507
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9904256860967436e-07,
+      "loss": 0.0702,
+      "step": 8508
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.980716595735257e-07,
+      "loss": 0.0722,
+      "step": 8509
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.971019091468997e-07,
+      "loss": 0.0735,
+      "step": 8510
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.961333174468152e-07,
+      "loss": 0.067,
+      "step": 8511
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9516588459014184e-07,
+      "loss": 0.0637,
+      "step": 8512
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9419961069361414e-07,
+      "loss": 0.073,
+      "step": 8513
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.932344958738299e-07,
+      "loss": 0.0689,
+      "step": 8514
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.922705402472426e-07,
+      "loss": 0.0687,
+      "step": 8515
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.913077439301649e-07,
+      "loss": 0.0682,
+      "step": 8516
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9034610703877596e-07,
+      "loss": 0.073,
+      "step": 8517
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8938562968911074e-07,
+      "loss": 0.0658,
+      "step": 8518
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8842631199705884e-07,
+      "loss": 0.0716,
+      "step": 8519
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.87468154078382e-07,
+      "loss": 0.074,
+      "step": 8520
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.865111560486934e-07,
+      "loss": 0.0699,
+      "step": 8521
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.855553180234661e-07,
+      "loss": 0.0691,
+      "step": 8522
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.846006401180402e-07,
+      "loss": 0.0674,
+      "step": 8523
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8364712244760907e-07,
+      "loss": 0.0651,
+      "step": 8524
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.826947651272261e-07,
+      "loss": 0.0689,
+      "step": 8525
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.817435682718096e-07,
+      "loss": 0.0705,
+      "step": 8526
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.807935319961342e-07,
+      "loss": 0.0658,
+      "step": 8527
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.798446564148339e-07,
+      "loss": 0.0763,
+      "step": 8528
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.788969416424071e-07,
+      "loss": 0.0754,
+      "step": 8529
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.779503877932067e-07,
+      "loss": 0.0719,
+      "step": 8530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7700499498144803e-07,
+      "loss": 0.0682,
+      "step": 8531
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7606076332120766e-07,
+      "loss": 0.0665,
+      "step": 8532
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7511769292642e-07,
+      "loss": 0.0709,
+      "step": 8533
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.741757839108773e-07,
+      "loss": 0.0701,
+      "step": 8534
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.732350363882387e-07,
+      "loss": 0.0806,
+      "step": 8535
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7229545047201555e-07,
+      "loss": 0.0689,
+      "step": 8536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.71357026275585e-07,
+      "loss": 0.0728,
+      "step": 8537
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.704197639121787e-07,
+      "loss": 0.0694,
+      "step": 8538
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6948366349489175e-07,
+      "loss": 0.0756,
+      "step": 8539
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6854872513667706e-07,
+      "loss": 0.0762,
+      "step": 8540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.676149489503511e-07,
+      "loss": 0.0692,
+      "step": 8541
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6668233504858486e-07,
+      "loss": 0.0686,
+      "step": 8542
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6575088354391054e-07,
+      "loss": 0.0695,
+      "step": 8543
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.648205945487249e-07,
+      "loss": 0.0684,
+      "step": 8544
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.638914681752759e-07,
+      "loss": 0.0751,
+      "step": 8545
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6296350453567943e-07,
+      "loss": 0.0702,
+      "step": 8546
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6203670374190704e-07,
+      "loss": 0.0694,
+      "step": 8547
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.611110659057893e-07,
+      "loss": 0.0773,
+      "step": 8548
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6018659113901676e-07,
+      "loss": 0.0683,
+      "step": 8549
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.592632795531437e-07,
+      "loss": 0.0696,
+      "step": 8550
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5834113125957747e-07,
+      "loss": 0.0762,
+      "step": 8551
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.574201463695903e-07,
+      "loss": 0.0687,
+      "step": 8552
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5650032499431107e-07,
+      "loss": 0.0752,
+      "step": 8553
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.55581667244731e-07,
+      "loss": 0.069,
+      "step": 8554
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.546641732316969e-07,
+      "loss": 0.0676,
+      "step": 8555
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5374784306591913e-07,
+      "loss": 0.0735,
+      "step": 8556
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5283267685796707e-07,
+      "loss": 0.066,
+      "step": 8557
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5191867471826567e-07,
+      "loss": 0.065,
+      "step": 8558
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.510058367571045e-07,
+      "loss": 0.0688,
+      "step": 8559
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5009416308462886e-07,
+      "loss": 0.0737,
+      "step": 8560
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.491836538108462e-07,
+      "loss": 0.0707,
+      "step": 8561
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.482743090456231e-07,
+      "loss": 0.0706,
+      "step": 8562
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4736612889868407e-07,
+      "loss": 0.0688,
+      "step": 8563
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4645911347961357e-07,
+      "loss": 0.0669,
+      "step": 8564
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.455532628978575e-07,
+      "loss": 0.0735,
+      "step": 8565
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4464857726271614e-07,
+      "loss": 0.0692,
+      "step": 8566
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4374505668335776e-07,
+      "loss": 0.0711,
+      "step": 8567
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.428427012688007e-07,
+      "loss": 0.0729,
+      "step": 8568
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4194151112793006e-07,
+      "loss": 0.0675,
+      "step": 8569
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4104148636948554e-07,
+      "loss": 0.065,
+      "step": 8570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4014262710206803e-07,
+      "loss": 0.0728,
+      "step": 8571
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3924493343413743e-07,
+      "loss": 0.0684,
+      "step": 8572
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3834840547401493e-07,
+      "loss": 0.0683,
+      "step": 8573
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.374530433298784e-07,
+      "loss": 0.07,
+      "step": 8574
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.365588471097658e-07,
+      "loss": 0.0661,
+      "step": 8575
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.356658169215743e-07,
+      "loss": 0.0649,
+      "step": 8576
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.34773952873062e-07,
+      "loss": 0.0652,
+      "step": 8577
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.338832550718429e-07,
+      "loss": 0.0688,
+      "step": 8578
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.329937236253944e-07,
+      "loss": 0.0694,
+      "step": 8579
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3210535864105166e-07,
+      "loss": 0.0643,
+      "step": 8580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.312181602260056e-07,
+      "loss": 0.0746,
+      "step": 8581
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.303321284873129e-07,
+      "loss": 0.0682,
+      "step": 8582
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.294472635318846e-07,
+      "loss": 0.0673,
+      "step": 8583
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2856356546648983e-07,
+      "loss": 0.078,
+      "step": 8584
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.276810343977621e-07,
+      "loss": 0.0708,
+      "step": 8585
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2679967043219185e-07,
+      "loss": 0.0734,
+      "step": 8586
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.259194736761251e-07,
+      "loss": 0.0678,
+      "step": 8587
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.250404442357735e-07,
+      "loss": 0.0722,
+      "step": 8588
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2416258221720345e-07,
+      "loss": 0.0708,
+      "step": 8589
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2328588772633895e-07,
+      "loss": 0.068,
+      "step": 8590
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.224103608689699e-07,
+      "loss": 0.0703,
+      "step": 8591
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.215360017507385e-07,
+      "loss": 0.0683,
+      "step": 8592
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.206628104771481e-07,
+      "loss": 0.07,
+      "step": 8593
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.197907871535633e-07,
+      "loss": 0.0678,
+      "step": 8594
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.189199318852065e-07,
+      "loss": 0.0719,
+      "step": 8595
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.180502447771572e-07,
+      "loss": 0.0703,
+      "step": 8596
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.171817259343568e-07,
+      "loss": 0.0768,
+      "step": 8597
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1631437546160383e-07,
+      "loss": 0.0677,
+      "step": 8598
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.154481934635556e-07,
+      "loss": 0.0734,
+      "step": 8599
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1458318004473075e-07,
+      "loss": 0.0685,
+      "step": 8600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1371933530950584e-07,
+      "loss": 0.0743,
+      "step": 8601
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.128566593621152e-07,
+      "loss": 0.0684,
+      "step": 8602
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.119951523066522e-07,
+      "loss": 0.0745,
+      "step": 8603
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1113481424707157e-07,
+      "loss": 0.0751,
+      "step": 8604
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1027564528718247e-07,
+      "loss": 0.0719,
+      "step": 8605
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.094176455306597e-07,
+      "loss": 0.0666,
+      "step": 8606
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.085608150810315e-07,
+      "loss": 0.0691,
+      "step": 8607
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0770515404168645e-07,
+      "loss": 0.068,
+      "step": 8608
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.068506625158729e-07,
+      "loss": 0.0669,
+      "step": 8609
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.059973406066963e-07,
+      "loss": 0.0679,
+      "step": 8610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.051451884171219e-07,
+      "loss": 0.0689,
+      "step": 8611
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.042942060499765e-07,
+      "loss": 0.0692,
+      "step": 8612
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.034443936079412e-07,
+      "loss": 0.0693,
+      "step": 8613
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.025957511935573e-07,
+      "loss": 0.0708,
+      "step": 8614
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.017482789092285e-07,
+      "loss": 0.0804,
+      "step": 8615
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.009019768572119e-07,
+      "loss": 0.0707,
+      "step": 8616
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0005684513962464e-07,
+      "loss": 0.0731,
+      "step": 8617
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.992128838584485e-07,
+      "loss": 0.0728,
+      "step": 8618
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9837009311551535e-07,
+      "loss": 0.063,
+      "step": 8619
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9752847301252165e-07,
+      "loss": 0.0666,
+      "step": 8620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9668802365102054e-07,
+      "loss": 0.0669,
+      "step": 8621
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9584874513242544e-07,
+      "loss": 0.0698,
+      "step": 8622
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.950106375580053e-07,
+      "loss": 0.0777,
+      "step": 8623
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.941737010288903e-07,
+      "loss": 0.0647,
+      "step": 8624
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.933379356460686e-07,
+      "loss": 0.0717,
+      "step": 8625
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.925033415103873e-07,
+      "loss": 0.0674,
+      "step": 8626
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9166991872255355e-07,
+      "loss": 0.0769,
+      "step": 8627
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9083766738313035e-07,
+      "loss": 0.0735,
+      "step": 8628
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.900065875925406e-07,
+      "loss": 0.0677,
+      "step": 8629
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.891766794510664e-07,
+      "loss": 0.0751,
+      "step": 8630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8834794305884874e-07,
+      "loss": 0.0659,
+      "step": 8631
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.875203785158831e-07,
+      "loss": 0.0746,
+      "step": 8632
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.866939859220308e-07,
+      "loss": 0.0658,
+      "step": 8633
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8586876537700645e-07,
+      "loss": 0.0701,
+      "step": 8634
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.850447169803849e-07,
+      "loss": 0.0696,
+      "step": 8635
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.842218408315989e-07,
+      "loss": 0.0743,
+      "step": 8636
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8340013702993995e-07,
+      "loss": 0.0696,
+      "step": 8637
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.825796056745589e-07,
+      "loss": 0.0676,
+      "step": 8638
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8176024686446425e-07,
+      "loss": 0.0668,
+      "step": 8639
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.809420606985236e-07,
+      "loss": 0.0678,
+      "step": 8640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.801250472754635e-07,
+      "loss": 0.0678,
+      "step": 8641
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7930920669386743e-07,
+      "loss": 0.0752,
+      "step": 8642
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.784945390521765e-07,
+      "loss": 0.07,
+      "step": 8643
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.776810444486944e-07,
+      "loss": 0.0647,
+      "step": 8644
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.768687229815803e-07,
+      "loss": 0.0725,
+      "step": 8645
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7605757474885255e-07,
+      "loss": 0.0722,
+      "step": 8646
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.752475998483872e-07,
+      "loss": 0.0682,
+      "step": 8647
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.744387983779195e-07,
+      "loss": 0.0825,
+      "step": 8648
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.736311704350425e-07,
+      "loss": 0.0633,
+      "step": 8649
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7282471611720817e-07,
+      "loss": 0.0676,
+      "step": 8650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.720194355217276e-07,
+      "loss": 0.0686,
+      "step": 8651
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7121532874576863e-07,
+      "loss": 0.0671,
+      "step": 8652
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7041239588635693e-07,
+      "loss": 0.0727,
+      "step": 8653
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6961063704038057e-07,
+      "loss": 0.063,
+      "step": 8654
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6881005230458115e-07,
+      "loss": 0.0665,
+      "step": 8655
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6801064177556123e-07,
+      "loss": 0.075,
+      "step": 8656
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6721240554978046e-07,
+      "loss": 0.0698,
+      "step": 8657
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.664153437235584e-07,
+      "loss": 0.0651,
+      "step": 8658
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.656194563930714e-07,
+      "loss": 0.0657,
+      "step": 8659
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6482474365435604e-07,
+      "loss": 0.0694,
+      "step": 8660
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6403120560330233e-07,
+      "loss": 0.0654,
+      "step": 8661
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6323884233566576e-07,
+      "loss": 0.0652,
+      "step": 8662
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.624476539470544e-07,
+      "loss": 0.0663,
+      "step": 8663
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.616576405329352e-07,
+      "loss": 0.0691,
+      "step": 8664
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6086880218863633e-07,
+      "loss": 0.0668,
+      "step": 8665
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.600811390093427e-07,
+      "loss": 0.0685,
+      "step": 8666
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.592946510900962e-07,
+      "loss": 0.0663,
+      "step": 8667
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5850933852579865e-07,
+      "loss": 0.0689,
+      "step": 8668
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5772520141120747e-07,
+      "loss": 0.06,
+      "step": 8669
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5694223984094044e-07,
+      "loss": 0.07,
+      "step": 8670
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.561604539094742e-07,
+      "loss": 0.0772,
+      "step": 8671
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.553798437111421e-07,
+      "loss": 0.0692,
+      "step": 8672
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.546004093401355e-07,
+      "loss": 0.0688,
+      "step": 8673
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.538221508905048e-07,
+      "loss": 0.0669,
+      "step": 8674
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5304506845615695e-07,
+      "loss": 0.0729,
+      "step": 8675
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5226916213085706e-07,
+      "loss": 0.0634,
+      "step": 8676
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5149443200823244e-07,
+      "loss": 0.0724,
+      "step": 8677
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.0704,
+      "step": 8678
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4994850074479104e-07,
+      "loss": 0.0737,
+      "step": 8679
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.49177299790514e-07,
+      "loss": 0.0703,
+      "step": 8680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4840727541198597e-07,
+      "loss": 0.0707,
+      "step": 8681
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4763842770212374e-07,
+      "loss": 0.0691,
+      "step": 8682
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4687075675369985e-07,
+      "loss": 0.0644,
+      "step": 8683
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.461042626593435e-07,
+      "loss": 0.0704,
+      "step": 8684
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4533894551154404e-07,
+      "loss": 0.0665,
+      "step": 8685
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4457480540264864e-07,
+      "loss": 0.077,
+      "step": 8686
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.438118424248592e-07,
+      "loss": 0.0725,
+      "step": 8687
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4305005667023985e-07,
+      "loss": 0.0705,
+      "step": 8688
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4228944823071144e-07,
+      "loss": 0.0637,
+      "step": 8689
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4153001719805057e-07,
+      "loss": 0.0693,
+      "step": 8690
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.40771763663894e-07,
+      "loss": 0.0699,
+      "step": 8691
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4001468771973623e-07,
+      "loss": 0.0655,
+      "step": 8692
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.392587894569309e-07,
+      "loss": 0.0644,
+      "step": 8693
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.38504068966684e-07,
+      "loss": 0.0726,
+      "step": 8694
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3775052634006702e-07,
+      "loss": 0.0734,
+      "step": 8695
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3699816166800284e-07,
+      "loss": 0.067,
+      "step": 8696
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3624697504127547e-07,
+      "loss": 0.0694,
+      "step": 8697
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.354969665505291e-07,
+      "loss": 0.0705,
+      "step": 8698
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3474813628626026e-07,
+      "loss": 0.0682,
+      "step": 8699
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.340004843388255e-07,
+      "loss": 0.0689,
+      "step": 8700
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3325401079844157e-07,
+      "loss": 0.0646,
+      "step": 8701
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3250871575518086e-07,
+      "loss": 0.0659,
+      "step": 8702
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3176459929897143e-07,
+      "loss": 0.0642,
+      "step": 8703
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3102166151960482e-07,
+      "loss": 0.0682,
+      "step": 8704
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3027990250672593e-07,
+      "loss": 0.0733,
+      "step": 8705
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2953932234983766e-07,
+      "loss": 0.0741,
+      "step": 8706
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2879992113830297e-07,
+      "loss": 0.0706,
+      "step": 8707
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2806169896134045e-07,
+      "loss": 0.067,
+      "step": 8708
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2732465590802554e-07,
+      "loss": 0.0689,
+      "step": 8709
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.265887920672949e-07,
+      "loss": 0.0654,
+      "step": 8710
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.258541075279419e-07,
+      "loss": 0.0703,
+      "step": 8711
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2512060237861455e-07,
+      "loss": 0.067,
+      "step": 8712
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2438827670782203e-07,
+      "loss": 0.0679,
+      "step": 8713
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2365713060392925e-07,
+      "loss": 0.0754,
+      "step": 8714
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.229271641551578e-07,
+      "loss": 0.0707,
+      "step": 8715
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2219837744959284e-07,
+      "loss": 0.0638,
+      "step": 8716
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2147077057516952e-07,
+      "loss": 0.0748,
+      "step": 8717
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2074434361968322e-07,
+      "loss": 0.0687,
+      "step": 8718
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2001909667079158e-07,
+      "loss": 0.0721,
+      "step": 8719
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.192950298160035e-07,
+      "loss": 0.0659,
+      "step": 8720
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1857214314268571e-07,
+      "loss": 0.0674,
+      "step": 8721
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.178504367380696e-07,
+      "loss": 0.068,
+      "step": 8722
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1712991068923662e-07,
+      "loss": 0.0737,
+      "step": 8723
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1641056508312718e-07,
+      "loss": 0.0739,
+      "step": 8724
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1569240000654523e-07,
+      "loss": 0.0679,
+      "step": 8725
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1497541554614255e-07,
+      "loss": 0.0735,
+      "step": 8726
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1425961178843545e-07,
+      "loss": 0.0685,
+      "step": 8727
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1354498881979714e-07,
+      "loss": 0.0692,
+      "step": 8728
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1283154672645522e-07,
+      "loss": 0.0631,
+      "step": 8729
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1211928559449756e-07,
+      "loss": 0.0723,
+      "step": 8730
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1140820550986874e-07,
+      "loss": 0.0647,
+      "step": 8731
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.106983065583701e-07,
+      "loss": 0.0681,
+      "step": 8732
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.09989588825662e-07,
+      "loss": 0.0693,
+      "step": 8733
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0928205239726053e-07,
+      "loss": 0.0657,
+      "step": 8734
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0857569735853956e-07,
+      "loss": 0.0753,
+      "step": 8735
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0787052379473204e-07,
+      "loss": 0.0677,
+      "step": 8736
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.071665317909266e-07,
+      "loss": 0.0639,
+      "step": 8737
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0646372143207083e-07,
+      "loss": 0.0625,
+      "step": 8738
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0576209280296688e-07,
+      "loss": 0.0686,
+      "step": 8739
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0506164598827815e-07,
+      "loss": 0.071,
+      "step": 8740
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.043623810725215e-07,
+      "loss": 0.0698,
+      "step": 8741
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0366429814007605e-07,
+      "loss": 0.0733,
+      "step": 8742
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0296739727517335e-07,
+      "loss": 0.0651,
+      "step": 8743
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0227167856190612e-07,
+      "loss": 0.0706,
+      "step": 8744
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0157714208422053e-07,
+      "loss": 0.0672,
+      "step": 8745
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0088378792592288e-07,
+      "loss": 0.0647,
+      "step": 8746
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0019161617067738e-07,
+      "loss": 0.0692,
+      "step": 8747
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.995006269020039e-07,
+      "loss": 0.0681,
+      "step": 8748
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9881082020327901e-07,
+      "loss": 0.0656,
+      "step": 8749
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9812219615774064e-07,
+      "loss": 0.0646,
+      "step": 8750
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9743475484847785e-07,
+      "loss": 0.0681,
+      "step": 8751
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9674849635844206e-07,
+      "loss": 0.0729,
+      "step": 8752
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9606342077043815e-07,
+      "loss": 0.0671,
+      "step": 8753
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9537952816713334e-07,
+      "loss": 0.0716,
+      "step": 8754
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9469681863104716e-07,
+      "loss": 0.0665,
+      "step": 8755
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9401529224455817e-07,
+      "loss": 0.0656,
+      "step": 8756
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.933349490899028e-07,
+      "loss": 0.0682,
+      "step": 8757
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.926557892491754e-07,
+      "loss": 0.0654,
+      "step": 8758
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.919778128043226e-07,
+      "loss": 0.071,
+      "step": 8759
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9130101983715455e-07,
+      "loss": 0.065,
+      "step": 8760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9062541042933703e-07,
+      "loss": 0.0673,
+      "step": 8761
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8995098466238816e-07,
+      "loss": 0.0721,
+      "step": 8762
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.892777426176906e-07,
+      "loss": 0.0631,
+      "step": 8763
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8860568437648053e-07,
+      "loss": 0.0665,
+      "step": 8764
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.879348100198486e-07,
+      "loss": 0.0701,
+      "step": 8765
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8726511962874783e-07,
+      "loss": 0.0744,
+      "step": 8766
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.865966132839847e-07,
+      "loss": 0.0707,
+      "step": 8767
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.859292910662236e-07,
+      "loss": 0.0689,
+      "step": 8768
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8526315305598785e-07,
+      "loss": 0.0707,
+      "step": 8769
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8459819933365541e-07,
+      "loss": 0.0772,
+      "step": 8770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.839344299794632e-07,
+      "loss": 0.0664,
+      "step": 8771
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8327184507350493e-07,
+      "loss": 0.0686,
+      "step": 8772
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8261044469572998e-07,
+      "loss": 0.066,
+      "step": 8773
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8195022892594449e-07,
+      "loss": 0.0702,
+      "step": 8774
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8129119784381478e-07,
+      "loss": 0.0684,
+      "step": 8775
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8063335152886273e-07,
+      "loss": 0.0672,
+      "step": 8776
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7997669006046604e-07,
+      "loss": 0.0694,
+      "step": 8777
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7932121351785903e-07,
+      "loss": 0.0627,
+      "step": 8778
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7866692198013623e-07,
+      "loss": 0.0662,
+      "step": 8779
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7801381552624565e-07,
+      "loss": 0.0692,
+      "step": 8780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7736189423499638e-07,
+      "loss": 0.0665,
+      "step": 8781
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7671115818504892e-07,
+      "loss": 0.0661,
+      "step": 8782
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7606160745492595e-07,
+      "loss": 0.066,
+      "step": 8783
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7541324212300482e-07,
+      "loss": 0.0713,
+      "step": 8784
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7476606226751958e-07,
+      "loss": 0.0667,
+      "step": 8785
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7412006796656e-07,
+      "loss": 0.0688,
+      "step": 8786
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7347525929807817e-07,
+      "loss": 0.0689,
+      "step": 8787
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.728316363398763e-07,
+      "loss": 0.0699,
+      "step": 8788
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7218919916961786e-07,
+      "loss": 0.0696,
+      "step": 8789
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.715479478648241e-07,
+      "loss": 0.0686,
+      "step": 8790
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.709078825028676e-07,
+      "loss": 0.0613,
+      "step": 8791
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7026900316098217e-07,
+      "loss": 0.0711,
+      "step": 8792
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6963130991625942e-07,
+      "loss": 0.0744,
+      "step": 8793
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6899480284564562e-07,
+      "loss": 0.0655,
+      "step": 8794
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6835948202594156e-07,
+      "loss": 0.0688,
+      "step": 8795
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6772534753381254e-07,
+      "loss": 0.0668,
+      "step": 8796
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6709239944577405e-07,
+      "loss": 0.0704,
+      "step": 8797
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.664606378381972e-07,
+      "loss": 0.0764,
+      "step": 8798
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.658300627873166e-07,
+      "loss": 0.0664,
+      "step": 8799
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6520067436922027e-07,
+      "loss": 0.0653,
+      "step": 8800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6457247265984967e-07,
+      "loss": 0.0652,
+      "step": 8801
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6394545773500968e-07,
+      "loss": 0.0669,
+      "step": 8802
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6331962967035764e-07,
+      "loss": 0.0747,
+      "step": 8803
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6269498854140752e-07,
+      "loss": 0.0633,
+      "step": 8804
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.620715344235324e-07,
+      "loss": 0.0702,
+      "step": 8805
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6144926739195989e-07,
+      "loss": 0.0662,
+      "step": 8806
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6082818752177654e-07,
+      "loss": 0.0721,
+      "step": 8807
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6020829488792355e-07,
+      "loss": 0.0777,
+      "step": 8808
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.595895895652e-07,
+      "loss": 0.0832,
+      "step": 8809
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5897207162826167e-07,
+      "loss": 0.0718,
+      "step": 8810
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5835574115162123e-07,
+      "loss": 0.0691,
+      "step": 8811
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5774059820964693e-07,
+      "loss": 0.0635,
+      "step": 8812
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5712664287656388e-07,
+      "loss": 0.0675,
+      "step": 8813
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5651387522645721e-07,
+      "loss": 0.0712,
+      "step": 8814
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.559022953332634e-07,
+      "loss": 0.0689,
+      "step": 8815
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5529190327078003e-07,
+      "loss": 0.0639,
+      "step": 8816
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5468269911265933e-07,
+      "loss": 0.0696,
+      "step": 8817
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5407468293240912e-07,
+      "loss": 0.069,
+      "step": 8818
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5346785480339633e-07,
+      "loss": 0.0673,
+      "step": 8819
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5286221479884343e-07,
+      "loss": 0.0663,
+      "step": 8820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.522577629918298e-07,
+      "loss": 0.0683,
+      "step": 8821
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5165449945529043e-07,
+      "loss": 0.0701,
+      "step": 8822
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5105242426201815e-07,
+      "loss": 0.0659,
+      "step": 8823
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5045153748466046e-07,
+      "loss": 0.0654,
+      "step": 8824
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4985183919572598e-07,
+      "loss": 0.0685,
+      "step": 8825
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4925332946757465e-07,
+      "loss": 0.0654,
+      "step": 8826
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4865600837242533e-07,
+      "loss": 0.0688,
+      "step": 8827
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4805987598235261e-07,
+      "loss": 0.0714,
+      "step": 8828
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4746493236929117e-07,
+      "loss": 0.0689,
+      "step": 8829
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4687117760502579e-07,
+      "loss": 0.0692,
+      "step": 8830
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.462786117612036e-07,
+      "loss": 0.0665,
+      "step": 8831
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4568723490932746e-07,
+      "loss": 0.07,
+      "step": 8832
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4509704712075247e-07,
+      "loss": 0.0688,
+      "step": 8833
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4450804846669387e-07,
+      "loss": 0.0674,
+      "step": 8834
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4392023901822481e-07,
+      "loss": 0.0702,
+      "step": 8835
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4333361884626974e-07,
+      "loss": 0.0652,
+      "step": 8836
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4274818802161527e-07,
+      "loss": 0.0618,
+      "step": 8837
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4216394661490163e-07,
+      "loss": 0.068,
+      "step": 8838
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.415808946966246e-07,
+      "loss": 0.0667,
+      "step": 8839
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4099903233713909e-07,
+      "loss": 0.0676,
+      "step": 8840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.404183596066533e-07,
+      "loss": 0.0747,
+      "step": 8841
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3983887657523676e-07,
+      "loss": 0.0704,
+      "step": 8842
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3926058331280912e-07,
+      "loss": 0.0612,
+      "step": 8843
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.386834798891512e-07,
+      "loss": 0.0673,
+      "step": 8844
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3810756637389844e-07,
+      "loss": 0.0672,
+      "step": 8845
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.37532842836543e-07,
+      "loss": 0.07,
+      "step": 8846
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3695930934643387e-07,
+      "loss": 0.0676,
+      "step": 8847
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.0757,
+      "step": 8848
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3581581278463095e-07,
+      "loss": 0.0796,
+      "step": 8849
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3524584985091572e-07,
+      "loss": 0.0715,
+      "step": 8850
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3467707724040491e-07,
+      "loss": 0.0707,
+      "step": 8851
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3410949502172808e-07,
+      "loss": 0.0712,
+      "step": 8852
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3354310326337383e-07,
+      "loss": 0.0779,
+      "step": 8853
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.329779020336841e-07,
+      "loss": 0.0726,
+      "step": 8854
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3241389140086103e-07,
+      "loss": 0.0755,
+      "step": 8855
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3185107143295684e-07,
+      "loss": 0.0748,
+      "step": 8856
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3128944219788497e-07,
+      "loss": 0.0667,
+      "step": 8857
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3072900376341459e-07,
+      "loss": 0.0719,
+      "step": 8858
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3016975619717042e-07,
+      "loss": 0.0689,
+      "step": 8859
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.296116995666341e-07,
+      "loss": 0.0731,
+      "step": 8860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2905483393914287e-07,
+      "loss": 0.0644,
+      "step": 8861
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2849915938189072e-07,
+      "loss": 0.0748,
+      "step": 8862
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2794467596192628e-07,
+      "loss": 0.0739,
+      "step": 8863
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.273913837461571e-07,
+      "loss": 0.0668,
+      "step": 8864
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.268392828013454e-07,
+      "loss": 0.0695,
+      "step": 8865
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2628837319411002e-07,
+      "loss": 0.0642,
+      "step": 8866
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2573865499092675e-07,
+      "loss": 0.0659,
+      "step": 8867
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2519012825812804e-07,
+      "loss": 0.0653,
+      "step": 8868
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2464279306189765e-07,
+      "loss": 0.0724,
+      "step": 8869
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2409664946828158e-07,
+      "loss": 0.0661,
+      "step": 8870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2355169754318053e-07,
+      "loss": 0.0689,
+      "step": 8871
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.230079373523496e-07,
+      "loss": 0.0682,
+      "step": 8872
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2246536896140192e-07,
+      "loss": 0.0681,
+      "step": 8873
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.219239924358062e-07,
+      "loss": 0.067,
+      "step": 8874
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2138380784088578e-07,
+      "loss": 0.0703,
+      "step": 8875
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.20844815241824e-07,
+      "loss": 0.0689,
+      "step": 8876
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2030701470365557e-07,
+      "loss": 0.0733,
+      "step": 8877
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1977040629127524e-07,
+      "loss": 0.0625,
+      "step": 8878
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1923499006943228e-07,
+      "loss": 0.0682,
+      "step": 8879
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1870076610273285e-07,
+      "loss": 0.0672,
+      "step": 8880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1816773445563646e-07,
+      "loss": 0.0636,
+      "step": 8881
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1763589519246388e-07,
+      "loss": 0.0688,
+      "step": 8882
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1710524837738713e-07,
+      "loss": 0.0732,
+      "step": 8883
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1657579407443608e-07,
+      "loss": 0.0734,
+      "step": 8884
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1604753234749855e-07,
+      "loss": 0.0652,
+      "step": 8885
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1552046326031685e-07,
+      "loss": 0.0712,
+      "step": 8886
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1499458687648791e-07,
+      "loss": 0.0758,
+      "step": 8887
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1446990325946649e-07,
+      "loss": 0.07,
+      "step": 8888
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1394641247256421e-07,
+      "loss": 0.0719,
+      "step": 8889
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1342411457894609e-07,
+      "loss": 0.069,
+      "step": 8890
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1290300964163613e-07,
+      "loss": 0.0638,
+      "step": 8891
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1238309772351297e-07,
+      "loss": 0.0689,
+      "step": 8892
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1186437888730972e-07,
+      "loss": 0.0605,
+      "step": 8893
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1134685319562077e-07,
+      "loss": 0.0695,
+      "step": 8894
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1083052071089063e-07,
+      "loss": 0.0633,
+      "step": 8895
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1031538149542054e-07,
+      "loss": 0.0701,
+      "step": 8896
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0980143561137191e-07,
+      "loss": 0.0687,
+      "step": 8897
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0928868312075958e-07,
+      "loss": 0.0675,
+      "step": 8898
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0877712408545294e-07,
+      "loss": 0.0701,
+      "step": 8899
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.082667585671815e-07,
+      "loss": 0.0694,
+      "step": 8900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0775758662752377e-07,
+      "loss": 0.0714,
+      "step": 8901
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0724960832792285e-07,
+      "loss": 0.0763,
+      "step": 8902
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0674282372967192e-07,
+      "loss": 0.0604,
+      "step": 8903
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0623723289392097e-07,
+      "loss": 0.0703,
+      "step": 8904
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0573283588167782e-07,
+      "loss": 0.0679,
+      "step": 8905
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0522963275380494e-07,
+      "loss": 0.0665,
+      "step": 8906
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.047276235710204e-07,
+      "loss": 0.0669,
+      "step": 8907
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0422680839390021e-07,
+      "loss": 0.0758,
+      "step": 8908
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0372718728287379e-07,
+      "loss": 0.0703,
+      "step": 8909
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0322876029822738e-07,
+      "loss": 0.0655,
+      "step": 8910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0273152750010285e-07,
+      "loss": 0.0615,
+      "step": 8911
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0223548894849999e-07,
+      "loss": 0.0675,
+      "step": 8912
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0174064470327206e-07,
+      "loss": 0.0622,
+      "step": 8913
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0124699482412903e-07,
+      "loss": 0.0638,
+      "step": 8914
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0075453937063662e-07,
+      "loss": 0.0694,
+      "step": 8915
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0026327840221728e-07,
+      "loss": 0.0772,
+      "step": 8916
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.977321197814694e-08,
+      "loss": 0.0663,
+      "step": 8917
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.928434015756161e-08,
+      "loss": 0.0661,
+      "step": 8918
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.879666299944857e-08,
+      "loss": 0.0699,
+      "step": 8919
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.831018056265518e-08,
+      "loss": 0.0745,
+      "step": 8920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.782489290588004e-08,
+      "loss": 0.0657,
+      "step": 8921
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.734080008768076e-08,
+      "loss": 0.0637,
+      "step": 8922
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.685790216647173e-08,
+      "loss": 0.0665,
+      "step": 8923
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.637619920051856e-08,
+      "loss": 0.0698,
+      "step": 8924
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.589569124794918e-08,
+      "loss": 0.0685,
+      "step": 8925
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.541637836674056e-08,
+      "loss": 0.0646,
+      "step": 8926
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.493826061472977e-08,
+      "loss": 0.0647,
+      "step": 8927
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.446133804960733e-08,
+      "loss": 0.0677,
+      "step": 8928
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.398561072892387e-08,
+      "loss": 0.0635,
+      "step": 8929
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.351107871008014e-08,
+      "loss": 0.0699,
+      "step": 8930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.303774205033589e-08,
+      "loss": 0.0697,
+      "step": 8931
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.256560080680543e-08,
+      "loss": 0.071,
+      "step": 8932
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.209465503646098e-08,
+      "loss": 0.0778,
+      "step": 8933
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.162490479612596e-08,
+      "loss": 0.0754,
+      "step": 8934
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.115635014248502e-08,
+      "loss": 0.0722,
+      "step": 8935
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.068899113207519e-08,
+      "loss": 0.0726,
+      "step": 8936
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.022282782128911e-08,
+      "loss": 0.0685,
+      "step": 8937
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.975786026637734e-08,
+      "loss": 0.0709,
+      "step": 8938
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.929408852344501e-08,
+      "loss": 0.0758,
+      "step": 8939
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.883151264845069e-08,
+      "loss": 0.0687,
+      "step": 8940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.837013269721306e-08,
+      "loss": 0.073,
+      "step": 8941
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.790994872540315e-08,
+      "loss": 0.0711,
+      "step": 8942
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.745096078854765e-08,
+      "loss": 0.0667,
+      "step": 8943
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.699316894203225e-08,
+      "loss": 0.0648,
+      "step": 8944
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.653657324109499e-08,
+      "loss": 0.0609,
+      "step": 8945
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.608117374083069e-08,
+      "loss": 0.0674,
+      "step": 8946
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.562697049618874e-08,
+      "loss": 0.0705,
+      "step": 8947
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.517396356197749e-08,
+      "loss": 0.0657,
+      "step": 8948
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.472215299285658e-08,
+      "loss": 0.0647,
+      "step": 8949
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.42715388433446e-08,
+      "loss": 0.0616,
+      "step": 8950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.382212116781475e-08,
+      "loss": 0.0648,
+      "step": 8951
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.337390002049473e-08,
+      "loss": 0.0689,
+      "step": 8952
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.292687545546907e-08,
+      "loss": 0.0701,
+      "step": 8953
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.248104752667796e-08,
+      "loss": 0.0643,
+      "step": 8954
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.203641628791614e-08,
+      "loss": 0.0753,
+      "step": 8955
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.159298179283515e-08,
+      "loss": 0.0718,
+      "step": 8956
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.115074409494328e-08,
+      "loss": 0.0675,
+      "step": 8957
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.070970324760009e-08,
+      "loss": 0.0714,
+      "step": 8958
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02698593040252e-08,
+      "loss": 0.0673,
+      "step": 8959
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.983121231729174e-08,
+      "loss": 0.0704,
+      "step": 8960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.939376234032847e-08,
+      "loss": 0.0707,
+      "step": 8961
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.895750942591984e-08,
+      "loss": 0.0666,
+      "step": 8962
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.852245362670707e-08,
+      "loss": 0.0714,
+      "step": 8963
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.808859499518374e-08,
+      "loss": 0.0717,
+      "step": 8964
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.765593358370349e-08,
+      "loss": 0.0643,
+      "step": 8965
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.722446944447126e-08,
+      "loss": 0.0653,
+      "step": 8966
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.679420262954984e-08,
+      "loss": 0.065,
+      "step": 8967
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.636513319085659e-08,
+      "loss": 0.0627,
+      "step": 8968
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.593726118016676e-08,
+      "loss": 0.0681,
+      "step": 8969
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.551058664910683e-08,
+      "loss": 0.0705,
+      "step": 8970
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.508510964916338e-08,
+      "loss": 0.0664,
+      "step": 8971
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.466083023167536e-08,
+      "loss": 0.0673,
+      "step": 8972
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.423774844783626e-08,
+      "loss": 0.0658,
+      "step": 8973
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.38158643486997e-08,
+      "loss": 0.0708,
+      "step": 8974
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.339517798517048e-08,
+      "loss": 0.0648,
+      "step": 8975
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.297568940801136e-08,
+      "loss": 0.0668,
+      "step": 8976
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.25573986678385e-08,
+      "loss": 0.0695,
+      "step": 8977
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.214030581512599e-08,
+      "loss": 0.0659,
+      "step": 8978
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.172441090020022e-08,
+      "loss": 0.0733,
+      "step": 8979
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.13097139732466e-08,
+      "loss": 0.0679,
+      "step": 8980
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.0896215084304e-08,
+      "loss": 0.0716,
+      "step": 8981
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.048391428326585e-08,
+      "loss": 0.0663,
+      "step": 8982
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.007281161988233e-08,
+      "loss": 0.0678,
+      "step": 8983
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.966290714375934e-08,
+      "loss": 0.0647,
+      "step": 8984
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.925420090435842e-08,
+      "loss": 0.0666,
+      "step": 8985
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.884669295099456e-08,
+      "loss": 0.0721,
+      "step": 8986
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.844038333283953e-08,
+      "loss": 0.0683,
+      "step": 8987
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.80352720989208e-08,
+      "loss": 0.0649,
+      "step": 8988
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.763135929812036e-08,
+      "loss": 0.0665,
+      "step": 8989
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.722864497917703e-08,
+      "loss": 0.0706,
+      "step": 8990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.682712919068301e-08,
+      "loss": 0.0668,
+      "step": 8991
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.642681198108735e-08,
+      "loss": 0.0601,
+      "step": 8992
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.602769339869475e-08,
+      "loss": 0.0666,
+      "step": 8993
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.562977349166222e-08,
+      "loss": 0.0681,
+      "step": 8994
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.523305230800691e-08,
+      "loss": 0.0667,
+      "step": 8995
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.483752989559722e-08,
+      "loss": 0.064,
+      "step": 8996
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.44432063021605e-08,
+      "loss": 0.0779,
+      "step": 8997
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.405008157527537e-08,
+      "loss": 0.0634,
+      "step": 8998
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.365815576237944e-08,
+      "loss": 0.067,
+      "step": 8999
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.326742891076376e-08,
+      "loss": 0.0755,
+      "step": 9000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.287790106757396e-08,
+      "loss": 0.0689,
+      "step": 9001
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.248957227981467e-08,
+      "loss": 0.0671,
+      "step": 9002
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.210244259433951e-08,
+      "loss": 0.0665,
+      "step": 9003
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.171651205786556e-08,
+      "loss": 0.0693,
+      "step": 9004
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.133178071695778e-08,
+      "loss": 0.0654,
+      "step": 9005
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.094824861804017e-08,
+      "loss": 0.0722,
+      "step": 9006
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.056591580739235e-08,
+      "loss": 0.0712,
+      "step": 9007
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.018478233114634e-08,
+      "loss": 0.0651,
+      "step": 9008
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.980484823529309e-08,
+      "loss": 0.0656,
+      "step": 9009
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.942611356567707e-08,
+      "loss": 0.0658,
+      "step": 9010
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9048578367996157e-08,
+      "loss": 0.0667,
+      "step": 9011
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8672242687807246e-08,
+      "loss": 0.0671,
+      "step": 9012
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.829710657051957e-08,
+      "loss": 0.0719,
+      "step": 9013
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.792317006139914e-08,
+      "loss": 0.0721,
+      "step": 9014
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7550433205566523e-08,
+      "loss": 0.0696,
+      "step": 9015
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.717889604799798e-08,
+      "loss": 0.0701,
+      "step": 9016
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6808558633524304e-08,
+      "loss": 0.0666,
+      "step": 9017
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6439421006833086e-08,
+      "loss": 0.0677,
+      "step": 9018
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.607148321246425e-08,
+      "loss": 0.0691,
+      "step": 9019
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5704745294815624e-08,
+      "loss": 0.07,
+      "step": 9020
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.533920729814068e-08,
+      "loss": 0.0641,
+      "step": 9021
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4974869266544165e-08,
+      "loss": 0.0684,
+      "step": 9022
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.461173124399088e-08,
+      "loss": 0.073,
+      "step": 9023
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.424979327429691e-08,
+      "loss": 0.0683,
+      "step": 9024
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3889055401137315e-08,
+      "loss": 0.069,
+      "step": 9025
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.352951766803727e-08,
+      "loss": 0.0719,
+      "step": 9026
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.317118011838318e-08,
+      "loss": 0.0701,
+      "step": 9027
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.281404279541269e-08,
+      "loss": 0.0706,
+      "step": 9028
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.245810574221799e-08,
+      "loss": 0.0695,
+      "step": 9029
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.210336900175028e-08,
+      "loss": 0.0744,
+      "step": 9030
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.174983261681088e-08,
+      "loss": 0.0652,
+      "step": 9031
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1397496630061216e-08,
+      "loss": 0.0647,
+      "step": 9032
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.104636108401506e-08,
+      "loss": 0.0695,
+      "step": 9033
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0696426021041856e-08,
+      "loss": 0.0722,
+      "step": 9034
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0347691483365604e-08,
+      "loss": 0.0708,
+      "step": 9035
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.000015751306819e-08,
+      "loss": 0.064,
+      "step": 9036
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.965382415208164e-08,
+      "loss": 0.0692,
+      "step": 9037
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9308691442196964e-08,
+      "loss": 0.0669,
+      "step": 9038
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.896475942506085e-08,
+      "loss": 0.0684,
+      "step": 9039
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8622028142172316e-08,
+      "loss": 0.0666,
+      "step": 9040
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.828049763488718e-08,
+      "loss": 0.0675,
+      "step": 9041
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7940167944415804e-08,
+      "loss": 0.0682,
+      "step": 9042
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.760103911182423e-08,
+      "loss": 0.0706,
+      "step": 9043
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.726311117803084e-08,
+      "loss": 0.0613,
+      "step": 9044
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.692638418381523e-08,
+      "loss": 0.0664,
+      "step": 9045
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6590858169806016e-08,
+      "loss": 0.0739,
+      "step": 9046
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6256533176488594e-08,
+      "loss": 0.0726,
+      "step": 9047
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.592340924420624e-08,
+      "loss": 0.0641,
+      "step": 9048
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5591486413154584e-08,
+      "loss": 0.0701,
+      "step": 9049
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.52607647233827e-08,
+      "loss": 0.0752,
+      "step": 9050
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.493124421479977e-08,
+      "loss": 0.0688,
+      "step": 9051
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.460292492716512e-08,
+      "loss": 0.0684,
+      "step": 9052
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4275806900094807e-08,
+      "loss": 0.0643,
+      "step": 9053
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3949890173061726e-08,
+      "loss": 0.0638,
+      "step": 9054
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.36251747853933e-08,
+      "loss": 0.0821,
+      "step": 9055
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3301660776268186e-08,
+      "loss": 0.0661,
+      "step": 9056
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2979348184725156e-08,
+      "loss": 0.0687,
+      "step": 9057
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2658237049655325e-08,
+      "loss": 0.073,
+      "step": 9058
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.233832740980437e-08,
+      "loss": 0.0623,
+      "step": 9059
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.201961930377474e-08,
+      "loss": 0.0726,
+      "step": 9060
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.170211277002345e-08,
+      "loss": 0.0744,
+      "step": 9061
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1385807846862084e-08,
+      "loss": 0.0627,
+      "step": 9062
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1070704572456764e-08,
+      "loss": 0.0731,
+      "step": 9063
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0756802984829315e-08,
+      "loss": 0.0662,
+      "step": 9064
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.044410312185609e-08,
+      "loss": 0.071,
+      "step": 9065
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.013260502127026e-08,
+      "loss": 0.0667,
+      "step": 9066
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.982230872065729e-08,
+      "loss": 0.0639,
+      "step": 9067
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.951321425745946e-08,
+      "loss": 0.0688,
+      "step": 9068
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9205321668972506e-08,
+      "loss": 0.0687,
+      "step": 9069
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.889863099234892e-08,
+      "loss": 0.0703,
+      "step": 9070
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.859314226459465e-08,
+      "loss": 0.0734,
+      "step": 9071
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.828885552257244e-08,
+      "loss": 0.0696,
+      "step": 9072
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7985770802997365e-08,
+      "loss": 0.0655,
+      "step": 9073
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.768388814244128e-08,
+      "loss": 0.0706,
+      "step": 9074
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.738320757733283e-08,
+      "loss": 0.071,
+      "step": 9075
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.708372914394964e-08,
+      "loss": 0.0709,
+      "step": 9076
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6785452878429494e-08,
+      "loss": 0.0775,
+      "step": 9077
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.648837881676581e-08,
+      "loss": 0.0755,
+      "step": 9078
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6192506994802146e-08,
+      "loss": 0.0678,
+      "step": 9079
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5897837448239935e-08,
+      "loss": 0.0661,
+      "step": 9080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.560437021263741e-08,
+      "loss": 0.0755,
+      "step": 9081
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.53121053234029e-08,
+      "loss": 0.0698,
+      "step": 9082
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5021042815804874e-08,
+      "loss": 0.0679,
+      "step": 9083
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.47311827249619e-08,
+      "loss": 0.0701,
+      "step": 9084
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4442525085850446e-08,
+      "loss": 0.071,
+      "step": 9085
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4155069933301535e-08,
+      "loss": 0.0682,
+      "step": 9086
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.386881730200076e-08,
+      "loss": 0.0704,
+      "step": 9087
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.358376722648826e-08,
+      "loss": 0.0704,
+      "step": 9088
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.329991974115987e-08,
+      "loss": 0.0707,
+      "step": 9089
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.301727488026485e-08,
+      "loss": 0.0683,
+      "step": 9090
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.273583267790925e-08,
+      "loss": 0.066,
+      "step": 9091
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2455593168052576e-08,
+      "loss": 0.0706,
+      "step": 9092
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.217655638451112e-08,
+      "loss": 0.0672,
+      "step": 9093
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1898722360952374e-08,
+      "loss": 0.0671,
+      "step": 9094
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.162209113090286e-08,
+      "loss": 0.0732,
+      "step": 9095
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.134666272774034e-08,
+      "loss": 0.0653,
+      "step": 9096
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.107243718470043e-08,
+      "loss": 0.0667,
+      "step": 9097
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.079941453487223e-08,
+      "loss": 0.0695,
+      "step": 9098
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.052759481119938e-08,
+      "loss": 0.0743,
+      "step": 9099
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.02569780464812e-08,
+      "loss": 0.0687,
+      "step": 9100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.998756427337157e-08,
+      "loss": 0.0708,
+      "step": 9101
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9719353524378936e-08,
+      "loss": 0.0719,
+      "step": 9102
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9452345831866292e-08,
+      "loss": 0.0625,
+      "step": 9103
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9186541228052313e-08,
+      "loss": 0.0642,
+      "step": 9104
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.892193974501023e-08,
+      "loss": 0.0702,
+      "step": 9105
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.865854141466784e-08,
+      "loss": 0.0693,
+      "step": 9106
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8396346268807497e-08,
+      "loss": 0.0667,
+      "step": 9107
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8135354339067223e-08,
+      "loss": 0.074,
+      "step": 9108
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7875565656939608e-08,
+      "loss": 0.0712,
+      "step": 9109
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7616980253771798e-08,
+      "loss": 0.07,
+      "step": 9110
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.73595981607655e-08,
+      "loss": 0.0679,
+      "step": 9111
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7103419408978092e-08,
+      "loss": 0.0728,
+      "step": 9112
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.684844402932041e-08,
+      "loss": 0.0722,
+      "step": 9113
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6594672052560054e-08,
+      "loss": 0.0697,
+      "step": 9114
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6342103509315874e-08,
+      "loss": 0.0692,
+      "step": 9115
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.609073843006682e-08,
+      "loss": 0.0711,
+      "step": 9116
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.584057684514085e-08,
+      "loss": 0.0675,
+      "step": 9117
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5591618784726047e-08,
+      "loss": 0.0704,
+      "step": 9118
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5343864278861707e-08,
+      "loss": 0.066,
+      "step": 9119
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.509731335744281e-08,
+      "loss": 0.0626,
+      "step": 9120
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4851966050218888e-08,
+      "loss": 0.0671,
+      "step": 9121
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4607822386795154e-08,
+      "loss": 0.0671,
+      "step": 9122
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4364882396630264e-08,
+      "loss": 0.0798,
+      "step": 9123
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4123146109039654e-08,
+      "loss": 0.0738,
+      "step": 9124
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3882613553191102e-08,
+      "loss": 0.0708,
+      "step": 9125
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3643284758109175e-08,
+      "loss": 0.0646,
+      "step": 9126
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3405159752672992e-08,
+      "loss": 0.0699,
+      "step": 9127
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.31682385656129e-08,
+      "loss": 0.0696,
+      "step": 9128
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2932521225519366e-08,
+      "loss": 0.0686,
+      "step": 9129
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2698007760834085e-08,
+      "loss": 0.0652,
+      "step": 9130
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.246469819985442e-08,
+      "loss": 0.0742,
+      "step": 9131
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.22325925707334e-08,
+      "loss": 0.0677,
+      "step": 9132
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2001690901476415e-08,
+      "loss": 0.0668,
+      "step": 9133
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.177199321994672e-08,
+      "loss": 0.0671,
+      "step": 9134
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1543499553858817e-08,
+      "loss": 0.0716,
+      "step": 9135
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.131620993078509e-08,
+      "loss": 0.0797,
+      "step": 9136
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1090124378150278e-08,
+      "loss": 0.0756,
+      "step": 9137
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0865242923235885e-08,
+      "loss": 0.064,
+      "step": 9138
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.064156559317576e-08,
+      "loss": 0.066,
+      "step": 9139
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.041909241496165e-08,
+      "loss": 0.0714,
+      "step": 9140
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0197823415436524e-08,
+      "loss": 0.0574,
+      "step": 9141
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9977758621299025e-08,
+      "loss": 0.069,
+      "step": 9142
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.975889805910569e-08,
+      "loss": 0.0712,
+      "step": 9143
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9541241755263174e-08,
+      "loss": 0.0663,
+      "step": 9144
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.932478973603491e-08,
+      "loss": 0.07,
+      "step": 9145
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9109542027540006e-08,
+      "loss": 0.0746,
+      "step": 9146
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.889549865574991e-08,
+      "loss": 0.0757,
+      "step": 9147
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8682659646491743e-08,
+      "loss": 0.0651,
+      "step": 9148
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8471025025449395e-08,
+      "loss": 0.064,
+      "step": 9149
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8260594818159116e-08,
+      "loss": 0.0724,
+      "step": 9150
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.805136905001059e-08,
+      "loss": 0.0618,
+      "step": 9151
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7843347746251403e-08,
+      "loss": 0.0725,
+      "step": 9152
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7636530931982586e-08,
+      "loss": 0.0717,
+      "step": 9153
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7430918632157513e-08,
+      "loss": 0.0699,
+      "step": 9154
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7226510871588553e-08,
+      "loss": 0.0804,
+      "step": 9155
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7023307674940424e-08,
+      "loss": 0.0712,
+      "step": 9156
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6821309066730183e-08,
+      "loss": 0.0638,
+      "step": 9157
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6620515071333887e-08,
+      "loss": 0.0705,
+      "step": 9158
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.642092571297993e-08,
+      "loss": 0.0694,
+      "step": 9159
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.622254101575016e-08,
+      "loss": 0.0697,
+      "step": 9160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6025361003583206e-08,
+      "loss": 0.0675,
+      "step": 9161
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.582938570027337e-08,
+      "loss": 0.0715,
+      "step": 9162
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.563461512946618e-08,
+      "loss": 0.0702,
+      "step": 9163
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.544104931466284e-08,
+      "loss": 0.0665,
+      "step": 9164
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5248688279222435e-08,
+      "loss": 0.0708,
+      "step": 9165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5057532046353075e-08,
+      "loss": 0.063,
+      "step": 9166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.486758063912186e-08,
+      "loss": 0.0658,
+      "step": 9167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4678834080450455e-08,
+      "loss": 0.075,
+      "step": 9168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4491292393110645e-08,
+      "loss": 0.0673,
+      "step": 9169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4304955599735437e-08,
+      "loss": 0.0699,
+      "step": 9170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4119823722806847e-08,
+      "loss": 0.0706,
+      "step": 9171
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3935896784663671e-08,
+      "loss": 0.0675,
+      "step": 9172
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.375317480750038e-08,
+      "loss": 0.0702,
+      "step": 9173
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3571657813364892e-08,
+      "loss": 0.0711,
+      "step": 9174
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3391345824158574e-08,
+      "loss": 0.061,
+      "step": 9175
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3212238861639581e-08,
+      "loss": 0.0667,
+      "step": 9176
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3034336947420623e-08,
+      "loss": 0.0679,
+      "step": 9177
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.285764010296564e-08,
+      "loss": 0.0721,
+      "step": 9178
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2682148349598689e-08,
+      "loss": 0.0651,
+      "step": 9179
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.250786170849283e-08,
+      "loss": 0.0671,
+      "step": 9180
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.233478020067902e-08,
+      "loss": 0.0716,
+      "step": 9181
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2162903847042772e-08,
+      "loss": 0.066,
+      "step": 9182
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1992232668323056e-08,
+      "loss": 0.0661,
+      "step": 9183
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1822766685112285e-08,
+      "loss": 0.0719,
+      "step": 9184
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.165450591786077e-08,
+      "loss": 0.0679,
+      "step": 9185
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1487450386871157e-08,
+      "loss": 0.0736,
+      "step": 9186
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1321600112300657e-08,
+      "loss": 0.0662,
+      "step": 9187
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1156955114162149e-08,
+      "loss": 0.0662,
+      "step": 9188
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0993515412321964e-08,
+      "loss": 0.0744,
+      "step": 9189
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0831281026500995e-08,
+      "loss": 0.0696,
+      "step": 9190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0670251976275803e-08,
+      "loss": 0.063,
+      "step": 9191
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0510428281076401e-08,
+      "loss": 0.0627,
+      "step": 9192
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0351809960188475e-08,
+      "loss": 0.0692,
+      "step": 9193
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0194397032750047e-08,
+      "loss": 0.0636,
+      "step": 9194
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0038189517757036e-08,
+      "loss": 0.0707,
+      "step": 9195
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.883187434057694e-09,
+      "loss": 0.0688,
+      "step": 9196
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.729390800354843e-09,
+      "loss": 0.0665,
+      "step": 9197
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.576799635205858e-09,
+      "loss": 0.0648,
+      "step": 9198
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.425413957023788e-09,
+      "loss": 0.0697,
+      "step": 9199
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.275233784075132e-09,
+      "loss": 0.0702,
+      "step": 9200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.126259134480953e-09,
+      "loss": 0.0729,
+      "step": 9201
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.97849002621798e-09,
+      "loss": 0.0673,
+      "step": 9202
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.831926477115283e-09,
+      "loss": 0.0664,
+      "step": 9203
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.686568504859827e-09,
+      "loss": 0.0724,
+      "step": 9204
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.542416126989805e-09,
+      "loss": 0.0639,
+      "step": 9205
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.399469360899082e-09,
+      "loss": 0.0728,
+      "step": 9206
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.257728223837191e-09,
+      "loss": 0.0734,
+      "step": 9207
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.11719273290601e-09,
+      "loss": 0.0655,
+      "step": 9208
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.977862905064193e-09,
+      "loss": 0.0647,
+      "step": 9209
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.83973875712385e-09,
+      "loss": 0.0648,
+      "step": 9210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.702820305751647e-09,
+      "loss": 0.0644,
+      "step": 9211
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.567107567467702e-09,
+      "loss": 0.0694,
+      "step": 9212
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.432600558647806e-09,
+      "loss": 0.0686,
+      "step": 9213
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.299299295523421e-09,
+      "loss": 0.0662,
+      "step": 9214
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.167203794178346e-09,
+      "loss": 0.0709,
+      "step": 9215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.036314070552053e-09,
+      "loss": 0.0695,
+      "step": 9216
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.906630140437464e-09,
+      "loss": 0.067,
+      "step": 9217
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.778152019484286e-09,
+      "loss": 0.0648,
+      "step": 9218
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.650879723194559e-09,
+      "loss": 0.0756,
+      "step": 9219
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.52481326692489e-09,
+      "loss": 0.0677,
+      "step": 9220
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.399952665887555e-09,
+      "loss": 0.0701,
+      "step": 9221
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.276297935149389e-09,
+      "loss": 0.0657,
+      "step": 9222
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.153849089629571e-09,
+      "loss": 0.0619,
+      "step": 9223
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.032606144104059e-09,
+      "loss": 0.072,
+      "step": 9224
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.912569113203371e-09,
+      "loss": 0.0676,
+      "step": 9225
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.793738011410366e-09,
+      "loss": 0.0725,
+      "step": 9226
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.676112853064686e-09,
+      "loss": 0.0684,
+      "step": 9227
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.5596936523583116e-09,
+      "loss": 0.0662,
+      "step": 9228
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.4444804233411144e-09,
+      "loss": 0.0696,
+      "step": 9229
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.330473179911977e-09,
+      "loss": 0.0648,
+      "step": 9230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.217671935831004e-09,
+      "loss": 0.0718,
+      "step": 9231
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.106076704706198e-09,
+      "loss": 0.0647,
+      "step": 9232
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.9956875000045646e-09,
+      "loss": 0.0701,
+      "step": 9233
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.88650433504656e-09,
+      "loss": 0.0717,
+      "step": 9234
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.778527223006091e-09,
+      "loss": 0.0665,
+      "step": 9235
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6717561769116235e-09,
+      "loss": 0.0668,
+      "step": 9236
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.566191209647297e-09,
+      "loss": 0.068,
+      "step": 9237
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.46183233395181e-09,
+      "loss": 0.0651,
+      "step": 9238
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.358679562416202e-09,
+      "loss": 0.0674,
+      "step": 9239
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.256732907487182e-09,
+      "loss": 0.0687,
+      "step": 9240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1559923814671335e-09,
+      "loss": 0.0683,
+      "step": 9241
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0564579965118865e-09,
+      "loss": 0.0676,
+      "step": 9242
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9581297646296145e-09,
+      "loss": 0.0786,
+      "step": 9243
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.8610076976874915e-09,
+      "loss": 0.066,
+      "step": 9244
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.76509180740392e-09,
+      "loss": 0.0665,
+      "step": 9245
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6703821053529766e-09,
+      "loss": 0.0658,
+      "step": 9246
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.576878602961076e-09,
+      "loss": 0.0613,
+      "step": 9247
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4845813115114147e-09,
+      "loss": 0.0711,
+      "step": 9248
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3934902421417504e-09,
+      "loss": 0.0678,
+      "step": 9249
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3036054058444013e-09,
+      "loss": 0.0701,
+      "step": 9250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2149268134629154e-09,
+      "loss": 0.075,
+      "step": 9251
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.127454475697622e-09,
+      "loss": 0.0727,
+      "step": 9252
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.041188403105633e-09,
+      "loss": 0.0683,
+      "step": 9253
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9561286060952877e-09,
+      "loss": 0.0646,
+      "step": 9254
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8722750949283783e-09,
+      "loss": 0.076,
+      "step": 9255
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7896278797256983e-09,
+      "loss": 0.0677,
+      "step": 9256
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.70818697045816e-09,
+      "loss": 0.0713,
+      "step": 9257
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6279523769534575e-09,
+      "loss": 0.0672,
+      "step": 9258
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5489241088927363e-09,
+      "loss": 0.063,
+      "step": 9259
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.471102175812812e-09,
+      "loss": 0.0714,
+      "step": 9260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.39448658710173e-09,
+      "loss": 0.0683,
+      "step": 9261
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.319077352006538e-09,
+      "loss": 0.074,
+      "step": 9262
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.244874479625514e-09,
+      "loss": 0.0692,
+      "step": 9263
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1718779789126064e-09,
+      "loss": 0.069,
+      "step": 9264
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1000878586752147e-09,
+      "loss": 0.0723,
+      "step": 9265
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0295041275764093e-09,
+      "loss": 0.0701,
+      "step": 9266
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.960126794133821e-09,
+      "loss": 0.0744,
+      "step": 9267
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8919558667174208e-09,
+      "loss": 0.0701,
+      "step": 9268
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.824991353552852e-09,
+      "loss": 0.0712,
+      "step": 9269
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.759233262721427e-09,
+      "loss": 0.0666,
+      "step": 9270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6946816021579104e-09,
+      "loss": 0.0642,
+      "step": 9271
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6313363796505167e-09,
+      "loss": 0.0672,
+      "step": 9272
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.569197602843131e-09,
+      "loss": 0.07,
+      "step": 9273
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5082652792342e-09,
+      "loss": 0.0756,
+      "step": 9274
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4485394161745103e-09,
+      "loss": 0.0636,
+      "step": 9275
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3900200208727399e-09,
+      "loss": 0.065,
+      "step": 9276
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.332707100388797e-09,
+      "loss": 0.0699,
+      "step": 9277
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2766006616393712e-09,
+      "loss": 0.0701,
+      "step": 9278
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.221700711393492e-09,
+      "loss": 0.0672,
+      "step": 9279
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1680072562758604e-09,
+      "loss": 0.0694,
+      "step": 9280
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.115520302765738e-09,
+      "loss": 0.0667,
+      "step": 9281
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0642398571958369e-09,
+      "loss": 0.0678,
+      "step": 9282
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.01416592575454e-09,
+      "loss": 0.0691,
+      "step": 9283
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.65298514482571e-10,
+      "loss": 0.0669,
+      "step": 9284
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.176376292785449e-10,
+      "loss": 0.072,
+      "step": 9285
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.711832758934169e-10,
+      "loss": 0.0662,
+      "step": 9286
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.259354599304825e-10,
+      "loss": 0.0674,
+      "step": 9287
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.818941868509289e-10,
+      "loss": 0.0753,
+      "step": 9288
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.390594619682834e-10,
+      "loss": 0.0669,
+      "step": 9289
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.974312904517444e-10,
+      "loss": 0.0681,
+      "step": 9290
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.570096773239609e-10,
+      "loss": 0.0654,
+      "step": 9291
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.177946274632529e-10,
+      "loss": 0.0652,
+      "step": 9292
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.797861456002806e-10,
+      "loss": 0.0736,
+      "step": 9293
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.429842363224858e-10,
+      "loss": 0.0743,
+      "step": 9294
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.0738890406965e-10,
+      "loss": 0.0735,
+      "step": 9295
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.730001531361161e-10,
+      "loss": 0.0694,
+      "step": 9296
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.39817987673008e-10,
+      "loss": 0.0648,
+      "step": 9297
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0784241168378977e-10,
+      "loss": 0.0686,
+      "step": 9298
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.770734290264866e-10,
+      "loss": 0.0689,
+      "step": 9299
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.4751104341479436e-10,
+      "loss": 0.0705,
+      "step": 9300
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1915525841363927e-10,
+      "loss": 0.0701,
+      "step": 9301
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9200607744583886e-10,
+      "loss": 0.0693,
+      "step": 9302
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6606350378877154e-10,
+      "loss": 0.0701,
+      "step": 9303
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.413275405699356e-10,
+      "loss": 0.0669,
+      "step": 9304
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1779819077583086e-10,
+      "loss": 0.068,
+      "step": 9305
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.954754572452977e-10,
+      "loss": 0.0704,
+      "step": 9306
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7435934267173714e-10,
+      "loss": 0.0713,
+      "step": 9307
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5444984960311104e-10,
+      "loss": 0.0739,
+      "step": 9308
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3574698044305224e-10,
+      "loss": 0.073,
+      "step": 9309
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1825073744642367e-10,
+      "loss": 0.0676,
+      "step": 9310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0196112272486958e-10,
+      "loss": 0.0668,
+      "step": 9311
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.68781382445949e-11,
+      "loss": 0.0706,
+      "step": 9312
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.300178582414497e-11,
+      "loss": 0.0687,
+      "step": 9313
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.033206713995654e-11,
+      "loss": 0.068,
+      "step": 9314
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.886898371969651e-11,
+      "loss": 0.0689,
+      "step": 9315
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.8612536947812975e-11,
+      "loss": 0.0653,
+      "step": 9316
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.9562728058873944e-11,
+      "loss": 0.0714,
+      "step": 9317
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1719558147559328e-11,
+      "loss": 0.0645,
+      "step": 9318
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5083028159779136e-11,
+      "loss": 0.0789,
+      "step": 9319
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.653138896004166e-12,
+      "loss": 0.0679,
+      "step": 9320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.429891011266009e-12,
+      "loss": 0.0693,
+      "step": 9321
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.4132850151570298e-12,
+      "loss": 0.0719,
+      "step": 9322
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.033212718303815e-13,
+      "loss": 0.0744,
+      "step": 9323
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.0723,
+      "step": 9324
+    },
+    {
+      "epoch": 3.0,
+      "step": 9324,
+      "total_flos": 3.269131628386976e+18,
+      "train_loss": 0.051072394966772315,
+      "train_runtime": 56917.8528,
+      "train_samples_per_second": 20.975,
+      "train_steps_per_second": 0.164
+    }
+  ],
+  "max_steps": 9324,
+  "num_train_epochs": 3,
+  "total_flos": 3.269131628386976e+18,
+  "trial_name": null,
+  "trial_params": null
+}