diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,55987 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 9327,
+  "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": 0.7724,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714287e-07,
+      "loss": 0.7677,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.142857142857143e-07,
+      "loss": 0.7772,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 0.7476,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 0.7682,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.285714285714286e-07,
+      "loss": 0.7659,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.745,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 0.7559,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.428571428571428e-07,
+      "loss": 0.7726,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 0.7367,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.857142857142857e-07,
+      "loss": 0.7416,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 0.7599,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.285714285714287e-07,
+      "loss": 0.7143,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.7219,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 0.7046,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 0.7072,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2142857142857144e-06,
+      "loss": 0.631,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2857142857142856e-06,
+      "loss": 0.6155,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3571428571428572e-06,
+      "loss": 0.6176,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.5966,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 0.5521,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5714285714285714e-06,
+      "loss": 0.5405,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.642857142857143e-06,
+      "loss": 0.4559,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 0.4414,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 0.4677,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8571428571428573e-06,
+      "loss": 0.456,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.928571428571429e-06,
+      "loss": 0.4148,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.4411,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0714285714285717e-06,
+      "loss": 0.4127,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.3886,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2142857142857146e-06,
+      "loss": 0.4031,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.3772,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3571428571428574e-06,
+      "loss": 0.4002,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.428571428571429e-06,
+      "loss": 0.3644,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 0.3522,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 0.3628,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.642857142857143e-06,
+      "loss": 0.3941,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7142857142857144e-06,
+      "loss": 0.3541,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.785714285714286e-06,
+      "loss": 0.3809,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.3648,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.928571428571429e-06,
+      "loss": 0.342,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 0.3216,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.071428571428572e-06,
+      "loss": 0.3609,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.142857142857143e-06,
+      "loss": 0.3202,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 0.3185,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.285714285714286e-06,
+      "loss": 0.3157,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.357142857142857e-06,
+      "loss": 0.3409,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 0.3159,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 0.3182,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.3194,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.642857142857143e-06,
+      "loss": 0.3264,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 0.3335,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.785714285714286e-06,
+      "loss": 0.334,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.857142857142858e-06,
+      "loss": 0.3194,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 0.2988,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.3065,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.071428571428572e-06,
+      "loss": 0.2915,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1428571428571435e-06,
+      "loss": 0.3074,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2142857142857145e-06,
+      "loss": 0.2993,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.2995,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.357142857142857e-06,
+      "loss": 0.2815,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.428571428571429e-06,
+      "loss": 0.2927,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-06,
+      "loss": 0.2883,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 0.3256,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 0.2982,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.714285714285715e-06,
+      "loss": 0.292,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.785714285714287e-06,
+      "loss": 0.2609,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 0.2808,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.928571428571429e-06,
+      "loss": 0.3034,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.272,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.071428571428571e-06,
+      "loss": 0.2855,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 0.2564,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.214285714285715e-06,
+      "loss": 0.2643,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.285714285714286e-06,
+      "loss": 0.2865,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.2863,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 0.296,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.2582,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.571428571428572e-06,
+      "loss": 0.2678,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6428571428571435e-06,
+      "loss": 0.2753,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.2646,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.785714285714286e-06,
+      "loss": 0.2983,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.857142857142858e-06,
+      "loss": 0.2391,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.928571428571429e-06,
+      "loss": 0.2848,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6e-06,
+      "loss": 0.2842,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.071428571428571e-06,
+      "loss": 0.2813,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.142857142857144e-06,
+      "loss": 0.2816,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.214285714285715e-06,
+      "loss": 0.2827,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 0.2846,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.357142857142858e-06,
+      "loss": 0.2788,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.3052,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.2538,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.571428571428572e-06,
+      "loss": 0.2668,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.642857142857143e-06,
+      "loss": 0.2576,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.714285714285714e-06,
+      "loss": 0.2717,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.785714285714287e-06,
+      "loss": 0.2641,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.2453,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.928571428571429e-06,
+      "loss": 0.2649,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-06,
+      "loss": 0.2647,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.0714285714285726e-06,
+      "loss": 0.2769,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.2834,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2142857142857145e-06,
+      "loss": 0.266,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.285714285714286e-06,
+      "loss": 0.2435,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.357142857142858e-06,
+      "loss": 0.2757,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.257,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.2885,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.571428571428572e-06,
+      "loss": 0.2449,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.642857142857143e-06,
+      "loss": 0.2485,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.714285714285716e-06,
+      "loss": 0.2525,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.785714285714287e-06,
+      "loss": 0.2727,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.2959,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.928571428571429e-06,
+      "loss": 0.2855,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.2641,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.071428571428572e-06,
+      "loss": 0.2616,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.142857142857143e-06,
+      "loss": 0.239,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.2786,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.285714285714287e-06,
+      "loss": 0.2581,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.357142857142858e-06,
+      "loss": 0.2622,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.428571428571429e-06,
+      "loss": 0.255,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.5e-06,
+      "loss": 0.2554,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.2501,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.642857142857144e-06,
+      "loss": 0.2776,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.714285714285715e-06,
+      "loss": 0.2291,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.785714285714286e-06,
+      "loss": 0.2626,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.857142857142858e-06,
+      "loss": 0.2596,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.2367,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9e-06,
+      "loss": 0.2565,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.071428571428573e-06,
+      "loss": 0.2609,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.2501,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.214285714285715e-06,
+      "loss": 0.2671,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.2479,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.357142857142859e-06,
+      "loss": 0.2346,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 0.2545,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.5e-06,
+      "loss": 0.2616,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.571428571428573e-06,
+      "loss": 0.2517,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 0.2527,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.2821,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.785714285714286e-06,
+      "loss": 0.2386,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.857142857142859e-06,
+      "loss": 0.2483,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.92857142857143e-06,
+      "loss": 0.2437,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.2653,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0071428571428572e-05,
+      "loss": 0.2157,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0142857142857143e-05,
+      "loss": 0.242,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0214285714285714e-05,
+      "loss": 0.2328,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 0.2279,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0357142857142859e-05,
+      "loss": 0.2384,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.042857142857143e-05,
+      "loss": 0.2353,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.2382,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0571428571428572e-05,
+      "loss": 0.2425,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0642857142857143e-05,
+      "loss": 0.2388,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.2348,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0785714285714287e-05,
+      "loss": 0.2313,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.2577,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0928571428571429e-05,
+      "loss": 0.2433,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.24,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1071428571428572e-05,
+      "loss": 0.221,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 0.2467,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1214285714285716e-05,
+      "loss": 0.2372,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1285714285714287e-05,
+      "loss": 0.2337,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1357142857142858e-05,
+      "loss": 0.2605,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.2676,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.15e-05,
+      "loss": 0.2604,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1571428571428573e-05,
+      "loss": 0.2437,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1642857142857145e-05,
+      "loss": 0.2409,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1714285714285716e-05,
+      "loss": 0.2261,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1785714285714287e-05,
+      "loss": 0.2557,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1857142857142858e-05,
+      "loss": 0.2243,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.192857142857143e-05,
+      "loss": 0.2518,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.2333,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2071428571428571e-05,
+      "loss": 0.2498,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.2385,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2214285714285717e-05,
+      "loss": 0.2672,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2285714285714288e-05,
+      "loss": 0.2206,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2357142857142859e-05,
+      "loss": 0.2656,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.242857142857143e-05,
+      "loss": 0.2223,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2651,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 0.2492,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2642857142857143e-05,
+      "loss": 0.2285,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2714285714285715e-05,
+      "loss": 0.2339,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2785714285714286e-05,
+      "loss": 0.2304,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.2162,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.292857142857143e-05,
+      "loss": 0.2424,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.2378,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3071428571428572e-05,
+      "loss": 0.2477,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 0.2433,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3214285714285716e-05,
+      "loss": 0.258,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3285714285714287e-05,
+      "loss": 0.2289,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3357142857142858e-05,
+      "loss": 0.2293,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3428571428571429e-05,
+      "loss": 0.253,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.2262,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.2434,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3642857142857145e-05,
+      "loss": 0.2425,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.2425,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3785714285714287e-05,
+      "loss": 0.2271,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3857142857142858e-05,
+      "loss": 0.224,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3928571428571429e-05,
+      "loss": 0.2419,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4e-05,
+      "loss": 0.2423,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4071428571428574e-05,
+      "loss": 0.2309,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4142857142857145e-05,
+      "loss": 0.244,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4214285714285716e-05,
+      "loss": 0.2484,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.2462,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4357142857142858e-05,
+      "loss": 0.2512,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4428571428571429e-05,
+      "loss": 0.2184,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.45e-05,
+      "loss": 0.2378,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4571428571428573e-05,
+      "loss": 0.2295,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4642857142857144e-05,
+      "loss": 0.2187,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4714285714285716e-05,
+      "loss": 0.244,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4785714285714287e-05,
+      "loss": 0.2401,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.2293,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.492857142857143e-05,
+      "loss": 0.2503,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2462,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5071428571428573e-05,
+      "loss": 0.2015,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5142857142857144e-05,
+      "loss": 0.2389,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5214285714285715e-05,
+      "loss": 0.2456,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5285714285714286e-05,
+      "loss": 0.223,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.535714285714286e-05,
+      "loss": 0.2112,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 0.2374,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.55e-05,
+      "loss": 0.2198,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5571428571428573e-05,
+      "loss": 0.2481,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5642857142857143e-05,
+      "loss": 0.2241,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.2242,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5785714285714288e-05,
+      "loss": 0.2215,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5857142857142857e-05,
+      "loss": 0.2249,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.592857142857143e-05,
+      "loss": 0.2324,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.2171,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.2386,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6142857142857145e-05,
+      "loss": 0.2462,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6214285714285717e-05,
+      "loss": 0.2322,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6285714285714287e-05,
+      "loss": 0.2746,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.635714285714286e-05,
+      "loss": 0.2417,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.2327,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.65e-05,
+      "loss": 0.2289,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 0.2001,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6642857142857147e-05,
+      "loss": 0.238,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6714285714285716e-05,
+      "loss": 0.2234,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.678571428571429e-05,
+      "loss": 0.2295,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6857142857142858e-05,
+      "loss": 0.2183,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.692857142857143e-05,
+      "loss": 0.2258,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7e-05,
+      "loss": 0.2235,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7071428571428573e-05,
+      "loss": 0.219,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.2177,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7214285714285718e-05,
+      "loss": 0.2358,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7285714285714287e-05,
+      "loss": 0.2493,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.735714285714286e-05,
+      "loss": 0.2265,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.742857142857143e-05,
+      "loss": 0.2279,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.2294,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.757142857142857e-05,
+      "loss": 0.2148,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7642857142857144e-05,
+      "loss": 0.221,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 0.2361,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7785714285714286e-05,
+      "loss": 0.2482,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.2496,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.792857142857143e-05,
+      "loss": 0.2528,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8e-05,
+      "loss": 0.2516,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8071428571428573e-05,
+      "loss": 0.2308,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8142857142857146e-05,
+      "loss": 0.2378,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8214285714285715e-05,
+      "loss": 0.2198,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.2389,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8357142857142857e-05,
+      "loss": 0.2336,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842857142857143e-05,
+      "loss": 0.2513,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 0.2124,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.2267,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8642857142857144e-05,
+      "loss": 0.212,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8714285714285717e-05,
+      "loss": 0.2259,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8785714285714286e-05,
+      "loss": 0.2534,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 0.2187,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.892857142857143e-05,
+      "loss": 0.2156,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9e-05,
+      "loss": 0.2317,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9071428571428574e-05,
+      "loss": 0.2235,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9142857142857146e-05,
+      "loss": 0.2263,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9214285714285716e-05,
+      "loss": 0.2407,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.2091,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9357142857142858e-05,
+      "loss": 0.2061,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.2408,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95e-05,
+      "loss": 0.2539,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9571428571428572e-05,
+      "loss": 0.2088,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.2413,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9714285714285718e-05,
+      "loss": 0.2149,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785714285714287e-05,
+      "loss": 0.2321,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985714285714286e-05,
+      "loss": 0.225,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.992857142857143e-05,
+      "loss": 0.2183,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2374,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999397078785e-05,
+      "loss": 0.2336,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999758831522e-05,
+      "loss": 0.2262,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999994573709515e-05,
+      "loss": 0.2358,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999990353262037e-05,
+      "loss": 0.2168,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999984926973296e-05,
+      "loss": 0.2274,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999978294843944e-05,
+      "loss": 0.2298,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997045687478e-05,
+      "loss": 0.2289,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999961413066752e-05,
+      "loss": 0.2275,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999995116342095e-05,
+      "loss": 0.2309,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999939707938612e-05,
+      "loss": 0.2538,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999927046621112e-05,
+      "loss": 0.2291,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999913179469988e-05,
+      "loss": 0.2089,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99998981064869e-05,
+      "loss": 0.2008,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999881827673673e-05,
+      "loss": 0.2305,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999986434303227e-05,
+      "loss": 0.2323,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999845652564795e-05,
+      "loss": 0.2345,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999982575627351e-05,
+      "loss": 0.2311,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999804654160803e-05,
+      "loss": 0.2269,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999782346229225e-05,
+      "loss": 0.2264,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999758832481466e-05,
+      "loss": 0.2153,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999973411292036e-05,
+      "loss": 0.2274,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999708187548893e-05,
+      "loss": 0.2579,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999968105637018e-05,
+      "loss": 0.2154,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999652719387504e-05,
+      "loss": 0.2448,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999623176604276e-05,
+      "loss": 0.2301,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999592428024057e-05,
+      "loss": 0.2423,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999560473650556e-05,
+      "loss": 0.2352,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999527313487633e-05,
+      "loss": 0.2169,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999949294753928e-05,
+      "loss": 0.2052,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999945737580964e-05,
+      "loss": 0.2161,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999420598303005e-05,
+      "loss": 0.2294,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999938261502381e-05,
+      "loss": 0.2288,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999343425976634e-05,
+      "loss": 0.2235,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999303031166203e-05,
+      "loss": 0.2169,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999261430597388e-05,
+      "loss": 0.2205,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999218624275204e-05,
+      "loss": 0.2186,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999174612204818e-05,
+      "loss": 0.2256,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999912939439153e-05,
+      "loss": 0.2285,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999082970840796e-05,
+      "loss": 0.2263,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999035341558218e-05,
+      "loss": 0.2175,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998986506549532e-05,
+      "loss": 0.2272,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998936465820627e-05,
+      "loss": 0.2189,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998885219377544e-05,
+      "loss": 0.2179,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998832767226458e-05,
+      "loss": 0.2279,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998779109373693e-05,
+      "loss": 0.2318,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998724245825724e-05,
+      "loss": 0.2306,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998668176589162e-05,
+      "loss": 0.2328,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998610901670766e-05,
+      "loss": 0.2161,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998552421077453e-05,
+      "loss": 0.23,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999849273481626e-05,
+      "loss": 0.2165,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998431842894398e-05,
+      "loss": 0.2182,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99983697453192e-05,
+      "loss": 0.2288,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999830644209816e-05,
+      "loss": 0.2317,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999824193323891e-05,
+      "loss": 0.224,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998176218749225e-05,
+      "loss": 0.2032,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998109298637033e-05,
+      "loss": 0.2315,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99980411729104e-05,
+      "loss": 0.241,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997971841577545e-05,
+      "loss": 0.2477,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997901304646827e-05,
+      "loss": 0.214,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999782956212675e-05,
+      "loss": 0.2391,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999775661402597e-05,
+      "loss": 0.2364,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997682460353277e-05,
+      "loss": 0.2632,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997607101117616e-05,
+      "loss": 0.2266,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997530536328074e-05,
+      "loss": 0.2087,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997452765993878e-05,
+      "loss": 0.2161,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997373790124417e-05,
+      "loss": 0.2262,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997293608729206e-05,
+      "loss": 0.2467,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997212221817914e-05,
+      "loss": 0.2258,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999712962940036e-05,
+      "loss": 0.2031,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99970458314865e-05,
+      "loss": 0.2256,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996960828086432e-05,
+      "loss": 0.2255,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999687461921042e-05,
+      "loss": 0.2109,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999678720486885e-05,
+      "loss": 0.2351,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996698585072266e-05,
+      "loss": 0.2285,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996608759831352e-05,
+      "loss": 0.1998,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996517729156943e-05,
+      "loss": 0.2265,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999642549306001e-05,
+      "loss": 0.2154,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999633205155168e-05,
+      "loss": 0.2146,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996237404643222e-05,
+      "loss": 0.2147,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996141552346043e-05,
+      "loss": 0.2105,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996044494671703e-05,
+      "loss": 0.1927,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995946231631912e-05,
+      "loss": 0.2171,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995846763238514e-05,
+      "loss": 0.2261,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99957460895035e-05,
+      "loss": 0.2399,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995644210439016e-05,
+      "loss": 0.2306,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999554112605734e-05,
+      "loss": 0.2228,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999543683637091e-05,
+      "loss": 0.2312,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995331341392302e-05,
+      "loss": 0.2142,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995224641134227e-05,
+      "loss": 0.2037,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995116735609563e-05,
+      "loss": 0.2093,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999500762483131e-05,
+      "loss": 0.233,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994897308812636e-05,
+      "loss": 0.2233,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994785787566836e-05,
+      "loss": 0.1966,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994673061107362e-05,
+      "loss": 0.2256,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994559129447805e-05,
+      "loss": 0.2248,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994443992601906e-05,
+      "loss": 0.2131,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994327650583543e-05,
+      "loss": 0.2364,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999421010340675e-05,
+      "loss": 0.2174,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99940913510857e-05,
+      "loss": 0.2166,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993971393634713e-05,
+      "loss": 0.2082,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993850231068252e-05,
+      "loss": 0.2007,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999372786340093e-05,
+      "loss": 0.2074,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993604290647503e-05,
+      "loss": 0.2066,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993479512822867e-05,
+      "loss": 0.2183,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993353529942076e-05,
+      "loss": 0.228,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993226342020315e-05,
+      "loss": 0.2025,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993097949072918e-05,
+      "loss": 0.2338,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999296835111538e-05,
+      "loss": 0.2054,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992837548163315e-05,
+      "loss": 0.2032,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992705540232504e-05,
+      "loss": 0.2052,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992572327338863e-05,
+      "loss": 0.2019,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992437909498456e-05,
+      "loss": 0.201,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992302286727487e-05,
+      "loss": 0.1892,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992165459042315e-05,
+      "loss": 0.2059,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999202742645944e-05,
+      "loss": 0.1962,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99918881889955e-05,
+      "loss": 0.2213,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999174774666729e-05,
+      "loss": 0.2101,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991606099491748e-05,
+      "loss": 0.2053,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999146324748595e-05,
+      "loss": 0.2218,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999131919066712e-05,
+      "loss": 0.1961,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999117392905263e-05,
+      "loss": 0.1984,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991027462660003e-05,
+      "loss": 0.2302,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990879791506893e-05,
+      "loss": 0.2163,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990730915611107e-05,
+      "loss": 0.22,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990580834990602e-05,
+      "loss": 0.202,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990429549663468e-05,
+      "loss": 0.1839,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990277059647956e-05,
+      "loss": 0.2121,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990123364962448e-05,
+      "loss": 0.2006,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989968465625482e-05,
+      "loss": 0.2194,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989812361655732e-05,
+      "loss": 0.1849,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998965505307202e-05,
+      "loss": 0.2084,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989496539893323e-05,
+      "loss": 0.2231,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998933682213875e-05,
+      "loss": 0.2071,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998917589982756e-05,
+      "loss": 0.2135,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989013772979157e-05,
+      "loss": 0.1979,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988850441613096e-05,
+      "loss": 0.2088,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988685905749064e-05,
+      "loss": 0.2154,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988520165406905e-05,
+      "loss": 0.2053,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988353220606612e-05,
+      "loss": 0.1828,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988185071368308e-05,
+      "loss": 0.1973,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988015717712267e-05,
+      "loss": 0.2201,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987845159658916e-05,
+      "loss": 0.2025,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998767339722882e-05,
+      "loss": 0.2,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987500430442694e-05,
+      "loss": 0.2052,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987326259321385e-05,
+      "loss": 0.1937,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987150883885908e-05,
+      "loss": 0.2133,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986974304157404e-05,
+      "loss": 0.2274,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986796520157162e-05,
+      "loss": 0.1849,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986617531906627e-05,
+      "loss": 0.2259,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998643733942738e-05,
+      "loss": 0.2155,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986255942741147e-05,
+      "loss": 0.2036,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986073341869807e-05,
+      "loss": 0.2068,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998588953683537e-05,
+      "loss": 0.2175,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998570452766001e-05,
+      "loss": 0.2178,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998551831436603e-05,
+      "loss": 0.209,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985330896975882e-05,
+      "loss": 0.2218,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998514227551217e-05,
+      "loss": 0.2204,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984952449997642e-05,
+      "loss": 0.2003,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984761420455184e-05,
+      "loss": 0.2177,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998456918690783e-05,
+      "loss": 0.203,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984375749378755e-05,
+      "loss": 0.2033,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99841811078913e-05,
+      "loss": 0.2026,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983985262468923e-05,
+      "loss": 0.1996,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983788213135242e-05,
+      "loss": 0.2042,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983589959914022e-05,
+      "loss": 0.2039,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983390502829168e-05,
+      "loss": 0.2144,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983189841904727e-05,
+      "loss": 0.2353,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982987977164902e-05,
+      "loss": 0.2083,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998278490863403e-05,
+      "loss": 0.2207,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982580636336598e-05,
+      "loss": 0.2019,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998237516029724e-05,
+      "loss": 0.2227,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982168480540736e-05,
+      "loss": 0.2092,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981960597092e-05,
+      "loss": 0.1928,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998175150997611e-05,
+      "loss": 0.2071,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981541219218267e-05,
+      "loss": 0.1926,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981329724843837e-05,
+      "loss": 0.1909,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998111702687832e-05,
+      "loss": 0.2017,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980903125347363e-05,
+      "loss": 0.2255,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980688020276763e-05,
+      "loss": 0.2185,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980471711692458e-05,
+      "loss": 0.2003,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998025419962053e-05,
+      "loss": 0.1963,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99800354840872e-05,
+      "loss": 0.1952,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979815565118853e-05,
+      "loss": 0.1964,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979594442742005e-05,
+      "loss": 0.2128,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979372116983318e-05,
+      "loss": 0.2001,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99791485878696e-05,
+      "loss": 0.1888,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997892385542781e-05,
+      "loss": 0.2111,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978697919685034e-05,
+      "loss": 0.2055,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978470780668532e-05,
+      "loss": 0.1887,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997824243840569e-05,
+      "loss": 0.2078,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978012892924034e-05,
+      "loss": 0.1761,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977782144251254e-05,
+      "loss": 0.1824,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997755019241517e-05,
+      "loss": 0.1913,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977317037443748e-05,
+      "loss": 0.1861,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977082679365106e-05,
+      "loss": 0.2163,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976847118207507e-05,
+      "loss": 0.2214,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976610353999353e-05,
+      "loss": 0.2022,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976372386769193e-05,
+      "loss": 0.1964,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976133216545725e-05,
+      "loss": 0.2068,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975892843357788e-05,
+      "loss": 0.2092,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975651267234367e-05,
+      "loss": 0.1884,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975408488204587e-05,
+      "loss": 0.1834,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997516450629773e-05,
+      "loss": 0.207,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997491932154322e-05,
+      "loss": 0.2159,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974672933970614e-05,
+      "loss": 0.2085,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974425343609626e-05,
+      "loss": 0.2271,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997417655049011e-05,
+      "loss": 0.1938,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973926554642073e-05,
+      "loss": 0.2098,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997367535609565e-05,
+      "loss": 0.2033,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973422954881137e-05,
+      "loss": 0.1729,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973169351028972e-05,
+      "loss": 0.1979,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997291454456973e-05,
+      "loss": 0.2205,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972658535534143e-05,
+      "loss": 0.1849,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972401323953075e-05,
+      "loss": 0.2024,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972142909857547e-05,
+      "loss": 0.1885,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997188329327872e-05,
+      "loss": 0.2029,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971622474247893e-05,
+      "loss": 0.2101,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971360452796523e-05,
+      "loss": 0.2071,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971097228956202e-05,
+      "loss": 0.1804,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970832802758673e-05,
+      "loss": 0.2134,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970567174235824e-05,
+      "loss": 0.2218,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997030034341968e-05,
+      "loss": 0.2053,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997003231034242e-05,
+      "loss": 0.2252,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969763075036364e-05,
+      "loss": 0.2148,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969492637533975e-05,
+      "loss": 0.2077,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969220997867868e-05,
+      "loss": 0.2147,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968948156070797e-05,
+      "loss": 0.2239,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996867411217566e-05,
+      "loss": 0.2178,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968398866215504e-05,
+      "loss": 0.2102,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996812241822352e-05,
+      "loss": 0.1983,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996784476823304e-05,
+      "loss": 0.1992,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967565916277555e-05,
+      "loss": 0.2018,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967285862390674e-05,
+      "loss": 0.2135,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967004606606177e-05,
+      "loss": 0.1968,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966722148957977e-05,
+      "loss": 0.202,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966438489480137e-05,
+      "loss": 0.1979,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966153628206854e-05,
+      "loss": 0.1927,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965867565172486e-05,
+      "loss": 0.2199,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965580300411523e-05,
+      "loss": 0.2063,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996529183395861e-05,
+      "loss": 0.1893,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996500216584852e-05,
+      "loss": 0.1904,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99647112961162e-05,
+      "loss": 0.1967,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964419224796707e-05,
+      "loss": 0.1977,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964125951925268e-05,
+      "loss": 0.2077,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996383147753725e-05,
+      "loss": 0.2109,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996353580166815e-05,
+      "loss": 0.213,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963238924353638e-05,
+      "loss": 0.189,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962940845629503e-05,
+      "loss": 0.1986,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962641565531694e-05,
+      "loss": 0.199,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962341084096293e-05,
+      "loss": 0.2163,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962039401359538e-05,
+      "loss": 0.1916,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99617365173578e-05,
+      "loss": 0.1777,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961432432127613e-05,
+      "loss": 0.2117,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996112714570564e-05,
+      "loss": 0.2036,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996082065812869e-05,
+      "loss": 0.1996,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960512969433727e-05,
+      "loss": 0.1986,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996020407965785e-05,
+      "loss": 0.1922,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959893988838303e-05,
+      "loss": 0.2088,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959582697012484e-05,
+      "loss": 0.1968,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959270204217928e-05,
+      "loss": 0.1826,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958956510492317e-05,
+      "loss": 0.2043,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958641615873473e-05,
+      "loss": 0.188,
+      "step": 542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958325520399374e-05,
+      "loss": 0.1976,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958008224108133e-05,
+      "loss": 0.1832,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995768972703801e-05,
+      "loss": 0.1891,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957370029227415e-05,
+      "loss": 0.1835,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957049130714893e-05,
+      "loss": 0.1765,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956727031539142e-05,
+      "loss": 0.207,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956403731739e-05,
+      "loss": 0.1896,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956079231353452e-05,
+      "loss": 0.2017,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955753530421635e-05,
+      "loss": 0.1948,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955426628982814e-05,
+      "loss": 0.2297,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955098527076412e-05,
+      "loss": 0.1789,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954769224741994e-05,
+      "loss": 0.1765,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954438722019267e-05,
+      "loss": 0.2041,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954107018948085e-05,
+      "loss": 0.1792,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953774115568444e-05,
+      "loss": 0.2016,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995344001192049e-05,
+      "loss": 0.2045,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953104708044512e-05,
+      "loss": 0.1843,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952768203980935e-05,
+      "loss": 0.2055,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952430499770345e-05,
+      "loss": 0.2148,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952091595453456e-05,
+      "loss": 0.2208,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995175149107114e-05,
+      "loss": 0.1986,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995141018666441e-05,
+      "loss": 0.1951,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951067682274413e-05,
+      "loss": 0.2102,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950723977942457e-05,
+      "loss": 0.1794,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950379073709987e-05,
+      "loss": 0.198,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995003296961859e-05,
+      "loss": 0.1683,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949685665710002e-05,
+      "loss": 0.1855,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949337162026104e-05,
+      "loss": 0.1919,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948987458608916e-05,
+      "loss": 0.2017,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994863655550061e-05,
+      "loss": 0.2095,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99482844527435e-05,
+      "loss": 0.1863,
+      "step": 573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994793115038004e-05,
+      "loss": 0.2162,
+      "step": 574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947576648452842e-05,
+      "loss": 0.1905,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994722094700464e-05,
+      "loss": 0.203,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946864046078337e-05,
+      "loss": 0.2056,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946505945716964e-05,
+      "loss": 0.2011,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99461466459637e-05,
+      "loss": 0.1982,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945786146861882e-05,
+      "loss": 0.2308,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945424448454966e-05,
+      "loss": 0.2032,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945061550786575e-05,
+      "loss": 0.201,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994469745390047e-05,
+      "loss": 0.2012,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944332157840553e-05,
+      "loss": 0.1917,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943965662650875e-05,
+      "loss": 0.2049,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943597968375626e-05,
+      "loss": 0.2159,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943229075059145e-05,
+      "loss": 0.2224,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942858982745915e-05,
+      "loss": 0.2163,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942487691480566e-05,
+      "loss": 0.2114,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942115201307867e-05,
+      "loss": 0.1861,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941741512272734e-05,
+      "loss": 0.196,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994136662442023e-05,
+      "loss": 0.1937,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994099053779556e-05,
+      "loss": 0.1756,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940613252444074e-05,
+      "loss": 0.2044,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940234768411267e-05,
+      "loss": 0.1788,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993985508574278e-05,
+      "loss": 0.2035,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993947420448439e-05,
+      "loss": 0.2152,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939092124682032e-05,
+      "loss": 0.2012,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938708846381777e-05,
+      "loss": 0.2024,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993832436962984e-05,
+      "loss": 0.181,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937938694472584e-05,
+      "loss": 0.1765,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937551820956523e-05,
+      "loss": 0.1971,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937163749128293e-05,
+      "loss": 0.1901,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99367744790347e-05,
+      "loss": 0.1956,
+      "step": 604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936384010722682e-05,
+      "loss": 0.2257,
+      "step": 605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935992344239324e-05,
+      "loss": 0.2194,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993559947963185e-05,
+      "loss": 0.2066,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935205416947637e-05,
+      "loss": 0.2089,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934810156234206e-05,
+      "loss": 0.2082,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934413697539214e-05,
+      "loss": 0.2142,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934016040910468e-05,
+      "loss": 0.1779,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933617186395917e-05,
+      "loss": 0.1997,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933217134043666e-05,
+      "loss": 0.1789,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932815883901944e-05,
+      "loss": 0.196,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932413436019145e-05,
+      "loss": 0.2058,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993200979044379e-05,
+      "loss": 0.1987,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931604947224556e-05,
+      "loss": 0.2057,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993119890641026e-05,
+      "loss": 0.1828,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930791668049865e-05,
+      "loss": 0.2049,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930383232192478e-05,
+      "loss": 0.2006,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929973598887347e-05,
+      "loss": 0.1873,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992956276818387e-05,
+      "loss": 0.1922,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929150740131588e-05,
+      "loss": 0.194,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992873751478018e-05,
+      "loss": 0.2025,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992832309217948e-05,
+      "loss": 0.2003,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992790747237945e-05,
+      "loss": 0.1829,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927490655430227e-05,
+      "loss": 0.1956,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927072641382056e-05,
+      "loss": 0.1807,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926653430285346e-05,
+      "loss": 0.1985,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992623302219065e-05,
+      "loss": 0.1845,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992581141714866e-05,
+      "loss": 0.2199,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925388615210222e-05,
+      "loss": 0.1882,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992496461642631e-05,
+      "loss": 0.205,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924539420848058e-05,
+      "loss": 0.1727,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924113028526735e-05,
+      "loss": 0.181,
+      "step": 635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923685439513757e-05,
+      "loss": 0.1903,
+      "step": 636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923256653860683e-05,
+      "loss": 0.1873,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922826671619223e-05,
+      "loss": 0.1742,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992239549284122e-05,
+      "loss": 0.1911,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992196311757867e-05,
+      "loss": 0.2154,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921529545883717e-05,
+      "loss": 0.1721,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921094777808632e-05,
+      "loss": 0.1689,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992065881340585e-05,
+      "loss": 0.1786,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920221652727933e-05,
+      "loss": 0.1824,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99197832958276e-05,
+      "loss": 0.184,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991934374275771e-05,
+      "loss": 0.1787,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991890299357127e-05,
+      "loss": 0.1793,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991846104832142e-05,
+      "loss": 0.1648,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991801790706146e-05,
+      "loss": 0.2066,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917573569844817e-05,
+      "loss": 0.2028,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917128036725075e-05,
+      "loss": 0.1805,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991668130775596e-05,
+      "loss": 0.1744,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916233382991342e-05,
+      "loss": 0.1971,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915784262485224e-05,
+      "loss": 0.1792,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915333946291774e-05,
+      "loss": 0.1828,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914882434465288e-05,
+      "loss": 0.1791,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914429727060213e-05,
+      "loss": 0.203,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913975824131136e-05,
+      "loss": 0.1941,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991352072573279e-05,
+      "loss": 0.196,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913064431920056e-05,
+      "loss": 0.1715,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912606942747958e-05,
+      "loss": 0.1996,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912148258271652e-05,
+      "loss": 0.1922,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911688378546456e-05,
+      "loss": 0.211,
+      "step": 663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911227303627824e-05,
+      "loss": 0.2124,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910765033571355e-05,
+      "loss": 0.2154,
+      "step": 665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910301568432783e-05,
+      "loss": 0.2081,
+      "step": 666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909836908268004e-05,
+      "loss": 0.1809,
+      "step": 667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909371053133045e-05,
+      "loss": 0.1948,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908904003084087e-05,
+      "loss": 0.1885,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990843575817744e-05,
+      "loss": 0.1746,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990796631846957e-05,
+      "loss": 0.2133,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990749568401708e-05,
+      "loss": 0.1889,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907023854876734e-05,
+      "loss": 0.1943,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990655083110541e-05,
+      "loss": 0.1905,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990607661276016e-05,
+      "loss": 0.2013,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905601199898167e-05,
+      "loss": 0.1906,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990512459257675e-05,
+      "loss": 0.1964,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904646790853387e-05,
+      "loss": 0.1857,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904167794785686e-05,
+      "loss": 0.1937,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903687604431417e-05,
+      "loss": 0.1852,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903206219848475e-05,
+      "loss": 0.1975,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902723641094912e-05,
+      "loss": 0.2029,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902239868228917e-05,
+      "loss": 0.1864,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901754901308824e-05,
+      "loss": 0.2062,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901268740393117e-05,
+      "loss": 0.2044,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900781385540418e-05,
+      "loss": 0.1776,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900292836809488e-05,
+      "loss": 0.1909,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989980309425925e-05,
+      "loss": 0.2072,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899312157948748e-05,
+      "loss": 0.1657,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898820027937188e-05,
+      "loss": 0.2014,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989832670428391e-05,
+      "loss": 0.2059,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897832187048404e-05,
+      "loss": 0.1886,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897336476290294e-05,
+      "loss": 0.1939,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896839572069365e-05,
+      "loss": 0.2066,
+      "step": 694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896341474445526e-05,
+      "loss": 0.1744,
+      "step": 695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895842183478848e-05,
+      "loss": 0.1833,
+      "step": 696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989534169922953e-05,
+      "loss": 0.1883,
+      "step": 697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989484002175793e-05,
+      "loss": 0.1744,
+      "step": 698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989433715112454e-05,
+      "loss": 0.1782,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893833087389995e-05,
+      "loss": 0.1815,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989332783061508e-05,
+      "loss": 0.209,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989282138086072e-05,
+      "loss": 0.2123,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892313738187985e-05,
+      "loss": 0.1942,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891804902658088e-05,
+      "loss": 0.1816,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989129487433239e-05,
+      "loss": 0.1763,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890783653272388e-05,
+      "loss": 0.1915,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989027123953973e-05,
+      "loss": 0.2038,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889757633196206e-05,
+      "loss": 0.1886,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889242834303744e-05,
+      "loss": 0.161,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888726842924423e-05,
+      "loss": 0.211,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888209659120465e-05,
+      "loss": 0.1977,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887691282954234e-05,
+      "loss": 0.1953,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887171714488234e-05,
+      "loss": 0.1693,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886650953785123e-05,
+      "loss": 0.1857,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886129000907696e-05,
+      "loss": 0.1954,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885605855918887e-05,
+      "loss": 0.2001,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988508151888178e-05,
+      "loss": 0.1933,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884555989859605e-05,
+      "loss": 0.1864,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884029268915733e-05,
+      "loss": 0.1868,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883501356113677e-05,
+      "loss": 0.2031,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882972251517092e-05,
+      "loss": 0.1849,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988244195518978e-05,
+      "loss": 0.1835,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881910467195694e-05,
+      "loss": 0.1904,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988137778759891e-05,
+      "loss": 0.2017,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880843916463674e-05,
+      "loss": 0.1853,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880308853854356e-05,
+      "loss": 0.1976,
+      "step": 726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879772599835478e-05,
+      "loss": 0.1831,
+      "step": 727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879235154471698e-05,
+      "loss": 0.1886,
+      "step": 728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987869651782783e-05,
+      "loss": 0.1794,
+      "step": 729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878156689968822e-05,
+      "loss": 0.1851,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877615670959773e-05,
+      "loss": 0.1806,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987707346086592e-05,
+      "loss": 0.1713,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987653005975264e-05,
+      "loss": 0.1819,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987598546768546e-05,
+      "loss": 0.1952,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875439684730053e-05,
+      "loss": 0.1889,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987489271095223e-05,
+      "loss": 0.2134,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874344546417946e-05,
+      "loss": 0.1644,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873795191193306e-05,
+      "loss": 0.2025,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873244645344548e-05,
+      "loss": 0.1742,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987269290893806e-05,
+      "loss": 0.168,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872139982040376e-05,
+      "loss": 0.1857,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871585864718162e-05,
+      "loss": 0.1784,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871030557038247e-05,
+      "loss": 0.1889,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870474059067586e-05,
+      "loss": 0.1807,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869916370873286e-05,
+      "loss": 0.1868,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869357492522593e-05,
+      "loss": 0.1922,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868797424082902e-05,
+      "loss": 0.1639,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868236165621747e-05,
+      "loss": 0.2219,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867673717206807e-05,
+      "loss": 0.1798,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867110078905902e-05,
+      "loss": 0.1595,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866545250787e-05,
+      "loss": 0.1843,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865979232918213e-05,
+      "loss": 0.1844,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986541202536779e-05,
+      "loss": 0.1886,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864843628204128e-05,
+      "loss": 0.2084,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986427404149577e-05,
+      "loss": 0.179,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863703265311393e-05,
+      "loss": 0.1998,
+      "step": 756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986313129971983e-05,
+      "loss": 0.1769,
+      "step": 757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862558144790047e-05,
+      "loss": 0.1818,
+      "step": 758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861983800591158e-05,
+      "loss": 0.2088,
+      "step": 759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986140826719242e-05,
+      "loss": 0.1883,
+      "step": 760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986083154466324e-05,
+      "loss": 0.1874,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860253633073148e-05,
+      "loss": 0.2004,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859674532491837e-05,
+      "loss": 0.1833,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859094242989142e-05,
+      "loss": 0.1928,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858512764635035e-05,
+      "loss": 0.1834,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857930097499632e-05,
+      "loss": 0.1912,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985734624165319e-05,
+      "loss": 0.1842,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856761197166122e-05,
+      "loss": 0.1855,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856174964108964e-05,
+      "loss": 0.2028,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855587542552414e-05,
+      "loss": 0.178,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98549989325673e-05,
+      "loss": 0.1794,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854409134224607e-05,
+      "loss": 0.1902,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853818147595446e-05,
+      "loss": 0.1896,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985322597275109e-05,
+      "loss": 0.186,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985263260976294e-05,
+      "loss": 0.1628,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852038058702544e-05,
+      "loss": 0.1932,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851442319641604e-05,
+      "loss": 0.2039,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985084539265195e-05,
+      "loss": 0.1839,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850247277805567e-05,
+      "loss": 0.1935,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849647975174574e-05,
+      "loss": 0.1757,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849047484831235e-05,
+      "loss": 0.1788,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848445806847967e-05,
+      "loss": 0.207,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847842941297318e-05,
+      "loss": 0.1727,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847238888251985e-05,
+      "loss": 0.1849,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984663364778481e-05,
+      "loss": 0.1836,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846027219968768e-05,
+      "loss": 0.1889,
+      "step": 786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845419604876995e-05,
+      "loss": 0.1903,
+      "step": 787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984481080258275e-05,
+      "loss": 0.1937,
+      "step": 788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844200813159455e-05,
+      "loss": 0.1764,
+      "step": 789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843589636680654e-05,
+      "loss": 0.1738,
+      "step": 790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842977273220052e-05,
+      "loss": 0.187,
+      "step": 791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984236372285149e-05,
+      "loss": 0.18,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841748985648952e-05,
+      "loss": 0.1795,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841133061686566e-05,
+      "loss": 0.1885,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98405159510386e-05,
+      "loss": 0.2104,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983989765377947e-05,
+      "loss": 0.193,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839278169983735e-05,
+      "loss": 0.199,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838657499726094e-05,
+      "loss": 0.1854,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838035643081387e-05,
+      "loss": 0.1844,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98374126001246e-05,
+      "loss": 0.1743,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836788370930862e-05,
+      "loss": 0.2007,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836162955575453e-05,
+      "loss": 0.1673,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983553635413378e-05,
+      "loss": 0.1744,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834908566681407e-05,
+      "loss": 0.1848,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834279593294026e-05,
+      "loss": 0.1818,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833649434047492e-05,
+      "loss": 0.1881,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833018089017786e-05,
+      "loss": 0.1988,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983238555828104e-05,
+      "loss": 0.1994,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831751841913526e-05,
+      "loss": 0.1696,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831116939991663e-05,
+      "loss": 0.1866,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983048085259201e-05,
+      "loss": 0.1726,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829843579791266e-05,
+      "loss": 0.1962,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829205121666274e-05,
+      "loss": 0.1947,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982856547829403e-05,
+      "loss": 0.173,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827924649751658e-05,
+      "loss": 0.1907,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827282636116437e-05,
+      "loss": 0.1811,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982663943746578e-05,
+      "loss": 0.1932,
+      "step": 817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982599505387725e-05,
+      "loss": 0.1732,
+      "step": 818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825349485428543e-05,
+      "loss": 0.1847,
+      "step": 819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982470273219751e-05,
+      "loss": 0.1955,
+      "step": 820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982405479426214e-05,
+      "loss": 0.1844,
+      "step": 821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982340567170056e-05,
+      "loss": 0.181,
+      "step": 822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822755364591047e-05,
+      "loss": 0.1844,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822103873012016e-05,
+      "loss": 0.178,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821451197042028e-05,
+      "loss": 0.1848,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820797336759785e-05,
+      "loss": 0.1968,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820142292244133e-05,
+      "loss": 0.1869,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819486063574055e-05,
+      "loss": 0.1976,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981882865082869e-05,
+      "loss": 0.21,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818170054087303e-05,
+      "loss": 0.2039,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817510273429322e-05,
+      "loss": 0.1907,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816849308934295e-05,
+      "loss": 0.2067,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816187160681926e-05,
+      "loss": 0.1928,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815523828752065e-05,
+      "loss": 0.1629,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814859313224693e-05,
+      "loss": 0.1921,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814193614179946e-05,
+      "loss": 0.167,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813526731698092e-05,
+      "loss": 0.1665,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981285866585955e-05,
+      "loss": 0.1752,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812189416744875e-05,
+      "loss": 0.1745,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811518984434774e-05,
+      "loss": 0.1898,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981084736901008e-05,
+      "loss": 0.2117,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981017457055179e-05,
+      "loss": 0.1868,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980950058914102e-05,
+      "loss": 0.1858,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980882542485906e-05,
+      "loss": 0.1709,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980814907778731e-05,
+      "loss": 0.1732,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980747154800733e-05,
+      "loss": 0.1956,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980679283560082e-05,
+      "loss": 0.2024,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806112940649623e-05,
+      "loss": 0.1883,
+      "step": 848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805431863235722e-05,
+      "loss": 0.1768,
+      "step": 849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804749603441248e-05,
+      "loss": 0.2046,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804066161348466e-05,
+      "loss": 0.1663,
+      "step": 851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980338153703979e-05,
+      "loss": 0.1706,
+      "step": 852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802695730597775e-05,
+      "loss": 0.1888,
+      "step": 853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802008742105118e-05,
+      "loss": 0.1907,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980132057164466e-05,
+      "loss": 0.1891,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980063121929938e-05,
+      "loss": 0.1785,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979994068515241e-05,
+      "loss": 0.1897,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799248969287015e-05,
+      "loss": 0.1704,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97985560717866e-05,
+      "loss": 0.178,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797861992734722e-05,
+      "loss": 0.1642,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797166732215078e-05,
+      "loss": 0.183,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796470290311497e-05,
+      "loss": 0.1962,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795772667107968e-05,
+      "loss": 0.1625,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979507386268861e-05,
+      "loss": 0.1746,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979437387713769e-05,
+      "loss": 0.1673,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793672710539605e-05,
+      "loss": 0.1788,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979297036297892e-05,
+      "loss": 0.2051,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979226683454032e-05,
+      "loss": 0.2014,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791562125308633e-05,
+      "loss": 0.1884,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790856235368843e-05,
+      "loss": 0.1833,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790149164806075e-05,
+      "loss": 0.1905,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789440913705577e-05,
+      "loss": 0.1951,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788731482152764e-05,
+      "loss": 0.1756,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788020870233172e-05,
+      "loss": 0.1689,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97873090780325e-05,
+      "loss": 0.1802,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786596105636573e-05,
+      "loss": 0.1799,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785881953131367e-05,
+      "loss": 0.1865,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785166620602994e-05,
+      "loss": 0.1845,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784450108137715e-05,
+      "loss": 0.1753,
+      "step": 879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783732415821927e-05,
+      "loss": 0.1923,
+      "step": 880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783013543742178e-05,
+      "loss": 0.186,
+      "step": 881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782293491985148e-05,
+      "loss": 0.1797,
+      "step": 882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781572260637666e-05,
+      "loss": 0.1948,
+      "step": 883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780849849786695e-05,
+      "loss": 0.1587,
+      "step": 884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780126259519357e-05,
+      "loss": 0.1899,
+      "step": 885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779401489922896e-05,
+      "loss": 0.1994,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778675541084717e-05,
+      "loss": 0.1892,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777948413092348e-05,
+      "loss": 0.1964,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777220106033477e-05,
+      "loss": 0.1854,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776490619995922e-05,
+      "loss": 0.1696,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775759955067652e-05,
+      "loss": 0.1811,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775028111336765e-05,
+      "loss": 0.1745,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774295088891518e-05,
+      "loss": 0.1672,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97735608878203e-05,
+      "loss": 0.1762,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772825508211643e-05,
+      "loss": 0.1711,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772088950154225e-05,
+      "loss": 0.1796,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771351213736856e-05,
+      "loss": 0.1809,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770612299048505e-05,
+      "loss": 0.1834,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769872206178267e-05,
+      "loss": 0.1803,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769130935215388e-05,
+      "loss": 0.1826,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976838848624925e-05,
+      "loss": 0.1744,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767644859369386e-05,
+      "loss": 0.1735,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766900054665465e-05,
+      "loss": 0.1829,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766154072227294e-05,
+      "loss": 0.1879,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765406912144833e-05,
+      "loss": 0.1826,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976465857450817e-05,
+      "loss": 0.1833,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976390905940755e-05,
+      "loss": 0.1821,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763158366933347e-05,
+      "loss": 0.1598,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976240649717609e-05,
+      "loss": 0.1847,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761653450226437e-05,
+      "loss": 0.1879,
+      "step": 910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760899226175195e-05,
+      "loss": 0.1688,
+      "step": 911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976014382511331e-05,
+      "loss": 0.1618,
+      "step": 912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975938724713187e-05,
+      "loss": 0.1865,
+      "step": 913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758629492322115e-05,
+      "loss": 0.176,
+      "step": 914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975787056077541e-05,
+      "loss": 0.1698,
+      "step": 915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975711045258327e-05,
+      "loss": 0.159,
+      "step": 916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975634916783736e-05,
+      "loss": 0.1931,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975558670662947e-05,
+      "loss": 0.1841,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754823069051544e-05,
+      "loss": 0.1934,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754058255195668e-05,
+      "loss": 0.1933,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975329226515406e-05,
+      "loss": 0.1805,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752525099019096e-05,
+      "loss": 0.1659,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751756756883274e-05,
+      "loss": 0.1748,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750987238839244e-05,
+      "loss": 0.1775,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975021654497981e-05,
+      "loss": 0.1853,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749444675397893e-05,
+      "loss": 0.1913,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748671630186574e-05,
+      "loss": 0.1707,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747897409439067e-05,
+      "loss": 0.174,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747122013248733e-05,
+      "loss": 0.1899,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746345441709075e-05,
+      "loss": 0.1746,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745567694913727e-05,
+      "loss": 0.2009,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744788772956482e-05,
+      "loss": 0.1753,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744008675931263e-05,
+      "loss": 0.1803,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743227403932135e-05,
+      "loss": 0.1746,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742444957053308e-05,
+      "loss": 0.1645,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741661335389137e-05,
+      "loss": 0.1932,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740876539034106e-05,
+      "loss": 0.1766,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740090568082857e-05,
+      "loss": 0.1829,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973930342263016e-05,
+      "loss": 0.1593,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973851510277094e-05,
+      "loss": 0.1736,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737725608600248e-05,
+      "loss": 0.1954,
+      "step": 941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736934940213287e-05,
+      "loss": 0.1712,
+      "step": 942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97361430977054e-05,
+      "loss": 0.1843,
+      "step": 943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973535008117207e-05,
+      "loss": 0.1776,
+      "step": 944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734555890708923e-05,
+      "loss": 0.203,
+      "step": 945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733760526411723e-05,
+      "loss": 0.1679,
+      "step": 946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732963988376384e-05,
+      "loss": 0.1688,
+      "step": 947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732166276698953e-05,
+      "loss": 0.1749,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731367391475618e-05,
+      "loss": 0.1913,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730567332802718e-05,
+      "loss": 0.1806,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729766100776726e-05,
+      "loss": 0.191,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728963695494253e-05,
+      "loss": 0.1743,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972816011705206e-05,
+      "loss": 0.1744,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972735536554705e-05,
+      "loss": 0.1831,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972654944107626e-05,
+      "loss": 0.1798,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725742343736868e-05,
+      "loss": 0.1688,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724934073626202e-05,
+      "loss": 0.1633,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724124630841725e-05,
+      "loss": 0.1706,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723314015481038e-05,
+      "loss": 0.1655,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97225022276419e-05,
+      "loss": 0.1911,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972168926742219e-05,
+      "loss": 0.1831,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720875134919942e-05,
+      "loss": 0.1667,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972005983023333e-05,
+      "loss": 0.1799,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719243353460662e-05,
+      "loss": 0.1755,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718425704700393e-05,
+      "loss": 0.1644,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971760688405112e-05,
+      "loss": 0.1903,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716786891611582e-05,
+      "loss": 0.1639,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971596572748065e-05,
+      "loss": 0.1923,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971514339175735e-05,
+      "loss": 0.1667,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714319884540844e-05,
+      "loss": 0.1949,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971349520593043e-05,
+      "loss": 0.1666,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971266935602555e-05,
+      "loss": 0.1509,
+      "step": 972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711842334925794e-05,
+      "loss": 0.1628,
+      "step": 973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711014142730886e-05,
+      "loss": 0.1759,
+      "step": 974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710184779540684e-05,
+      "loss": 0.1746,
+      "step": 975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970935424545521e-05,
+      "loss": 0.1648,
+      "step": 976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708522540574604e-05,
+      "loss": 0.1756,
+      "step": 977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707689664999158e-05,
+      "loss": 0.1618,
+      "step": 978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706855618829308e-05,
+      "loss": 0.1913,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970602040216562e-05,
+      "loss": 0.1849,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705184015108817e-05,
+      "loss": 0.1865,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704346457759745e-05,
+      "loss": 0.1721,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703507730219407e-05,
+      "loss": 0.1673,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702667832588935e-05,
+      "loss": 0.1789,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701826764969613e-05,
+      "loss": 0.1634,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700984527462852e-05,
+      "loss": 0.1876,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970014112017022e-05,
+      "loss": 0.1935,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969929654319342e-05,
+      "loss": 0.1838,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698450796634288e-05,
+      "loss": 0.1659,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697603880594813e-05,
+      "loss": 0.1831,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696755795177114e-05,
+      "loss": 0.1752,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695906540483462e-05,
+      "loss": 0.1666,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969505611661626e-05,
+      "loss": 0.1785,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694204523678065e-05,
+      "loss": 0.1876,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693351761771552e-05,
+      "loss": 0.1883,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692497830999562e-05,
+      "loss": 0.1845,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691642731465057e-05,
+      "loss": 0.1989,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969078646327115e-05,
+      "loss": 0.1631,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96899290265211e-05,
+      "loss": 0.1651,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96890704213183e-05,
+      "loss": 0.1568,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688210647766274e-05,
+      "loss": 0.188,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687349705968705e-05,
+      "loss": 0.1846,
+      "step": 1002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686487596029407e-05,
+      "loss": 0.1919,
+      "step": 1003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968562431805234e-05,
+      "loss": 0.1792,
+      "step": 1004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684759872141595e-05,
+      "loss": 0.2001,
+      "step": 1005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683894258401417e-05,
+      "loss": 0.1714,
+      "step": 1006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683027476936184e-05,
+      "loss": 0.1929,
+      "step": 1007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682159527850415e-05,
+      "loss": 0.1762,
+      "step": 1008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968129041124877e-05,
+      "loss": 0.1776,
+      "step": 1009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968042012723605e-05,
+      "loss": 0.2067,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679548675917206e-05,
+      "loss": 0.1808,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967867605739731e-05,
+      "loss": 0.184,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677802271781592e-05,
+      "loss": 0.1857,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676927319175416e-05,
+      "loss": 0.1826,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676051199684286e-05,
+      "loss": 0.1925,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675173913413852e-05,
+      "loss": 0.1738,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674295460469897e-05,
+      "loss": 0.1825,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673415840958347e-05,
+      "loss": 0.1677,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672535054985277e-05,
+      "loss": 0.1701,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671653102656888e-05,
+      "loss": 0.1909,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670769984079537e-05,
+      "loss": 0.1792,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966988569935971e-05,
+      "loss": 0.1716,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966900024860404e-05,
+      "loss": 0.1725,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668113631919296e-05,
+      "loss": 0.1762,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667225849412387e-05,
+      "loss": 0.1927,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666336901190374e-05,
+      "loss": 0.1649,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665446787360444e-05,
+      "loss": 0.1552,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664555508029933e-05,
+      "loss": 0.1603,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663663063306316e-05,
+      "loss": 0.1683,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662769453297203e-05,
+      "loss": 0.1705,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661874678110357e-05,
+      "loss": 0.1716,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660978737853663e-05,
+      "loss": 0.1931,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966008163263517e-05,
+      "loss": 0.1952,
+      "step": 1033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659183362563044e-05,
+      "loss": 0.1618,
+      "step": 1034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965828392774561e-05,
+      "loss": 0.1691,
+      "step": 1035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657383328291325e-05,
+      "loss": 0.1852,
+      "step": 1036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656481564308778e-05,
+      "loss": 0.1544,
+      "step": 1037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655578635906717e-05,
+      "loss": 0.1635,
+      "step": 1038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965467454319402e-05,
+      "loss": 0.1951,
+      "step": 1039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96537692862797e-05,
+      "loss": 0.1788,
+      "step": 1040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652862865272926e-05,
+      "loss": 0.1904,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965195528028299e-05,
+      "loss": 0.2001,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651046531419335e-05,
+      "loss": 0.1797,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650136618791543e-05,
+      "loss": 0.1799,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649225542509332e-05,
+      "loss": 0.1455,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964831330268257e-05,
+      "loss": 0.1946,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647399899421256e-05,
+      "loss": 0.1903,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646485332835526e-05,
+      "loss": 0.1869,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964556960303567e-05,
+      "loss": 0.1657,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644652710132108e-05,
+      "loss": 0.179,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96437346542354e-05,
+      "loss": 0.1616,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642815435456252e-05,
+      "loss": 0.1791,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964189505390551e-05,
+      "loss": 0.1669,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964097350969415e-05,
+      "loss": 0.1871,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640050802933304e-05,
+      "loss": 0.1793,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963912693373423e-05,
+      "loss": 0.1842,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638201902208335e-05,
+      "loss": 0.1467,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637275708467157e-05,
+      "loss": 0.1544,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636348352622393e-05,
+      "loss": 0.1769,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635419834785858e-05,
+      "loss": 0.192,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634490155069516e-05,
+      "loss": 0.1704,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633559313585477e-05,
+      "loss": 0.1844,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963262731044598e-05,
+      "loss": 0.1737,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631694145763417e-05,
+      "loss": 0.1903,
+      "step": 1064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963075981965031e-05,
+      "loss": 0.1707,
+      "step": 1065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962982433221932e-05,
+      "loss": 0.1748,
+      "step": 1066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628887683583255e-05,
+      "loss": 0.1851,
+      "step": 1067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962794987385506e-05,
+      "loss": 0.1885,
+      "step": 1068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627010903147824e-05,
+      "loss": 0.1645,
+      "step": 1069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626070771574766e-05,
+      "loss": 0.1552,
+      "step": 1070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962512947924925e-05,
+      "loss": 0.1786,
+      "step": 1071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962418702628479e-05,
+      "loss": 0.1609,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962324341279502e-05,
+      "loss": 0.2014,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622298638893738e-05,
+      "loss": 0.1835,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621352704694854e-05,
+      "loss": 0.1777,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620405610312446e-05,
+      "loss": 0.1722,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961945735586071e-05,
+      "loss": 0.1792,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961850794145399e-05,
+      "loss": 0.159,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961755736720678e-05,
+      "loss": 0.1766,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961660563323369e-05,
+      "loss": 0.1629,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615652739649496e-05,
+      "loss": 0.1785,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614698686569103e-05,
+      "loss": 0.1899,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613743474107542e-05,
+      "loss": 0.1708,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961278710238001e-05,
+      "loss": 0.203,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961182957150182e-05,
+      "loss": 0.1714,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610870881588443e-05,
+      "loss": 0.1789,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609911032755476e-05,
+      "loss": 0.1736,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608950025118665e-05,
+      "loss": 0.1671,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607987858793892e-05,
+      "loss": 0.1934,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960702453389718e-05,
+      "loss": 0.1687,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606060050544688e-05,
+      "loss": 0.1895,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960509440885272e-05,
+      "loss": 0.1617,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604127608937717e-05,
+      "loss": 0.1671,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603159650916255e-05,
+      "loss": 0.1698,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602190534905063e-05,
+      "loss": 0.1714,
+      "step": 1095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601220261020995e-05,
+      "loss": 0.1916,
+      "step": 1096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600248829381052e-05,
+      "loss": 0.1667,
+      "step": 1097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599276240102373e-05,
+      "loss": 0.1574,
+      "step": 1098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959830249330224e-05,
+      "loss": 0.1812,
+      "step": 1099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959732758909807e-05,
+      "loss": 0.1689,
+      "step": 1100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959635152760742e-05,
+      "loss": 0.1706,
+      "step": 1101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595374308947984e-05,
+      "loss": 0.1814,
+      "step": 1102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594395933237607e-05,
+      "loss": 0.1957,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593416400594263e-05,
+      "loss": 0.1901,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592435711136063e-05,
+      "loss": 0.1868,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959145386498127e-05,
+      "loss": 0.1768,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590470862248274e-05,
+      "loss": 0.1761,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589486703055613e-05,
+      "loss": 0.1571,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588501387521958e-05,
+      "loss": 0.1912,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587514915766124e-05,
+      "loss": 0.1662,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586527287907065e-05,
+      "loss": 0.1819,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585538504063874e-05,
+      "loss": 0.1867,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958454856435578e-05,
+      "loss": 0.1415,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958355746890215e-05,
+      "loss": 0.1807,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958256521782251e-05,
+      "loss": 0.1562,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581571811236494e-05,
+      "loss": 0.1698,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580577249263894e-05,
+      "loss": 0.1672,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579581532024646e-05,
+      "loss": 0.1541,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578584659638812e-05,
+      "loss": 0.1778,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95775866322266e-05,
+      "loss": 0.1792,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576587449908356e-05,
+      "loss": 0.1842,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575587112804572e-05,
+      "loss": 0.1624,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957458562103586e-05,
+      "loss": 0.1904,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573582974722994e-05,
+      "loss": 0.1997,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572579173986875e-05,
+      "loss": 0.1766,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571574218948543e-05,
+      "loss": 0.1734,
+      "step": 1126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570568109729186e-05,
+      "loss": 0.1666,
+      "step": 1127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956956084645012e-05,
+      "loss": 0.1867,
+      "step": 1128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568552429232804e-05,
+      "loss": 0.1692,
+      "step": 1129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956754285819884e-05,
+      "loss": 0.171,
+      "step": 1130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566532133469964e-05,
+      "loss": 0.1669,
+      "step": 1131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565520255168054e-05,
+      "loss": 0.1937,
+      "step": 1132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564507223415133e-05,
+      "loss": 0.1599,
+      "step": 1133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563493038333348e-05,
+      "loss": 0.1627,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562477700044997e-05,
+      "loss": 0.1466,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561461208672515e-05,
+      "loss": 0.1808,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560443564338474e-05,
+      "loss": 0.153,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559424767165584e-05,
+      "loss": 0.1986,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558404817276697e-05,
+      "loss": 0.1774,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557383714794804e-05,
+      "loss": 0.1939,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955636145984303e-05,
+      "loss": 0.1656,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955533805254465e-05,
+      "loss": 0.1851,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554313493023068e-05,
+      "loss": 0.1723,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553287781401825e-05,
+      "loss": 0.1467,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552260917804613e-05,
+      "loss": 0.1724,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551232902355252e-05,
+      "loss": 0.1609,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95502037351777e-05,
+      "loss": 0.1723,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549173416396063e-05,
+      "loss": 0.1693,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954814194613458e-05,
+      "loss": 0.1967,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547109324517634e-05,
+      "loss": 0.1683,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954607555166974e-05,
+      "loss": 0.1722,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545040627715554e-05,
+      "loss": 0.1634,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954400455277987e-05,
+      "loss": 0.1902,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542967326987625e-05,
+      "loss": 0.1727,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541928950463893e-05,
+      "loss": 0.1834,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540889423333885e-05,
+      "loss": 0.1883,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953984874572295e-05,
+      "loss": 0.1749,
+      "step": 1157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538806917756573e-05,
+      "loss": 0.178,
+      "step": 1158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537763939560394e-05,
+      "loss": 0.1682,
+      "step": 1159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953671981126017e-05,
+      "loss": 0.1546,
+      "step": 1160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953567453298181e-05,
+      "loss": 0.1632,
+      "step": 1161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534628104851357e-05,
+      "loss": 0.1747,
+      "step": 1162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533580526994993e-05,
+      "loss": 0.1632,
+      "step": 1163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532531799539042e-05,
+      "loss": 0.1601,
+      "step": 1164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531481922609962e-05,
+      "loss": 0.1639,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953043089633435e-05,
+      "loss": 0.157,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952937872083895e-05,
+      "loss": 0.1681,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528325396250628e-05,
+      "loss": 0.1753,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527270922696408e-05,
+      "loss": 0.1996,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952621530030344e-05,
+      "loss": 0.177,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952515852919901e-05,
+      "loss": 0.1651,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952410060951055e-05,
+      "loss": 0.1639,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523041541365636e-05,
+      "loss": 0.1683,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521981324891964e-05,
+      "loss": 0.1843,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520919960217386e-05,
+      "loss": 0.1845,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519857447469883e-05,
+      "loss": 0.1591,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518793786777583e-05,
+      "loss": 0.1899,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951772897826874e-05,
+      "loss": 0.1839,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516663022071752e-05,
+      "loss": 0.1873,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515595918315164e-05,
+      "loss": 0.1755,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951452766712765e-05,
+      "loss": 0.1763,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513458268638018e-05,
+      "loss": 0.1814,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512387722975225e-05,
+      "loss": 0.1611,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511316030268364e-05,
+      "loss": 0.1671,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951024319064666e-05,
+      "loss": 0.1691,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509169204239486e-05,
+      "loss": 0.1792,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508094071176345e-05,
+      "loss": 0.1695,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507017791586877e-05,
+      "loss": 0.1694,
+      "step": 1188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950594036560087e-05,
+      "loss": 0.169,
+      "step": 1189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504861793348236e-05,
+      "loss": 0.1802,
+      "step": 1190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503782074959047e-05,
+      "loss": 0.1781,
+      "step": 1191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950270121056349e-05,
+      "loss": 0.1758,
+      "step": 1192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950161920029191e-05,
+      "loss": 0.1854,
+      "step": 1193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500536044274767e-05,
+      "loss": 0.1742,
+      "step": 1194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499451742642684e-05,
+      "loss": 0.1823,
+      "step": 1195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498366295526404e-05,
+      "loss": 0.1812,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497279703056814e-05,
+      "loss": 0.1694,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496191965364948e-05,
+      "loss": 0.1514,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495103082581962e-05,
+      "loss": 0.1912,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949401305483916e-05,
+      "loss": 0.1793,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492921882267988e-05,
+      "loss": 0.1767,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491829565000017e-05,
+      "loss": 0.1941,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490736103166964e-05,
+      "loss": 0.1763,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489641496900687e-05,
+      "loss": 0.1765,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488545746333172e-05,
+      "loss": 0.1748,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948744885159656e-05,
+      "loss": 0.1673,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486350812823106e-05,
+      "loss": 0.1778,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948525163014523e-05,
+      "loss": 0.179,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484151303695466e-05,
+      "loss": 0.1686,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483049833606496e-05,
+      "loss": 0.1654,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481947220011146e-05,
+      "loss": 0.1773,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948084346304237e-05,
+      "loss": 0.1845,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479738562833264e-05,
+      "loss": 0.1588,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947863251951706e-05,
+      "loss": 0.1747,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477525333227135e-05,
+      "loss": 0.1712,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476417004096995e-05,
+      "loss": 0.1572,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475307532260287e-05,
+      "loss": 0.1742,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474196917850793e-05,
+      "loss": 0.1852,
+      "step": 1218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473085161002438e-05,
+      "loss": 0.1741,
+      "step": 1219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471972261849285e-05,
+      "loss": 0.1834,
+      "step": 1220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470858220525528e-05,
+      "loss": 0.1905,
+      "step": 1221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469743037165504e-05,
+      "loss": 0.1624,
+      "step": 1222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468626711903687e-05,
+      "loss": 0.167,
+      "step": 1223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467509244874688e-05,
+      "loss": 0.1777,
+      "step": 1224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466390636213253e-05,
+      "loss": 0.1517,
+      "step": 1225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465270886054277e-05,
+      "loss": 0.1639,
+      "step": 1226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946414999453278e-05,
+      "loss": 0.1885,
+      "step": 1227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946302796178392e-05,
+      "loss": 0.1624,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461904787942995e-05,
+      "loss": 0.1718,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460780473145452e-05,
+      "loss": 0.1925,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459655017526856e-05,
+      "loss": 0.1616,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458528421222925e-05,
+      "loss": 0.1659,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457400684369507e-05,
+      "loss": 0.176,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945627180710259e-05,
+      "loss": 0.1685,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455141789558297e-05,
+      "loss": 0.1781,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454010631872892e-05,
+      "loss": 0.1705,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452878334182773e-05,
+      "loss": 0.1748,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945174489662448e-05,
+      "loss": 0.1508,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450610319334684e-05,
+      "loss": 0.1666,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94494746024502e-05,
+      "loss": 0.1642,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448337746107978e-05,
+      "loss": 0.1705,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447199750445102e-05,
+      "loss": 0.1823,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94460606155988e-05,
+      "loss": 0.1703,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944492034170643e-05,
+      "loss": 0.1642,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443778928905493e-05,
+      "loss": 0.1555,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442636377333626e-05,
+      "loss": 0.1568,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441492687128602e-05,
+      "loss": 0.1717,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944034785842833e-05,
+      "loss": 0.1836,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439201891370864e-05,
+      "loss": 0.168,
+      "step": 1249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438054786094385e-05,
+      "loss": 0.1673,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436906542737215e-05,
+      "loss": 0.1686,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435757161437815e-05,
+      "loss": 0.1675,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434606642334786e-05,
+      "loss": 0.1644,
+      "step": 1253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433454985566856e-05,
+      "loss": 0.1766,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432302191272903e-05,
+      "loss": 0.1609,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943114825959193e-05,
+      "loss": 0.1716,
+      "step": 1256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942999319066309e-05,
+      "loss": 0.1807,
+      "step": 1257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9428836984625658e-05,
+      "loss": 0.1665,
+      "step": 1258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942767964161906e-05,
+      "loss": 0.1603,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942652116178285e-05,
+      "loss": 0.1504,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942536154525673e-05,
+      "loss": 0.1767,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942420079218052e-05,
+      "loss": 0.1612,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423038902694197e-05,
+      "loss": 0.1817,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421875876937858e-05,
+      "loss": 0.1812,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942071171505176e-05,
+      "loss": 0.1815,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419546417176268e-05,
+      "loss": 0.1748,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418379983451902e-05,
+      "loss": 0.1669,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941721241401932e-05,
+      "loss": 0.1521,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416043709019316e-05,
+      "loss": 0.1498,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414873868592804e-05,
+      "loss": 0.1745,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941370289288086e-05,
+      "loss": 0.1629,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412530782024683e-05,
+      "loss": 0.1973,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411357536165605e-05,
+      "loss": 0.1811,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410183155445107e-05,
+      "loss": 0.1705,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94090076400048e-05,
+      "loss": 0.1571,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940783098998643e-05,
+      "loss": 0.1537,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406653205531887e-05,
+      "loss": 0.1699,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940547428678319e-05,
+      "loss": 0.1677,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404294233882494e-05,
+      "loss": 0.1807,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403113046972103e-05,
+      "loss": 0.1607,
+      "step": 1280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401930726194448e-05,
+      "loss": 0.1555,
+      "step": 1281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940074727169209e-05,
+      "loss": 0.1714,
+      "step": 1282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399562683607748e-05,
+      "loss": 0.1738,
+      "step": 1283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939837696208425e-05,
+      "loss": 0.1734,
+      "step": 1284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939719010726459e-05,
+      "loss": 0.1625,
+      "step": 1285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396002119291876e-05,
+      "loss": 0.1707,
+      "step": 1286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394812998309358e-05,
+      "loss": 0.1887,
+      "step": 1287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393622744460432e-05,
+      "loss": 0.1675,
+      "step": 1288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939243135788862e-05,
+      "loss": 0.1536,
+      "step": 1289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391238838737587e-05,
+      "loss": 0.1737,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390045187151125e-05,
+      "loss": 0.1606,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938885040327318e-05,
+      "loss": 0.1615,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387654487247817e-05,
+      "loss": 0.1597,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386457439219245e-05,
+      "loss": 0.1712,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385259259331812e-05,
+      "loss": 0.1645,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384059947729997e-05,
+      "loss": 0.1516,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382859504558423e-05,
+      "loss": 0.1747,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381657929961837e-05,
+      "loss": 0.163,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380455224085137e-05,
+      "loss": 0.1808,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379251387073347e-05,
+      "loss": 0.167,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937804641907163e-05,
+      "loss": 0.1759,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376840320225285e-05,
+      "loss": 0.1619,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375633090679754e-05,
+      "loss": 0.1709,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374424730580602e-05,
+      "loss": 0.1762,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937321524007355e-05,
+      "loss": 0.1862,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937200461930443e-05,
+      "loss": 0.1695,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370792868419234e-05,
+      "loss": 0.1844,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369579987564076e-05,
+      "loss": 0.1568,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368365976885206e-05,
+      "loss": 0.1755,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367150836529024e-05,
+      "loss": 0.1588,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936593456664205e-05,
+      "loss": 0.1623,
+      "step": 1311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364717167370947e-05,
+      "loss": 0.1603,
+      "step": 1312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363498638862515e-05,
+      "loss": 0.1715,
+      "step": 1313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362278981263694e-05,
+      "loss": 0.1642,
+      "step": 1314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936105819472155e-05,
+      "loss": 0.1696,
+      "step": 1315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359836279383293e-05,
+      "loss": 0.1725,
+      "step": 1316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358613235396263e-05,
+      "loss": 0.1915,
+      "step": 1317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357389062907946e-05,
+      "loss": 0.1598,
+      "step": 1318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356163762065952e-05,
+      "loss": 0.1608,
+      "step": 1319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935493733301804e-05,
+      "loss": 0.1773,
+      "step": 1320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9353709775912092e-05,
+      "loss": 0.1783,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352481090896135e-05,
+      "loss": 0.1596,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351251278118326e-05,
+      "loss": 0.1718,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935002033772696e-05,
+      "loss": 0.1513,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934878826987048e-05,
+      "loss": 0.1903,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934755507469744e-05,
+      "loss": 0.162,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346320752356554e-05,
+      "loss": 0.1548,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345085302996653e-05,
+      "loss": 0.1731,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934384872676672e-05,
+      "loss": 0.1588,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342611023815863e-05,
+      "loss": 0.1774,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934137219429333e-05,
+      "loss": 0.1833,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340132238348504e-05,
+      "loss": 0.1686,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933889115613091e-05,
+      "loss": 0.1765,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337648947790194e-05,
+      "loss": 0.1706,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336405613476152e-05,
+      "loss": 0.1697,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335161153338707e-05,
+      "loss": 0.1701,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333915567527923e-05,
+      "loss": 0.1669,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332668856194003e-05,
+      "loss": 0.169,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933142101948727e-05,
+      "loss": 0.1893,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330172057558206e-05,
+      "loss": 0.1746,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328921970557407e-05,
+      "loss": 0.167,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327670758635616e-05,
+      "loss": 0.158,
+      "step": 1342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932641842194371e-05,
+      "loss": 0.1774,
+      "step": 1343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325164960632704e-05,
+      "loss": 0.1708,
+      "step": 1344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932391037485374e-05,
+      "loss": 0.1772,
+      "step": 1345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322654664758102e-05,
+      "loss": 0.1697,
+      "step": 1346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321397830497214e-05,
+      "loss": 0.1683,
+      "step": 1347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320139872222628e-05,
+      "loss": 0.1732,
+      "step": 1348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318880790086032e-05,
+      "loss": 0.1814,
+      "step": 1349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317620584239255e-05,
+      "loss": 0.164,
+      "step": 1350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931635925483425e-05,
+      "loss": 0.1788,
+      "step": 1351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315096802023124e-05,
+      "loss": 0.1531,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313833225958105e-05,
+      "loss": 0.1746,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312568526791556e-05,
+      "loss": 0.1462,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311302704675984e-05,
+      "loss": 0.1834,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931003575976403e-05,
+      "loss": 0.1549,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308767692208457e-05,
+      "loss": 0.1759,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307498502162184e-05,
+      "loss": 0.1712,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306228189778255e-05,
+      "loss": 0.1995,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304956755209848e-05,
+      "loss": 0.1635,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303684198610272e-05,
+      "loss": 0.1551,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302410520132983e-05,
+      "loss": 0.1825,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930113571993157e-05,
+      "loss": 0.1795,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299859798159745e-05,
+      "loss": 0.1721,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929858275497137e-05,
+      "loss": 0.1661,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297304590520437e-05,
+      "loss": 0.1591,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296025304961068e-05,
+      "loss": 0.1712,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929474489844753e-05,
+      "loss": 0.1705,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929346337113421e-05,
+      "loss": 0.1587,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292180723175656e-05,
+      "loss": 0.1729,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929089695472652e-05,
+      "loss": 0.1669,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289612065941612e-05,
+      "loss": 0.1521,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928832605697587e-05,
+      "loss": 0.163,
+      "step": 1373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287038927984358e-05,
+      "loss": 0.1586,
+      "step": 1374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285750679122293e-05,
+      "loss": 0.1835,
+      "step": 1375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284461310545018e-05,
+      "loss": 0.175,
+      "step": 1376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283170822408002e-05,
+      "loss": 0.1755,
+      "step": 1377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281879214866863e-05,
+      "loss": 0.1678,
+      "step": 1378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928058648807735e-05,
+      "loss": 0.175,
+      "step": 1379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927929264219534e-05,
+      "loss": 0.1873,
+      "step": 1380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277997677376855e-05,
+      "loss": 0.1729,
+      "step": 1381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276701593778045e-05,
+      "loss": 0.1677,
+      "step": 1382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92754043915552e-05,
+      "loss": 0.163,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274106070864737e-05,
+      "loss": 0.1665,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927280663186322e-05,
+      "loss": 0.1581,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271506074707337e-05,
+      "loss": 0.155,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927020439955391e-05,
+      "loss": 0.1631,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926890160655991e-05,
+      "loss": 0.1652,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926759769588243e-05,
+      "loss": 0.1705,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266292667678698e-05,
+      "loss": 0.1688,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926498652210608e-05,
+      "loss": 0.1723,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926367925932208e-05,
+      "loss": 0.1727,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926237087948433e-05,
+      "loss": 0.1554,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92610613827506e-05,
+      "loss": 0.16,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259750769278804e-05,
+      "loss": 0.1598,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925843903922697e-05,
+      "loss": 0.1676,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925712619275327e-05,
+      "loss": 0.1682,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925581223001602e-05,
+      "loss": 0.177,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254497151173665e-05,
+      "loss": 0.1736,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253180956384778e-05,
+      "loss": 0.1812,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925186364580807e-05,
+      "loss": 0.1823,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250545219602393e-05,
+      "loss": 0.184,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249225677926726e-05,
+      "loss": 0.1616,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247905020940182e-05,
+      "loss": 0.1644,
+      "step": 1404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246583248802016e-05,
+      "loss": 0.1759,
+      "step": 1405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924526036167161e-05,
+      "loss": 0.1685,
+      "step": 1406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243936359708488e-05,
+      "loss": 0.1716,
+      "step": 1407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242611243072294e-05,
+      "loss": 0.1868,
+      "step": 1408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241285011922826e-05,
+      "loss": 0.1879,
+      "step": 1409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239957666420005e-05,
+      "loss": 0.1671,
+      "step": 1410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923862920672388e-05,
+      "loss": 0.1998,
+      "step": 1411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237299632994654e-05,
+      "loss": 0.1884,
+      "step": 1412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9235968945392645e-05,
+      "loss": 0.1904,
+      "step": 1413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923463714407831e-05,
+      "loss": 0.1736,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9233304229212254e-05,
+      "loss": 0.1631,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92319702009552e-05,
+      "loss": 0.1734,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230635059468006e-05,
+      "loss": 0.1807,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229298804911676e-05,
+      "loss": 0.1639,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227961437447336e-05,
+      "loss": 0.1786,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226622957236257e-05,
+      "loss": 0.1498,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225283364439833e-05,
+      "loss": 0.1647,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223942659219602e-05,
+      "loss": 0.1729,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222600841737227e-05,
+      "loss": 0.1636,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221257912154518e-05,
+      "loss": 0.1753,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219913870633407e-05,
+      "loss": 0.1655,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218568717335962e-05,
+      "loss": 0.1911,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921722245242439e-05,
+      "loss": 0.1738,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215875076061026e-05,
+      "loss": 0.1704,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214526588408345e-05,
+      "loss": 0.1822,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213176989628954e-05,
+      "loss": 0.1644,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211826279885594e-05,
+      "loss": 0.1902,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9210474459341137e-05,
+      "loss": 0.1671,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920912152815859e-05,
+      "loss": 0.1803,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207767486501097e-05,
+      "loss": 0.1554,
+      "step": 1434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206412334531938e-05,
+      "loss": 0.1801,
+      "step": 1435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920505607241452e-05,
+      "loss": 0.191,
+      "step": 1436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9203698700312382e-05,
+      "loss": 0.1779,
+      "step": 1437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202340218389207e-05,
+      "loss": 0.1845,
+      "step": 1438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200980626808806e-05,
+      "loss": 0.1742,
+      "step": 1439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199619925735122e-05,
+      "loss": 0.1758,
+      "step": 1440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198258115332237e-05,
+      "loss": 0.1726,
+      "step": 1441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196895195764363e-05,
+      "loss": 0.1732,
+      "step": 1442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195531167195845e-05,
+      "loss": 0.1984,
+      "step": 1443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194166029791165e-05,
+      "loss": 0.162,
+      "step": 1444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192799783714938e-05,
+      "loss": 0.1767,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919143242913191e-05,
+      "loss": 0.1633,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190063966206966e-05,
+      "loss": 0.1683,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188694395105113e-05,
+      "loss": 0.1904,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187323715991504e-05,
+      "loss": 0.1681,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9185951929031423e-05,
+      "loss": 0.1732,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184579034390288e-05,
+      "loss": 0.178,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918320503223364e-05,
+      "loss": 0.185,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918182992272717e-05,
+      "loss": 0.1799,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180453706036695e-05,
+      "loss": 0.1732,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179076382328157e-05,
+      "loss": 0.1715,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177697951767646e-05,
+      "loss": 0.1698,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176318414521377e-05,
+      "loss": 0.1727,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174937770755702e-05,
+      "loss": 0.1661,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173556020637102e-05,
+      "loss": 0.1584,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172173164332197e-05,
+      "loss": 0.1773,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917078920200774e-05,
+      "loss": 0.1473,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169404133830607e-05,
+      "loss": 0.1696,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168017959967823e-05,
+      "loss": 0.1662,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166630680586534e-05,
+      "loss": 0.1526,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165242295854028e-05,
+      "loss": 0.1559,
+      "step": 1465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163852805937717e-05,
+      "loss": 0.1598,
+      "step": 1466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162462211005154e-05,
+      "loss": 0.1645,
+      "step": 1467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161070511224028e-05,
+      "loss": 0.1981,
+      "step": 1468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915967770676215e-05,
+      "loss": 0.1592,
+      "step": 1469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915828379778747e-05,
+      "loss": 0.1614,
+      "step": 1470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915688878446808e-05,
+      "loss": 0.1622,
+      "step": 1471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155492666972182e-05,
+      "loss": 0.1556,
+      "step": 1472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915409544546814e-05,
+      "loss": 0.1911,
+      "step": 1473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152697120124426e-05,
+      "loss": 0.1459,
+      "step": 1474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915129769110966e-05,
+      "loss": 0.1633,
+      "step": 1475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149897158592597e-05,
+      "loss": 0.1956,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148495522742112e-05,
+      "loss": 0.1713,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147092783727222e-05,
+      "loss": 0.1654,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145688941717074e-05,
+      "loss": 0.1528,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144283996880952e-05,
+      "loss": 0.1805,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914287794938827e-05,
+      "loss": 0.1679,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914147079940857e-05,
+      "loss": 0.1615,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914006254711154e-05,
+      "loss": 0.1783,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138653192666987e-05,
+      "loss": 0.1409,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913724273624486e-05,
+      "loss": 0.1552,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135831178015238e-05,
+      "loss": 0.1571,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913441851814833e-05,
+      "loss": 0.2006,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913300475681448e-05,
+      "loss": 0.1771,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131589894184173e-05,
+      "loss": 0.1657,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913017393042801e-05,
+      "loss": 0.1716,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128756865716738e-05,
+      "loss": 0.171,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127338700221232e-05,
+      "loss": 0.1699,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91259194341125e-05,
+      "loss": 0.1829,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124499067561686e-05,
+      "loss": 0.1639,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912307760074006e-05,
+      "loss": 0.1686,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121655033819028e-05,
+      "loss": 0.1636,
+      "step": 1496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120231366970135e-05,
+      "loss": 0.1643,
+      "step": 1497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118806600365048e-05,
+      "loss": 0.1566,
+      "step": 1498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911738073417557e-05,
+      "loss": 0.1584,
+      "step": 1499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115953768573643e-05,
+      "loss": 0.1675,
+      "step": 1500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114525703731334e-05,
+      "loss": 0.165,
+      "step": 1501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113096539820844e-05,
+      "loss": 0.183,
+      "step": 1502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111666277014508e-05,
+      "loss": 0.1623,
+      "step": 1503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110234915484795e-05,
+      "loss": 0.1602,
+      "step": 1504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108802455404304e-05,
+      "loss": 0.1708,
+      "step": 1505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107368896945768e-05,
+      "loss": 0.1614,
+      "step": 1506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105934240282047e-05,
+      "loss": 0.191,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910449848558614e-05,
+      "loss": 0.1829,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910306163303118e-05,
+      "loss": 0.1647,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910162368279043e-05,
+      "loss": 0.1479,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100184635037277e-05,
+      "loss": 0.1707,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909874448994525e-05,
+      "loss": 0.1565,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909730324768801e-05,
+      "loss": 0.1805,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095860908439348e-05,
+      "loss": 0.1664,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094417472373183e-05,
+      "loss": 0.1764,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092972939663578e-05,
+      "loss": 0.1681,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091527310484713e-05,
+      "loss": 0.1614,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090080585010912e-05,
+      "loss": 0.1867,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908863276341663e-05,
+      "loss": 0.1484,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908718384587645e-05,
+      "loss": 0.1953,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085733832565082e-05,
+      "loss": 0.1867,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084282723657384e-05,
+      "loss": 0.1557,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908283051932833e-05,
+      "loss": 0.1624,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081377219753036e-05,
+      "loss": 0.1638,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907992282510675e-05,
+      "loss": 0.1951,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907846733556484e-05,
+      "loss": 0.1454,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907701075130283e-05,
+      "loss": 0.1604,
+      "step": 1527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075553072496346e-05,
+      "loss": 0.1702,
+      "step": 1528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907409429932117e-05,
+      "loss": 0.1625,
+      "step": 1529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072634431953204e-05,
+      "loss": 0.1656,
+      "step": 1530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9071173470568487e-05,
+      "loss": 0.1469,
+      "step": 1531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069711415343183e-05,
+      "loss": 0.1851,
+      "step": 1532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9068248266453597e-05,
+      "loss": 0.1749,
+      "step": 1533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906678402407616e-05,
+      "loss": 0.1731,
+      "step": 1534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065318688387438e-05,
+      "loss": 0.1824,
+      "step": 1535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063852259564127e-05,
+      "loss": 0.1722,
+      "step": 1536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062384737783056e-05,
+      "loss": 0.1461,
+      "step": 1537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9060916123221186e-05,
+      "loss": 0.1594,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9059446416055602e-05,
+      "loss": 0.1701,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057975616463533e-05,
+      "loss": 0.1524,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056503724622335e-05,
+      "loss": 0.1581,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055030740709494e-05,
+      "loss": 0.1694,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053556664902628e-05,
+      "loss": 0.196,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052081497379483e-05,
+      "loss": 0.1649,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905060523831795e-05,
+      "loss": 0.1741,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904912788789604e-05,
+      "loss": 0.1389,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047649446291893e-05,
+      "loss": 0.1662,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904616991368379e-05,
+      "loss": 0.1705,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044689290250135e-05,
+      "loss": 0.1789,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043207576169473e-05,
+      "loss": 0.1693,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904172477162048e-05,
+      "loss": 0.1498,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040240876781944e-05,
+      "loss": 0.1887,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038755891832806e-05,
+      "loss": 0.1335,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903726981695214e-05,
+      "loss": 0.1696,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035782652319133e-05,
+      "loss": 0.1949,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903429439811312e-05,
+      "loss": 0.1749,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032805054513555e-05,
+      "loss": 0.1688,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9031314621700035e-05,
+      "loss": 0.1863,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902982309985228e-05,
+      "loss": 0.1772,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028330489150147e-05,
+      "loss": 0.1686,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026836789773616e-05,
+      "loss": 0.1647,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025342001902806e-05,
+      "loss": 0.1714,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023846125717967e-05,
+      "loss": 0.1628,
+      "step": 1563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022349161399475e-05,
+      "loss": 0.1592,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902085110912784e-05,
+      "loss": 0.1644,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901935196908371e-05,
+      "loss": 0.1579,
+      "step": 1566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017851741447846e-05,
+      "loss": 0.1716,
+      "step": 1567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9016350426401163e-05,
+      "loss": 0.173,
+      "step": 1568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014848024124692e-05,
+      "loss": 0.1696,
+      "step": 1569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9013344534799598e-05,
+      "loss": 0.1678,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011839958607176e-05,
+      "loss": 0.1597,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010334295728858e-05,
+      "loss": 0.1566,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008827546346206e-05,
+      "loss": 0.2042,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007319710640902e-05,
+      "loss": 0.1918,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005810788794774e-05,
+      "loss": 0.1644,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004300780989772e-05,
+      "loss": 0.1637,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002789687407976e-05,
+      "loss": 0.1551,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001277508231607e-05,
+      "loss": 0.1835,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8999764243643002e-05,
+      "loss": 0.1684,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998249893824644e-05,
+      "loss": 0.161,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996734458959134e-05,
+      "loss": 0.163,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995217939229215e-05,
+      "loss": 0.1591,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899370033481775e-05,
+      "loss": 0.1571,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992181645907738e-05,
+      "loss": 0.1571,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8990661872682315e-05,
+      "loss": 0.1362,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898914101532474e-05,
+      "loss": 0.1747,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89876190740184e-05,
+      "loss": 0.1898,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986096048946826e-05,
+      "loss": 0.1829,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898457194029366e-05,
+      "loss": 0.1483,
+      "step": 1589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983046748242685e-05,
+      "loss": 0.1577,
+      "step": 1590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981520472977827e-05,
+      "loss": 0.1626,
+      "step": 1591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979993114683123e-05,
+      "loss": 0.159,
+      "step": 1592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897846467354275e-05,
+      "loss": 0.1809,
+      "step": 1593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976935149741012e-05,
+      "loss": 0.1694,
+      "step": 1594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897540454346235e-05,
+      "loss": 0.1803,
+      "step": 1595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897387285489132e-05,
+      "loss": 0.155,
+      "step": 1596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972340084212636e-05,
+      "loss": 0.1631,
+      "step": 1597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897080623161111e-05,
+      "loss": 0.1848,
+      "step": 1598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8969271297271714e-05,
+      "loss": 0.1652,
+      "step": 1599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967735281379528e-05,
+      "loss": 0.1747,
+      "step": 1600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966198184119775e-05,
+      "loss": 0.1389,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8964660005677803e-05,
+      "loss": 0.178,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8963120746239096e-05,
+      "loss": 0.1571,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961580405989255e-05,
+      "loss": 0.1591,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896003898511403e-05,
+      "loss": 0.1825,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895849648379929e-05,
+      "loss": 0.1689,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956952902231035e-05,
+      "loss": 0.1603,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955408240595396e-05,
+      "loss": 0.1755,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953862499078634e-05,
+      "loss": 0.1677,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8952315677867145e-05,
+      "loss": 0.1578,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950767777147452e-05,
+      "loss": 0.1632,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89492187971062e-05,
+      "loss": 0.1729,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947668737930178e-05,
+      "loss": 0.1501,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946117599806298e-05,
+      "loss": 0.1429,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944565382921602e-05,
+      "loss": 0.1741,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894301208746326e-05,
+      "loss": 0.1761,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894145771361858e-05,
+      "loss": 0.1589,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939902261574992e-05,
+      "loss": 0.1546,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893834573152006e-05,
+      "loss": 0.1633,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936788123641476e-05,
+      "loss": 0.1858,
+      "step": 1620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935229438127064e-05,
+      "loss": 0.1743,
+      "step": 1621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933669675164775e-05,
+      "loss": 0.1732,
+      "step": 1622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932108834942697e-05,
+      "loss": 0.1677,
+      "step": 1623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930546917649034e-05,
+      "loss": 0.143,
+      "step": 1624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892898392347214e-05,
+      "loss": 0.1856,
+      "step": 1625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927419852600478e-05,
+      "loss": 0.1782,
+      "step": 1626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925854705222653e-05,
+      "loss": 0.1541,
+      "step": 1627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924288481527397e-05,
+      "loss": 0.1724,
+      "step": 1628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922721181703575e-05,
+      "loss": 0.1622,
+      "step": 1629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892115280594017e-05,
+      "loss": 0.1685,
+      "step": 1630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8919583354426315e-05,
+      "loss": 0.188,
+      "step": 1631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8918012827351253e-05,
+      "loss": 0.1608,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891644122490437e-05,
+      "loss": 0.1675,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914868547275173e-05,
+      "loss": 0.1568,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.89132947946533e-05,
+      "loss": 0.1739,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911719967228528e-05,
+      "loss": 0.1603,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910144065190753e-05,
+      "loss": 0.16,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890856708873e-05,
+      "loss": 0.1693,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890698903803643e-05,
+      "loss": 0.1666,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905409913300338e-05,
+      "loss": 0.1662,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903829714712126e-05,
+      "loss": 0.169,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902248442462356e-05,
+      "loss": 0.1694,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.89006660967417e-05,
+      "loss": 0.1616,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899082677740962e-05,
+      "loss": 0.1678,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897498185651077e-05,
+      "loss": 0.1654,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889591262066311e-05,
+      "loss": 0.1695,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889432598296826e-05,
+      "loss": 0.172,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892738272757843e-05,
+      "loss": 0.177,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891149490223313e-05,
+      "loss": 0.1686,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888955963555626e-05,
+      "loss": 0.179,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887968708948388e-05,
+      "loss": 0.1615,
+      "step": 1651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886376710591542e-05,
+      "loss": 0.1753,
+      "step": 1652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884783640677686e-05,
+      "loss": 0.1843,
+      "step": 1653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883189499398925e-05,
+      "loss": 0.1585,
+      "step": 1654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881594286947484e-05,
+      "loss": 0.1636,
+      "step": 1655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879998003515722e-05,
+      "loss": 0.1645,
+      "step": 1656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8878400649296123e-05,
+      "loss": 0.1745,
+      "step": 1657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887680222448131e-05,
+      "loss": 0.1669,
+      "step": 1658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875202729264018e-05,
+      "loss": 0.1571,
+      "step": 1659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887360216383713e-05,
+      "loss": 0.172,
+      "step": 1660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872000528393642e-05,
+      "loss": 0.17,
+      "step": 1661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870397823126686e-05,
+      "loss": 0.1625,
+      "step": 1662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886879404822953e-05,
+      "loss": 0.1436,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886718920389556e-05,
+      "loss": 0.1577,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865583290318294e-05,
+      "loss": 0.1781,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863976307691378e-05,
+      "loss": 0.1776,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886236825620859e-05,
+      "loss": 0.1557,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860759136063844e-05,
+      "loss": 0.1493,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885914894745116e-05,
+      "loss": 0.1587,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857537690564713e-05,
+      "loss": 0.1522,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855925365598792e-05,
+      "loss": 0.1451,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854311972747816e-05,
+      "loss": 0.1635,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852697512206334e-05,
+      "loss": 0.168,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885108198416903e-05,
+      "loss": 0.1585,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8849465388830702e-05,
+      "loss": 0.1756,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847847726386295e-05,
+      "loss": 0.1826,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884622899703087e-05,
+      "loss": 0.1534,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844609200959617e-05,
+      "loss": 0.1638,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842988338367866e-05,
+      "loss": 0.1517,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841366409451057e-05,
+      "loss": 0.1657,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883974341440478e-05,
+      "loss": 0.1635,
+      "step": 1681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838119353424735e-05,
+      "loss": 0.1689,
+      "step": 1682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883649422670676e-05,
+      "loss": 0.1485,
+      "step": 1683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883486803444682e-05,
+      "loss": 0.1562,
+      "step": 1684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883324077684101e-05,
+      "loss": 0.1737,
+      "step": 1685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883161245408555e-05,
+      "loss": 0.158,
+      "step": 1686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829983066376793e-05,
+      "loss": 0.1573,
+      "step": 1687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882835261391121e-05,
+      "loss": 0.1657,
+      "step": 1688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826721096885412e-05,
+      "loss": 0.1513,
+      "step": 1689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825088515496138e-05,
+      "loss": 0.17,
+      "step": 1690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823454869940243e-05,
+      "loss": 0.1501,
+      "step": 1691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882182016041473e-05,
+      "loss": 0.1612,
+      "step": 1692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820184387116713e-05,
+      "loss": 0.1559,
+      "step": 1693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8818547550243436e-05,
+      "loss": 0.1826,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881690964999228e-05,
+      "loss": 0.1467,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8815270686560757e-05,
+      "loss": 0.165,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881363066014649e-05,
+      "loss": 0.1759,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811989570947246e-05,
+      "loss": 0.1557,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810347419160912e-05,
+      "loss": 0.1617,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8808704204985503e-05,
+      "loss": 0.1633,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880705992861917e-05,
+      "loss": 0.1595,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8805414590260188e-05,
+      "loss": 0.1658,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880376819010695e-05,
+      "loss": 0.1664,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802120728357996e-05,
+      "loss": 0.1535,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8800472205211977e-05,
+      "loss": 0.1727,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879882262086768e-05,
+      "loss": 0.1689,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797171975524024e-05,
+      "loss": 0.1761,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795520269380043e-05,
+      "loss": 0.1586,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8793867502634914e-05,
+      "loss": 0.1346,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879221367548793e-05,
+      "loss": 0.1557,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879055878813852e-05,
+      "loss": 0.1684,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788902840786233e-05,
+      "loss": 0.1571,
+      "step": 1712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787245833630753e-05,
+      "loss": 0.1547,
+      "step": 1713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878558776687189e-05,
+      "loss": 0.1735,
+      "step": 1714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783928640709577e-05,
+      "loss": 0.1565,
+      "step": 1715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782268455343883e-05,
+      "loss": 0.145,
+      "step": 1716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780607210974992e-05,
+      "loss": 0.1784,
+      "step": 1717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8778944907803236e-05,
+      "loss": 0.1525,
+      "step": 1718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8777281546029054e-05,
+      "loss": 0.1552,
+      "step": 1719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775617125853023e-05,
+      "loss": 0.1534,
+      "step": 1720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773951647475847e-05,
+      "loss": 0.1673,
+      "step": 1721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8772285111098355e-05,
+      "loss": 0.1586,
+      "step": 1722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770617516921507e-05,
+      "loss": 0.1622,
+      "step": 1723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768948865146385e-05,
+      "loss": 0.1615,
+      "step": 1724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8767279155974207e-05,
+      "loss": 0.1425,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876560838960631e-05,
+      "loss": 0.1483,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876393656624416e-05,
+      "loss": 0.1601,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762263686089363e-05,
+      "loss": 0.1422,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876058974934363e-05,
+      "loss": 0.1584,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758914756208815e-05,
+      "loss": 0.1671,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87572387068869e-05,
+      "loss": 0.1583,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755561601579984e-05,
+      "loss": 0.1558,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875388344049031e-05,
+      "loss": 0.1815,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875220422382022e-05,
+      "loss": 0.1417,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750523951772216e-05,
+      "loss": 0.1594,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874884262454891e-05,
+      "loss": 0.1442,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747160242353036e-05,
+      "loss": 0.1444,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874547680538747e-05,
+      "loss": 0.1508,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874379231385521e-05,
+      "loss": 0.1722,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874210676795937e-05,
+      "loss": 0.1617,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874042016790321e-05,
+      "loss": 0.1628,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87387325138901e-05,
+      "loss": 0.151,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737043806123548e-05,
+      "loss": 0.1632,
+      "step": 1743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8735354044807183e-05,
+      "loss": 0.1475,
+      "step": 1744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733663230144766e-05,
+      "loss": 0.1557,
+      "step": 1745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731971362340183e-05,
+      "loss": 0.1726,
+      "step": 1746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730278441597448e-05,
+      "loss": 0.1637,
+      "step": 1747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728584468120695e-05,
+      "loss": 0.1501,
+      "step": 1748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726889442114195e-05,
+      "loss": 0.1546,
+      "step": 1749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872519336378234e-05,
+      "loss": 0.16,
+      "step": 1750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872349623332965e-05,
+      "loss": 0.1525,
+      "step": 1751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721798050960774e-05,
+      "loss": 0.1674,
+      "step": 1752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720098816880484e-05,
+      "loss": 0.1678,
+      "step": 1753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871839853129368e-05,
+      "loss": 0.1565,
+      "step": 1754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871669719440539e-05,
+      "loss": 0.1623,
+      "step": 1755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8714994806420773e-05,
+      "loss": 0.1476,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8713291367545107e-05,
+      "loss": 0.1779,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.87115868779838e-05,
+      "loss": 0.1608,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870988133794238e-05,
+      "loss": 0.1782,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708174747626523e-05,
+      "loss": 0.1682,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706467107242002e-05,
+      "loss": 0.1471,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704758416994743e-05,
+      "loss": 0.1667,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703048677090777e-05,
+      "loss": 0.1618,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870133788773628e-05,
+      "loss": 0.1623,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699626049137544e-05,
+      "loss": 0.164,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697913161500985e-05,
+      "loss": 0.1559,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696199225033157e-05,
+      "loss": 0.1639,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869448423994073e-05,
+      "loss": 0.166,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8692768206430502e-05,
+      "loss": 0.1593,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691051124709403e-05,
+      "loss": 0.1713,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689332994984492e-05,
+      "loss": 0.1681,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687613817462938e-05,
+      "loss": 0.1634,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868589359235205e-05,
+      "loss": 0.1469,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684172319859258e-05,
+      "loss": 0.1475,
+      "step": 1774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8682450000192127e-05,
+      "loss": 0.1668,
+      "step": 1775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680726633558337e-05,
+      "loss": 0.1539,
+      "step": 1776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679002220165697e-05,
+      "loss": 0.1509,
+      "step": 1777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677276760222152e-05,
+      "loss": 0.1591,
+      "step": 1778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675550253935752e-05,
+      "loss": 0.168,
+      "step": 1779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86738227015147e-05,
+      "loss": 0.1595,
+      "step": 1780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672094103167306e-05,
+      "loss": 0.1462,
+      "step": 1781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867036445910201e-05,
+      "loss": 0.16,
+      "step": 1782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866863376952738e-05,
+      "loss": 0.1591,
+      "step": 1783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666902034652112e-05,
+      "loss": 0.15,
+      "step": 1784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665169254685027e-05,
+      "loss": 0.1573,
+      "step": 1785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866343542983507e-05,
+      "loss": 0.1499,
+      "step": 1786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866170056031131e-05,
+      "loss": 0.176,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659964646322946e-05,
+      "loss": 0.1393,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8658227688079304e-05,
+      "loss": 0.1886,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865648968578983e-05,
+      "loss": 0.1471,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865475063966411e-05,
+      "loss": 0.1853,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865301054991183e-05,
+      "loss": 0.1632,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651269416742833e-05,
+      "loss": 0.1477,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649527240367058e-05,
+      "loss": 0.1537,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647784020994593e-05,
+      "loss": 0.1533,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646039758835642e-05,
+      "loss": 0.1649,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644294454100533e-05,
+      "loss": 0.1514,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864254810699972e-05,
+      "loss": 0.1552,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640800717743793e-05,
+      "loss": 0.1458,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8639052286543455e-05,
+      "loss": 0.1635,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637302813609536e-05,
+      "loss": 0.148,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635552299152996e-05,
+      "loss": 0.1626,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633800743384928e-05,
+      "loss": 0.171,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863204814651653e-05,
+      "loss": 0.1677,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630294508759142e-05,
+      "loss": 0.1736,
+      "step": 1805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862853983032423e-05,
+      "loss": 0.1747,
+      "step": 1806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626784111423375e-05,
+      "loss": 0.1696,
+      "step": 1807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862502735226829e-05,
+      "loss": 0.1738,
+      "step": 1808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623269553070814e-05,
+      "loss": 0.1503,
+      "step": 1809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621510714042908e-05,
+      "loss": 0.1638,
+      "step": 1810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8619750835396663e-05,
+      "loss": 0.1484,
+      "step": 1811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617989917344292e-05,
+      "loss": 0.1611,
+      "step": 1812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861622796009813e-05,
+      "loss": 0.151,
+      "step": 1813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8614464963870643e-05,
+      "loss": 0.1571,
+      "step": 1814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8612700928874425e-05,
+      "loss": 0.159,
+      "step": 1815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861093585532219e-05,
+      "loss": 0.1804,
+      "step": 1816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860916974342677e-05,
+      "loss": 0.1661,
+      "step": 1817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607402593401135e-05,
+      "loss": 0.1643,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860563440545838e-05,
+      "loss": 0.1534,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603865179811716e-05,
+      "loss": 0.1611,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8602094916674485e-05,
+      "loss": 0.1541,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860032361626015e-05,
+      "loss": 0.1622,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859855127878231e-05,
+      "loss": 0.1608,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859677790445467e-05,
+      "loss": 0.1621,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859500349349108e-05,
+      "loss": 0.137,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85932280461055e-05,
+      "loss": 0.1641,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591451562512025e-05,
+      "loss": 0.1557,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589674042924867e-05,
+      "loss": 0.1537,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587895487558373e-05,
+      "loss": 0.1484,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8586115896627e-05,
+      "loss": 0.1493,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584335270345348e-05,
+      "loss": 0.1749,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8582553608928125e-05,
+      "loss": 0.1631,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580770912590175e-05,
+      "loss": 0.1612,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8578987181546463e-05,
+      "loss": 0.1482,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577202416012078e-05,
+      "loss": 0.1596,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575416616202234e-05,
+      "loss": 0.1593,
+      "step": 1836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8573629782332272e-05,
+      "loss": 0.1628,
+      "step": 1837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8571841914617655e-05,
+      "loss": 0.1468,
+      "step": 1838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570053013273974e-05,
+      "loss": 0.1627,
+      "step": 1839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8568263078516934e-05,
+      "loss": 0.1545,
+      "step": 1840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566472110562387e-05,
+      "loss": 0.162,
+      "step": 1841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564680109626284e-05,
+      "loss": 0.1558,
+      "step": 1842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8562887075924712e-05,
+      "loss": 0.1724,
+      "step": 1843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856109300967389e-05,
+      "loss": 0.1558,
+      "step": 1844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855929791109015e-05,
+      "loss": 0.1436,
+      "step": 1845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557501780389956e-05,
+      "loss": 0.1611,
+      "step": 1846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555704617789887e-05,
+      "loss": 0.1574,
+      "step": 1847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8553906423506657e-05,
+      "loss": 0.1648,
+      "step": 1848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552107197757095e-05,
+      "loss": 0.1548,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550306940758163e-05,
+      "loss": 0.1481,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8548505652726944e-05,
+      "loss": 0.1574,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546703333880647e-05,
+      "loss": 0.1719,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85448999844366e-05,
+      "loss": 0.1495,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854309560461226e-05,
+      "loss": 0.161,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85412901946252e-05,
+      "loss": 0.1603,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8539483754693137e-05,
+      "loss": 0.1641,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537676285033886e-05,
+      "loss": 0.1547,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853586778586541e-05,
+      "loss": 0.153,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853405825740578e-05,
+      "loss": 0.1595,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532247699873197e-05,
+      "loss": 0.1613,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530436113485986e-05,
+      "loss": 0.158,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528623498462593e-05,
+      "loss": 0.1634,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526809855021598e-05,
+      "loss": 0.1581,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524995183381694e-05,
+      "loss": 0.1832,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85231794837617e-05,
+      "loss": 0.1567,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521362756380562e-05,
+      "loss": 0.1682,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519545001457348e-05,
+      "loss": 0.1644,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517726219211254e-05,
+      "loss": 0.1487,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8515906409861594e-05,
+      "loss": 0.1635,
+      "step": 1869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514085573627803e-05,
+      "loss": 0.1849,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512263710729453e-05,
+      "loss": 0.1455,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8510440821386234e-05,
+      "loss": 0.1526,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850861690581795e-05,
+      "loss": 0.1382,
+      "step": 1873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506791964244536e-05,
+      "loss": 0.1714,
+      "step": 1874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8504965996886063e-05,
+      "loss": 0.1474,
+      "step": 1875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85031390039627e-05,
+      "loss": 0.1742,
+      "step": 1876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850131098569476e-05,
+      "loss": 0.1665,
+      "step": 1877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499481942302672e-05,
+      "loss": 0.1618,
+      "step": 1878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8497651874006993e-05,
+      "loss": 0.1479,
+      "step": 1879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8495820781028394e-05,
+      "loss": 0.1847,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493988663587685e-05,
+      "loss": 0.1514,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849215552190578e-05,
+      "loss": 0.1577,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8490321356203733e-05,
+      "loss": 0.1644,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488486166702717e-05,
+      "loss": 0.1814,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848664995362402e-05,
+      "loss": 0.1734,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848481271718907e-05,
+      "loss": 0.1644,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84829744576194e-05,
+      "loss": 0.1716,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481135175136683e-05,
+      "loss": 0.1462,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84792948699627e-05,
+      "loss": 0.1651,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477453542319365e-05,
+      "loss": 0.1597,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847561119242872e-05,
+      "loss": 0.1631,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847376782051291e-05,
+      "loss": 0.1578,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471923426794228e-05,
+      "loss": 0.1459,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847007801149508e-05,
+      "loss": 0.1523,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8468231574837983e-05,
+      "loss": 0.1593,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466384117045594e-05,
+      "loss": 0.1426,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464535638340687e-05,
+      "loss": 0.1657,
+      "step": 1897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462686138946164e-05,
+      "loss": 0.1527,
+      "step": 1898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8460835619085038e-05,
+      "loss": 0.1668,
+      "step": 1899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845898407898046e-05,
+      "loss": 0.1762,
+      "step": 1900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8457131518855688e-05,
+      "loss": 0.1507,
+      "step": 1901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455277938934118e-05,
+      "loss": 0.1572,
+      "step": 1902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845342333943926e-05,
+      "loss": 0.16,
+      "step": 1903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845156772059475e-05,
+      "loss": 0.1643,
+      "step": 1904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8449711082624348e-05,
+      "loss": 0.1538,
+      "step": 1905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844785342575193e-05,
+      "loss": 0.1519,
+      "step": 1906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8445994750201506e-05,
+      "loss": 0.1602,
+      "step": 1907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8444135056197205e-05,
+      "loss": 0.1722,
+      "step": 1908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8442274343963268e-05,
+      "loss": 0.1555,
+      "step": 1909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844041261372407e-05,
+      "loss": 0.1594,
+      "step": 1910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438549865704112e-05,
+      "loss": 0.1576,
+      "step": 1911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.843668610012801e-05,
+      "loss": 0.1678,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434821317220502e-05,
+      "loss": 0.1498,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843295551720645e-05,
+      "loss": 0.1616,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8431088700310846e-05,
+      "loss": 0.1526,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429220866758794e-05,
+      "loss": 0.1542,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842735201677553e-05,
+      "loss": 0.1536,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84254821505864e-05,
+      "loss": 0.1808,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842361126841689e-05,
+      "loss": 0.1624,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842173937049259e-05,
+      "loss": 0.1705,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841986645703923e-05,
+      "loss": 0.1768,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8417992528282645e-05,
+      "loss": 0.1676,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416117584448808e-05,
+      "loss": 0.16,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8414241625763804e-05,
+      "loss": 0.1592,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412364652453845e-05,
+      "loss": 0.1774,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410486664745265e-05,
+      "loss": 0.153,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840860766286452e-05,
+      "loss": 0.158,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406727647038187e-05,
+      "loss": 0.1583,
+      "step": 1928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404846617492968e-05,
+      "loss": 0.1523,
+      "step": 1929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402964574455683e-05,
+      "loss": 0.1527,
+      "step": 1930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840108151815328e-05,
+      "loss": 0.1646,
+      "step": 1931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839919744881282e-05,
+      "loss": 0.1593,
+      "step": 1932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397312366661498e-05,
+      "loss": 0.1737,
+      "step": 1933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8395426271926624e-05,
+      "loss": 0.1543,
+      "step": 1934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839353916483563e-05,
+      "loss": 0.1563,
+      "step": 1935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839165104561607e-05,
+      "loss": 0.1516,
+      "step": 1936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8389761914495626e-05,
+      "loss": 0.1393,
+      "step": 1937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387871771702094e-05,
+      "loss": 0.1605,
+      "step": 1938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385980617463398e-05,
+      "loss": 0.1511,
+      "step": 1939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838408845200758e-05,
+      "loss": 0.1664,
+      "step": 1940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.83821952755628e-05,
+      "loss": 0.1652,
+      "step": 1941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380301088357354e-05,
+      "loss": 0.1734,
+      "step": 1942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8378405890619647e-05,
+      "loss": 0.1423,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376509682578208e-05,
+      "loss": 0.1438,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8374612464461692e-05,
+      "loss": 0.1406,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8372714236498876e-05,
+      "loss": 0.1588,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8370814998918652e-05,
+      "loss": 0.1503,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8368914751950044e-05,
+      "loss": 0.1559,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8367013495822185e-05,
+      "loss": 0.1494,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365111230764343e-05,
+      "loss": 0.1685,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363207957005894e-05,
+      "loss": 0.1628,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836130367477635e-05,
+      "loss": 0.1618,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835939838430533e-05,
+      "loss": 0.1742,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357492085822588e-05,
+      "loss": 0.1638,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835558477955799e-05,
+      "loss": 0.1547,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835367646574153e-05,
+      "loss": 0.1588,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835176714460332e-05,
+      "loss": 0.1847,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8349856816373595e-05,
+      "loss": 0.1545,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834794548128271e-05,
+      "loss": 0.1445,
+      "step": 1959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346033139561133e-05,
+      "loss": 0.1714,
+      "step": 1960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344119791439476e-05,
+      "loss": 0.1465,
+      "step": 1961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834220543714845e-05,
+      "loss": 0.1574,
+      "step": 1962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83402900769189e-05,
+      "loss": 0.1571,
+      "step": 1963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338373710981787e-05,
+      "loss": 0.1373,
+      "step": 1964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336456339568193e-05,
+      "loss": 0.1523,
+      "step": 1965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8334537962909328e-05,
+      "loss": 0.142,
+      "step": 1966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332618581236506e-05,
+      "loss": 0.1547,
+      "step": 1967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833069819478119e-05,
+      "loss": 0.1761,
+      "step": 1968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328776803774938e-05,
+      "loss": 0.1494,
+      "step": 1969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8326854408449445e-05,
+      "loss": 0.1602,
+      "step": 1970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8324931009036513e-05,
+      "loss": 0.1692,
+      "step": 1971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8323006605768084e-05,
+      "loss": 0.1644,
+      "step": 1972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321081198876205e-05,
+      "loss": 0.1608,
+      "step": 1973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.831915478859305e-05,
+      "loss": 0.1559,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8317227375150915e-05,
+      "loss": 0.1399,
+      "step": 1975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315298958782216e-05,
+      "loss": 0.1874,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831336953971949e-05,
+      "loss": 0.1683,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831143911819539e-05,
+      "loss": 0.1627,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83095076944427e-05,
+      "loss": 0.1599,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8307575268694317e-05,
+      "loss": 0.168,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305641841183258e-05,
+      "loss": 0.1594,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8303707412142673e-05,
+      "loss": 0.1469,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301771981805815e-05,
+      "loss": 0.153,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299835550406068e-05,
+      "loss": 0.1649,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8297898118176937e-05,
+      "loss": 0.1644,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295959685352044e-05,
+      "loss": 0.1652,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294020252165137e-05,
+      "loss": 0.1899,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8292079818850077e-05,
+      "loss": 0.1772,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290138385640854e-05,
+      "loss": 0.1461,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828819595277157e-05,
+      "loss": 0.1558,
+      "step": 1990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828625252047645e-05,
+      "loss": 0.1577,
+      "step": 1991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828430808898985e-05,
+      "loss": 0.1644,
+      "step": 1992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8282362658546228e-05,
+      "loss": 0.1745,
+      "step": 1993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280416229380177e-05,
+      "loss": 0.1685,
+      "step": 1994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827846880172641e-05,
+      "loss": 0.1566,
+      "step": 1995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276520375819745e-05,
+      "loss": 0.138,
+      "step": 1996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8274570951895142e-05,
+      "loss": 0.1506,
+      "step": 1997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8272620530187666e-05,
+      "loss": 0.1542,
+      "step": 1998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827066911093251e-05,
+      "loss": 0.1815,
+      "step": 1999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826871669436498e-05,
+      "loss": 0.1699,
+      "step": 2000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826676328072051e-05,
+      "loss": 0.1621,
+      "step": 2001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8264808870234647e-05,
+      "loss": 0.1688,
+      "step": 2002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262853463143066e-05,
+      "loss": 0.1591,
+      "step": 2003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826089705968156e-05,
+      "loss": 0.1555,
+      "step": 2004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8258939660086036e-05,
+      "loss": 0.1409,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825698126459253e-05,
+      "loss": 0.145,
+      "step": 2006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255021873437185e-05,
+      "loss": 0.1638,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253061486856283e-05,
+      "loss": 0.1404,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825110010508621e-05,
+      "loss": 0.1485,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824913772836348e-05,
+      "loss": 0.1715,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824717435692472e-05,
+      "loss": 0.1666,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245209991006692e-05,
+      "loss": 0.1593,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824324463084626e-05,
+      "loss": 0.1542,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824127827668041e-05,
+      "loss": 0.1406,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8239310928746264e-05,
+      "loss": 0.1412,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823734258728105e-05,
+      "loss": 0.169,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8235373252522115e-05,
+      "loss": 0.1565,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8233402924706935e-05,
+      "loss": 0.1578,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231431604073098e-05,
+      "loss": 0.1485,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229459290858314e-05,
+      "loss": 0.1477,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822748598530041e-05,
+      "loss": 0.1576,
+      "step": 2021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225511687637342e-05,
+      "loss": 0.1537,
+      "step": 2022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223536398107177e-05,
+      "loss": 0.1545,
+      "step": 2023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221560116948103e-05,
+      "loss": 0.1505,
+      "step": 2024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8219582844398426e-05,
+      "loss": 0.1754,
+      "step": 2025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217604580696577e-05,
+      "loss": 0.1457,
+      "step": 2026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215625326081102e-05,
+      "loss": 0.1664,
+      "step": 2027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821364508079067e-05,
+      "loss": 0.1549,
+      "step": 2028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211663845064062e-05,
+      "loss": 0.1628,
+      "step": 2029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820968161914019e-05,
+      "loss": 0.1527,
+      "step": 2030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207698403258078e-05,
+      "loss": 0.1709,
+      "step": 2031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820571419765687e-05,
+      "loss": 0.1502,
+      "step": 2032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203729002575826e-05,
+      "loss": 0.1653,
+      "step": 2033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8201742818254336e-05,
+      "loss": 0.1292,
+      "step": 2034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199755644931894e-05,
+      "loss": 0.1551,
+      "step": 2035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8197767482848133e-05,
+      "loss": 0.1479,
+      "step": 2036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195778332242784e-05,
+      "loss": 0.1636,
+      "step": 2037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193788193355713e-05,
+      "loss": 0.1563,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191797066426895e-05,
+      "loss": 0.1696,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8189804951696435e-05,
+      "loss": 0.1882,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818781184940454e-05,
+      "loss": 0.1674,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8185817759791563e-05,
+      "loss": 0.1568,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8183822683097944e-05,
+      "loss": 0.1564,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181826619564266e-05,
+      "loss": 0.1607,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817982956943122e-05,
+      "loss": 0.1551,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817783153293962e-05,
+      "loss": 0.1776,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81758325103304e-05,
+      "loss": 0.1834,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81738325018446e-05,
+      "loss": 0.1532,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171831507723406e-05,
+      "loss": 0.1537,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8169829528208095e-05,
+      "loss": 0.1526,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167826563540073e-05,
+      "loss": 0.1576,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8165822613960876e-05,
+      "loss": 0.1593,
+      "step": 2052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816381767971214e-05,
+      "loss": 0.1636,
+      "step": 2053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816181176103563e-05,
+      "loss": 0.1466,
+      "step": 2054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159804858173232e-05,
+      "loss": 0.1518,
+      "step": 2055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157796971366945e-05,
+      "loss": 0.1586,
+      "step": 2056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155788100858883e-05,
+      "loss": 0.157,
+      "step": 2057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153778246891292e-05,
+      "loss": 0.174,
+      "step": 2058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8151767409706527e-05,
+      "loss": 0.1833,
+      "step": 2059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149755589547058e-05,
+      "loss": 0.1799,
+      "step": 2060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8147742786655484e-05,
+      "loss": 0.1571,
+      "step": 2061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145729001274518e-05,
+      "loss": 0.1528,
+      "step": 2062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814371423364699e-05,
+      "loss": 0.1431,
+      "step": 2063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141698484015843e-05,
+      "loss": 0.169,
+      "step": 2064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139681752624156e-05,
+      "loss": 0.1418,
+      "step": 2065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137664039715104e-05,
+      "loss": 0.1674,
+      "step": 2066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8135645345531996e-05,
+      "loss": 0.1571,
+      "step": 2067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8133625670318256e-05,
+      "loss": 0.1597,
+      "step": 2068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8131605014317423e-05,
+      "loss": 0.1648,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812958337777316e-05,
+      "loss": 0.1696,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812756076092924e-05,
+      "loss": 0.1447,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125537164029562e-05,
+      "loss": 0.1611,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123512587318134e-05,
+      "loss": 0.149,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121487031039096e-05,
+      "loss": 0.1442,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811946049543669e-05,
+      "loss": 0.1499,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811743298075529e-05,
+      "loss": 0.1536,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8115404487239386e-05,
+      "loss": 0.171,
+      "step": 2077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113375015133575e-05,
+      "loss": 0.1586,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811134456468258e-05,
+      "loss": 0.1623,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8109313136131244e-05,
+      "loss": 0.1544,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107280729724522e-05,
+      "loss": 0.152,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8105247345707498e-05,
+      "loss": 0.1825,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8103212984325356e-05,
+      "loss": 0.1683,
+      "step": 2083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8101177645823414e-05,
+      "loss": 0.1577,
+      "step": 2084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8099141330447102e-05,
+      "loss": 0.1495,
+      "step": 2085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809710403844196e-05,
+      "loss": 0.1537,
+      "step": 2086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8095065770053665e-05,
+      "loss": 0.1428,
+      "step": 2087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809302652552799e-05,
+      "loss": 0.1536,
+      "step": 2088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090986305110844e-05,
+      "loss": 0.1517,
+      "step": 2089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808894510904824e-05,
+      "loss": 0.1674,
+      "step": 2090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086902937586316e-05,
+      "loss": 0.1564,
+      "step": 2091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084859790971324e-05,
+      "loss": 0.1635,
+      "step": 2092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8082815669449637e-05,
+      "loss": 0.1539,
+      "step": 2093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808077057326774e-05,
+      "loss": 0.1751,
+      "step": 2094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8078724502672247e-05,
+      "loss": 0.1512,
+      "step": 2095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8076677457909877e-05,
+      "loss": 0.1678,
+      "step": 2096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807462943922747e-05,
+      "loss": 0.1549,
+      "step": 2097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072580446871987e-05,
+      "loss": 0.1607,
+      "step": 2098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8070530481090504e-05,
+      "loss": 0.1425,
+      "step": 2099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068479542130215e-05,
+      "loss": 0.1507,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806642763023843e-05,
+      "loss": 0.1516,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806437474566258e-05,
+      "loss": 0.1522,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8062320888650203e-05,
+      "loss": 0.1542,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806026605944897e-05,
+      "loss": 0.1365,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805821025830666e-05,
+      "loss": 0.1683,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056153485471167e-05,
+      "loss": 0.1575,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054095741190502e-05,
+      "loss": 0.1507,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052037025712804e-05,
+      "loss": 0.1989,
+      "step": 2108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049977339286323e-05,
+      "loss": 0.163,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8047916682159418e-05,
+      "loss": 0.1617,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8045855054580573e-05,
+      "loss": 0.1483,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043792456798387e-05,
+      "loss": 0.1686,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804172888906158e-05,
+      "loss": 0.1579,
+      "step": 2113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8039664351618986e-05,
+      "loss": 0.1494,
+      "step": 2114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037598844719554e-05,
+      "loss": 0.1455,
+      "step": 2115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803553236861235e-05,
+      "loss": 0.1629,
+      "step": 2116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033464923546563e-05,
+      "loss": 0.1485,
+      "step": 2117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803139650977149e-05,
+      "loss": 0.1604,
+      "step": 2118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8029327127536552e-05,
+      "loss": 0.1601,
+      "step": 2119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802725677709128e-05,
+      "loss": 0.1825,
+      "step": 2120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025185458685333e-05,
+      "loss": 0.1427,
+      "step": 2121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802311317256847e-05,
+      "loss": 0.1709,
+      "step": 2122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8021039918990584e-05,
+      "loss": 0.1601,
+      "step": 2123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8018965698201675e-05,
+      "loss": 0.1579,
+      "step": 2124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016890510451856e-05,
+      "loss": 0.1597,
+      "step": 2125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014814355991364e-05,
+      "loss": 0.159,
+      "step": 2126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8012737235070554e-05,
+      "loss": 0.1498,
+      "step": 2127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8010659147939896e-05,
+      "loss": 0.1611,
+      "step": 2128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8008580094849964e-05,
+      "loss": 0.1838,
+      "step": 2129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800650007605147e-05,
+      "loss": 0.1557,
+      "step": 2130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8004419091795226e-05,
+      "loss": 0.1644,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800233714233217e-05,
+      "loss": 0.1584,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8000254227913346e-05,
+      "loss": 0.1512,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998170348789927e-05,
+      "loss": 0.1538,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799608550521319e-05,
+      "loss": 0.1568,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799399969743454e-05,
+      "loss": 0.1765,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799191292570549e-05,
+      "loss": 0.1703,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798982519027767e-05,
+      "loss": 0.1752,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987736491402833e-05,
+      "loss": 0.1497,
+      "step": 2139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7985646829332837e-05,
+      "loss": 0.1589,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983556204319666e-05,
+      "loss": 0.161,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981464616615417e-05,
+      "loss": 0.1652,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79793720664723e-05,
+      "loss": 0.1586,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7977278554142643e-05,
+      "loss": 0.1485,
+      "step": 2144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7975184079878893e-05,
+      "loss": 0.1639,
+      "step": 2145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973088643933613e-05,
+      "loss": 0.1437,
+      "step": 2146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970992246559476e-05,
+      "loss": 0.1443,
+      "step": 2147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796889488800927e-05,
+      "loss": 0.1567,
+      "step": 2148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966796568535915e-05,
+      "loss": 0.1602,
+      "step": 2149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7964697288392424e-05,
+      "loss": 0.1465,
+      "step": 2150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7962597047831945e-05,
+      "loss": 0.1438,
+      "step": 2151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796049584710773e-05,
+      "loss": 0.1424,
+      "step": 2152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795839368647315e-05,
+      "loss": 0.1515,
+      "step": 2153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956290566181696e-05,
+      "loss": 0.1676,
+      "step": 2154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795418648648697e-05,
+      "loss": 0.1422,
+      "step": 2155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952081447642688e-05,
+      "loss": 0.1516,
+      "step": 2156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7949975449902686e-05,
+      "loss": 0.1594,
+      "step": 2157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7947868493520918e-05,
+      "loss": 0.1669,
+      "step": 2158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794576057875144e-05,
+      "loss": 0.1574,
+      "step": 2159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7943651705848444e-05,
+      "loss": 0.1668,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794154187506622e-05,
+      "loss": 0.144,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939431086659183e-05,
+      "loss": 0.1412,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793731934088186e-05,
+      "loss": 0.1491,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935206637988903e-05,
+      "loss": 0.156,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933092978235055e-05,
+      "loss": 0.1489,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7930978361875197e-05,
+      "loss": 0.1389,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792886278916432e-05,
+      "loss": 0.1453,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792674626035753e-05,
+      "loss": 0.1759,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924628775710043e-05,
+      "loss": 0.1685,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922510335477198e-05,
+      "loss": 0.1588,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792039093991444e-05,
+      "loss": 0.1335,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7918270589277344e-05,
+      "loss": 0.1589,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791614928382158e-05,
+      "loss": 0.1507,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914027023802953e-05,
+      "loss": 0.1505,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791190380947737e-05,
+      "loss": 0.1508,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7909779641100852e-05,
+      "loss": 0.1478,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790765451892955e-05,
+      "loss": 0.1716,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905528443219717e-05,
+      "loss": 0.161,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790340141422772e-05,
+      "loss": 0.1365,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790127343221005e-05,
+      "loss": 0.1613,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899144497423307e-05,
+      "loss": 0.1545,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7897014610124205e-05,
+      "loss": 0.1612,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894883770569576e-05,
+      "loss": 0.1555,
+      "step": 2183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789275197901637e-05,
+      "loss": 0.1547,
+      "step": 2184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789061923572164e-05,
+      "loss": 0.161,
+      "step": 2185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7888485540942568e-05,
+      "loss": 0.1429,
+      "step": 2186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7886350894936433e-05,
+      "loss": 0.1556,
+      "step": 2187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7884215297960653e-05,
+      "loss": 0.1446,
+      "step": 2188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788207875027274e-05,
+      "loss": 0.1585,
+      "step": 2189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879941252130334e-05,
+      "loss": 0.1444,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7877802803791176e-05,
+      "loss": 0.1796,
+      "step": 2191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7875663405513135e-05,
+      "loss": 0.1519,
+      "step": 2192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787352305755418e-05,
+      "loss": 0.1588,
+      "step": 2193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7871381760172417e-05,
+      "loss": 0.161,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786923951362604e-05,
+      "loss": 0.1757,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7867096318173377e-05,
+      "loss": 0.158,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786495217407286e-05,
+      "loss": 0.1597,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862807081583042e-05,
+      "loss": 0.1473,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860661040962588e-05,
+      "loss": 0.1475,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7858514052470273e-05,
+      "loss": 0.1386,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7856366116364994e-05,
+      "loss": 0.1508,
+      "step": 2201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7854217232905755e-05,
+      "loss": 0.1442,
+      "step": 2202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785206740235168e-05,
+      "loss": 0.1595,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7849916624962e-05,
+      "loss": 0.1383,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847764900996068e-05,
+      "loss": 0.1523,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784561223071335e-05,
+      "loss": 0.1613,
+      "step": 2206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7843458614373425e-05,
+      "loss": 0.1506,
+      "step": 2207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784130405223598e-05,
+      "loss": 0.1624,
+      "step": 2208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7839148544560824e-05,
+      "loss": 0.1622,
+      "step": 2209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7836992091607877e-05,
+      "loss": 0.164,
+      "step": 2210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783483469363717e-05,
+      "loss": 0.1491,
+      "step": 2211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783267635090886e-05,
+      "loss": 0.1748,
+      "step": 2212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78305170636832e-05,
+      "loss": 0.1624,
+      "step": 2213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7828356832220574e-05,
+      "loss": 0.1647,
+      "step": 2214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782619565678146e-05,
+      "loss": 0.1674,
+      "step": 2215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824033537626477e-05,
+      "loss": 0.1633,
+      "step": 2216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782187047501633e-05,
+      "loss": 0.1528,
+      "step": 2217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819706469211856e-05,
+      "loss": 0.1598,
+      "step": 2218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7817541520474e-05,
+      "loss": 0.1623,
+      "step": 2219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7815375629063818e-05,
+      "loss": 0.1539,
+      "step": 2220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7813208795242486e-05,
+      "loss": 0.1609,
+      "step": 2221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7811041019271285e-05,
+      "loss": 0.1678,
+      "step": 2222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780887230141162e-05,
+      "loss": 0.1537,
+      "step": 2223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7806702641925002e-05,
+      "loss": 0.1534,
+      "step": 2224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7804532041073054e-05,
+      "loss": 0.1594,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7802360499117527e-05,
+      "loss": 0.1718,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780018801632026e-05,
+      "loss": 0.1458,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798014592943232e-05,
+      "loss": 0.1724,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7795840229248518e-05,
+      "loss": 0.1453,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7793664925498316e-05,
+      "loss": 0.1478,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7791488681954928e-05,
+      "loss": 0.1606,
+      "step": 2231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778931149888078e-05,
+      "loss": 0.166,
+      "step": 2232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77871333765384e-05,
+      "loss": 0.1595,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778495431519044e-05,
+      "loss": 0.1444,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7782774315099662e-05,
+      "loss": 0.1631,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778059337652893e-05,
+      "loss": 0.1412,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7778411499741246e-05,
+      "loss": 0.155,
+      "step": 2237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7776228684999696e-05,
+      "loss": 0.1728,
+      "step": 2238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77740449325675e-05,
+      "loss": 0.1541,
+      "step": 2239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771860242707983e-05,
+      "loss": 0.1534,
+      "step": 2240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769674615684586e-05,
+      "loss": 0.1378,
+      "step": 2241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7767488051760858e-05,
+      "loss": 0.1443,
+      "step": 2242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765300551200464e-05,
+      "loss": 0.1525,
+      "step": 2243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7763112114267183e-05,
+      "loss": 0.1539,
+      "step": 2244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776092274122491e-05,
+      "loss": 0.1505,
+      "step": 2245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775873243233764e-05,
+      "loss": 0.1675,
+      "step": 2246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7756541187869496e-05,
+      "loss": 0.1563,
+      "step": 2247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775434900808471e-05,
+      "loss": 0.151,
+      "step": 2248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752155893247617e-05,
+      "loss": 0.1543,
+      "step": 2249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7749961843622682e-05,
+      "loss": 0.1687,
+      "step": 2250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774776685947446e-05,
+      "loss": 0.1507,
+      "step": 2251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774557094106764e-05,
+      "loss": 0.1841,
+      "step": 2252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7743374088667015e-05,
+      "loss": 0.1517,
+      "step": 2253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7741176302537492e-05,
+      "loss": 0.1389,
+      "step": 2254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773897758294408e-05,
+      "loss": 0.1534,
+      "step": 2255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7736777930151924e-05,
+      "loss": 0.1619,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7734577344426254e-05,
+      "loss": 0.1593,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732375826032434e-05,
+      "loss": 0.16,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7730173375235932e-05,
+      "loss": 0.1551,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727969992302325e-05,
+      "loss": 0.1436,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725765677497307e-05,
+      "loss": 0.1441,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772356043108669e-05,
+      "loss": 0.1624,
+      "step": 2262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7721354253336388e-05,
+      "loss": 0.156,
+      "step": 2263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719147144512426e-05,
+      "loss": 0.1611,
+      "step": 2264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771693910488095e-05,
+      "loss": 0.1402,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7714730134708217e-05,
+      "loss": 0.1435,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7712520234260594e-05,
+      "loss": 0.1606,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771030940380456e-05,
+      "loss": 0.1644,
+      "step": 2268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77080976436067e-05,
+      "loss": 0.1539,
+      "step": 2269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705884953933732e-05,
+      "loss": 0.1548,
+      "step": 2270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7703671335052456e-05,
+      "loss": 0.1643,
+      "step": 2271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701456787229805e-05,
+      "loss": 0.1293,
+      "step": 2272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769924131073282e-05,
+      "loss": 0.1427,
+      "step": 2273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697024905828654e-05,
+      "loss": 0.1496,
+      "step": 2274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7694807572784568e-05,
+      "loss": 0.1689,
+      "step": 2275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7692589311867937e-05,
+      "loss": 0.1681,
+      "step": 2276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769037012334625e-05,
+      "loss": 0.1669,
+      "step": 2277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7688150007487105e-05,
+      "loss": 0.1393,
+      "step": 2278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685928964558212e-05,
+      "loss": 0.1489,
+      "step": 2279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683706994827394e-05,
+      "loss": 0.1592,
+      "step": 2280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768148409856259e-05,
+      "loss": 0.1633,
+      "step": 2281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767926027603184e-05,
+      "loss": 0.1431,
+      "step": 2282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677035527503306e-05,
+      "loss": 0.1509,
+      "step": 2283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674809853245256e-05,
+      "loss": 0.1364,
+      "step": 2284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7672583253526073e-05,
+      "loss": 0.1368,
+      "step": 2285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670355728614244e-05,
+      "loss": 0.1709,
+      "step": 2286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668127278778386e-05,
+      "loss": 0.1634,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7665897904287202e-05,
+      "loss": 0.1613,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663667605409523e-05,
+      "loss": 0.155,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7661436382414293e-05,
+      "loss": 0.1421,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659204235570552e-05,
+      "loss": 0.1606,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656971165147475e-05,
+      "loss": 0.1495,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7654737171414325e-05,
+      "loss": 0.1535,
+      "step": 2293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652502254640492e-05,
+      "loss": 0.1625,
+      "step": 2294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650266415095468e-05,
+      "loss": 0.1623,
+      "step": 2295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7648029653048864e-05,
+      "loss": 0.1593,
+      "step": 2296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764579196877039e-05,
+      "loss": 0.1418,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643553362529888e-05,
+      "loss": 0.1482,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7641313834597286e-05,
+      "loss": 0.1631,
+      "step": 2299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639073385242648e-05,
+      "loss": 0.155,
+      "step": 2300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7636832014736125e-05,
+      "loss": 0.1406,
+      "step": 2301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7634589723348e-05,
+      "loss": 0.1723,
+      "step": 2302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632346511348654e-05,
+      "loss": 0.1466,
+      "step": 2303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630102379008583e-05,
+      "loss": 0.1333,
+      "step": 2304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762785732659839e-05,
+      "loss": 0.1634,
+      "step": 2305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625611354388807e-05,
+      "loss": 0.154,
+      "step": 2306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762336446265065e-05,
+      "loss": 0.1716,
+      "step": 2307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621116651654857e-05,
+      "loss": 0.1702,
+      "step": 2308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761886792167249e-05,
+      "loss": 0.1607,
+      "step": 2309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76166182729747e-05,
+      "loss": 0.1678,
+      "step": 2310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614367705832763e-05,
+      "loss": 0.1685,
+      "step": 2311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612116220518066e-05,
+      "loss": 0.1541,
+      "step": 2312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7609863817302095e-05,
+      "loss": 0.1707,
+      "step": 2313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760761049645646e-05,
+      "loss": 0.1524,
+      "step": 2314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7605356258252874e-05,
+      "loss": 0.1449,
+      "step": 2315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7603101102963164e-05,
+      "loss": 0.1454,
+      "step": 2316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600845030859263e-05,
+      "loss": 0.1506,
+      "step": 2317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7598588042213224e-05,
+      "loss": 0.1495,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75963301372972e-05,
+      "loss": 0.1524,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759407131638345e-05,
+      "loss": 0.1385,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759181157974437e-05,
+      "loss": 0.1809,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7589550927652434e-05,
+      "loss": 0.1486,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758728936038025e-05,
+      "loss": 0.1591,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758502687820052e-05,
+      "loss": 0.1546,
+      "step": 2324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7582763481386072e-05,
+      "loss": 0.1509,
+      "step": 2325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7580499170209834e-05,
+      "loss": 0.1676,
+      "step": 2326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757823394494484e-05,
+      "loss": 0.1614,
+      "step": 2327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575967805864245e-05,
+      "loss": 0.1465,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757370075324131e-05,
+      "loss": 0.1397,
+      "step": 2329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571432787349403e-05,
+      "loss": 0.1472,
+      "step": 2330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756916390846201e-05,
+      "loss": 0.1527,
+      "step": 2331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566894116852715e-05,
+      "loss": 0.1628,
+      "step": 2332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564623412795226e-05,
+      "loss": 0.1461,
+      "step": 2333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756235179656335e-05,
+      "loss": 0.1484,
+      "step": 2334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560079268431006e-05,
+      "loss": 0.1604,
+      "step": 2335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7557805828672234e-05,
+      "loss": 0.1393,
+      "step": 2336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555531477561167e-05,
+      "loss": 0.1728,
+      "step": 2337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755325621537206e-05,
+      "loss": 0.1605,
+      "step": 2338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7550980042379268e-05,
+      "loss": 0.1509,
+      "step": 2339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754870295885727e-05,
+      "loss": 0.1437,
+      "step": 2340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546424965080637e-05,
+      "loss": 0.1452,
+      "step": 2341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544146061324067e-05,
+      "loss": 0.1435,
+      "step": 2342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754186624786236e-05,
+      "loss": 0.1597,
+      "step": 2343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753958552497042e-05,
+      "loss": 0.145,
+      "step": 2344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753730389292327e-05,
+      "loss": 0.1675,
+      "step": 2345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7535021351996032e-05,
+      "loss": 0.1645,
+      "step": 2346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7532737902463956e-05,
+      "loss": 0.1598,
+      "step": 2347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530453544602384e-05,
+      "loss": 0.1255,
+      "step": 2348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7528168278686774e-05,
+      "loss": 0.1535,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752588210499269e-05,
+      "loss": 0.1479,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523595023795814e-05,
+      "loss": 0.1782,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521307035371927e-05,
+      "loss": 0.1457,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751901813999693e-05,
+      "loss": 0.1597,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516728337946825e-05,
+      "loss": 0.1611,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751443762949772e-05,
+      "loss": 0.1451,
+      "step": 2355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751214601492585e-05,
+      "loss": 0.1365,
+      "step": 2356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750985349450754e-05,
+      "loss": 0.163,
+      "step": 2357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7507560068519228e-05,
+      "loss": 0.1762,
+      "step": 2358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7505265737237475e-05,
+      "loss": 0.1536,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7502970500938934e-05,
+      "loss": 0.1482,
+      "step": 2360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750067435990038e-05,
+      "loss": 0.1582,
+      "step": 2361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7498377314398686e-05,
+      "loss": 0.1398,
+      "step": 2362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749607936471084e-05,
+      "loss": 0.1665,
+      "step": 2363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7493780511113945e-05,
+      "loss": 0.1796,
+      "step": 2364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74914807538852e-05,
+      "loss": 0.1517,
+      "step": 2365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748918009330192e-05,
+      "loss": 0.1592,
+      "step": 2366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748687852964153e-05,
+      "loss": 0.1414,
+      "step": 2367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7484576063181563e-05,
+      "loss": 0.1539,
+      "step": 2368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748227269419966e-05,
+      "loss": 0.1515,
+      "step": 2369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747996842297357e-05,
+      "loss": 0.143,
+      "step": 2370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7477663249781152e-05,
+      "loss": 0.1575,
+      "step": 2371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7475357174900372e-05,
+      "loss": 0.1603,
+      "step": 2372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7473050198609308e-05,
+      "loss": 0.1614,
+      "step": 2373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7470742321186146e-05,
+      "loss": 0.1507,
+      "step": 2374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468433542909177e-05,
+      "loss": 0.1506,
+      "step": 2375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7466123864056807e-05,
+      "loss": 0.1581,
+      "step": 2376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7463813284907544e-05,
+      "loss": 0.1665,
+      "step": 2377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461501805740005e-05,
+      "loss": 0.1593,
+      "step": 2378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7459189426832923e-05,
+      "loss": 0.1453,
+      "step": 2379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7456876148465133e-05,
+      "loss": 0.169,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745456197091558e-05,
+      "loss": 0.164,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7452246894463317e-05,
+      "loss": 0.1451,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74499309193875e-05,
+      "loss": 0.173,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744761404596741e-05,
+      "loss": 0.1462,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744529627448242e-05,
+      "loss": 0.1445,
+      "step": 2385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7442977605212016e-05,
+      "loss": 0.1466,
+      "step": 2386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7440658038435793e-05,
+      "loss": 0.1624,
+      "step": 2387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743833757443346e-05,
+      "loss": 0.1616,
+      "step": 2388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7436016213484816e-05,
+      "loss": 0.1581,
+      "step": 2389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7433693955869793e-05,
+      "loss": 0.1357,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431370801868407e-05,
+      "loss": 0.1627,
+      "step": 2391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429046751760807e-05,
+      "loss": 0.1642,
+      "step": 2392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426721805827228e-05,
+      "loss": 0.1445,
+      "step": 2393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7424395964348025e-05,
+      "loss": 0.1425,
+      "step": 2394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7422069227603654e-05,
+      "loss": 0.1507,
+      "step": 2395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7419741595874686e-05,
+      "loss": 0.162,
+      "step": 2396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74174130694418e-05,
+      "loss": 0.1434,
+      "step": 2397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7415083648585774e-05,
+      "loss": 0.1489,
+      "step": 2398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74127533335875e-05,
+      "loss": 0.1456,
+      "step": 2399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741042212472798e-05,
+      "loss": 0.1647,
+      "step": 2400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740809002228832e-05,
+      "loss": 0.1354,
+      "step": 2401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7405757026549734e-05,
+      "loss": 0.1606,
+      "step": 2402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7403423137793544e-05,
+      "loss": 0.1403,
+      "step": 2403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401088356301183e-05,
+      "loss": 0.1352,
+      "step": 2404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398752682354186e-05,
+      "loss": 0.1609,
+      "step": 2405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7396416116234197e-05,
+      "loss": 0.1442,
+      "step": 2406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394078658222978e-05,
+      "loss": 0.1693,
+      "step": 2407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391740308602378e-05,
+      "loss": 0.1677,
+      "step": 2408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7389401067654374e-05,
+      "loss": 0.1452,
+      "step": 2409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387060935661036e-05,
+      "loss": 0.1582,
+      "step": 2410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738471991290455e-05,
+      "loss": 0.1415,
+      "step": 2411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7382377999667202e-05,
+      "loss": 0.158,
+      "step": 2412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380035196231397e-05,
+      "loss": 0.1639,
+      "step": 2413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377691502879636e-05,
+      "loss": 0.1548,
+      "step": 2414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737534691989453e-05,
+      "loss": 0.155,
+      "step": 2415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7373001447558803e-05,
+      "loss": 0.1485,
+      "step": 2416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737065508615528e-05,
+      "loss": 0.1612,
+      "step": 2417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368307835966892e-05,
+      "loss": 0.135,
+      "step": 2418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365959697276683e-05,
+      "loss": 0.1754,
+      "step": 2419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7363610670367805e-05,
+      "loss": 0.1822,
+      "step": 2420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736126075552351e-05,
+      "loss": 0.135,
+      "step": 2421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735890995302716e-05,
+      "loss": 0.1675,
+      "step": 2422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356558263162228e-05,
+      "loss": 0.1459,
+      "step": 2423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7354205686212285e-05,
+      "loss": 0.1545,
+      "step": 2424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735185222246102e-05,
+      "loss": 0.168,
+      "step": 2425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349497872192224e-05,
+      "loss": 0.1474,
+      "step": 2426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347142635689787e-05,
+      "loss": 0.1423,
+      "step": 2427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344786513237726e-05,
+      "loss": 0.1549,
+      "step": 2428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734242950512014e-05,
+      "loss": 0.1638,
+      "step": 2429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340071611621255e-05,
+      "loss": 0.1484,
+      "step": 2430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7337712833025392e-05,
+      "loss": 0.149,
+      "step": 2431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7335353169616987e-05,
+      "loss": 0.1488,
+      "step": 2432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733299262168057e-05,
+      "loss": 0.1593,
+      "step": 2433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330631189500794e-05,
+      "loss": 0.1627,
+      "step": 2434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7328268873362405e-05,
+      "loss": 0.1439,
+      "step": 2435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325905673550262e-05,
+      "loss": 0.1465,
+      "step": 2436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7323541590349333e-05,
+      "loss": 0.1513,
+      "step": 2437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732117662404469e-05,
+      "loss": 0.151,
+      "step": 2438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73188107749215e-05,
+      "loss": 0.1377,
+      "step": 2439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7316444043265063e-05,
+      "loss": 0.181,
+      "step": 2440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731407642936076e-05,
+      "loss": 0.1414,
+      "step": 2441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311707933494088e-05,
+      "loss": 0.1364,
+      "step": 2442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730933855595065e-05,
+      "loss": 0.156,
+      "step": 2443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306968297016156e-05,
+      "loss": 0.1495,
+      "step": 2444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304597156976427e-05,
+      "loss": 0.1385,
+      "step": 2445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730222513611738e-05,
+      "loss": 0.1598,
+      "step": 2446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7299852234725046e-05,
+      "loss": 0.1594,
+      "step": 2447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729747845308556e-05,
+      "loss": 0.1487,
+      "step": 2448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295103791485158e-05,
+      "loss": 0.1536,
+      "step": 2449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7292728250210186e-05,
+      "loss": 0.1535,
+      "step": 2450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7290351829547106e-05,
+      "loss": 0.1527,
+      "step": 2451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7287974529782466e-05,
+      "loss": 0.1706,
+      "step": 2452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728559635120294e-05,
+      "loss": 0.1421,
+      "step": 2453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7283217294095295e-05,
+      "loss": 0.1283,
+      "step": 2454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728083735874641e-05,
+      "loss": 0.1613,
+      "step": 2455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278456545443262e-05,
+      "loss": 0.1699,
+      "step": 2456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7276074854472944e-05,
+      "loss": 0.1698,
+      "step": 2457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727369228612265e-05,
+      "loss": 0.1781,
+      "step": 2458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727130884067968e-05,
+      "loss": 0.1584,
+      "step": 2459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268924518431437e-05,
+      "loss": 0.1459,
+      "step": 2460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266539319665438e-05,
+      "loss": 0.1483,
+      "step": 2461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264153244669297e-05,
+      "loss": 0.1561,
+      "step": 2462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726176629373074e-05,
+      "loss": 0.1444,
+      "step": 2463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259378467137592e-05,
+      "loss": 0.1585,
+      "step": 2464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256989765177788e-05,
+      "loss": 0.1619,
+      "step": 2465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725460018813937e-05,
+      "loss": 0.1463,
+      "step": 2466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7252209736310483e-05,
+      "loss": 0.1338,
+      "step": 2467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7249818409979374e-05,
+      "loss": 0.1315,
+      "step": 2468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72474262094344e-05,
+      "loss": 0.1493,
+      "step": 2469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724503313496403e-05,
+      "loss": 0.1602,
+      "step": 2470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7242639186856823e-05,
+      "loss": 0.1407,
+      "step": 2471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7240244365401453e-05,
+      "loss": 0.1489,
+      "step": 2472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72378486708867e-05,
+      "loss": 0.1482,
+      "step": 2473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7235452103601447e-05,
+      "loss": 0.1611,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233054663834683e-05,
+      "loss": 0.159,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230656351875495e-05,
+      "loss": 0.1431,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722825716801309e-05,
+      "loss": 0.159,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7225857112536766e-05,
+      "loss": 0.1369,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223456185735933e-05,
+      "loss": 0.1713,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7221054387900108e-05,
+      "loss": 0.1653,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721865171931891e-05,
+      "loss": 0.1452,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7216248180282056e-05,
+      "loss": 0.1627,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213843771079383e-05,
+      "loss": 0.1519,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7211438492000822e-05,
+      "loss": 0.1721,
+      "step": 2484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209032343336408e-05,
+      "loss": 0.1548,
+      "step": 2485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720662532537629e-05,
+      "loss": 0.189,
+      "step": 2486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7204217438410715e-05,
+      "loss": 0.1645,
+      "step": 2487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201808682730033e-05,
+      "loss": 0.1575,
+      "step": 2488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199399058624706e-05,
+      "loss": 0.1722,
+      "step": 2489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7196988566385296e-05,
+      "loss": 0.1476,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719457720630247e-05,
+      "loss": 0.1488,
+      "step": 2491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7192164978667e-05,
+      "loss": 0.148,
+      "step": 2492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7189751883769764e-05,
+      "loss": 0.149,
+      "step": 2493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187337921901737e-05,
+      "loss": 0.134,
+      "step": 2494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7184923093354013e-05,
+      "loss": 0.143,
+      "step": 2495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182507398417775e-05,
+      "loss": 0.149,
+      "step": 2496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7180090837384328e-05,
+      "loss": 0.1634,
+      "step": 2497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7177673410545058e-05,
+      "loss": 0.1361,
+      "step": 2498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717525511819148e-05,
+      "loss": 0.1379,
+      "step": 2499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7172835960615192e-05,
+      "loss": 0.1498,
+      "step": 2500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7170415938107914e-05,
+      "loss": 0.1436,
+      "step": 2501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7167995050961458e-05,
+      "loss": 0.1561,
+      "step": 2502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716557329946775e-05,
+      "loss": 0.1383,
+      "step": 2503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716315068391881e-05,
+      "loss": 0.1401,
+      "step": 2504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7160727204606772e-05,
+      "loss": 0.1661,
+      "step": 2505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7158302861823864e-05,
+      "loss": 0.1461,
+      "step": 2506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715587765586243e-05,
+      "loss": 0.1586,
+      "step": 2507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71534515870149e-05,
+      "loss": 0.1474,
+      "step": 2508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151024655573833e-05,
+      "loss": 0.1689,
+      "step": 2509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714859686183187e-05,
+      "loss": 0.1511,
+      "step": 2510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714616820608177e-05,
+      "loss": 0.1598,
+      "step": 2511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714373868861639e-05,
+      "loss": 0.1468,
+      "step": 2512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7141308309728686e-05,
+      "loss": 0.1538,
+      "step": 2513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713887706971173e-05,
+      "loss": 0.1548,
+      "step": 2514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7136444968858685e-05,
+      "loss": 0.1533,
+      "step": 2515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713401200746283e-05,
+      "loss": 0.1513,
+      "step": 2516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713157818581754e-05,
+      "loss": 0.141,
+      "step": 2517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129143504216297e-05,
+      "loss": 0.1503,
+      "step": 2518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7126707962952678e-05,
+      "loss": 0.1675,
+      "step": 2519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712427156232038e-05,
+      "loss": 0.1574,
+      "step": 2520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712183430261319e-05,
+      "loss": 0.1573,
+      "step": 2521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7119396184125e-05,
+      "loss": 0.1364,
+      "step": 2522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7116957207149816e-05,
+      "loss": 0.1456,
+      "step": 2523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114517371981736e-05,
+      "loss": 0.1655,
+      "step": 2524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7112076678914965e-05,
+      "loss": 0.1552,
+      "step": 2525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109635128243816e-05,
+      "loss": 0.1486,
+      "step": 2526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71071927202627e-05,
+      "loss": 0.1506,
+      "step": 2527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7104749455266128e-05,
+      "loss": 0.144,
+      "step": 2528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710230533354872e-05,
+      "loss": 0.1721,
+      "step": 2529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709986035540521e-05,
+      "loss": 0.1606,
+      "step": 2530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097414521130412e-05,
+      "loss": 0.1441,
+      "step": 2531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7094967831019255e-05,
+      "loss": 0.1488,
+      "step": 2532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709252028536678e-05,
+      "loss": 0.1503,
+      "step": 2533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7090071884468112e-05,
+      "loss": 0.1652,
+      "step": 2534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7087622628618498e-05,
+      "loss": 0.1482,
+      "step": 2535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085172518113276e-05,
+      "loss": 0.1539,
+      "step": 2536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7082721553247893e-05,
+      "loss": 0.1502,
+      "step": 2537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080269734317895e-05,
+      "loss": 0.139,
+      "step": 2538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7077817061618927e-05,
+      "loss": 0.1606,
+      "step": 2539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075363535446756e-05,
+      "loss": 0.1441,
+      "step": 2540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7072909156097227e-05,
+      "loss": 0.156,
+      "step": 2541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7070453923866303e-05,
+      "loss": 0.1473,
+      "step": 2542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067997839050046e-05,
+      "loss": 0.1561,
+      "step": 2543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065540901944627e-05,
+      "loss": 0.1455,
+      "step": 2544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063083112846306e-05,
+      "loss": 0.1471,
+      "step": 2545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060624472051458e-05,
+      "loss": 0.1568,
+      "step": 2546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705816497985655e-05,
+      "loss": 0.1534,
+      "step": 2547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055704636558166e-05,
+      "loss": 0.1385,
+      "step": 2548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705324344245298e-05,
+      "loss": 0.1406,
+      "step": 2549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050781397837776e-05,
+      "loss": 0.1327,
+      "step": 2550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048318503009435e-05,
+      "loss": 0.1689,
+      "step": 2551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045854758264944e-05,
+      "loss": 0.1756,
+      "step": 2552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043390163901397e-05,
+      "loss": 0.1664,
+      "step": 2553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7040924720215978e-05,
+      "loss": 0.1448,
+      "step": 2554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038458427505983e-05,
+      "loss": 0.158,
+      "step": 2555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703599128606881e-05,
+      "loss": 0.1568,
+      "step": 2556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703352329620195e-05,
+      "loss": 0.1614,
+      "step": 2557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031054458203016e-05,
+      "loss": 0.1453,
+      "step": 2558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70285847723697e-05,
+      "loss": 0.1475,
+      "step": 2559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026114238999817e-05,
+      "loss": 0.1446,
+      "step": 2560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702364285839127e-05,
+      "loss": 0.1522,
+      "step": 2561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021170630842062e-05,
+      "loss": 0.1459,
+      "step": 2562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701869755665031e-05,
+      "loss": 0.1542,
+      "step": 2563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016223636114233e-05,
+      "loss": 0.1461,
+      "step": 2564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701374886953214e-05,
+      "loss": 0.1337,
+      "step": 2565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011273257202446e-05,
+      "loss": 0.1527,
+      "step": 2566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7008796799423683e-05,
+      "loss": 0.17,
+      "step": 2567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006319496494463e-05,
+      "loss": 0.1558,
+      "step": 2568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700384134871351e-05,
+      "loss": 0.1589,
+      "step": 2569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001362356379655e-05,
+      "loss": 0.1728,
+      "step": 2570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699888251979182e-05,
+      "loss": 0.1453,
+      "step": 2571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6996401839249034e-05,
+      "loss": 0.1519,
+      "step": 2572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6993920315050433e-05,
+      "loss": 0.1468,
+      "step": 2573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699143794749525e-05,
+      "loss": 0.1523,
+      "step": 2574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698895473688281e-05,
+      "loss": 0.1445,
+      "step": 2575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986470683512558e-05,
+      "loss": 0.1561,
+      "step": 2576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983985787684033e-05,
+      "loss": 0.162,
+      "step": 2577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981500049696866e-05,
+      "loss": 0.1616,
+      "step": 2578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979013469850802e-05,
+      "loss": 0.1591,
+      "step": 2579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976526048445687e-05,
+      "loss": 0.1521,
+      "step": 2580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974037785781462e-05,
+      "loss": 0.1541,
+      "step": 2581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6971548682158167e-05,
+      "loss": 0.164,
+      "step": 2582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6969058737875955e-05,
+      "loss": 0.1373,
+      "step": 2583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966567953235074e-05,
+      "loss": 0.1403,
+      "step": 2584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696407632853587e-05,
+      "loss": 0.1446,
+      "step": 2585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961583864078798e-05,
+      "loss": 0.1609,
+      "step": 2586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6959090560164407e-05,
+      "loss": 0.1404,
+      "step": 2587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695659641709335e-05,
+      "loss": 0.1386,
+      "step": 2588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6954101435166382e-05,
+      "loss": 0.1766,
+      "step": 2589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695160561468436e-05,
+      "loss": 0.1606,
+      "step": 2590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6949108955948237e-05,
+      "loss": 0.1419,
+      "step": 2591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946611459259073e-05,
+      "loss": 0.1557,
+      "step": 2592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944113124918025e-05,
+      "loss": 0.1424,
+      "step": 2593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694161395322636e-05,
+      "loss": 0.1557,
+      "step": 2594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6939113944485427e-05,
+      "loss": 0.1494,
+      "step": 2595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6936613098996697e-05,
+      "loss": 0.1435,
+      "step": 2596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934111417061728e-05,
+      "loss": 0.1714,
+      "step": 2597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6931608898982187e-05,
+      "loss": 0.1567,
+      "step": 2598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6929105545059834e-05,
+      "loss": 0.1473,
+      "step": 2599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926601355596535e-05,
+      "loss": 0.1727,
+      "step": 2600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692409633089426e-05,
+      "loss": 0.1723,
+      "step": 2601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692159047125507e-05,
+      "loss": 0.1495,
+      "step": 2602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6919083776981133e-05,
+      "loss": 0.1594,
+      "step": 2603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691657624837472e-05,
+      "loss": 0.1413,
+      "step": 2604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69140678857382e-05,
+      "loss": 0.1524,
+      "step": 2605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6911558689374035e-05,
+      "loss": 0.144,
+      "step": 2606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909048659584805e-05,
+      "loss": 0.163,
+      "step": 2607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690653779667317e-05,
+      "loss": 0.1427,
+      "step": 2608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904026100941903e-05,
+      "loss": 0.1439,
+      "step": 2609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6901513572693883e-05,
+      "loss": 0.1441,
+      "step": 2610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689900021223207e-05,
+      "loss": 0.1428,
+      "step": 2611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6896486019859543e-05,
+      "loss": 0.1657,
+      "step": 2612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689397099587947e-05,
+      "loss": 0.1632,
+      "step": 2613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689145514059513e-05,
+      "loss": 0.1571,
+      "step": 2614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6888938454309883e-05,
+      "loss": 0.1536,
+      "step": 2615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688642093732721e-05,
+      "loss": 0.1238,
+      "step": 2616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6883902589950687e-05,
+      "loss": 0.1479,
+      "step": 2617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6881383412483986e-05,
+      "loss": 0.1455,
+      "step": 2618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687886340523087e-05,
+      "loss": 0.1451,
+      "step": 2619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876342568495224e-05,
+      "loss": 0.1646,
+      "step": 2620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6873820902581015e-05,
+      "loss": 0.1687,
+      "step": 2621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871298407792317e-05,
+      "loss": 0.1521,
+      "step": 2622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6868775084433306e-05,
+      "loss": 0.1354,
+      "step": 2623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686625093280825e-05,
+      "loss": 0.1509,
+      "step": 2624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6863725953221525e-05,
+      "loss": 0.1471,
+      "step": 2625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861200145977608e-05,
+      "loss": 0.1507,
+      "step": 2626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6858673511381063e-05,
+      "loss": 0.1477,
+      "step": 2627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6856146049736567e-05,
+      "loss": 0.1611,
+      "step": 2628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6853617761348894e-05,
+      "loss": 0.148,
+      "step": 2629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685108864652291e-05,
+      "loss": 0.1546,
+      "step": 2630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684855870556359e-05,
+      "loss": 0.1673,
+      "step": 2631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6846027938776005e-05,
+      "loss": 0.1621,
+      "step": 2632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843496346465327e-05,
+      "loss": 0.1634,
+      "step": 2633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6840963928936824e-05,
+      "loss": 0.1568,
+      "step": 2634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6838430686495863e-05,
+      "loss": 0.1595,
+      "step": 2635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6835896619447922e-05,
+      "loss": 0.1465,
+      "step": 2636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6833361728098555e-05,
+      "loss": 0.145,
+      "step": 2637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6830826012753446e-05,
+      "loss": 0.1684,
+      "step": 2638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682828947371835e-05,
+      "loss": 0.1512,
+      "step": 2639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6825752111299143e-05,
+      "loss": 0.1663,
+      "step": 2640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823213925801788e-05,
+      "loss": 0.1256,
+      "step": 2641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682067491753234e-05,
+      "loss": 0.1598,
+      "step": 2642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681813508679698e-05,
+      "loss": 0.1563,
+      "step": 2643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681559443390196e-05,
+      "loss": 0.1427,
+      "step": 2644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6813052959153647e-05,
+      "loss": 0.1618,
+      "step": 2645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6810510662858503e-05,
+      "loss": 0.1415,
+      "step": 2646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807967545323086e-05,
+      "loss": 0.145,
+      "step": 2647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680542360685406e-05,
+      "loss": 0.1439,
+      "step": 2648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6802878847758185e-05,
+      "loss": 0.1487,
+      "step": 2649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680033326834231e-05,
+      "loss": 0.1494,
+      "step": 2650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6797786868913403e-05,
+      "loss": 0.1334,
+      "step": 2651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6795239649778508e-05,
+      "loss": 0.1554,
+      "step": 2652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679269161124479e-05,
+      "loss": 0.1557,
+      "step": 2653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6790142753619493e-05,
+      "loss": 0.1533,
+      "step": 2654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678759307720998e-05,
+      "loss": 0.1441,
+      "step": 2655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678504258232369e-05,
+      "loss": 0.1458,
+      "step": 2656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6782491269268182e-05,
+      "loss": 0.1459,
+      "step": 2657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67799391383511e-05,
+      "loss": 0.1282,
+      "step": 2658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677738618988019e-05,
+      "loss": 0.1498,
+      "step": 2659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67748324241633e-05,
+      "loss": 0.137,
+      "step": 2660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677227784150837e-05,
+      "loss": 0.1641,
+      "step": 2661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6769722442223447e-05,
+      "loss": 0.1443,
+      "step": 2662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676716622661667e-05,
+      "loss": 0.1441,
+      "step": 2663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6764609194996278e-05,
+      "loss": 0.1564,
+      "step": 2664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676205134767061e-05,
+      "loss": 0.1511,
+      "step": 2665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6759492684948096e-05,
+      "loss": 0.1586,
+      "step": 2666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675693320713728e-05,
+      "loss": 0.1477,
+      "step": 2667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754372914546786e-05,
+      "loss": 0.1577,
+      "step": 2668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6751811807485353e-05,
+      "loss": 0.1764,
+      "step": 2669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749249886261804e-05,
+      "loss": 0.1332,
+      "step": 2670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674668715118507e-05,
+      "loss": 0.1563,
+      "step": 2671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674412360256417e-05,
+      "loss": 0.1563,
+      "step": 2672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6741559240708238e-05,
+      "loss": 0.1518,
+      "step": 2673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6738994065926486e-05,
+      "loss": 0.1579,
+      "step": 2674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673642807852824e-05,
+      "loss": 0.1491,
+      "step": 2675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6733861278822913e-05,
+      "loss": 0.1571,
+      "step": 2676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731293667120026e-05,
+      "loss": 0.1573,
+      "step": 2677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672872524372919e-05,
+      "loss": 0.1623,
+      "step": 2678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6726156008960113e-05,
+      "loss": 0.1544,
+      "step": 2679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672358596312261e-05,
+      "loss": 0.1612,
+      "step": 2680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6721015106526583e-05,
+      "loss": 0.1634,
+      "step": 2681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671844343948204e-05,
+      "loss": 0.1578,
+      "step": 2682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715870962299083e-05,
+      "loss": 0.1639,
+      "step": 2683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671329767528791e-05,
+      "loss": 0.1593,
+      "step": 2684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671072357875882e-05,
+      "loss": 0.1423,
+      "step": 2685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6708148673022213e-05,
+      "loss": 0.1446,
+      "step": 2686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6705572958388576e-05,
+      "loss": 0.1636,
+      "step": 2687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6702996435168498e-05,
+      "loss": 0.1668,
+      "step": 2688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6700419103672674e-05,
+      "loss": 0.1413,
+      "step": 2689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669784096421189e-05,
+      "loss": 0.1565,
+      "step": 2690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695262017097016e-05,
+      "loss": 0.1459,
+      "step": 2691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669268226263905e-05,
+      "loss": 0.1494,
+      "step": 2692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669010170114906e-05,
+      "loss": 0.1408,
+      "step": 2693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668752033293822e-05,
+      "loss": 0.1501,
+      "step": 2694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6684938158317807e-05,
+      "loss": 0.1508,
+      "step": 2695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668235517759919e-05,
+      "loss": 0.1655,
+      "step": 2696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679771391093828e-05,
+      "loss": 0.1531,
+      "step": 2697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6677186799113293e-05,
+      "loss": 0.1543,
+      "step": 2698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674601401969247e-05,
+      "loss": 0.1482,
+      "step": 2699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6672015199973447e-05,
+      "loss": 0.1461,
+      "step": 2700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666942819343774e-05,
+      "loss": 0.1743,
+      "step": 2701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666684038267409e-05,
+      "loss": 0.1608,
+      "step": 2702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664251767994543e-05,
+      "loss": 0.1319,
+      "step": 2703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6661662349711236e-05,
+      "loss": 0.1598,
+      "step": 2704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6659072128136425e-05,
+      "loss": 0.1568,
+      "step": 2705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665648110358244e-05,
+      "loss": 0.1439,
+      "step": 2706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6653889276361724e-05,
+      "loss": 0.1668,
+      "step": 2707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665129664678681e-05,
+      "loss": 0.1656,
+      "step": 2708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648703215170323e-05,
+      "loss": 0.1571,
+      "step": 2709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646108981824998e-05,
+      "loss": 0.1449,
+      "step": 2710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664351394706365e-05,
+      "loss": 0.1359,
+      "step": 2711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640918111199208e-05,
+      "loss": 0.1553,
+      "step": 2712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663832147454468e-05,
+      "loss": 0.1599,
+      "step": 2713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6635724037413184e-05,
+      "loss": 0.13,
+      "step": 2714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6633125800117934e-05,
+      "loss": 0.1413,
+      "step": 2715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663052676297223e-05,
+      "loss": 0.1411,
+      "step": 2716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6627926926289476e-05,
+      "loss": 0.168,
+      "step": 2717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662532629038317e-05,
+      "loss": 0.1674,
+      "step": 2718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6622724855566918e-05,
+      "loss": 0.1604,
+      "step": 2719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66201226221544e-05,
+      "loss": 0.1635,
+      "step": 2720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6617519590459407e-05,
+      "loss": 0.1397,
+      "step": 2721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6614915760795832e-05,
+      "loss": 0.1602,
+      "step": 2722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612311133477644e-05,
+      "loss": 0.1449,
+      "step": 2723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6609705708818924e-05,
+      "loss": 0.1381,
+      "step": 2724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660709948713385e-05,
+      "loss": 0.1655,
+      "step": 2725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604492468736688e-05,
+      "loss": 0.1497,
+      "step": 2726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66018846539418e-05,
+      "loss": 0.1453,
+      "step": 2727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659927604306365e-05,
+      "loss": 0.1572,
+      "step": 2728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65966666364168e-05,
+      "loss": 0.1412,
+      "step": 2729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594056434315894e-05,
+      "loss": 0.1756,
+      "step": 2730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6591445437075687e-05,
+      "loss": 0.1538,
+      "step": 2731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588833645011025e-05,
+      "loss": 0.1465,
+      "step": 2732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658622105843685e-05,
+      "loss": 0.1485,
+      "step": 2733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583607677668192e-05,
+      "loss": 0.1575,
+      "step": 2734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580993503020192e-05,
+      "loss": 0.1373,
+      "step": 2735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657837853480807e-05,
+      "loss": 0.1631,
+      "step": 2736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575762773347153e-05,
+      "loss": 0.1326,
+      "step": 2737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6573146218952865e-05,
+      "loss": 0.167,
+      "step": 2738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6570528871940713e-05,
+      "loss": 0.1632,
+      "step": 2739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567910732626316e-05,
+      "loss": 0.1376,
+      "step": 2740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656529180132538e-05,
+      "loss": 0.1497,
+      "step": 2741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6562672078353702e-05,
+      "loss": 0.1651,
+      "step": 2742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560051564027175e-05,
+      "loss": 0.1729,
+      "step": 2743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6557430258661802e-05,
+      "loss": 0.1372,
+      "step": 2744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655480816257367e-05,
+      "loss": 0.1367,
+      "step": 2745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552185276078957e-05,
+      "loss": 0.1694,
+      "step": 2746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6549561599493943e-05,
+      "loss": 0.1594,
+      "step": 2747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6546937133135002e-05,
+      "loss": 0.1602,
+      "step": 2748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654431187731861e-05,
+      "loss": 0.1424,
+      "step": 2749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6541685832361324e-05,
+      "loss": 0.1506,
+      "step": 2750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6539058998579804e-05,
+      "loss": 0.1507,
+      "step": 2751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6536431376290803e-05,
+      "loss": 0.1657,
+      "step": 2752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6533802965811178e-05,
+      "loss": 0.1622,
+      "step": 2753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6531173767457873e-05,
+      "loss": 0.161,
+      "step": 2754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528543781547922e-05,
+      "loss": 0.1474,
+      "step": 2755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6525913008398463e-05,
+      "loss": 0.1394,
+      "step": 2756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6523281448326725e-05,
+      "loss": 0.1655,
+      "step": 2757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6520649101650035e-05,
+      "loss": 0.1328,
+      "step": 2758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651801596868581e-05,
+      "loss": 0.1617,
+      "step": 2759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515382049751566e-05,
+      "loss": 0.1524,
+      "step": 2760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651274734516491e-05,
+      "loss": 0.162,
+      "step": 2761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510111855243546e-05,
+      "loss": 0.1667,
+      "step": 2762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6507475580305276e-05,
+      "loss": 0.1626,
+      "step": 2763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504838520667994e-05,
+      "loss": 0.1539,
+      "step": 2764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650220067664968e-05,
+      "loss": 0.1486,
+      "step": 2765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6499562048568422e-05,
+      "loss": 0.1412,
+      "step": 2766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64969226367424e-05,
+      "loss": 0.1588,
+      "step": 2767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649428244148988e-05,
+      "loss": 0.148,
+      "step": 2768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6491641463129225e-05,
+      "loss": 0.1447,
+      "step": 2769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6488999701978905e-05,
+      "loss": 0.1495,
+      "step": 2770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648635715835747e-05,
+      "loss": 0.1401,
+      "step": 2771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6483713832583566e-05,
+      "loss": 0.1562,
+      "step": 2772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648106972497594e-05,
+      "loss": 0.1565,
+      "step": 2773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6478424835853427e-05,
+      "loss": 0.1296,
+      "step": 2774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475779165534966e-05,
+      "loss": 0.1435,
+      "step": 2775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6473132714339575e-05,
+      "loss": 0.1414,
+      "step": 2776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647048548258638e-05,
+      "loss": 0.1326,
+      "step": 2777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6467837470594588e-05,
+      "loss": 0.1664,
+      "step": 2778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646518867868352e-05,
+      "loss": 0.1548,
+      "step": 2779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6462539107172565e-05,
+      "loss": 0.1365,
+      "step": 2780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6459888756381227e-05,
+      "loss": 0.1509,
+      "step": 2781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6457237626629096e-05,
+      "loss": 0.1448,
+      "step": 2782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6454585718235857e-05,
+      "loss": 0.1468,
+      "step": 2783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6451933031521282e-05,
+      "loss": 0.1328,
+      "step": 2784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644927956680525e-05,
+      "loss": 0.1408,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6446625324407728e-05,
+      "loss": 0.1421,
+      "step": 2786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644397030464877e-05,
+      "loss": 0.1338,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441314507848537e-05,
+      "loss": 0.1433,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6438657934327266e-05,
+      "loss": 0.1515,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643600058440531e-05,
+      "loss": 0.1468,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433342458403093e-05,
+      "loss": 0.1553,
+      "step": 2791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6430683556641147e-05,
+      "loss": 0.1642,
+      "step": 2792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6428023879440095e-05,
+      "loss": 0.1563,
+      "step": 2793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6425363427120652e-05,
+      "loss": 0.1383,
+      "step": 2794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6422702200003625e-05,
+      "loss": 0.1338,
+      "step": 2795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6420040198409917e-05,
+      "loss": 0.1528,
+      "step": 2796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6417377422660523e-05,
+      "loss": 0.167,
+      "step": 2797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6414713873076533e-05,
+      "loss": 0.1382,
+      "step": 2798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641204954997913e-05,
+      "loss": 0.1434,
+      "step": 2799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640938445368958e-05,
+      "loss": 0.1477,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6406718584529267e-05,
+      "loss": 0.1371,
+      "step": 2801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6404051942819645e-05,
+      "loss": 0.1543,
+      "step": 2802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6401384528882264e-05,
+      "loss": 0.1511,
+      "step": 2803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639871634303878e-05,
+      "loss": 0.1483,
+      "step": 2804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639604738561093e-05,
+      "loss": 0.143,
+      "step": 2805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639337765692055e-05,
+      "loss": 0.1464,
+      "step": 2806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639070715728957e-05,
+      "loss": 0.1302,
+      "step": 2807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388035887040003e-05,
+      "loss": 0.1563,
+      "step": 2808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6385363846493966e-05,
+      "loss": 0.1414,
+      "step": 2809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382691035973666e-05,
+      "loss": 0.1529,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.63800174558014e-05,
+      "loss": 0.1471,
+      "step": 2811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377343106299563e-05,
+      "loss": 0.1466,
+      "step": 2812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6374667987790633e-05,
+      "loss": 0.1459,
+      "step": 2813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6371992100597194e-05,
+      "loss": 0.1405,
+      "step": 2814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369315445041912e-05,
+      "loss": 0.1393,
+      "step": 2815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6366638021447546e-05,
+      "loss": 0.1548,
+      "step": 2816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6363959830136963e-05,
+      "loss": 0.1652,
+      "step": 2817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6361280871433097e-05,
+      "loss": 0.1524,
+      "step": 2818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6358601145658996e-05,
+      "loss": 0.1521,
+      "step": 2819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355920653137793e-05,
+      "loss": 0.1621,
+      "step": 2820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635323939419271e-05,
+      "loss": 0.1562,
+      "step": 2821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6350557369147067e-05,
+      "loss": 0.1627,
+      "step": 2822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634787457832427e-05,
+      "loss": 0.1344,
+      "step": 2823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345191022047826e-05,
+      "loss": 0.1457,
+      "step": 2824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6342506700641325e-05,
+      "loss": 0.158,
+      "step": 2825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633982161442846e-05,
+      "loss": 0.1565,
+      "step": 2826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337135763733002e-05,
+      "loss": 0.1448,
+      "step": 2827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633444914887883e-05,
+      "loss": 0.1605,
+      "step": 2828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6331761770189903e-05,
+      "loss": 0.1369,
+      "step": 2829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632907362799028e-05,
+      "loss": 0.1326,
+      "step": 2830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6326384722604105e-05,
+      "loss": 0.1501,
+      "step": 2831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6323695054355616e-05,
+      "loss": 0.1711,
+      "step": 2832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632100462356915e-05,
+      "loss": 0.1596,
+      "step": 2833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631831343056913e-05,
+      "loss": 0.1529,
+      "step": 2834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6315621475680067e-05,
+      "loss": 0.1417,
+      "step": 2835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631292875922657e-05,
+      "loss": 0.1315,
+      "step": 2836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631023528153334e-05,
+      "loss": 0.1411,
+      "step": 2837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630754104292517e-05,
+      "loss": 0.1672,
+      "step": 2838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6304846043726937e-05,
+      "loss": 0.1587,
+      "step": 2839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630215028426362e-05,
+      "loss": 0.1426,
+      "step": 2840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629945376486028e-05,
+      "loss": 0.1618,
+      "step": 2841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629675648584208e-05,
+      "loss": 0.1424,
+      "step": 2842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629405844753427e-05,
+      "loss": 0.148,
+      "step": 2843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6291359650262185e-05,
+      "loss": 0.1484,
+      "step": 2844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6288660094351264e-05,
+      "loss": 0.1637,
+      "step": 2845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285959780127022e-05,
+      "loss": 0.1406,
+      "step": 2846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283258707915086e-05,
+      "loss": 0.1467,
+      "step": 2847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6280556878041154e-05,
+      "loss": 0.1466,
+      "step": 2848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627785429083103e-05,
+      "loss": 0.1337,
+      "step": 2849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6275150946610594e-05,
+      "loss": 0.1524,
+      "step": 2850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6272446845705836e-05,
+      "loss": 0.1417,
+      "step": 2851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6269741988442828e-05,
+      "loss": 0.1459,
+      "step": 2852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6267036375147728e-05,
+      "loss": 0.1407,
+      "step": 2853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626433000614679e-05,
+      "loss": 0.1532,
+      "step": 2854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6261622881766364e-05,
+      "loss": 0.1571,
+      "step": 2855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6258915002332883e-05,
+      "loss": 0.1477,
+      "step": 2856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625620636817288e-05,
+      "loss": 0.1677,
+      "step": 2857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253496979612965e-05,
+      "loss": 0.1521,
+      "step": 2858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250786836979854e-05,
+      "loss": 0.1552,
+      "step": 2859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6248075940600346e-05,
+      "loss": 0.1459,
+      "step": 2860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245364290801332e-05,
+      "loss": 0.1683,
+      "step": 2861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6242651887909797e-05,
+      "loss": 0.1442,
+      "step": 2862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6239938732252808e-05,
+      "loss": 0.1612,
+      "step": 2863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237224824157535e-05,
+      "loss": 0.1515,
+      "step": 2864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234510163951225e-05,
+      "loss": 0.1502,
+      "step": 2865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6231794751961234e-05,
+      "loss": 0.1406,
+      "step": 2866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622907858851499e-05,
+      "loss": 0.1398,
+      "step": 2867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6226361673940017e-05,
+      "loss": 0.1464,
+      "step": 2868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223644008563942e-05,
+      "loss": 0.1588,
+      "step": 2869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220925592714463e-05,
+      "loss": 0.1475,
+      "step": 2870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6218206426719386e-05,
+      "loss": 0.1378,
+      "step": 2871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6215486510906597e-05,
+      "loss": 0.1374,
+      "step": 2872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621276584560407e-05,
+      "loss": 0.1415,
+      "step": 2873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621004443113988e-05,
+      "loss": 0.1531,
+      "step": 2874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620732226784218e-05,
+      "loss": 0.1356,
+      "step": 2875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.620459935603923e-05,
+      "loss": 0.1539,
+      "step": 2876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6201875696059363e-05,
+      "loss": 0.1734,
+      "step": 2877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6199151288231016e-05,
+      "loss": 0.1367,
+      "step": 2878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6196426132882702e-05,
+      "loss": 0.1382,
+      "step": 2879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6193700230343035e-05,
+      "loss": 0.1477,
+      "step": 2880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6190973580940716e-05,
+      "loss": 0.1594,
+      "step": 2881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6188246185004536e-05,
+      "loss": 0.1302,
+      "step": 2882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6185518042863376e-05,
+      "loss": 0.165,
+      "step": 2883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6182789154846205e-05,
+      "loss": 0.149,
+      "step": 2884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618005952128209e-05,
+      "loss": 0.1623,
+      "step": 2885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177329142500172e-05,
+      "loss": 0.1506,
+      "step": 2886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6174598018829702e-05,
+      "loss": 0.1473,
+      "step": 2887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6171866150600002e-05,
+      "loss": 0.1392,
+      "step": 2888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61691335381405e-05,
+      "loss": 0.1408,
+      "step": 2889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61664001817807e-05,
+      "loss": 0.1291,
+      "step": 2890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6163666081850203e-05,
+      "loss": 0.1537,
+      "step": 2891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61609312386787e-05,
+      "loss": 0.138,
+      "step": 2892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615819565259597e-05,
+      "loss": 0.1666,
+      "step": 2893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6155459323931877e-05,
+      "loss": 0.1645,
+      "step": 2894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6152722253016384e-05,
+      "loss": 0.1466,
+      "step": 2895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614998444017954e-05,
+      "loss": 0.1479,
+      "step": 2896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6147245885751474e-05,
+      "loss": 0.1481,
+      "step": 2897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6144506590062424e-05,
+      "loss": 0.1648,
+      "step": 2898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6141766553442697e-05,
+      "loss": 0.1515,
+      "step": 2899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61390257762227e-05,
+      "loss": 0.1541,
+      "step": 2900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6136284258732932e-05,
+      "loss": 0.1443,
+      "step": 2901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613354200130397e-05,
+      "loss": 0.1572,
+      "step": 2902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6130799004266493e-05,
+      "loss": 0.1628,
+      "step": 2903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6128055267951256e-05,
+      "loss": 0.1446,
+      "step": 2904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6125310792689123e-05,
+      "loss": 0.1584,
+      "step": 2905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6122565578811018e-05,
+      "loss": 0.1465,
+      "step": 2906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6119819626647985e-05,
+      "loss": 0.1336,
+      "step": 2907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611707293653113e-05,
+      "loss": 0.1465,
+      "step": 2908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114325508791676e-05,
+      "loss": 0.1543,
+      "step": 2909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6111577343760905e-05,
+      "loss": 0.1316,
+      "step": 2910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610882844177021e-05,
+      "loss": 0.1503,
+      "step": 2911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106078803151065e-05,
+      "loss": 0.1563,
+      "step": 2912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610332842823503e-05,
+      "loss": 0.1585,
+      "step": 2913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6100577317353755e-05,
+      "loss": 0.1403,
+      "step": 2914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097825470838987e-05,
+      "loss": 0.1513,
+      "step": 2915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609507288902255e-05,
+      "loss": 0.1332,
+      "step": 2916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092319572236367e-05,
+      "loss": 0.1352,
+      "step": 2917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608956552081244e-05,
+      "loss": 0.1509,
+      "step": 2918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6086810735082866e-05,
+      "loss": 0.1453,
+      "step": 2919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608405521537983e-05,
+      "loss": 0.152,
+      "step": 2920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6081298962035602e-05,
+      "loss": 0.1449,
+      "step": 2921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6078541975382546e-05,
+      "loss": 0.1565,
+      "step": 2922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6075784255753107e-05,
+      "loss": 0.1367,
+      "step": 2923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6073025803479826e-05,
+      "loss": 0.1469,
+      "step": 2924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6070266618895324e-05,
+      "loss": 0.1322,
+      "step": 2925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606750670233232e-05,
+      "loss": 0.15,
+      "step": 2926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6064746054123618e-05,
+      "loss": 0.1353,
+      "step": 2927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6061984674602104e-05,
+      "loss": 0.1533,
+      "step": 2928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6059222564100755e-05,
+      "loss": 0.133,
+      "step": 2929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6056459722952645e-05,
+      "loss": 0.1731,
+      "step": 2930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605369615149093e-05,
+      "loss": 0.1547,
+      "step": 2931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6050931850048843e-05,
+      "loss": 0.14,
+      "step": 2932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6048166818959718e-05,
+      "loss": 0.1463,
+      "step": 2933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6045401058556983e-05,
+      "loss": 0.146,
+      "step": 2934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6042634569174137e-05,
+      "loss": 0.1548,
+      "step": 2935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039867351144778e-05,
+      "loss": 0.1385,
+      "step": 2936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6037099404802585e-05,
+      "loss": 0.1541,
+      "step": 2937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6034330730481334e-05,
+      "loss": 0.1518,
+      "step": 2938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6031561328514885e-05,
+      "loss": 0.1421,
+      "step": 2939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602879119923718e-05,
+      "loss": 0.1361,
+      "step": 2940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026020342982248e-05,
+      "loss": 0.1312,
+      "step": 2941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602324876008422e-05,
+      "loss": 0.1468,
+      "step": 2942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6020476450877303e-05,
+      "loss": 0.148,
+      "step": 2943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601770341569579e-05,
+      "loss": 0.1341,
+      "step": 2944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6014929654874064e-05,
+      "loss": 0.1688,
+      "step": 2945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012155168746608e-05,
+      "loss": 0.1363,
+      "step": 2946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6009379957647968e-05,
+      "loss": 0.1297,
+      "step": 2947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6006604021912798e-05,
+      "loss": 0.1508,
+      "step": 2948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600382736187583e-05,
+      "loss": 0.145,
+      "step": 2949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600104997787189e-05,
+      "loss": 0.1436,
+      "step": 2950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599827187023588e-05,
+      "loss": 0.1286,
+      "step": 2951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59954930393028e-05,
+      "loss": 0.1573,
+      "step": 2952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599271348540773e-05,
+      "loss": 0.1463,
+      "step": 2953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598993320888584e-05,
+      "loss": 0.1412,
+      "step": 2954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5987152210072398e-05,
+      "loss": 0.1564,
+      "step": 2955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598437048930273e-05,
+      "loss": 0.1364,
+      "step": 2956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981588046912286e-05,
+      "loss": 0.1454,
+      "step": 2957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5978804883236577e-05,
+      "loss": 0.1601,
+      "step": 2958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976020998611206e-05,
+      "loss": 0.1731,
+      "step": 2959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597323639337187e-05,
+      "loss": 0.1487,
+      "step": 2960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970451067854344e-05,
+      "loss": 0.1321,
+      "step": 2961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5967665022394498e-05,
+      "loss": 0.1555,
+      "step": 2962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5964878257328284e-05,
+      "loss": 0.1425,
+      "step": 2963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5962090772991744e-05,
+      "loss": 0.1257,
+      "step": 2964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5959302569721e-05,
+      "loss": 0.1582,
+      "step": 2965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5956513647852273e-05,
+      "loss": 0.1468,
+      "step": 2966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5953724007721856e-05,
+      "loss": 0.1553,
+      "step": 2967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595093364966614e-05,
+      "loss": 0.1509,
+      "step": 2968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5948142574021593e-05,
+      "loss": 0.1482,
+      "step": 2969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594535078112478e-05,
+      "loss": 0.1574,
+      "step": 2970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5942558271312345e-05,
+      "loss": 0.1432,
+      "step": 2971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593976504492102e-05,
+      "loss": 0.1505,
+      "step": 2972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5936971102287625e-05,
+      "loss": 0.1558,
+      "step": 2973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5934176443749067e-05,
+      "loss": 0.1521,
+      "step": 2974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5931381069642333e-05,
+      "loss": 0.1586,
+      "step": 2975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592858498030451e-05,
+      "loss": 0.1339,
+      "step": 2976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5925788176072752e-05,
+      "loss": 0.1613,
+      "step": 2977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5922990657284315e-05,
+      "loss": 0.1442,
+      "step": 2978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5920192424276535e-05,
+      "loss": 0.1432,
+      "step": 2979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5917393477386836e-05,
+      "loss": 0.1502,
+      "step": 2980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5914593816952723e-05,
+      "loss": 0.1438,
+      "step": 2981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5911793443311796e-05,
+      "loss": 0.1438,
+      "step": 2982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590899235680173e-05,
+      "loss": 0.148,
+      "step": 2983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59061905577603e-05,
+      "loss": 0.1556,
+      "step": 2984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590338804652535e-05,
+      "loss": 0.1523,
+      "step": 2985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5900584823434827e-05,
+      "loss": 0.1416,
+      "step": 2986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5897780888826746e-05,
+      "loss": 0.1463,
+      "step": 2987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5894976243039225e-05,
+      "loss": 0.1416,
+      "step": 2988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589217088641046e-05,
+      "loss": 0.1431,
+      "step": 2989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588936481927873e-05,
+      "loss": 0.1468,
+      "step": 2990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886558041982402e-05,
+      "loss": 0.1587,
+      "step": 2991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588375055485993e-05,
+      "loss": 0.1418,
+      "step": 2992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588094235824985e-05,
+      "loss": 0.1355,
+      "step": 2993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5878133452490794e-05,
+      "loss": 0.1408,
+      "step": 2994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5875323837921465e-05,
+      "loss": 0.1619,
+      "step": 2995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5872513514880662e-05,
+      "loss": 0.1672,
+      "step": 2996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586970248370726e-05,
+      "loss": 0.1567,
+      "step": 2997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5866890744740232e-05,
+      "loss": 0.1385,
+      "step": 2998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5864078298318623e-05,
+      "loss": 0.1544,
+      "step": 2999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5861265144781577e-05,
+      "loss": 0.1523,
+      "step": 3000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5858451284468307e-05,
+      "loss": 0.1474,
+      "step": 3001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585563671771813e-05,
+      "loss": 0.1562,
+      "step": 3002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585282144487043e-05,
+      "loss": 0.1368,
+      "step": 3003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585000546626469e-05,
+      "loss": 0.143,
+      "step": 3004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5847188782240473e-05,
+      "loss": 0.1491,
+      "step": 3005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584437139313742e-05,
+      "loss": 0.1488,
+      "step": 3006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5841553299295272e-05,
+      "loss": 0.133,
+      "step": 3007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838734501053844e-05,
+      "loss": 0.1825,
+      "step": 3008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583591499875303e-05,
+      "loss": 0.1616,
+      "step": 3009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5833094792732836e-05,
+      "loss": 0.1375,
+      "step": 3010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5830273883333316e-05,
+      "loss": 0.1379,
+      "step": 3011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827452270894637e-05,
+      "loss": 0.1813,
+      "step": 3012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582462995575704e-05,
+      "loss": 0.1405,
+      "step": 3013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582180693826085e-05,
+      "loss": 0.1544,
+      "step": 3014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5818983218746478e-05,
+      "loss": 0.1487,
+      "step": 3015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581615879755442e-05,
+      "loss": 0.1377,
+      "step": 3016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581333367502526e-05,
+      "loss": 0.1486,
+      "step": 3017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581050785149966e-05,
+      "loss": 0.1556,
+      "step": 3018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5807681327318372e-05,
+      "loss": 0.1659,
+      "step": 3019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804854102822223e-05,
+      "loss": 0.1436,
+      "step": 3020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5802026178352145e-05,
+      "loss": 0.1396,
+      "step": 3021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5799197554249132e-05,
+      "loss": 0.1422,
+      "step": 3022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5796368230854272e-05,
+      "loss": 0.1376,
+      "step": 3023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5793538208508742e-05,
+      "loss": 0.1657,
+      "step": 3024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579070748755379e-05,
+      "loss": 0.1615,
+      "step": 3025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5787876068330764e-05,
+      "loss": 0.1647,
+      "step": 3026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5785043951181082e-05,
+      "loss": 0.1541,
+      "step": 3027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578221113644626e-05,
+      "loss": 0.1792,
+      "step": 3028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5779377624467885e-05,
+      "loss": 0.1491,
+      "step": 3029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5776543415587633e-05,
+      "loss": 0.1486,
+      "step": 3030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577370851014727e-05,
+      "loss": 0.1431,
+      "step": 3031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577087290848864e-05,
+      "loss": 0.132,
+      "step": 3032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5768036610953664e-05,
+      "loss": 0.1501,
+      "step": 3033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576519961788437e-05,
+      "loss": 0.1437,
+      "step": 3034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5762361929622837e-05,
+      "loss": 0.1554,
+      "step": 3035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5759523546511258e-05,
+      "loss": 0.1464,
+      "step": 3036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5756684468891897e-05,
+      "loss": 0.1396,
+      "step": 3037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753844697107098e-05,
+      "loss": 0.1542,
+      "step": 3038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5751004231499288e-05,
+      "loss": 0.1587,
+      "step": 3039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574816307241099e-05,
+      "loss": 0.1515,
+      "step": 3040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5745321220184802e-05,
+      "loss": 0.1562,
+      "step": 3041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5742478675163407e-05,
+      "loss": 0.1553,
+      "step": 3042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5739635437689568e-05,
+      "loss": 0.144,
+      "step": 3043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736791508106134e-05,
+      "loss": 0.1612,
+      "step": 3044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733946886756042e-05,
+      "loss": 0.1456,
+      "step": 3045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5731101573982307e-05,
+      "loss": 0.1501,
+      "step": 3046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572825557012803e-05,
+      "loss": 0.1558,
+      "step": 3047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572540887553639e-05,
+      "loss": 0.1424,
+      "step": 3048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722561490550655e-05,
+      "loss": 0.1626,
+      "step": 3049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571971341551418e-05,
+      "loss": 0.1346,
+      "step": 3050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716864650770392e-05,
+      "loss": 0.1507,
+      "step": 3051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5714015196662812e-05,
+      "loss": 0.1444,
+      "step": 3052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5711165053535034e-05,
+      "loss": 0.1609,
+      "step": 3053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5708314221730744e-05,
+      "loss": 0.144,
+      "step": 3054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5705462701593704e-05,
+      "loss": 0.1419,
+      "step": 3055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570261049346777e-05,
+      "loss": 0.147,
+      "step": 3056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5699757597696864e-05,
+      "loss": 0.15,
+      "step": 3057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5696904014625003e-05,
+      "loss": 0.1315,
+      "step": 3058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5694049744596286e-05,
+      "loss": 0.1488,
+      "step": 3059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691194787954896e-05,
+      "loss": 0.1429,
+      "step": 3060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568833914504509e-05,
+      "loss": 0.1576,
+      "step": 3061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5685482816211216e-05,
+      "loss": 0.1542,
+      "step": 3062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56826258017977e-05,
+      "loss": 0.1654,
+      "step": 3063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5679768102149054e-05,
+      "loss": 0.1433,
+      "step": 3064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5676909717609877e-05,
+      "loss": 0.1552,
+      "step": 3065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567405064852484e-05,
+      "loss": 0.1404,
+      "step": 3066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56711908952387e-05,
+      "loss": 0.1678,
+      "step": 3067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5668330458096298e-05,
+      "loss": 0.1489,
+      "step": 3068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5665469337442564e-05,
+      "loss": 0.1668,
+      "step": 3069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662607533622504e-05,
+      "loss": 0.1402,
+      "step": 3070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5659745046981196e-05,
+      "loss": 0.1489,
+      "step": 3071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565688187786382e-05,
+      "loss": 0.1356,
+      "step": 3072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5654018026615624e-05,
+      "loss": 0.1467,
+      "step": 3073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565115349358195e-05,
+      "loss": 0.1414,
+      "step": 3074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564828827910821e-05,
+      "loss": 0.1526,
+      "step": 3075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56454223835399e-05,
+      "loss": 0.1422,
+      "step": 3076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5642555807222614e-05,
+      "loss": 0.1512,
+      "step": 3077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5639688550502006e-05,
+      "loss": 0.155,
+      "step": 3078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5636820613723824e-05,
+      "loss": 0.1489,
+      "step": 3079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5633951997233897e-05,
+      "loss": 0.1745,
+      "step": 3080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563108270137814e-05,
+      "loss": 0.146,
+      "step": 3081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562821272650253e-05,
+      "loss": 0.1459,
+      "step": 3082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5625342072953155e-05,
+      "loss": 0.138,
+      "step": 3083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562247074107617e-05,
+      "loss": 0.1488,
+      "step": 3084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5619598731217804e-05,
+      "loss": 0.1468,
+      "step": 3085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5616726043724383e-05,
+      "loss": 0.1262,
+      "step": 3086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56138526789423e-05,
+      "loss": 0.1381,
+      "step": 3087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5610978637218045e-05,
+      "loss": 0.1587,
+      "step": 3088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5608103918898182e-05,
+      "loss": 0.1529,
+      "step": 3089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5605228524329353e-05,
+      "loss": 0.1471,
+      "step": 3090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602352453858292e-05,
+      "loss": 0.1438,
+      "step": 3091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5599475707831797e-05,
+      "loss": 0.154,
+      "step": 3092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5596598286596766e-05,
+      "loss": 0.164,
+      "step": 3093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5593720190500167e-05,
+      "loss": 0.1329,
+      "step": 3094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5590841419889057e-05,
+      "loss": 0.1401,
+      "step": 3095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558796197511057e-05,
+      "loss": 0.1452,
+      "step": 3096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5585081856511916e-05,
+      "loss": 0.1706,
+      "step": 3097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5582201064440397e-05,
+      "loss": 0.1523,
+      "step": 3098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557931959924339e-05,
+      "loss": 0.1553,
+      "step": 3099
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5576437461268356e-05,
+      "loss": 0.1399,
+      "step": 3100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557355465086283e-05,
+      "loss": 0.1464,
+      "step": 3101
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557067116837444e-05,
+      "loss": 0.1244,
+      "step": 3102
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5567787014150883e-05,
+      "loss": 0.1421,
+      "step": 3103
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5564902188539945e-05,
+      "loss": 0.1629,
+      "step": 3104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5562016691889493e-05,
+      "loss": 0.1628,
+      "step": 3105
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559130524547467e-05,
+      "loss": 0.1526,
+      "step": 3106
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55562436868619e-05,
+      "loss": 0.1397,
+      "step": 3107
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5553356179180894e-05,
+      "loss": 0.1615,
+      "step": 3108
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555046800185264e-05,
+      "loss": 0.1371,
+      "step": 3109
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547579155225404e-05,
+      "loss": 0.1196,
+      "step": 3110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5544689639647535e-05,
+      "loss": 0.1077,
+      "step": 3111
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5541799455467466e-05,
+      "loss": 0.1231,
+      "step": 3112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5538908603033706e-05,
+      "loss": 0.1345,
+      "step": 3113
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5536017082694846e-05,
+      "loss": 0.1243,
+      "step": 3114
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553312489479956e-05,
+      "loss": 0.1073,
+      "step": 3115
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530232039696596e-05,
+      "loss": 0.1184,
+      "step": 3116
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552733851773479e-05,
+      "loss": 0.1194,
+      "step": 3117
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5524444329263056e-05,
+      "loss": 0.1178,
+      "step": 3118
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5521549474630387e-05,
+      "loss": 0.1067,
+      "step": 3119
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5518653954185853e-05,
+      "loss": 0.0965,
+      "step": 3120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551575776827861e-05,
+      "loss": 0.1144,
+      "step": 3121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5512860917257897e-05,
+      "loss": 0.1203,
+      "step": 3122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550996340147302e-05,
+      "loss": 0.1169,
+      "step": 3123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5507065221273385e-05,
+      "loss": 0.1205,
+      "step": 3124
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550416637700846e-05,
+      "loss": 0.108,
+      "step": 3125
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5501266869027795e-05,
+      "loss": 0.1105,
+      "step": 3126
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5498366697681036e-05,
+      "loss": 0.1201,
+      "step": 3127
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549546586331789e-05,
+      "loss": 0.1106,
+      "step": 3128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492564366288152e-05,
+      "loss": 0.1259,
+      "step": 3129
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489662206941704e-05,
+      "loss": 0.1122,
+      "step": 3130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548675938562849e-05,
+      "loss": 0.1197,
+      "step": 3131
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5483855902698556e-05,
+      "loss": 0.1207,
+      "step": 3132
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5480951758502008e-05,
+      "loss": 0.1104,
+      "step": 3133
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5478046953389046e-05,
+      "loss": 0.1095,
+      "step": 3134
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5475141487709935e-05,
+      "loss": 0.1125,
+      "step": 3135
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547223536181504e-05,
+      "loss": 0.103,
+      "step": 3136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5469328576054782e-05,
+      "loss": 0.113,
+      "step": 3137
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5466421130779684e-05,
+      "loss": 0.1085,
+      "step": 3138
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5463513026340327e-05,
+      "loss": 0.1232,
+      "step": 3139
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460604263087393e-05,
+      "loss": 0.1187,
+      "step": 3140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545769484137163e-05,
+      "loss": 0.0996,
+      "step": 3141
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545478476154387e-05,
+      "loss": 0.1039,
+      "step": 3142
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5451874023955012e-05,
+      "loss": 0.1158,
+      "step": 3143
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5448962628956058e-05,
+      "loss": 0.1051,
+      "step": 3144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5446050576898075e-05,
+      "loss": 0.1181,
+      "step": 3145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54431378681322e-05,
+      "loss": 0.1212,
+      "step": 3146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5440224503009673e-05,
+      "loss": 0.1112,
+      "step": 3147
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437310481881793e-05,
+      "loss": 0.1096,
+      "step": 3148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5434395805099946e-05,
+      "loss": 0.1179,
+      "step": 3149
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5431480473015593e-05,
+      "loss": 0.1174,
+      "step": 3150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5428564485980283e-05,
+      "loss": 0.108,
+      "step": 3151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5425647844345636e-05,
+      "loss": 0.1144,
+      "step": 3152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542273054846335e-05,
+      "loss": 0.1444,
+      "step": 3153
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541981259868521e-05,
+      "loss": 0.1173,
+      "step": 3154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541689399536307e-05,
+      "loss": 0.1196,
+      "step": 3155
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541397473884887e-05,
+      "loss": 0.1154,
+      "step": 3156
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541105482949462e-05,
+      "loss": 0.1076,
+      "step": 3157
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408134267652432e-05,
+      "loss": 0.1009,
+      "step": 3158
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540521305367446e-05,
+      "loss": 0.103,
+      "step": 3159
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540229118791297e-05,
+      "loss": 0.1068,
+      "step": 3160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5399368670720288e-05,
+      "loss": 0.1149,
+      "step": 3161
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539644550244882e-05,
+      "loss": 0.109,
+      "step": 3162
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539352168345106e-05,
+      "loss": 0.1123,
+      "step": 3163
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5390597214079574e-05,
+      "loss": 0.1126,
+      "step": 3164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5387672094687e-05,
+      "loss": 0.1081,
+      "step": 3165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538474632562607e-05,
+      "loss": 0.1254,
+      "step": 3166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538181990724958e-05,
+      "loss": 0.1184,
+      "step": 3167
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378892839910414e-05,
+      "loss": 0.1043,
+      "step": 3168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5375965123961528e-05,
+      "loss": 0.116,
+      "step": 3169
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537303675975596e-05,
+      "loss": 0.1141,
+      "step": 3170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5370107747646817e-05,
+      "loss": 0.1081,
+      "step": 3171
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53671780879873e-05,
+      "loss": 0.1011,
+      "step": 3172
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536424778113068e-05,
+      "loss": 0.1112,
+      "step": 3173
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5361316827430303e-05,
+      "loss": 0.1317,
+      "step": 3174
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358385227239594e-05,
+      "loss": 0.0994,
+      "step": 3175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5355452980912062e-05,
+      "loss": 0.116,
+      "step": 3176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352520088801285e-05,
+      "loss": 0.1181,
+      "step": 3177
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349586551260932e-05,
+      "loss": 0.1028,
+      "step": 3178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346652368644736e-05,
+      "loss": 0.1138,
+      "step": 3179
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5343717541306505e-05,
+      "loss": 0.1145,
+      "step": 3180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5340782069600146e-05,
+      "loss": 0.1061,
+      "step": 3181
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5337845953879625e-05,
+      "loss": 0.106,
+      "step": 3182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533490919449899e-05,
+      "loss": 0.1287,
+      "step": 3183
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5331971791812375e-05,
+      "loss": 0.1164,
+      "step": 3184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329033746173975e-05,
+      "loss": 0.1076,
+      "step": 3185
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532609505793808e-05,
+      "loss": 0.1063,
+      "step": 3186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5323155727459037e-05,
+      "loss": 0.1152,
+      "step": 3187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53202157550913e-05,
+      "loss": 0.1278,
+      "step": 3188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5317275141189372e-05,
+      "loss": 0.1142,
+      "step": 3189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314333886107848e-05,
+      "loss": 0.117,
+      "step": 3190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5311391990201396e-05,
+      "loss": 0.1154,
+      "step": 3191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530844945382476e-05,
+      "loss": 0.1164,
+      "step": 3192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5305506277332774e-05,
+      "loss": 0.1051,
+      "step": 3193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5302562461080326e-05,
+      "loss": 0.1097,
+      "step": 3194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5299618005422398e-05,
+      "loss": 0.1213,
+      "step": 3195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296672910714052e-05,
+      "loss": 0.1137,
+      "step": 3196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529372717731041e-05,
+      "loss": 0.106,
+      "step": 3197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290780805566684e-05,
+      "loss": 0.1184,
+      "step": 3198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5287833795838164e-05,
+      "loss": 0.1166,
+      "step": 3199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528488614848021e-05,
+      "loss": 0.1273,
+      "step": 3200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5281937863848257e-05,
+      "loss": 0.1174,
+      "step": 3201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527898894229783e-05,
+      "loss": 0.1289,
+      "step": 3202
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5276039384184516e-05,
+      "loss": 0.1201,
+      "step": 3203
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527308918986399e-05,
+      "loss": 0.1216,
+      "step": 3204
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5270138359692e-05,
+      "loss": 0.1273,
+      "step": 3205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5267186894024363e-05,
+      "loss": 0.1171,
+      "step": 3206
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526423479321699e-05,
+      "loss": 0.101,
+      "step": 3207
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526128205762584e-05,
+      "loss": 0.1168,
+      "step": 3208
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5258328687606985e-05,
+      "loss": 0.1183,
+      "step": 3209
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5255374683516544e-05,
+      "loss": 0.1307,
+      "step": 3210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5252420045710729e-05,
+      "loss": 0.1205,
+      "step": 3211
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5249464774545818e-05,
+      "loss": 0.1117,
+      "step": 3212
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5246508870378172e-05,
+      "loss": 0.1111,
+      "step": 3213
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243552333564229e-05,
+      "loss": 0.1243,
+      "step": 3214
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5240595164460497e-05,
+      "loss": 0.1236,
+      "step": 3215
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237637363423564e-05,
+      "loss": 0.1133,
+      "step": 3216
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5234678930810098e-05,
+      "loss": 0.1147,
+      "step": 3217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5231719866976837e-05,
+      "loss": 0.1051,
+      "step": 3218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228760172280597e-05,
+      "loss": 0.1101,
+      "step": 3219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225799847078273e-05,
+      "loss": 0.1158,
+      "step": 3220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522283889172683e-05,
+      "loss": 0.1036,
+      "step": 3221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5219877306583315e-05,
+      "loss": 0.1237,
+      "step": 3222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5216915092004847e-05,
+      "loss": 0.1134,
+      "step": 3223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5213952248348623e-05,
+      "loss": 0.1121,
+      "step": 3224
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210988775971917e-05,
+      "loss": 0.1201,
+      "step": 3225
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5208024675232077e-05,
+      "loss": 0.1179,
+      "step": 3226
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.520505994648652e-05,
+      "loss": 0.1149,
+      "step": 3227
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202094590092753e-05,
+      "loss": 0.1092,
+      "step": 3228
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199128606408353e-05,
+      "loss": 0.0935,
+      "step": 3229
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519616199579096e-05,
+      "loss": 0.1105,
+      "step": 3230
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5193194758598312e-05,
+      "loss": 0.1214,
+      "step": 3231
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5190226895188209e-05,
+      "loss": 0.1143,
+      "step": 3232
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187258405918523e-05,
+      "loss": 0.1177,
+      "step": 3233
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5184289291147209e-05,
+      "loss": 0.1082,
+      "step": 3234
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5181319551232298e-05,
+      "loss": 0.1096,
+      "step": 3235
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5178349186531896e-05,
+      "loss": 0.1243,
+      "step": 3236
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175378197404173e-05,
+      "loss": 0.1139,
+      "step": 3237
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5172406584207393e-05,
+      "loss": 0.121,
+      "step": 3238
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516943434729988e-05,
+      "loss": 0.108,
+      "step": 3239
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5166461487040043e-05,
+      "loss": 0.1143,
+      "step": 3240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163488003786358e-05,
+      "loss": 0.1268,
+      "step": 3241
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160513897897381e-05,
+      "loss": 0.1229,
+      "step": 3242
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157539169731747e-05,
+      "loss": 0.1287,
+      "step": 3243
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5154563819648156e-05,
+      "loss": 0.1146,
+      "step": 3244
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151587848005389e-05,
+      "loss": 0.1092,
+      "step": 3245
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148611255162305e-05,
+      "loss": 0.1048,
+      "step": 3246
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514563404147783e-05,
+      "loss": 0.13,
+      "step": 3247
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5142656207310971e-05,
+      "loss": 0.1101,
+      "step": 3248
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5139677753020807e-05,
+      "loss": 0.1118,
+      "step": 3249
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5136698678966494e-05,
+      "loss": 0.1112,
+      "step": 3250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5133718985507262e-05,
+      "loss": 0.1233,
+      "step": 3251
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513073867300241e-05,
+      "loss": 0.124,
+      "step": 3252
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5127757741811321e-05,
+      "loss": 0.1169,
+      "step": 3253
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124776192293449e-05,
+      "loss": 0.1082,
+      "step": 3254
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121794024808321e-05,
+      "loss": 0.1102,
+      "step": 3255
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118811239715538e-05,
+      "loss": 0.112,
+      "step": 3256
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115827837374778e-05,
+      "loss": 0.123,
+      "step": 3257
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5112843818145795e-05,
+      "loss": 0.1087,
+      "step": 3258
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109859182388408e-05,
+      "loss": 0.1181,
+      "step": 3259
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5106873930462525e-05,
+      "loss": 0.1111,
+      "step": 3260
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103888062728114e-05,
+      "loss": 0.1221,
+      "step": 3261
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5100901579545227e-05,
+      "loss": 0.1145,
+      "step": 3262
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097914481273983e-05,
+      "loss": 0.1211,
+      "step": 3263
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5094926768274585e-05,
+      "loss": 0.125,
+      "step": 3264
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.50919384409073e-05,
+      "loss": 0.1035,
+      "step": 3265
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5088949499532473e-05,
+      "loss": 0.1037,
+      "step": 3266
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5085959944510524e-05,
+      "loss": 0.1087,
+      "step": 3267
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082969776201948e-05,
+      "loss": 0.1233,
+      "step": 3268
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507997899496731e-05,
+      "loss": 0.1037,
+      "step": 3269
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507698760116725e-05,
+      "loss": 0.1168,
+      "step": 3270
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5073995595162489e-05,
+      "loss": 0.1232,
+      "step": 3271
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5071002977313808e-05,
+      "loss": 0.1081,
+      "step": 3272
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068009747982073e-05,
+      "loss": 0.1228,
+      "step": 3273
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506501590752822e-05,
+      "loss": 0.1131,
+      "step": 3274
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5062021456313262e-05,
+      "loss": 0.1103,
+      "step": 3275
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5059026394698277e-05,
+      "loss": 0.1031,
+      "step": 3276
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056030723044429e-05,
+      "loss": 0.1183,
+      "step": 3277
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5053034441712941e-05,
+      "loss": 0.1198,
+      "step": 3278
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5050037551065124e-05,
+      "loss": 0.114,
+      "step": 3279
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5047040051462352e-05,
+      "loss": 0.1267,
+      "step": 3280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5044041943266077e-05,
+      "loss": 0.1183,
+      "step": 3281
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041043226837824e-05,
+      "loss": 0.1085,
+      "step": 3282
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038043902539192e-05,
+      "loss": 0.1147,
+      "step": 3283
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503504397073185e-05,
+      "loss": 0.112,
+      "step": 3284
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5032043431777544e-05,
+      "loss": 0.1149,
+      "step": 3285
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5029042286038091e-05,
+      "loss": 0.1273,
+      "step": 3286
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5026040533875384e-05,
+      "loss": 0.1165,
+      "step": 3287
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023038175651383e-05,
+      "loss": 0.1209,
+      "step": 3288
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5020035211728127e-05,
+      "loss": 0.1121,
+      "step": 3289
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017031642467728e-05,
+      "loss": 0.1049,
+      "step": 3290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5014027468232369e-05,
+      "loss": 0.1258,
+      "step": 3291
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5011022689384301e-05,
+      "loss": 0.1031,
+      "step": 3292
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5008017306285858e-05,
+      "loss": 0.1038,
+      "step": 3293
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005011319299444e-05,
+      "loss": 0.1248,
+      "step": 3294
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5002004728787523e-05,
+      "loss": 0.1183,
+      "step": 3295
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4998997535112654e-05,
+      "loss": 0.1139,
+      "step": 3296
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995989738637455e-05,
+      "loss": 0.1133,
+      "step": 3297
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992981339724613e-05,
+      "loss": 0.1085,
+      "step": 3298
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4989972338736897e-05,
+      "loss": 0.106,
+      "step": 3299
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4986962736037147e-05,
+      "loss": 0.1209,
+      "step": 3300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4983952531988273e-05,
+      "loss": 0.12,
+      "step": 3301
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980941726953253e-05,
+      "loss": 0.1086,
+      "step": 3302
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4977930321295148e-05,
+      "loss": 0.1235,
+      "step": 3303
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4974918315377089e-05,
+      "loss": 0.1052,
+      "step": 3304
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4971905709562268e-05,
+      "loss": 0.1127,
+      "step": 3305
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968892504213963e-05,
+      "loss": 0.1128,
+      "step": 3306
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965878699695517e-05,
+      "loss": 0.1145,
+      "step": 3307
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4962864296370352e-05,
+      "loss": 0.121,
+      "step": 3308
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959849294601951e-05,
+      "loss": 0.1205,
+      "step": 3309
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4956833694753878e-05,
+      "loss": 0.1158,
+      "step": 3310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4953817497189774e-05,
+      "loss": 0.1127,
+      "step": 3311
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4950800702273332e-05,
+      "loss": 0.116,
+      "step": 3312
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4947783310368335e-05,
+      "loss": 0.1193,
+      "step": 3313
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494476532183864e-05,
+      "loss": 0.1151,
+      "step": 3314
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4941746737048159e-05,
+      "loss": 0.1176,
+      "step": 3315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.493872755636089e-05,
+      "loss": 0.117,
+      "step": 3316
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.49357077801409e-05,
+      "loss": 0.1016,
+      "step": 3317
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4932687408752325e-05,
+      "loss": 0.1282,
+      "step": 3318
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4929666442559374e-05,
+      "loss": 0.1185,
+      "step": 3319
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4926644881926326e-05,
+      "loss": 0.1011,
+      "step": 3320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923622727217539e-05,
+      "loss": 0.1059,
+      "step": 3321
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920599978797435e-05,
+      "loss": 0.1105,
+      "step": 3322
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4917576637030505e-05,
+      "loss": 0.1251,
+      "step": 3323
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491455270228132e-05,
+      "loss": 0.1235,
+      "step": 3324
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491152817491452e-05,
+      "loss": 0.1238,
+      "step": 3325
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4908503055294813e-05,
+      "loss": 0.1228,
+      "step": 3326
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905477343786985e-05,
+      "loss": 0.1117,
+      "step": 3327
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4902451040755886e-05,
+      "loss": 0.1095,
+      "step": 3328
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489942414656644e-05,
+      "loss": 0.1242,
+      "step": 3329
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4896396661583641e-05,
+      "loss": 0.1147,
+      "step": 3330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489336858617256e-05,
+      "loss": 0.1069,
+      "step": 3331
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890339920698334e-05,
+      "loss": 0.1134,
+      "step": 3332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4887310665526173e-05,
+      "loss": 0.1137,
+      "step": 3333
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884280821021356e-05,
+      "loss": 0.1035,
+      "step": 3334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4881250387549233e-05,
+      "loss": 0.1152,
+      "step": 3335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4878219365475228e-05,
+      "loss": 0.1199,
+      "step": 3336
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4875187755164837e-05,
+      "loss": 0.1181,
+      "step": 3337
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872155556983621e-05,
+      "loss": 0.1276,
+      "step": 3338
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869122771297217e-05,
+      "loss": 0.106,
+      "step": 3339
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486608939847133e-05,
+      "loss": 0.102,
+      "step": 3340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4863055438871741e-05,
+      "loss": 0.1099,
+      "step": 3341
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486002089286429e-05,
+      "loss": 0.1089,
+      "step": 3342
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4856985760814904e-05,
+      "loss": 0.1204,
+      "step": 3343
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853950043089566e-05,
+      "loss": 0.1058,
+      "step": 3344
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850913740054339e-05,
+      "loss": 0.1215,
+      "step": 3345
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4847876852075351e-05,
+      "loss": 0.1081,
+      "step": 3346
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4844839379518805e-05,
+      "loss": 0.1072,
+      "step": 3347
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4841801322750971e-05,
+      "loss": 0.1241,
+      "step": 3348
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838762682138188e-05,
+      "loss": 0.1045,
+      "step": 3349
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4835723458046875e-05,
+      "loss": 0.1113,
+      "step": 3350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.483268365084351e-05,
+      "loss": 0.1125,
+      "step": 3351
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829643260894644e-05,
+      "loss": 0.1131,
+      "step": 3352
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4826602288566903e-05,
+      "loss": 0.1065,
+      "step": 3353
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4823560734226985e-05,
+      "loss": 0.1164,
+      "step": 3354
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820518598241643e-05,
+      "loss": 0.1154,
+      "step": 3355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4817475880977718e-05,
+      "loss": 0.1265,
+      "step": 3356
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814432582802111e-05,
+      "loss": 0.113,
+      "step": 3357
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.48113887040818e-05,
+      "loss": 0.1082,
+      "step": 3358
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808344245183823e-05,
+      "loss": 0.1144,
+      "step": 3359
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4805299206475297e-05,
+      "loss": 0.1151,
+      "step": 3360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4802253588323405e-05,
+      "loss": 0.106,
+      "step": 3361
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.47992073910954e-05,
+      "loss": 0.1089,
+      "step": 3362
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4796160615158606e-05,
+      "loss": 0.1207,
+      "step": 3363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4793113260880416e-05,
+      "loss": 0.1127,
+      "step": 3364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790065328628296e-05,
+      "loss": 0.1171,
+      "step": 3365
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4787016818769773e-05,
+      "loss": 0.1097,
+      "step": 3366
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783967731672453e-05,
+      "loss": 0.1107,
+      "step": 3367
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780918067704006e-05,
+      "loss": 0.1189,
+      "step": 3368
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4777867827232176e-05,
+      "loss": 0.1213,
+      "step": 3369
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4774817010624769e-05,
+      "loss": 0.1182,
+      "step": 3370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771765618249669e-05,
+      "loss": 0.1029,
+      "step": 3371
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768713650474829e-05,
+      "loss": 0.113,
+      "step": 3372
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.476566110766826e-05,
+      "loss": 0.1155,
+      "step": 3373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762607990198057e-05,
+      "loss": 0.1311,
+      "step": 3374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759554298432375e-05,
+      "loss": 0.1195,
+      "step": 3375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756500032739446e-05,
+      "loss": 0.1122,
+      "step": 3376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753445193487558e-05,
+      "loss": 0.1267,
+      "step": 3377
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750389781045083e-05,
+      "loss": 0.1161,
+      "step": 3378
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747333795780456e-05,
+      "loss": 0.1112,
+      "step": 3379
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744277238062175e-05,
+      "loss": 0.1224,
+      "step": 3380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4741220108258815e-05,
+      "loss": 0.1134,
+      "step": 3381
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738162406739021e-05,
+      "loss": 0.1118,
+      "step": 3382
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.47351041338715e-05,
+      "loss": 0.1228,
+      "step": 3383
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732045290025032e-05,
+      "loss": 0.1128,
+      "step": 3384
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728985875568467e-05,
+      "loss": 0.1144,
+      "step": 3385
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472592589087072e-05,
+      "loss": 0.1215,
+      "step": 3386
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722865336300783e-05,
+      "loss": 0.106,
+      "step": 3387
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.47198042122277e-05,
+      "loss": 0.1096,
+      "step": 3388
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716742519020603e-05,
+      "loss": 0.134,
+      "step": 3389
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713680257048682e-05,
+      "loss": 0.1194,
+      "step": 3390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4710617426681195e-05,
+      "loss": 0.1128,
+      "step": 3391
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4707554028287472e-05,
+      "loss": 0.123,
+      "step": 3392
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4704490062236914e-05,
+      "loss": 0.1178,
+      "step": 3393
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701425528898983e-05,
+      "loss": 0.1172,
+      "step": 3394
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4698360428643215e-05,
+      "loss": 0.1211,
+      "step": 3395
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469529476183921e-05,
+      "loss": 0.1226,
+      "step": 3396
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692228528856647e-05,
+      "loss": 0.1117,
+      "step": 3397
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689161730065256e-05,
+      "loss": 0.1428,
+      "step": 3398
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4686094365834848e-05,
+      "loss": 0.109,
+      "step": 3399
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4683026436535302e-05,
+      "loss": 0.1196,
+      "step": 3400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467995794253656e-05,
+      "loss": 0.1242,
+      "step": 3401
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4676888884208631e-05,
+      "loss": 0.1157,
+      "step": 3402
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673819261921598e-05,
+      "loss": 0.1171,
+      "step": 3403
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4670749076045607e-05,
+      "loss": 0.1101,
+      "step": 3404
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466767832695088e-05,
+      "loss": 0.116,
+      "step": 3405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664607015007695e-05,
+      "loss": 0.1185,
+      "step": 3406
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4661535140586404e-05,
+      "loss": 0.1133,
+      "step": 3407
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465846270405743e-05,
+      "loss": 0.1063,
+      "step": 3408
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655389705791257e-05,
+      "loss": 0.1276,
+      "step": 3409
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652316146158441e-05,
+      "loss": 0.111,
+      "step": 3410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4649242025529607e-05,
+      "loss": 0.1178,
+      "step": 3411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646167344275444e-05,
+      "loss": 0.1097,
+      "step": 3412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464309210276671e-05,
+      "loss": 0.1228,
+      "step": 3413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4640016301374229e-05,
+      "loss": 0.1106,
+      "step": 3414
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4636939940468898e-05,
+      "loss": 0.1064,
+      "step": 3415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633863020421677e-05,
+      "loss": 0.1139,
+      "step": 3416
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463078554160359e-05,
+      "loss": 0.1159,
+      "step": 3417
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627707504385736e-05,
+      "loss": 0.114,
+      "step": 3418
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4624628909139277e-05,
+      "loss": 0.1172,
+      "step": 3419
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621549756235443e-05,
+      "loss": 0.1139,
+      "step": 3420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618470046045531e-05,
+      "loss": 0.1225,
+      "step": 3421
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4615389778940908e-05,
+      "loss": 0.112,
+      "step": 3422
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4612308955293e-05,
+      "loss": 0.1083,
+      "step": 3423
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460922757547331e-05,
+      "loss": 0.1221,
+      "step": 3424
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4606145639853403e-05,
+      "loss": 0.1248,
+      "step": 3425
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603063148804916e-05,
+      "loss": 0.1227,
+      "step": 3426
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4599980102699542e-05,
+      "loss": 0.107,
+      "step": 3427
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4596896501909051e-05,
+      "loss": 0.116,
+      "step": 3428
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4593812346805276e-05,
+      "loss": 0.1281,
+      "step": 3429
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4590727637760118e-05,
+      "loss": 0.122,
+      "step": 3430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587642375145545e-05,
+      "loss": 0.1182,
+      "step": 3431
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458455655933359e-05,
+      "loss": 0.1023,
+      "step": 3432
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581470190696355e-05,
+      "loss": 0.1148,
+      "step": 3433
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4578383269606004e-05,
+      "loss": 0.117,
+      "step": 3434
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4575295796434774e-05,
+      "loss": 0.1099,
+      "step": 3435
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4572207771554968e-05,
+      "loss": 0.1199,
+      "step": 3436
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4569119195338948e-05,
+      "loss": 0.1155,
+      "step": 3437
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4566030068159147e-05,
+      "loss": 0.1217,
+      "step": 3438
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562940390388072e-05,
+      "loss": 0.1231,
+      "step": 3439
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4559850162398284e-05,
+      "loss": 0.1117,
+      "step": 3440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556759384562418e-05,
+      "loss": 0.1107,
+      "step": 3441
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455366805725317e-05,
+      "loss": 0.1181,
+      "step": 3442
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550576180843308e-05,
+      "loss": 0.1239,
+      "step": 3443
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547483755705666e-05,
+      "loss": 0.1157,
+      "step": 3444
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544390782213138e-05,
+      "loss": 0.1114,
+      "step": 3445
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4541297260738687e-05,
+      "loss": 0.1144,
+      "step": 3446
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538203191655347e-05,
+      "loss": 0.1203,
+      "step": 3447
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453510857533621e-05,
+      "loss": 0.107,
+      "step": 3448
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453201341215444e-05,
+      "loss": 0.1131,
+      "step": 3449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528917702483264e-05,
+      "loss": 0.1132,
+      "step": 3450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4525821446695977e-05,
+      "loss": 0.1045,
+      "step": 3451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4522724645165939e-05,
+      "loss": 0.1121,
+      "step": 3452
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4519627298266574e-05,
+      "loss": 0.1157,
+      "step": 3453
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4516529406371371e-05,
+      "loss": 0.1166,
+      "step": 3454
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4513430969853895e-05,
+      "loss": 0.1189,
+      "step": 3455
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451033198908776e-05,
+      "loss": 0.112,
+      "step": 3456
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4507232464446657e-05,
+      "loss": 0.1193,
+      "step": 3457
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4504132396304344e-05,
+      "loss": 0.1238,
+      "step": 3458
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4501031785034634e-05,
+      "loss": 0.1097,
+      "step": 3459
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4497930631011415e-05,
+      "loss": 0.117,
+      "step": 3460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4494828934608636e-05,
+      "loss": 0.1226,
+      "step": 3461
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4491726696200315e-05,
+      "loss": 0.1141,
+      "step": 3462
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488623916160531e-05,
+      "loss": 0.1124,
+      "step": 3463
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485520594863431e-05,
+      "loss": 0.1065,
+      "step": 3464
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448241673268323e-05,
+      "loss": 0.1138,
+      "step": 3465
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4479312329994197e-05,
+      "loss": 0.1302,
+      "step": 3466
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4476207387170682e-05,
+      "loss": 0.1171,
+      "step": 3467
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473101904587086e-05,
+      "loss": 0.1145,
+      "step": 3468
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4469995882617887e-05,
+      "loss": 0.1218,
+      "step": 3469
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4466889321637618e-05,
+      "loss": 0.1277,
+      "step": 3470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446378222202088e-05,
+      "loss": 0.1133,
+      "step": 3471
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446067458414235e-05,
+      "loss": 0.1075,
+      "step": 3472
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4457566408376748e-05,
+      "loss": 0.1102,
+      "step": 3473
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4454457695098878e-05,
+      "loss": 0.1154,
+      "step": 3474
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4451348444683598e-05,
+      "loss": 0.0983,
+      "step": 3475
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444823865750584e-05,
+      "loss": 0.1221,
+      "step": 3476
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4445128333940588e-05,
+      "loss": 0.1119,
+      "step": 3477
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4442017474362904e-05,
+      "loss": 0.1096,
+      "step": 3478
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4438906079147905e-05,
+      "loss": 0.1128,
+      "step": 3479
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4435794148670781e-05,
+      "loss": 0.1303,
+      "step": 3480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432681683306777e-05,
+      "loss": 0.1092,
+      "step": 3481
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429568683431207e-05,
+      "loss": 0.1046,
+      "step": 3482
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426455149419455e-05,
+      "loss": 0.1012,
+      "step": 3483
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442334108164696e-05,
+      "loss": 0.1163,
+      "step": 3484
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4420226480489228e-05,
+      "loss": 0.1128,
+      "step": 3485
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4417111346321834e-05,
+      "loss": 0.1286,
+      "step": 3486
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4413995679520416e-05,
+      "loss": 0.1157,
+      "step": 3487
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441087948046067e-05,
+      "loss": 0.1181,
+      "step": 3488
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440776274951836e-05,
+      "loss": 0.1136,
+      "step": 3489
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4404645487069317e-05,
+      "loss": 0.1254,
+      "step": 3490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401527693489436e-05,
+      "loss": 0.1081,
+      "step": 3491
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4398409369154669e-05,
+      "loss": 0.1221,
+      "step": 3492
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4395290514441041e-05,
+      "loss": 0.1092,
+      "step": 3493
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4392171129724632e-05,
+      "loss": 0.1091,
+      "step": 3494
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389051215381595e-05,
+      "loss": 0.1226,
+      "step": 3495
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4385930771788138e-05,
+      "loss": 0.1215,
+      "step": 3496
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4382809799320545e-05,
+      "loss": 0.1112,
+      "step": 3497
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437968829835515e-05,
+      "loss": 0.1148,
+      "step": 3498
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4376566269268355e-05,
+      "loss": 0.1132,
+      "step": 3499
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373443712436632e-05,
+      "loss": 0.1105,
+      "step": 3500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4370320628236512e-05,
+      "loss": 0.1102,
+      "step": 3501
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436719701704459e-05,
+      "loss": 0.1161,
+      "step": 3502
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364072879237521e-05,
+      "loss": 0.1185,
+      "step": 3503
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360948215192025e-05,
+      "loss": 0.1146,
+      "step": 3504
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357823025284895e-05,
+      "loss": 0.1088,
+      "step": 3505
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354697309892974e-05,
+      "loss": 0.1161,
+      "step": 3506
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351571069393175e-05,
+      "loss": 0.108,
+      "step": 3507
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348444304162477e-05,
+      "loss": 0.111,
+      "step": 3508
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345317014577913e-05,
+      "loss": 0.1216,
+      "step": 3509
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4342189201016588e-05,
+      "loss": 0.1172,
+      "step": 3510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4339060863855666e-05,
+      "loss": 0.1115,
+      "step": 3511
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335932003472376e-05,
+      "loss": 0.1187,
+      "step": 3512
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332802620244009e-05,
+      "loss": 0.1137,
+      "step": 3513
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329672714547917e-05,
+      "loss": 0.1223,
+      "step": 3514
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4326542286761523e-05,
+      "loss": 0.1147,
+      "step": 3515
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4323411337262301e-05,
+      "loss": 0.1203,
+      "step": 3516
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4320279866427798e-05,
+      "loss": 0.116,
+      "step": 3517
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4317147874635617e-05,
+      "loss": 0.116,
+      "step": 3518
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4314015362263432e-05,
+      "loss": 0.1128,
+      "step": 3519
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4310882329688968e-05,
+      "loss": 0.1173,
+      "step": 3520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4307748777290025e-05,
+      "loss": 0.1241,
+      "step": 3521
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4304614705444455e-05,
+      "loss": 0.1086,
+      "step": 3522
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301480114530187e-05,
+      "loss": 0.1052,
+      "step": 3523
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429834500492519e-05,
+      "loss": 0.1156,
+      "step": 3524
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4295209377007517e-05,
+      "loss": 0.1172,
+      "step": 3525
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4292073231155277e-05,
+      "loss": 0.1134,
+      "step": 3526
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4288936567746631e-05,
+      "loss": 0.1217,
+      "step": 3527
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4285799387159819e-05,
+      "loss": 0.1171,
+      "step": 3528
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282661689773137e-05,
+      "loss": 0.109,
+      "step": 3529
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4279523475964936e-05,
+      "loss": 0.1044,
+      "step": 3530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276384746113636e-05,
+      "loss": 0.1089,
+      "step": 3531
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4273245500597721e-05,
+      "loss": 0.1136,
+      "step": 3532
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4270105739795735e-05,
+      "loss": 0.1157,
+      "step": 3533
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426696546408628e-05,
+      "loss": 0.1157,
+      "step": 3534
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4263824673848023e-05,
+      "loss": 0.1119,
+      "step": 3535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.42606833694597e-05,
+      "loss": 0.1153,
+      "step": 3536
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.42575415513001e-05,
+      "loss": 0.1219,
+      "step": 3537
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4254399219748074e-05,
+      "loss": 0.1156,
+      "step": 3538
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425125637518254e-05,
+      "loss": 0.1107,
+      "step": 3539
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4248113017982474e-05,
+      "loss": 0.1193,
+      "step": 3540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244969148526921e-05,
+      "loss": 0.1076,
+      "step": 3541
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4241824767194975e-05,
+      "loss": 0.1205,
+      "step": 3542
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238679874365802e-05,
+      "loss": 0.1042,
+      "step": 3543
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235534470418627e-05,
+      "loss": 0.1267,
+      "step": 3544
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4232388555732733e-05,
+      "loss": 0.1192,
+      "step": 3545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422924213068747e-05,
+      "loss": 0.0994,
+      "step": 3546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4226095195662248e-05,
+      "loss": 0.1282,
+      "step": 3547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422294775103654e-05,
+      "loss": 0.1246,
+      "step": 3548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219799797189872e-05,
+      "loss": 0.1113,
+      "step": 3549
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421665133450184e-05,
+      "loss": 0.1146,
+      "step": 3550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213502363352104e-05,
+      "loss": 0.1101,
+      "step": 3551
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210352884120374e-05,
+      "loss": 0.115,
+      "step": 3552
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.420720289718643e-05,
+      "loss": 0.1271,
+      "step": 3553
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204052402930109e-05,
+      "loss": 0.119,
+      "step": 3554
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4200901401731316e-05,
+      "loss": 0.1229,
+      "step": 3555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4197749893970006e-05,
+      "loss": 0.1073,
+      "step": 3556
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4194597880026203e-05,
+      "loss": 0.1084,
+      "step": 3557
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4191445360279991e-05,
+      "loss": 0.1167,
+      "step": 3558
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4188292335111517e-05,
+      "loss": 0.124,
+      "step": 3559
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4185138804900983e-05,
+      "loss": 0.1107,
+      "step": 3560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181984770028654e-05,
+      "loss": 0.1077,
+      "step": 3561
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417883023087486e-05,
+      "loss": 0.1141,
+      "step": 3562
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4175675187819987e-05,
+      "loss": 0.1152,
+      "step": 3563
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417251964124448e-05,
+      "loss": 0.1161,
+      "step": 3564
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169363591528856e-05,
+      "loss": 0.1145,
+      "step": 3565
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166207039053678e-05,
+      "loss": 0.1121,
+      "step": 3566
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416304998419958e-05,
+      "loss": 0.1135,
+      "step": 3567
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415989242734725e-05,
+      "loss": 0.1177,
+      "step": 3568
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156734368877445e-05,
+      "loss": 0.1176,
+      "step": 3569
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4153575809170974e-05,
+      "loss": 0.1199,
+      "step": 3570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4150416748608708e-05,
+      "loss": 0.1216,
+      "step": 3571
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147257187571582e-05,
+      "loss": 0.1067,
+      "step": 3572
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4144097126440588e-05,
+      "loss": 0.1137,
+      "step": 3573
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140936565596781e-05,
+      "loss": 0.1308,
+      "step": 3574
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4137775505421272e-05,
+      "loss": 0.1113,
+      "step": 3575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134613946295236e-05,
+      "loss": 0.13,
+      "step": 3576
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4131451888599915e-05,
+      "loss": 0.1178,
+      "step": 3577
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412828933271659e-05,
+      "loss": 0.1126,
+      "step": 3578
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4125126279026625e-05,
+      "loss": 0.111,
+      "step": 3579
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4121962727911428e-05,
+      "loss": 0.1136,
+      "step": 3580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411879867975248e-05,
+      "loss": 0.1139,
+      "step": 3581
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4115634134931312e-05,
+      "loss": 0.1091,
+      "step": 3582
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112469093829517e-05,
+      "loss": 0.1152,
+      "step": 3583
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4109303556828753e-05,
+      "loss": 0.1175,
+      "step": 3584
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410613752431073e-05,
+      "loss": 0.1218,
+      "step": 3585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102970996657224e-05,
+      "loss": 0.1105,
+      "step": 3586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4099803974250064e-05,
+      "loss": 0.1244,
+      "step": 3587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4096636457471151e-05,
+      "loss": 0.1102,
+      "step": 3588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4093468446702427e-05,
+      "loss": 0.1115,
+      "step": 3589
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4090299942325914e-05,
+      "loss": 0.1056,
+      "step": 3590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4087130944723681e-05,
+      "loss": 0.1228,
+      "step": 3591
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4083961454277857e-05,
+      "loss": 0.1118,
+      "step": 3592
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080791471370635e-05,
+      "loss": 0.1097,
+      "step": 3593
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077620996384263e-05,
+      "loss": 0.1203,
+      "step": 3594
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4074450029701048e-05,
+      "loss": 0.1027,
+      "step": 3595
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071278571703363e-05,
+      "loss": 0.1156,
+      "step": 3596
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4068106622773634e-05,
+      "loss": 0.1145,
+      "step": 3597
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4064934183294351e-05,
+      "loss": 0.1191,
+      "step": 3598
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4061761253648056e-05,
+      "loss": 0.1112,
+      "step": 3599
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058587834217356e-05,
+      "loss": 0.1309,
+      "step": 3600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4055413925384916e-05,
+      "loss": 0.1085,
+      "step": 3601
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405223952753346e-05,
+      "loss": 0.1119,
+      "step": 3602
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4049064641045767e-05,
+      "loss": 0.1074,
+      "step": 3603
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4045889266304682e-05,
+      "loss": 0.1229,
+      "step": 3604
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4042713403693105e-05,
+      "loss": 0.1074,
+      "step": 3605
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403953705359399e-05,
+      "loss": 0.1189,
+      "step": 3606
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4036360216390359e-05,
+      "loss": 0.1141,
+      "step": 3607
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403318289246529e-05,
+      "loss": 0.1073,
+      "step": 3608
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4030005082201915e-05,
+      "loss": 0.1208,
+      "step": 3609
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4026826785983427e-05,
+      "loss": 0.0955,
+      "step": 3610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4023648004193083e-05,
+      "loss": 0.1145,
+      "step": 3611
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4020468737214192e-05,
+      "loss": 0.1195,
+      "step": 3612
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401728898543012e-05,
+      "loss": 0.1142,
+      "step": 3613
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40141087492243e-05,
+      "loss": 0.1145,
+      "step": 3614
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010928028980212e-05,
+      "loss": 0.1196,
+      "step": 3615
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4007746825081411e-05,
+      "loss": 0.1163,
+      "step": 3616
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400456513791149e-05,
+      "loss": 0.1071,
+      "step": 3617
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4001382967854115e-05,
+      "loss": 0.1121,
+      "step": 3618
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3998200315293006e-05,
+      "loss": 0.1088,
+      "step": 3619
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995017180611941e-05,
+      "loss": 0.1146,
+      "step": 3620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991833564194752e-05,
+      "loss": 0.1208,
+      "step": 3621
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3988649466425334e-05,
+      "loss": 0.1123,
+      "step": 3622
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985464887687647e-05,
+      "loss": 0.1035,
+      "step": 3623
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398227982836569e-05,
+      "loss": 0.113,
+      "step": 3624
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3979094288843534e-05,
+      "loss": 0.1163,
+      "step": 3625
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3975908269505309e-05,
+      "loss": 0.122,
+      "step": 3626
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972721770735196e-05,
+      "loss": 0.1178,
+      "step": 3627
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3969534792917436e-05,
+      "loss": 0.123,
+      "step": 3628
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3966347336436327e-05,
+      "loss": 0.1161,
+      "step": 3629
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3963159401676229e-05,
+      "loss": 0.1244,
+      "step": 3630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3959970989021558e-05,
+      "loss": 0.1146,
+      "step": 3631
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3956782098856784e-05,
+      "loss": 0.1165,
+      "step": 3632
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953592731566435e-05,
+      "loss": 0.1058,
+      "step": 3633
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3950402887535101e-05,
+      "loss": 0.124,
+      "step": 3634
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3947212567147426e-05,
+      "loss": 0.1086,
+      "step": 3635
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3944021770788111e-05,
+      "loss": 0.115,
+      "step": 3636
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3940830498841922e-05,
+      "loss": 0.129,
+      "step": 3637
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937638751693669e-05,
+      "loss": 0.113,
+      "step": 3638
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393444652972823e-05,
+      "loss": 0.1341,
+      "step": 3639
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931253833330534e-05,
+      "loss": 0.1119,
+      "step": 3640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3928060662885572e-05,
+      "loss": 0.1125,
+      "step": 3641
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.392486701877839e-05,
+      "loss": 0.1165,
+      "step": 3642
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921672901394088e-05,
+      "loss": 0.1081,
+      "step": 3643
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918478311117831e-05,
+      "loss": 0.115,
+      "step": 3644
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3915283248334834e-05,
+      "loss": 0.1169,
+      "step": 3645
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912087713430374e-05,
+      "loss": 0.1278,
+      "step": 3646
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3908891706789775e-05,
+      "loss": 0.1135,
+      "step": 3647
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905695228798433e-05,
+      "loss": 0.1203,
+      "step": 3648
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902498279841787e-05,
+      "loss": 0.1163,
+      "step": 3649
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3899300860305343e-05,
+      "loss": 0.1203,
+      "step": 3650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3896102970574657e-05,
+      "loss": 0.138,
+      "step": 3651
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3892904611035344e-05,
+      "loss": 0.1053,
+      "step": 3652
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3889705782073078e-05,
+      "loss": 0.1126,
+      "step": 3653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3886506484073583e-05,
+      "loss": 0.1087,
+      "step": 3654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883306717422649e-05,
+      "loss": 0.1121,
+      "step": 3655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3880106482506116e-05,
+      "loss": 0.1126,
+      "step": 3656
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876905779709877e-05,
+      "loss": 0.1199,
+      "step": 3657
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873704609419894e-05,
+      "loss": 0.1066,
+      "step": 3658
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870502972022175e-05,
+      "loss": 0.1054,
+      "step": 3659
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867300867902782e-05,
+      "loss": 0.1151,
+      "step": 3660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3864098297447845e-05,
+      "loss": 0.1121,
+      "step": 3661
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3860895261043541e-05,
+      "loss": 0.1187,
+      "step": 3662
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3857691759076106e-05,
+      "loss": 0.1139,
+      "step": 3663
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385448779193183e-05,
+      "loss": 0.123,
+      "step": 3664
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3851283359997064e-05,
+      "loss": 0.1019,
+      "step": 3665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3848078463658207e-05,
+      "loss": 0.1172,
+      "step": 3666
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844873103301726e-05,
+      "loss": 0.1243,
+      "step": 3667
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841667279314132e-05,
+      "loss": 0.1145,
+      "step": 3668
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3838460992081998e-05,
+      "loss": 0.1017,
+      "step": 3669
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3835254241991955e-05,
+      "loss": 0.1182,
+      "step": 3670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832047029430679e-05,
+      "loss": 0.118,
+      "step": 3671
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3828839354784917e-05,
+      "loss": 0.1212,
+      "step": 3672
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382563121844146e-05,
+      "loss": 0.1058,
+      "step": 3673
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3822422620787159e-05,
+      "loss": 0.1242,
+      "step": 3674
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381921356220892e-05,
+      "loss": 0.1153,
+      "step": 3675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3816004043093706e-05,
+      "loss": 0.1055,
+      "step": 3676
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3812794063828537e-05,
+      "loss": 0.1123,
+      "step": 3677
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380958362480048e-05,
+      "loss": 0.1109,
+      "step": 3678
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3806372726396666e-05,
+      "loss": 0.1157,
+      "step": 3679
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3803161369004283e-05,
+      "loss": 0.1065,
+      "step": 3680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799949553010563e-05,
+      "loss": 0.1186,
+      "step": 3681
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796737278802805e-05,
+      "loss": 0.1178,
+      "step": 3682
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3793524546768358e-05,
+      "loss": 0.1128,
+      "step": 3683
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3790311357294626e-05,
+      "loss": 0.1145,
+      "step": 3684
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3787097710769069e-05,
+      "loss": 0.1156,
+      "step": 3685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.37838836075792e-05,
+      "loss": 0.1175,
+      "step": 3686
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3780669048112593e-05,
+      "loss": 0.1043,
+      "step": 3687
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777454032756874e-05,
+      "loss": 0.12,
+      "step": 3688
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3774238561899717e-05,
+      "loss": 0.1161,
+      "step": 3689
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3771022635928866e-05,
+      "loss": 0.1113,
+      "step": 3690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3767806255232103e-05,
+      "loss": 0.1199,
+      "step": 3691
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3764589420197277e-05,
+      "loss": 0.1145,
+      "step": 3692
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3761372131212285e-05,
+      "loss": 0.1232,
+      "step": 3693
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3758154388665083e-05,
+      "loss": 0.1233,
+      "step": 3694
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754936192943681e-05,
+      "loss": 0.1204,
+      "step": 3695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3751717544436143e-05,
+      "loss": 0.1181,
+      "step": 3696
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748498443530582e-05,
+      "loss": 0.1171,
+      "step": 3697
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3745278890615176e-05,
+      "loss": 0.0991,
+      "step": 3698
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3742058886078156e-05,
+      "loss": 0.1144,
+      "step": 3699
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3738838430307793e-05,
+      "loss": 0.1093,
+      "step": 3700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373561752369243e-05,
+      "loss": 0.1187,
+      "step": 3701
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373239616662046e-05,
+      "loss": 0.1192,
+      "step": 3702
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3729174359480323e-05,
+      "loss": 0.1113,
+      "step": 3703
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725952102660518e-05,
+      "loss": 0.1203,
+      "step": 3704
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3722729396549604e-05,
+      "loss": 0.1174,
+      "step": 3705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3719506241536183e-05,
+      "loss": 0.1184,
+      "step": 3706
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3716282638008918e-05,
+      "loss": 0.1092,
+      "step": 3707
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3713058586356524e-05,
+      "loss": 0.1152,
+      "step": 3708
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3709834086967775e-05,
+      "loss": 0.1213,
+      "step": 3709
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3706609140231488e-05,
+      "loss": 0.1044,
+      "step": 3710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3703383746536547e-05,
+      "loss": 0.117,
+      "step": 3711
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700157906271882e-05,
+      "loss": 0.1161,
+      "step": 3712
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696931619826479e-05,
+      "loss": 0.1104,
+      "step": 3713
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3693704887589374e-05,
+      "loss": 0.1189,
+      "step": 3714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3690477709949661e-05,
+      "loss": 0.1148,
+      "step": 3715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3687250087296488e-05,
+      "loss": 0.1308,
+      "step": 3716
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3684022020019061e-05,
+      "loss": 0.1195,
+      "step": 3717
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3680793508506622e-05,
+      "loss": 0.112,
+      "step": 3718
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3677564553148487e-05,
+      "loss": 0.1185,
+      "step": 3719
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3674335154334017e-05,
+      "loss": 0.1126,
+      "step": 3720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3671105312452623e-05,
+      "loss": 0.1115,
+      "step": 3721
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667875027893776e-05,
+      "loss": 0.1248,
+      "step": 3722
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3664644301046997e-05,
+      "loss": 0.1108,
+      "step": 3723
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3661413132301859e-05,
+      "loss": 0.1142,
+      "step": 3724
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365818152204799e-05,
+      "loss": 0.1106,
+      "step": 3725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3654949470675071e-05,
+      "loss": 0.1204,
+      "step": 3726
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651716978572841e-05,
+      "loss": 0.108,
+      "step": 3727
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3648484046131086e-05,
+      "loss": 0.1195,
+      "step": 3728
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3645250673739643e-05,
+      "loss": 0.1059,
+      "step": 3729
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3642016861788405e-05,
+      "loss": 0.1005,
+      "step": 3730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3638782610667327e-05,
+      "loss": 0.1026,
+      "step": 3731
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635547920766402e-05,
+      "loss": 0.1123,
+      "step": 3732
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3632312792475682e-05,
+      "loss": 0.1156,
+      "step": 3733
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3629077226185271e-05,
+      "loss": 0.1139,
+      "step": 3734
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3625841222285338e-05,
+      "loss": 0.1006,
+      "step": 3735
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622604781166085e-05,
+      "loss": 0.1165,
+      "step": 3736
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3619367903217774e-05,
+      "loss": 0.1199,
+      "step": 3737
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3616130588830732e-05,
+      "loss": 0.1146,
+      "step": 3738
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3612892838395315e-05,
+      "loss": 0.1105,
+      "step": 3739
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609654652301953e-05,
+      "loss": 0.1088,
+      "step": 3740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3606416030941119e-05,
+      "loss": 0.1164,
+      "step": 3741
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3603176974703338e-05,
+      "loss": 0.118,
+      "step": 3742
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3599937483979191e-05,
+      "loss": 0.1121,
+      "step": 3743
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596697559159306e-05,
+      "loss": 0.114,
+      "step": 3744
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3593457200634372e-05,
+      "loss": 0.1148,
+      "step": 3745
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3590216408795124e-05,
+      "loss": 0.1155,
+      "step": 3746
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3586975184032345e-05,
+      "loss": 0.1099,
+      "step": 3747
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3583733526736881e-05,
+      "loss": 0.1217,
+      "step": 3748
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3580491437299628e-05,
+      "loss": 0.1145,
+      "step": 3749
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.357724891611152e-05,
+      "loss": 0.1219,
+      "step": 3750
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574005963563564e-05,
+      "loss": 0.127,
+      "step": 3751
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3570762580046804e-05,
+      "loss": 0.118,
+      "step": 3752
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3567518765952344e-05,
+      "loss": 0.1139,
+      "step": 3753
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3564274521671333e-05,
+      "loss": 0.1154,
+      "step": 3754
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561029847594978e-05,
+      "loss": 0.1129,
+      "step": 3755
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3557784744114536e-05,
+      "loss": 0.103,
+      "step": 3756
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3554539211621314e-05,
+      "loss": 0.1053,
+      "step": 3757
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551293250506675e-05,
+      "loss": 0.116,
+      "step": 3758
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3548046861162028e-05,
+      "loss": 0.1152,
+      "step": 3759
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3544800043978836e-05,
+      "loss": 0.1223,
+      "step": 3760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3541552799348614e-05,
+      "loss": 0.1192,
+      "step": 3761
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3538305127662931e-05,
+      "loss": 0.1194,
+      "step": 3762
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535057029313406e-05,
+      "loss": 0.1171,
+      "step": 3763
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3531808504691703e-05,
+      "loss": 0.1123,
+      "step": 3764
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3528559554189546e-05,
+      "loss": 0.1073,
+      "step": 3765
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3525310178198707e-05,
+      "loss": 0.1115,
+      "step": 3766
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3522060377111012e-05,
+      "loss": 0.1102,
+      "step": 3767
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.351881015131833e-05,
+      "loss": 0.1214,
+      "step": 3768
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3515559501212594e-05,
+      "loss": 0.1139,
+      "step": 3769
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.351230842718578e-05,
+      "loss": 0.1118,
+      "step": 3770
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3509056929629909e-05,
+      "loss": 0.1238,
+      "step": 3771
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3505805008937066e-05,
+      "loss": 0.1144,
+      "step": 3772
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502552665499379e-05,
+      "loss": 0.1166,
+      "step": 3773
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499299899709037e-05,
+      "loss": 0.1167,
+      "step": 3774
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3496046711958263e-05,
+      "loss": 0.1161,
+      "step": 3775
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492793102639343e-05,
+      "loss": 0.1288,
+      "step": 3776
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3489539072144615e-05,
+      "loss": 0.1184,
+      "step": 3777
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3486284620866456e-05,
+      "loss": 0.1203,
+      "step": 3778
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348302974919731e-05,
+      "loss": 0.1173,
+      "step": 3779
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3479774457529656e-05,
+      "loss": 0.1116,
+      "step": 3780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3476518746256038e-05,
+      "loss": 0.1148,
+      "step": 3781
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3473262615769035e-05,
+      "loss": 0.1119,
+      "step": 3782
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3470006066461291e-05,
+      "loss": 0.12,
+      "step": 3783
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466749098725495e-05,
+      "loss": 0.1204,
+      "step": 3784
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463491712954384e-05,
+      "loss": 0.1113,
+      "step": 3785
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3460233909540747e-05,
+      "loss": 0.1223,
+      "step": 3786
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3456975688877424e-05,
+      "loss": 0.1299,
+      "step": 3787
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3453717051357306e-05,
+      "loss": 0.1107,
+      "step": 3788
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3450457997373332e-05,
+      "loss": 0.1094,
+      "step": 3789
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447198527318493e-05,
+      "loss": 0.119,
+      "step": 3790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344393864158583e-05,
+      "loss": 0.1309,
+      "step": 3791
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3440678340568436e-05,
+      "loss": 0.1098,
+      "step": 3792
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3437417624659447e-05,
+      "loss": 0.1162,
+      "step": 3793
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3434156494252057e-05,
+      "loss": 0.1119,
+      "step": 3794
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3430894949739508e-05,
+      "loss": 0.1089,
+      "step": 3795
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342763299151509e-05,
+      "loss": 0.1168,
+      "step": 3796
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342437061997214e-05,
+      "loss": 0.1099,
+      "step": 3797
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3421107835504055e-05,
+      "loss": 0.122,
+      "step": 3798
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417844638504273e-05,
+      "loss": 0.1123,
+      "step": 3799
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3414581029366283e-05,
+      "loss": 0.1123,
+      "step": 3800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411317008483624e-05,
+      "loss": 0.11,
+      "step": 3801
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3408052576249885e-05,
+      "loss": 0.1264,
+      "step": 3802
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.340478773305871e-05,
+      "loss": 0.1102,
+      "step": 3803
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3401522479303784e-05,
+      "loss": 0.1284,
+      "step": 3804
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3398256815378845e-05,
+      "loss": 0.1196,
+      "step": 3805
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3394990741677682e-05,
+      "loss": 0.1229,
+      "step": 3806
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3391724258594132e-05,
+      "loss": 0.1176,
+      "step": 3807
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.338845736652208e-05,
+      "loss": 0.1051,
+      "step": 3808
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.338519006585546e-05,
+      "loss": 0.112,
+      "step": 3809
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381922356988261e-05,
+      "loss": 0.1067,
+      "step": 3810
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3378654240314516e-05,
+      "loss": 0.1129,
+      "step": 3811
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3375385716228309e-05,
+      "loss": 0.1158,
+      "step": 3812
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337211678512377e-05,
+      "loss": 0.1092,
+      "step": 3813
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3368847447395083e-05,
+      "loss": 0.1062,
+      "step": 3814
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3365577703436477e-05,
+      "loss": 0.1148,
+      "step": 3815
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336230755364223e-05,
+      "loss": 0.1159,
+      "step": 3816
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3359036998406678e-05,
+      "loss": 0.1261,
+      "step": 3817
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335576603812419e-05,
+      "loss": 0.1139,
+      "step": 3818
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352494673189194e-05,
+      "loss": 0.1101,
+      "step": 3819
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349222903996168e-05,
+      "loss": 0.1038,
+      "step": 3820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3345950730939637e-05,
+      "loss": 0.1281,
+      "step": 3821
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342678154414169e-05,
+      "loss": 0.1223,
+      "step": 3822
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3339405174814383e-05,
+      "loss": 0.1236,
+      "step": 3823
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336131792534957e-05,
+      "loss": 0.1193,
+      "step": 3824
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3332858007970606e-05,
+      "loss": 0.1043,
+      "step": 3825
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329583821516094e-05,
+      "loss": 0.1354,
+      "step": 3826
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326309233566236e-05,
+      "loss": 0.1153,
+      "step": 3827
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3323034244515905e-05,
+      "loss": 0.1079,
+      "step": 3828
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3319758854760001e-05,
+      "loss": 0.1147,
+      "step": 3829
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3316483064693489e-05,
+      "loss": 0.0989,
+      "step": 3830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3313206874711378e-05,
+      "loss": 0.1129,
+      "step": 3831
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309930285208727e-05,
+      "loss": 0.1115,
+      "step": 3832
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3306653296580639e-05,
+      "loss": 0.1051,
+      "step": 3833
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3303375909222264e-05,
+      "loss": 0.1048,
+      "step": 3834
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3300098123528811e-05,
+      "loss": 0.1236,
+      "step": 3835
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3296819939895522e-05,
+      "loss": 0.1274,
+      "step": 3836
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3293541358717699e-05,
+      "loss": 0.1135,
+      "step": 3837
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290262380390682e-05,
+      "loss": 0.1149,
+      "step": 3838
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.328698300530987e-05,
+      "loss": 0.1103,
+      "step": 3839
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3283703233870702e-05,
+      "loss": 0.1094,
+      "step": 3840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3280423066468661e-05,
+      "loss": 0.1058,
+      "step": 3841
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3277142503499296e-05,
+      "loss": 0.1167,
+      "step": 3842
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3273861545358177e-05,
+      "loss": 0.1255,
+      "step": 3843
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3270580192440947e-05,
+      "loss": 0.0968,
+      "step": 3844
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267298445143282e-05,
+      "loss": 0.1101,
+      "step": 3845
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3264016303860907e-05,
+      "loss": 0.1181,
+      "step": 3846
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3260733768989598e-05,
+      "loss": 0.1188,
+      "step": 3847
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3257450840925175e-05,
+      "loss": 0.1081,
+      "step": 3848
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.325416752006351e-05,
+      "loss": 0.1266,
+      "step": 3849
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.325088380680052e-05,
+      "loss": 0.0983,
+      "step": 3850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3247599701532166e-05,
+      "loss": 0.1167,
+      "step": 3851
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244315204654462e-05,
+      "loss": 0.1197,
+      "step": 3852
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241030316563466e-05,
+      "loss": 0.1167,
+      "step": 3853
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3237745037655283e-05,
+      "loss": 0.1124,
+      "step": 3854
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3234459368326066e-05,
+      "loss": 0.1207,
+      "step": 3855
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323117330897202e-05,
+      "loss": 0.1173,
+      "step": 3856
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3227886859989383e-05,
+      "loss": 0.1152,
+      "step": 3857
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3224600021774454e-05,
+      "loss": 0.109,
+      "step": 3858
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221312794723572e-05,
+      "loss": 0.1119,
+      "step": 3859
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.321802517923313e-05,
+      "loss": 0.0966,
+      "step": 3860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3214737175699555e-05,
+      "loss": 0.1105,
+      "step": 3861
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211448784519332e-05,
+      "loss": 0.1314,
+      "step": 3862
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320816000608899e-05,
+      "loss": 0.1215,
+      "step": 3863
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3204870840805104e-05,
+      "loss": 0.1117,
+      "step": 3864
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3201581289064293e-05,
+      "loss": 0.119,
+      "step": 3865
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3198291351263229e-05,
+      "loss": 0.1073,
+      "step": 3866
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3195001027798624e-05,
+      "loss": 0.1149,
+      "step": 3867
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191710319067237e-05,
+      "loss": 0.1238,
+      "step": 3868
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.318841922546588e-05,
+      "loss": 0.1089,
+      "step": 3869
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3185127747391404e-05,
+      "loss": 0.1149,
+      "step": 3870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3181835885240712e-05,
+      "loss": 0.1184,
+      "step": 3871
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3178543639410747e-05,
+      "loss": 0.1117,
+      "step": 3872
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3175251010298506e-05,
+      "loss": 0.1204,
+      "step": 3873
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3171957998301028e-05,
+      "loss": 0.1134,
+      "step": 3874
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3168664603815395e-05,
+      "loss": 0.1065,
+      "step": 3875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316537082723874e-05,
+      "loss": 0.1315,
+      "step": 3876
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316207666896824e-05,
+      "loss": 0.1243,
+      "step": 3877
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158782129401123e-05,
+      "loss": 0.1113,
+      "step": 3878
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155487208934652e-05,
+      "loss": 0.1133,
+      "step": 3879
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3152191907966146e-05,
+      "loss": 0.111,
+      "step": 3880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3148896226892962e-05,
+      "loss": 0.1111,
+      "step": 3881
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145600166112517e-05,
+      "loss": 0.1084,
+      "step": 3882
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3142303726022257e-05,
+      "loss": 0.107,
+      "step": 3883
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3139006907019679e-05,
+      "loss": 0.1112,
+      "step": 3884
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3135709709502332e-05,
+      "loss": 0.114,
+      "step": 3885
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3132412133867805e-05,
+      "loss": 0.1177,
+      "step": 3886
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3129114180513728e-05,
+      "loss": 0.1325,
+      "step": 3887
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312581584983779e-05,
+      "loss": 0.1208,
+      "step": 3888
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3122517142237718e-05,
+      "loss": 0.1173,
+      "step": 3889
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3119218058111279e-05,
+      "loss": 0.1227,
+      "step": 3890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115918597856292e-05,
+      "loss": 0.1165,
+      "step": 3891
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311261876187062e-05,
+      "loss": 0.1094,
+      "step": 3892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3109318550552172e-05,
+      "loss": 0.1305,
+      "step": 3893
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3106017964298903e-05,
+      "loss": 0.1125,
+      "step": 3894
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102717003508807e-05,
+      "loss": 0.1137,
+      "step": 3895
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099415668579934e-05,
+      "loss": 0.1209,
+      "step": 3896
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3096113959910366e-05,
+      "loss": 0.1107,
+      "step": 3897
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3092811877898241e-05,
+      "loss": 0.1251,
+      "step": 3898
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3089509422941737e-05,
+      "loss": 0.1288,
+      "step": 3899
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3086206595439085e-05,
+      "loss": 0.1038,
+      "step": 3900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082903395788541e-05,
+      "loss": 0.1241,
+      "step": 3901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.307959982438843e-05,
+      "loss": 0.1175,
+      "step": 3902
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3076295881637105e-05,
+      "loss": 0.1064,
+      "step": 3903
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3072991567932969e-05,
+      "loss": 0.108,
+      "step": 3904
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306968688367447e-05,
+      "loss": 0.119,
+      "step": 3905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3066381829260106e-05,
+      "loss": 0.1351,
+      "step": 3906
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3063076405088413e-05,
+      "loss": 0.1227,
+      "step": 3907
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059770611557969e-05,
+      "loss": 0.1139,
+      "step": 3908
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3056464449067401e-05,
+      "loss": 0.1092,
+      "step": 3909
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3053157918015384e-05,
+      "loss": 0.1218,
+      "step": 3910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049851018800633e-05,
+      "loss": 0.12,
+      "step": 3911
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3046543751821903e-05,
+      "loss": 0.1242,
+      "step": 3912
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043236117478002e-05,
+      "loss": 0.1138,
+      "step": 3913
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3039928116167782e-05,
+      "loss": 0.1158,
+      "step": 3914
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3036619748290129e-05,
+      "loss": 0.1048,
+      "step": 3915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033311014243983e-05,
+      "loss": 0.1129,
+      "step": 3916
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3030001914428328e-05,
+      "loss": 0.1166,
+      "step": 3917
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026692449242184e-05,
+      "loss": 0.1236,
+      "step": 3918
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3023382619084623e-05,
+      "loss": 0.1158,
+      "step": 3919
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302007242435476e-05,
+      "loss": 0.1162,
+      "step": 3920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301676186545175e-05,
+      "loss": 0.1247,
+      "step": 3921
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3013450942774797e-05,
+      "loss": 0.1306,
+      "step": 3922
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301013965672314e-05,
+      "loss": 0.1091,
+      "step": 3923
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3006828007696072e-05,
+      "loss": 0.1101,
+      "step": 3924
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3003515996092927e-05,
+      "loss": 0.1204,
+      "step": 3925
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000203622313081e-05,
+      "loss": 0.1105,
+      "step": 3926
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996890886755953e-05,
+      "loss": 0.1248,
+      "step": 3927
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2993577789821007e-05,
+      "loss": 0.1184,
+      "step": 3928
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299026433190775e-05,
+      "loss": 0.1221,
+      "step": 3929
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2986950513415732e-05,
+      "loss": 0.1017,
+      "step": 3930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298363633474455e-05,
+      "loss": 0.1199,
+      "step": 3931
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2980321796293838e-05,
+      "loss": 0.1186,
+      "step": 3932
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.297700689846328e-05,
+      "loss": 0.1187,
+      "step": 3933
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973691641652599e-05,
+      "loss": 0.1049,
+      "step": 3934
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2970376026261563e-05,
+      "loss": 0.1177,
+      "step": 3935
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2967060052689987e-05,
+      "loss": 0.1136,
+      "step": 3936
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963743721337715e-05,
+      "loss": 0.1038,
+      "step": 3937
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2960427032604654e-05,
+      "loss": 0.1148,
+      "step": 3938
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2957109986890743e-05,
+      "loss": 0.1149,
+      "step": 3939
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2953792584595961e-05,
+      "loss": 0.1149,
+      "step": 3940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2950474826120335e-05,
+      "loss": 0.114,
+      "step": 3941
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294715671186394e-05,
+      "loss": 0.1141,
+      "step": 3942
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943838242226885e-05,
+      "loss": 0.1181,
+      "step": 3943
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940519417609323e-05,
+      "loss": 0.132,
+      "step": 3944
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2937200238411454e-05,
+      "loss": 0.1292,
+      "step": 3945
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2933880705033519e-05,
+      "loss": 0.1125,
+      "step": 3946
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29305608178758e-05,
+      "loss": 0.1207,
+      "step": 3947
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2927240577338624e-05,
+      "loss": 0.123,
+      "step": 3948
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292391998382236e-05,
+      "loss": 0.1187,
+      "step": 3949
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2920599037727422e-05,
+      "loss": 0.1168,
+      "step": 3950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2917277739454255e-05,
+      "loss": 0.1069,
+      "step": 3951
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2913956089403364e-05,
+      "loss": 0.1224,
+      "step": 3952
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2910634087975287e-05,
+      "loss": 0.1273,
+      "step": 3953
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2907311735570599e-05,
+      "loss": 0.1203,
+      "step": 3954
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2903989032589925e-05,
+      "loss": 0.1176,
+      "step": 3955
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2900665979433935e-05,
+      "loss": 0.1096,
+      "step": 3956
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897342576503337e-05,
+      "loss": 0.1141,
+      "step": 3957
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2894018824198877e-05,
+      "loss": 0.1051,
+      "step": 3958
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2890694722921347e-05,
+      "loss": 0.1096,
+      "step": 3959
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2887370273071582e-05,
+      "loss": 0.1121,
+      "step": 3960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2884045475050462e-05,
+      "loss": 0.1213,
+      "step": 3961
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2880720329258903e-05,
+      "loss": 0.115,
+      "step": 3962
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.287739483609786e-05,
+      "loss": 0.1138,
+      "step": 3963
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2874068995968345e-05,
+      "loss": 0.1093,
+      "step": 3964
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2870742809271392e-05,
+      "loss": 0.1196,
+      "step": 3965
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2867416276408093e-05,
+      "loss": 0.124,
+      "step": 3966
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2864089397779573e-05,
+      "loss": 0.1135,
+      "step": 3967
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2860762173787005e-05,
+      "loss": 0.1184,
+      "step": 3968
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2857434604831594e-05,
+      "loss": 0.1104,
+      "step": 3969
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854106691314594e-05,
+      "loss": 0.1093,
+      "step": 3970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2850778433637303e-05,
+      "loss": 0.1184,
+      "step": 3971
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284744983220105e-05,
+      "loss": 0.107,
+      "step": 3972
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2844120887407216e-05,
+      "loss": 0.1175,
+      "step": 3973
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284079159965722e-05,
+      "loss": 0.1125,
+      "step": 3974
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283746196935252e-05,
+      "loss": 0.1209,
+      "step": 3975
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2834131996894617e-05,
+      "loss": 0.1263,
+      "step": 3976
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830801682685052e-05,
+      "loss": 0.127,
+      "step": 3977
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2827471027125411e-05,
+      "loss": 0.1113,
+      "step": 3978
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824140030617318e-05,
+      "loss": 0.1062,
+      "step": 3979
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820808693562437e-05,
+      "loss": 0.1232,
+      "step": 3980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2817477016362477e-05,
+      "loss": 0.1113,
+      "step": 3981
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2814144999419182e-05,
+      "loss": 0.1157,
+      "step": 3982
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810812643134345e-05,
+      "loss": 0.1157,
+      "step": 3983
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2807479947909794e-05,
+      "loss": 0.1125,
+      "step": 3984
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2804146914147397e-05,
+      "loss": 0.1233,
+      "step": 3985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800813542249073e-05,
+      "loss": 0.1047,
+      "step": 3986
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2797479832616763e-05,
+      "loss": 0.1175,
+      "step": 3987
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2794145785652469e-05,
+      "loss": 0.1212,
+      "step": 3988
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790811401758221e-05,
+      "loss": 0.1123,
+      "step": 3989
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2787476681336095e-05,
+      "loss": 0.1173,
+      "step": 3990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2784141624788203e-05,
+      "loss": 0.1139,
+      "step": 3991
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.27808062325167e-05,
+      "loss": 0.1117,
+      "step": 3992
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2777470504923786e-05,
+      "loss": 0.1194,
+      "step": 3993
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2774134442411691e-05,
+      "loss": 0.1196,
+      "step": 3994
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2770798045382698e-05,
+      "loss": 0.1093,
+      "step": 3995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2767461314239118e-05,
+      "loss": 0.1225,
+      "step": 3996
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2764124249383314e-05,
+      "loss": 0.1176,
+      "step": 3997
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276078685121768e-05,
+      "loss": 0.1171,
+      "step": 3998
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2757449120144654e-05,
+      "loss": 0.1238,
+      "step": 3999
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2754111056566714e-05,
+      "loss": 0.1037,
+      "step": 4000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2750772660886376e-05,
+      "loss": 0.098,
+      "step": 4001
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2747433933506202e-05,
+      "loss": 0.117,
+      "step": 4002
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2744094874828785e-05,
+      "loss": 0.1102,
+      "step": 4003
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2740755485256768e-05,
+      "loss": 0.1071,
+      "step": 4004
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737415765192826e-05,
+      "loss": 0.111,
+      "step": 4005
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2734075715039675e-05,
+      "loss": 0.1086,
+      "step": 4006
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2730735335200076e-05,
+      "loss": 0.1145,
+      "step": 4007
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2727394626076823e-05,
+      "loss": 0.0969,
+      "step": 4008
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2724053588072756e-05,
+      "loss": 0.1224,
+      "step": 4009
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2720712221590747e-05,
+      "loss": 0.123,
+      "step": 4010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717370527033721e-05,
+      "loss": 0.1177,
+      "step": 4011
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2714028504804623e-05,
+      "loss": 0.1168,
+      "step": 4012
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2710686155306452e-05,
+      "loss": 0.1175,
+      "step": 4013
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270734347894225e-05,
+      "loss": 0.1187,
+      "step": 4014
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704000476115079e-05,
+      "loss": 0.1159,
+      "step": 4015
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270065714722806e-05,
+      "loss": 0.1067,
+      "step": 4016
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2697313492684344e-05,
+      "loss": 0.1152,
+      "step": 4017
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2693969512887122e-05,
+      "loss": 0.1273,
+      "step": 4018
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2690625208239628e-05,
+      "loss": 0.1133,
+      "step": 4019
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2687280579145129e-05,
+      "loss": 0.1086,
+      "step": 4020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268393562600694e-05,
+      "loss": 0.1089,
+      "step": 4021
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2680590349228404e-05,
+      "loss": 0.1109,
+      "step": 4022
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2677244749212907e-05,
+      "loss": 0.1141,
+      "step": 4023
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673898826363882e-05,
+      "loss": 0.1233,
+      "step": 4024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2670552581084794e-05,
+      "loss": 0.1233,
+      "step": 4025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2667206013779143e-05,
+      "loss": 0.1143,
+      "step": 4026
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2663859124850475e-05,
+      "loss": 0.1165,
+      "step": 4027
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.266051191470237e-05,
+      "loss": 0.1277,
+      "step": 4028
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2657164383738454e-05,
+      "loss": 0.1187,
+      "step": 4029
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653816532362382e-05,
+      "loss": 0.1236,
+      "step": 4030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.265046836097785e-05,
+      "loss": 0.1264,
+      "step": 4031
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647119869988602e-05,
+      "loss": 0.1027,
+      "step": 4032
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2643771059798407e-05,
+      "loss": 0.1253,
+      "step": 4033
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264042193081108e-05,
+      "loss": 0.1166,
+      "step": 4034
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2637072483430473e-05,
+      "loss": 0.1327,
+      "step": 4035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2633722718060481e-05,
+      "loss": 0.1105,
+      "step": 4036
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2630372635105028e-05,
+      "loss": 0.1156,
+      "step": 4037
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2627022234968083e-05,
+      "loss": 0.1081,
+      "step": 4038
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2623671518053648e-05,
+      "loss": 0.1292,
+      "step": 4039
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2620320484765775e-05,
+      "loss": 0.1186,
+      "step": 4040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616969135508537e-05,
+      "loss": 0.1024,
+      "step": 4041
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2613617470686057e-05,
+      "loss": 0.1116,
+      "step": 4042
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2610265490702496e-05,
+      "loss": 0.1034,
+      "step": 4043
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2606913195962047e-05,
+      "loss": 0.1166,
+      "step": 4044
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2603560586868944e-05,
+      "loss": 0.1199,
+      "step": 4045
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2600207663827457e-05,
+      "loss": 0.1223,
+      "step": 4046
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2596854427241903e-05,
+      "loss": 0.109,
+      "step": 4047
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259350087751662e-05,
+      "loss": 0.1131,
+      "step": 4048
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2590147015055997e-05,
+      "loss": 0.12,
+      "step": 4049
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258679284026446e-05,
+      "loss": 0.1157,
+      "step": 4050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258343835354647e-05,
+      "loss": 0.1058,
+      "step": 4051
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2580083555306516e-05,
+      "loss": 0.1206,
+      "step": 4052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2576728445949144e-05,
+      "loss": 0.1183,
+      "step": 4053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2573373025878923e-05,
+      "loss": 0.1187,
+      "step": 4054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2570017295500466e-05,
+      "loss": 0.1054,
+      "step": 4055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2566661255218416e-05,
+      "loss": 0.1103,
+      "step": 4056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563304905437463e-05,
+      "loss": 0.1047,
+      "step": 4057
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2559948246562332e-05,
+      "loss": 0.1246,
+      "step": 4058
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2556591278997779e-05,
+      "loss": 0.1128,
+      "step": 4059
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2553234003148603e-05,
+      "loss": 0.1153,
+      "step": 4060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549876419419637e-05,
+      "loss": 0.1213,
+      "step": 4061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2546518528215755e-05,
+      "loss": 0.1127,
+      "step": 4062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2543160329941862e-05,
+      "loss": 0.1212,
+      "step": 4063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2539801825002909e-05,
+      "loss": 0.1176,
+      "step": 4064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2536443013803876e-05,
+      "loss": 0.1189,
+      "step": 4065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2533083896749782e-05,
+      "loss": 0.1153,
+      "step": 4066
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2529724474245684e-05,
+      "loss": 0.1179,
+      "step": 4067
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252636474669668e-05,
+      "loss": 0.1158,
+      "step": 4068
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2523004714507891e-05,
+      "loss": 0.1196,
+      "step": 4069
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2519644378084491e-05,
+      "loss": 0.1188,
+      "step": 4070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251628373783168e-05,
+      "loss": 0.1172,
+      "step": 4071
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512922794154703e-05,
+      "loss": 0.1275,
+      "step": 4072
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2509561547458834e-05,
+      "loss": 0.1103,
+      "step": 4073
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2506199998149384e-05,
+      "loss": 0.1158,
+      "step": 4074
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502838146631704e-05,
+      "loss": 0.1228,
+      "step": 4075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499475993311185e-05,
+      "loss": 0.1183,
+      "step": 4076
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2496113538593244e-05,
+      "loss": 0.1283,
+      "step": 4077
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249275078288334e-05,
+      "loss": 0.1191,
+      "step": 4078
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2489387726586977e-05,
+      "loss": 0.1253,
+      "step": 4079
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2486024370109675e-05,
+      "loss": 0.1175,
+      "step": 4080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2482660713857007e-05,
+      "loss": 0.1162,
+      "step": 4081
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2479296758234579e-05,
+      "loss": 0.1173,
+      "step": 4082
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475932503648028e-05,
+      "loss": 0.129,
+      "step": 4083
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2472567950503029e-05,
+      "loss": 0.1191,
+      "step": 4084
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2469203099205297e-05,
+      "loss": 0.1071,
+      "step": 4085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246583795016058e-05,
+      "loss": 0.1276,
+      "step": 4086
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2462472503774663e-05,
+      "loss": 0.1103,
+      "step": 4087
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2459106760453361e-05,
+      "loss": 0.1073,
+      "step": 4088
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2455740720602532e-05,
+      "loss": 0.1178,
+      "step": 4089
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.245237438462807e-05,
+      "loss": 0.119,
+      "step": 4090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.24490077529359e-05,
+      "loss": 0.1098,
+      "step": 4091
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2445640825931979e-05,
+      "loss": 0.1091,
+      "step": 4092
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2442273604022319e-05,
+      "loss": 0.1141,
+      "step": 4093
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438906087612941e-05,
+      "loss": 0.1132,
+      "step": 4094
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435538277109919e-05,
+      "loss": 0.102,
+      "step": 4095
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243217017291936e-05,
+      "loss": 0.1042,
+      "step": 4096
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428801775447406e-05,
+      "loss": 0.1191,
+      "step": 4097
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2425433085100224e-05,
+      "loss": 0.1106,
+      "step": 4098
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.242206410228403e-05,
+      "loss": 0.1274,
+      "step": 4099
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2418694827405075e-05,
+      "loss": 0.1157,
+      "step": 4100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2415325260869634e-05,
+      "loss": 0.1045,
+      "step": 4101
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2411955403084022e-05,
+      "loss": 0.1106,
+      "step": 4102
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2408585254454595e-05,
+      "loss": 0.1244,
+      "step": 4103
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2405214815387743e-05,
+      "loss": 0.1174,
+      "step": 4104
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240184408628988e-05,
+      "loss": 0.1225,
+      "step": 4105
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2398473067567469e-05,
+      "loss": 0.1137,
+      "step": 4106
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2395101759626998e-05,
+      "loss": 0.1204,
+      "step": 4107
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2391730162874995e-05,
+      "loss": 0.1176,
+      "step": 4108
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2388358277718022e-05,
+      "loss": 0.1112,
+      "step": 4109
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2384986104562676e-05,
+      "loss": 0.1079,
+      "step": 4110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2381613643815587e-05,
+      "loss": 0.0991,
+      "step": 4111
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2378240895883418e-05,
+      "loss": 0.1224,
+      "step": 4112
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237486786117287e-05,
+      "loss": 0.1121,
+      "step": 4113
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2371494540090681e-05,
+      "loss": 0.106,
+      "step": 4114
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368120933043621e-05,
+      "loss": 0.1109,
+      "step": 4115
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236474704043849e-05,
+      "loss": 0.1246,
+      "step": 4116
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2361372862682127e-05,
+      "loss": 0.1133,
+      "step": 4117
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357998400181408e-05,
+      "loss": 0.1186,
+      "step": 4118
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2354623653343234e-05,
+      "loss": 0.118,
+      "step": 4119
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2351248622574552e-05,
+      "loss": 0.1276,
+      "step": 4120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2347873308282334e-05,
+      "loss": 0.1122,
+      "step": 4121
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2344497710873594e-05,
+      "loss": 0.1048,
+      "step": 4122
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234112183075537e-05,
+      "loss": 0.1316,
+      "step": 4123
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2337745668334742e-05,
+      "loss": 0.1102,
+      "step": 4124
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2334369224018825e-05,
+      "loss": 0.1164,
+      "step": 4125
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2330992498214762e-05,
+      "loss": 0.1172,
+      "step": 4126
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2327615491329735e-05,
+      "loss": 0.1173,
+      "step": 4127
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2324238203770955e-05,
+      "loss": 0.1248,
+      "step": 4128
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2320860635945672e-05,
+      "loss": 0.1133,
+      "step": 4129
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2317482788261167e-05,
+      "loss": 0.1219,
+      "step": 4130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2314104661124756e-05,
+      "loss": 0.1127,
+      "step": 4131
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2310726254943783e-05,
+      "loss": 0.1284,
+      "step": 4132
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2307347570125639e-05,
+      "loss": 0.1161,
+      "step": 4133
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2303968607077734e-05,
+      "loss": 0.1229,
+      "step": 4134
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2300589366207519e-05,
+      "loss": 0.1012,
+      "step": 4135
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2297209847922475e-05,
+      "loss": 0.1182,
+      "step": 4136
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293830052630124e-05,
+      "loss": 0.1216,
+      "step": 4137
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2290449980738014e-05,
+      "loss": 0.1225,
+      "step": 4138
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287069632653723e-05,
+      "loss": 0.1163,
+      "step": 4139
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283689008784876e-05,
+      "loss": 0.1024,
+      "step": 4140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.228030810953912e-05,
+      "loss": 0.1116,
+      "step": 4141
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2276926935324133e-05,
+      "loss": 0.1301,
+      "step": 4142
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2273545486547638e-05,
+      "loss": 0.119,
+      "step": 4143
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2270163763617383e-05,
+      "loss": 0.1132,
+      "step": 4144
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.226678176694115e-05,
+      "loss": 0.1112,
+      "step": 4145
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2263399496926753e-05,
+      "loss": 0.1223,
+      "step": 4146
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2260016953982043e-05,
+      "loss": 0.1073,
+      "step": 4147
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2256634138514898e-05,
+      "loss": 0.11,
+      "step": 4148
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2253251050933234e-05,
+      "loss": 0.1207,
+      "step": 4149
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2249867691645001e-05,
+      "loss": 0.1258,
+      "step": 4150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2246484061058175e-05,
+      "loss": 0.1143,
+      "step": 4151
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.224310015958077e-05,
+      "loss": 0.1137,
+      "step": 4152
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2239715987620828e-05,
+      "loss": 0.1078,
+      "step": 4153
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2236331545586434e-05,
+      "loss": 0.107,
+      "step": 4154
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2232946833885693e-05,
+      "loss": 0.105,
+      "step": 4155
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2229561852926745e-05,
+      "loss": 0.119,
+      "step": 4156
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2226176603117775e-05,
+      "loss": 0.1154,
+      "step": 4157
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2222791084866983e-05,
+      "loss": 0.1006,
+      "step": 4158
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221940529858261e-05,
+      "loss": 0.1059,
+      "step": 4159
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2216019244672929e-05,
+      "loss": 0.1237,
+      "step": 4160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2212632923546246e-05,
+      "loss": 0.1083,
+      "step": 4161
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.22092463356109e-05,
+      "loss": 0.1187,
+      "step": 4162
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2205859481275255e-05,
+      "loss": 0.1137,
+      "step": 4163
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2202472360947717e-05,
+      "loss": 0.1073,
+      "step": 4164
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199084975036715e-05,
+      "loss": 0.1139,
+      "step": 4165
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219569732395072e-05,
+      "loss": 0.1077,
+      "step": 4166
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2192309408098222e-05,
+      "loss": 0.1157,
+      "step": 4167
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2188921227887755e-05,
+      "loss": 0.1073,
+      "step": 4168
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218553278372788e-05,
+      "loss": 0.1108,
+      "step": 4169
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2182144076027189e-05,
+      "loss": 0.111,
+      "step": 4170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2178755105194307e-05,
+      "loss": 0.0926,
+      "step": 4171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217536587163789e-05,
+      "loss": 0.1215,
+      "step": 4172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217197637576663e-05,
+      "loss": 0.1085,
+      "step": 4173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216858661798924e-05,
+      "loss": 0.1097,
+      "step": 4174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2165196598714475e-05,
+      "loss": 0.1008,
+      "step": 4175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2161806318351118e-05,
+      "loss": 0.1214,
+      "step": 4176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2158415777307983e-05,
+      "loss": 0.126,
+      "step": 4177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2155024975993917e-05,
+      "loss": 0.114,
+      "step": 4178
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2151633914817798e-05,
+      "loss": 0.1311,
+      "step": 4179
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2148242594188528e-05,
+      "loss": 0.1184,
+      "step": 4180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2144851014515055e-05,
+      "loss": 0.1168,
+      "step": 4181
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2141459176206345e-05,
+      "loss": 0.1103,
+      "step": 4182
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138067079671405e-05,
+      "loss": 0.1071,
+      "step": 4183
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2134674725319263e-05,
+      "loss": 0.1215,
+      "step": 4184
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2131282113558984e-05,
+      "loss": 0.1046,
+      "step": 4185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127889244799668e-05,
+      "loss": 0.1333,
+      "step": 4186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2124496119450437e-05,
+      "loss": 0.1119,
+      "step": 4187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121102737920452e-05,
+      "loss": 0.115,
+      "step": 4188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2117709100618897e-05,
+      "loss": 0.1238,
+      "step": 4189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114315207954996e-05,
+      "loss": 0.1174,
+      "step": 4190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2110921060337996e-05,
+      "loss": 0.1195,
+      "step": 4191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2107526658177179e-05,
+      "loss": 0.1266,
+      "step": 4192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2104132001881854e-05,
+      "loss": 0.1351,
+      "step": 4193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100737091861367e-05,
+      "loss": 0.1115,
+      "step": 4194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097341928525088e-05,
+      "loss": 0.1169,
+      "step": 4195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209394651228242e-05,
+      "loss": 0.1125,
+      "step": 4196
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20905508435428e-05,
+      "loss": 0.1099,
+      "step": 4197
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2087154922715688e-05,
+      "loss": 0.1121,
+      "step": 4198
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2083758750210579e-05,
+      "loss": 0.1034,
+      "step": 4199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2080362326436999e-05,
+      "loss": 0.1116,
+      "step": 4200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2076965651804507e-05,
+      "loss": 0.1012,
+      "step": 4201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207356872672268e-05,
+      "loss": 0.11,
+      "step": 4202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207017155160114e-05,
+      "loss": 0.1074,
+      "step": 4203
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206677412684953e-05,
+      "loss": 0.1138,
+      "step": 4204
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206337645287753e-05,
+      "loss": 0.1074,
+      "step": 4205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059978530094841e-05,
+      "loss": 0.1183,
+      "step": 4206
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20565803589112e-05,
+      "loss": 0.1152,
+      "step": 4207
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2053181939736378e-05,
+      "loss": 0.1232,
+      "step": 4208
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2049783272980162e-05,
+      "loss": 0.1215,
+      "step": 4209
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2046384359052384e-05,
+      "loss": 0.1037,
+      "step": 4210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2042985198362896e-05,
+      "loss": 0.1094,
+      "step": 4211
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.203958579132159e-05,
+      "loss": 0.1026,
+      "step": 4212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203618613833837e-05,
+      "loss": 0.1024,
+      "step": 4213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2032786239823189e-05,
+      "loss": 0.1111,
+      "step": 4214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202938609618602e-05,
+      "loss": 0.1231,
+      "step": 4215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2025985707836862e-05,
+      "loss": 0.1284,
+      "step": 4216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2022585075185753e-05,
+      "loss": 0.1194,
+      "step": 4217
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2019184198642753e-05,
+      "loss": 0.1026,
+      "step": 4218
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2015783078617959e-05,
+      "loss": 0.1171,
+      "step": 4219
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012381715521485e-05,
+      "loss": 0.1157,
+      "step": 4220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2008980109763488e-05,
+      "loss": 0.1062,
+      "step": 4221
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2005578261754145e-05,
+      "loss": 0.1049,
+      "step": 4222
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2002176171903667e-05,
+      "loss": 0.1135,
+      "step": 4223
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1998773840622292e-05,
+      "loss": 0.1163,
+      "step": 4224
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1995371268320286e-05,
+      "loss": 0.1148,
+      "step": 4225
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1991968455407947e-05,
+      "loss": 0.0964,
+      "step": 4226
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.19885654022956e-05,
+      "loss": 0.1094,
+      "step": 4227
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.19851621093936e-05,
+      "loss": 0.1053,
+      "step": 4228
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1981758577112329e-05,
+      "loss": 0.1101,
+      "step": 4229
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1978354805862202e-05,
+      "loss": 0.1178,
+      "step": 4230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1974950796053659e-05,
+      "loss": 0.136,
+      "step": 4231
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1971546548097171e-05,
+      "loss": 0.1224,
+      "step": 4232
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1968142062403231e-05,
+      "loss": 0.1136,
+      "step": 4233
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1964737339382375e-05,
+      "loss": 0.1084,
+      "step": 4234
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1961332379445154e-05,
+      "loss": 0.1155,
+      "step": 4235
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1957927183002151e-05,
+      "loss": 0.1024,
+      "step": 4236
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1954521750463986e-05,
+      "loss": 0.1131,
+      "step": 4237
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1951116082241292e-05,
+      "loss": 0.1023,
+      "step": 4238
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1947710178744743e-05,
+      "loss": 0.1124,
+      "step": 4239
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1944304040385038e-05,
+      "loss": 0.1092,
+      "step": 4240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1940897667572901e-05,
+      "loss": 0.1029,
+      "step": 4241
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193749106071909e-05,
+      "loss": 0.1155,
+      "step": 4242
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934084220234385e-05,
+      "loss": 0.1058,
+      "step": 4243
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1930677146529603e-05,
+      "loss": 0.1136,
+      "step": 4244
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1927269840015576e-05,
+      "loss": 0.1138,
+      "step": 4245
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1923862301103174e-05,
+      "loss": 0.1146,
+      "step": 4246
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1920454530203295e-05,
+      "loss": 0.1072,
+      "step": 4247
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1917046527726861e-05,
+      "loss": 0.1121,
+      "step": 4248
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1913638294084824e-05,
+      "loss": 0.1014,
+      "step": 4249
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191022982968816e-05,
+      "loss": 0.119,
+      "step": 4250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190682113494788e-05,
+      "loss": 0.1126,
+      "step": 4251
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1903412210275016e-05,
+      "loss": 0.1182,
+      "step": 4252
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1900003056080633e-05,
+      "loss": 0.1085,
+      "step": 4253
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189659367277582e-05,
+      "loss": 0.1134,
+      "step": 4254
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1893184060771696e-05,
+      "loss": 0.1066,
+      "step": 4255
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1889774220479405e-05,
+      "loss": 0.1082,
+      "step": 4256
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1886364152310119e-05,
+      "loss": 0.1145,
+      "step": 4257
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1882953856675041e-05,
+      "loss": 0.1161,
+      "step": 4258
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1879543333985402e-05,
+      "loss": 0.1048,
+      "step": 4259
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1876132584652446e-05,
+      "loss": 0.1202,
+      "step": 4260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1872721609087467e-05,
+      "loss": 0.1133,
+      "step": 4261
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1869310407701775e-05,
+      "loss": 0.1165,
+      "step": 4262
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1865898980906698e-05,
+      "loss": 0.1105,
+      "step": 4263
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1862487329113606e-05,
+      "loss": 0.1083,
+      "step": 4264
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1859075452733892e-05,
+      "loss": 0.1153,
+      "step": 4265
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185566335217897e-05,
+      "loss": 0.1094,
+      "step": 4266
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852251027860289e-05,
+      "loss": 0.1239,
+      "step": 4267
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1848838480189317e-05,
+      "loss": 0.114,
+      "step": 4268
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1845425709577562e-05,
+      "loss": 0.1211,
+      "step": 4269
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1842012716436544e-05,
+      "loss": 0.1163,
+      "step": 4270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838599501177817e-05,
+      "loss": 0.1298,
+      "step": 4271
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.183518606421296e-05,
+      "loss": 0.1011,
+      "step": 4272
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1831772405953583e-05,
+      "loss": 0.1187,
+      "step": 4273
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1828358526811317e-05,
+      "loss": 0.1111,
+      "step": 4274
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1824944427197822e-05,
+      "loss": 0.1119,
+      "step": 4275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1821530107524785e-05,
+      "loss": 0.1136,
+      "step": 4276
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181811556820392e-05,
+      "loss": 0.109,
+      "step": 4277
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1814700809646967e-05,
+      "loss": 0.1234,
+      "step": 4278
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1811285832265689e-05,
+      "loss": 0.1084,
+      "step": 4279
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807870636471884e-05,
+      "loss": 0.112,
+      "step": 4280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1804455222677364e-05,
+      "loss": 0.1179,
+      "step": 4281
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801039591293978e-05,
+      "loss": 0.1132,
+      "step": 4282
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17976237427336e-05,
+      "loss": 0.0979,
+      "step": 4283
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1794207677408119e-05,
+      "loss": 0.1271,
+      "step": 4284
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1790791395729466e-05,
+      "loss": 0.1151,
+      "step": 4285
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1787374898109588e-05,
+      "loss": 0.1075,
+      "step": 4286
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1783958184960465e-05,
+      "loss": 0.0967,
+      "step": 4287
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178054125669409e-05,
+      "loss": 0.1163,
+      "step": 4288
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17771241137225e-05,
+      "loss": 0.1173,
+      "step": 4289
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177370675645774e-05,
+      "loss": 0.1168,
+      "step": 4290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1770289185311901e-05,
+      "loss": 0.1174,
+      "step": 4291
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1766871400697075e-05,
+      "loss": 0.1227,
+      "step": 4292
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17634534030254e-05,
+      "loss": 0.1009,
+      "step": 4293
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760035192709037e-05,
+      "loss": 0.0981,
+      "step": 4294
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1756616770160156e-05,
+      "loss": 0.1028,
+      "step": 4295
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1753198135790975e-05,
+      "loss": 0.1268,
+      "step": 4296
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1749779290013725e-05,
+      "loss": 0.1284,
+      "step": 4297
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1746360233240663e-05,
+      "loss": 0.1168,
+      "step": 4298
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1742940965884075e-05,
+      "loss": 0.1192,
+      "step": 4299
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173952148835627e-05,
+      "loss": 0.1,
+      "step": 4300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736101801069586e-05,
+      "loss": 0.1103,
+      "step": 4301
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173268190443638e-05,
+      "loss": 0.1195,
+      "step": 4302
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1729261798869039e-05,
+      "loss": 0.1178,
+      "step": 4303
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1725841484779972e-05,
+      "loss": 0.1131,
+      "step": 4304
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1722420962581617e-05,
+      "loss": 0.1201,
+      "step": 4305
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1719000232686435e-05,
+      "loss": 0.1095,
+      "step": 4306
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1715579295506911e-05,
+      "loss": 0.1173,
+      "step": 4307
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1712158151455557e-05,
+      "loss": 0.1128,
+      "step": 4308
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1708736800944911e-05,
+      "loss": 0.1098,
+      "step": 4309
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1705315244387529e-05,
+      "loss": 0.1225,
+      "step": 4310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1701893482196e-05,
+      "loss": 0.1012,
+      "step": 4311
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1698471514782936e-05,
+      "loss": 0.121,
+      "step": 4312
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1695049342560969e-05,
+      "loss": 0.1136,
+      "step": 4313
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691626965942759e-05,
+      "loss": 0.1153,
+      "step": 4314
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688204385340995e-05,
+      "loss": 0.1162,
+      "step": 4315
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1684781601168385e-05,
+      "loss": 0.1124,
+      "step": 4316
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681358613837656e-05,
+      "loss": 0.1177,
+      "step": 4317
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1677935423761576e-05,
+      "loss": 0.1125,
+      "step": 4318
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674512031352922e-05,
+      "loss": 0.1049,
+      "step": 4319
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671088437024504e-05,
+      "loss": 0.1236,
+      "step": 4320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1667664641189152e-05,
+      "loss": 0.1174,
+      "step": 4321
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1664240644259721e-05,
+      "loss": 0.1095,
+      "step": 4322
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1660816446649095e-05,
+      "loss": 0.1068,
+      "step": 4323
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657392048770173e-05,
+      "loss": 0.1076,
+      "step": 4324
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1653967451035887e-05,
+      "loss": 0.1122,
+      "step": 4325
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650542653859185e-05,
+      "loss": 0.1294,
+      "step": 4326
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164711765765305e-05,
+      "loss": 0.1095,
+      "step": 4327
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643692462830481e-05,
+      "loss": 0.116,
+      "step": 4328
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1640267069804496e-05,
+      "loss": 0.1165,
+      "step": 4329
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1636841478988152e-05,
+      "loss": 0.1129,
+      "step": 4330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.163341569079452e-05,
+      "loss": 0.105,
+      "step": 4331
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162998970563669e-05,
+      "loss": 0.1198,
+      "step": 4332
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1626563523927788e-05,
+      "loss": 0.1128,
+      "step": 4333
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1623137146080954e-05,
+      "loss": 0.1176,
+      "step": 4334
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1619710572509354e-05,
+      "loss": 0.1098,
+      "step": 4335
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1616283803626184e-05,
+      "loss": 0.1213,
+      "step": 4336
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1612856839844655e-05,
+      "loss": 0.1237,
+      "step": 4337
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1609429681578009e-05,
+      "loss": 0.1113,
+      "step": 4338
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1606002329239498e-05,
+      "loss": 0.1138,
+      "step": 4339
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602574783242415e-05,
+      "loss": 0.1134,
+      "step": 4340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1599147044000066e-05,
+      "loss": 0.1257,
+      "step": 4341
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159571911192578e-05,
+      "loss": 0.1033,
+      "step": 4342
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159229098743291e-05,
+      "loss": 0.115,
+      "step": 4343
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588862670934841e-05,
+      "loss": 0.1073,
+      "step": 4344
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1585434162844971e-05,
+      "loss": 0.1184,
+      "step": 4345
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158200546357672e-05,
+      "loss": 0.1141,
+      "step": 4346
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1578576573543541e-05,
+      "loss": 0.1276,
+      "step": 4347
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15751474931589e-05,
+      "loss": 0.1117,
+      "step": 4348
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1571718222836292e-05,
+      "loss": 0.1013,
+      "step": 4349
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156828876298923e-05,
+      "loss": 0.1111,
+      "step": 4350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1564859114031258e-05,
+      "loss": 0.1202,
+      "step": 4351
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1561429276375934e-05,
+      "loss": 0.1096,
+      "step": 4352
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1557999250436845e-05,
+      "loss": 0.1184,
+      "step": 4353
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1554569036627591e-05,
+      "loss": 0.1268,
+      "step": 4354
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1551138635361812e-05,
+      "loss": 0.1211,
+      "step": 4355
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1547708047053153e-05,
+      "loss": 0.1086,
+      "step": 4356
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154427727211529e-05,
+      "loss": 0.1112,
+      "step": 4357
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1540846310961925e-05,
+      "loss": 0.1267,
+      "step": 4358
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1537415164006771e-05,
+      "loss": 0.1139,
+      "step": 4359
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1533983831663575e-05,
+      "loss": 0.106,
+      "step": 4360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15305523143461e-05,
+      "loss": 0.1081,
+      "step": 4361
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1527120612468135e-05,
+      "loss": 0.1281,
+      "step": 4362
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523688726443486e-05,
+      "loss": 0.1053,
+      "step": 4363
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1520256656685984e-05,
+      "loss": 0.1015,
+      "step": 4364
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1516824403609485e-05,
+      "loss": 0.1056,
+      "step": 4365
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513391967627866e-05,
+      "loss": 0.1095,
+      "step": 4366
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1509959349155022e-05,
+      "loss": 0.0921,
+      "step": 4367
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1506526548604873e-05,
+      "loss": 0.1132,
+      "step": 4368
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.150309356639136e-05,
+      "loss": 0.1101,
+      "step": 4369
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149966040292845e-05,
+      "loss": 0.1035,
+      "step": 4370
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496227058630122e-05,
+      "loss": 0.1032,
+      "step": 4371
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149279353391039e-05,
+      "loss": 0.1091,
+      "step": 4372
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148935982918328e-05,
+      "loss": 0.1049,
+      "step": 4373
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1485925944862842e-05,
+      "loss": 0.109,
+      "step": 4374
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148249188136315e-05,
+      "loss": 0.1199,
+      "step": 4375
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1479057639098294e-05,
+      "loss": 0.1143,
+      "step": 4376
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.14756232184824e-05,
+      "loss": 0.1166,
+      "step": 4377
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147218861992959e-05,
+      "loss": 0.1017,
+      "step": 4378
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1468753843854032e-05,
+      "loss": 0.1079,
+      "step": 4379
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1465318890669907e-05,
+      "loss": 0.1096,
+      "step": 4380
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146188376079141e-05,
+      "loss": 0.1226,
+      "step": 4381
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145844845463277e-05,
+      "loss": 0.1183,
+      "step": 4382
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1455012972608224e-05,
+      "loss": 0.1108,
+      "step": 4383
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1451577315132043e-05,
+      "loss": 0.1198,
+      "step": 4384
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144814148261851e-05,
+      "loss": 0.1115,
+      "step": 4385
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1444705475481929e-05,
+      "loss": 0.1193,
+      "step": 4386
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1441269294136636e-05,
+      "loss": 0.1044,
+      "step": 4387
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1437832938996978e-05,
+      "loss": 0.1034,
+      "step": 4388
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434396410477322e-05,
+      "loss": 0.1198,
+      "step": 4389
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1430959708992061e-05,
+      "loss": 0.116,
+      "step": 4390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1427522834955608e-05,
+      "loss": 0.1103,
+      "step": 4391
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1424085788782395e-05,
+      "loss": 0.12,
+      "step": 4392
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1420648570886874e-05,
+      "loss": 0.1127,
+      "step": 4393
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141721118168352e-05,
+      "loss": 0.1118,
+      "step": 4394
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413773621586834e-05,
+      "loss": 0.1144,
+      "step": 4395
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1410335891011322e-05,
+      "loss": 0.111,
+      "step": 4396
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1406897990371526e-05,
+      "loss": 0.1089,
+      "step": 4397
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1403459920081998e-05,
+      "loss": 0.1139,
+      "step": 4398
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1400021680557326e-05,
+      "loss": 0.1139,
+      "step": 4399
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1396583272212092e-05,
+      "loss": 0.1274,
+      "step": 4400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1393144695460924e-05,
+      "loss": 0.1128,
+      "step": 4401
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1389705950718461e-05,
+      "loss": 0.1064,
+      "step": 4402
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386267038399354e-05,
+      "loss": 0.1098,
+      "step": 4403
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1382827958918288e-05,
+      "loss": 0.1114,
+      "step": 4404
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1379388712689958e-05,
+      "loss": 0.1073,
+      "step": 4405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137594930012909e-05,
+      "loss": 0.1152,
+      "step": 4406
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1372509721650414e-05,
+      "loss": 0.1145,
+      "step": 4407
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136906997766869e-05,
+      "loss": 0.1187,
+      "step": 4408
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1365630068598704e-05,
+      "loss": 0.1138,
+      "step": 4409
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1362189994855246e-05,
+      "loss": 0.1068,
+      "step": 4410
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1358749756853142e-05,
+      "loss": 0.1123,
+      "step": 4411
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1355309355007225e-05,
+      "loss": 0.1131,
+      "step": 4412
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1351868789732359e-05,
+      "loss": 0.1083,
+      "step": 4413
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1348428061443417e-05,
+      "loss": 0.107,
+      "step": 4414
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1344987170555297e-05,
+      "loss": 0.1188,
+      "step": 4415
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134154611748292e-05,
+      "loss": 0.1068,
+      "step": 4416
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1338104902641221e-05,
+      "loss": 0.1099,
+      "step": 4417
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1334663526445153e-05,
+      "loss": 0.1183,
+      "step": 4418
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1331221989309693e-05,
+      "loss": 0.1131,
+      "step": 4419
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1327780291649841e-05,
+      "loss": 0.1094,
+      "step": 4420
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1324338433880606e-05,
+      "loss": 0.1232,
+      "step": 4421
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1320896416417026e-05,
+      "loss": 0.1098,
+      "step": 4422
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1317454239674148e-05,
+      "loss": 0.1155,
+      "step": 4423
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1314011904067052e-05,
+      "loss": 0.1101,
+      "step": 4424
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1310569410010825e-05,
+      "loss": 0.114,
+      "step": 4425
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307126757920577e-05,
+      "loss": 0.124,
+      "step": 4426
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130368394821144e-05,
+      "loss": 0.1128,
+      "step": 4427
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1300240981298557e-05,
+      "loss": 0.1311,
+      "step": 4428
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1296797857597103e-05,
+      "loss": 0.1091,
+      "step": 4429
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1293354577522264e-05,
+      "loss": 0.1048,
+      "step": 4430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1289911141489241e-05,
+      "loss": 0.1094,
+      "step": 4431
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1286467549913258e-05,
+      "loss": 0.1152,
+      "step": 4432
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283023803209558e-05,
+      "loss": 0.1147,
+      "step": 4433
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127957990179341e-05,
+      "loss": 0.1275,
+      "step": 4434
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1276135846080085e-05,
+      "loss": 0.1009,
+      "step": 4435
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1272691636484885e-05,
+      "loss": 0.1133,
+      "step": 4436
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1269247273423129e-05,
+      "loss": 0.1234,
+      "step": 4437
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265802757310154e-05,
+      "loss": 0.1052,
+      "step": 4438
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1262358088561307e-05,
+      "loss": 0.1202,
+      "step": 4439
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1258913267591969e-05,
+      "loss": 0.111,
+      "step": 4440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1255468294817529e-05,
+      "loss": 0.1184,
+      "step": 4441
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1252023170653393e-05,
+      "loss": 0.1018,
+      "step": 4442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1248577895514989e-05,
+      "loss": 0.105,
+      "step": 4443
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1245132469817767e-05,
+      "loss": 0.1011,
+      "step": 4444
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1241686893977193e-05,
+      "loss": 0.1147,
+      "step": 4445
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238241168408741e-05,
+      "loss": 0.1159,
+      "step": 4446
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234795293527915e-05,
+      "loss": 0.1118,
+      "step": 4447
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1231349269750235e-05,
+      "loss": 0.12,
+      "step": 4448
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1227903097491237e-05,
+      "loss": 0.1128,
+      "step": 4449
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224456777166472e-05,
+      "loss": 0.1103,
+      "step": 4450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1221010309191514e-05,
+      "loss": 0.1152,
+      "step": 4451
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1217563693981954e-05,
+      "loss": 0.1052,
+      "step": 4452
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1214116931953396e-05,
+      "loss": 0.1125,
+      "step": 4453
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1210670023521467e-05,
+      "loss": 0.1136,
+      "step": 4454
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.120722296910181e-05,
+      "loss": 0.1157,
+      "step": 4455
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1203775769110088e-05,
+      "loss": 0.1121,
+      "step": 4456
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1200328423961974e-05,
+      "loss": 0.1209,
+      "step": 4457
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1196880934073167e-05,
+      "loss": 0.1271,
+      "step": 4458
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119343329985938e-05,
+      "loss": 0.1272,
+      "step": 4459
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.118998552173634e-05,
+      "loss": 0.1083,
+      "step": 4460
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1186537600119797e-05,
+      "loss": 0.1089,
+      "step": 4461
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1183089535425516e-05,
+      "loss": 0.1181,
+      "step": 4462
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1179641328069282e-05,
+      "loss": 0.1303,
+      "step": 4463
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1176192978466888e-05,
+      "loss": 0.1127,
+      "step": 4464
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1172744487034157e-05,
+      "loss": 0.1151,
+      "step": 4465
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116929585418692e-05,
+      "loss": 0.1201,
+      "step": 4466
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165847080341029e-05,
+      "loss": 0.1123,
+      "step": 4467
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1162398165912348e-05,
+      "loss": 0.0991,
+      "step": 4468
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1158949111316763e-05,
+      "loss": 0.1219,
+      "step": 4469
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115549991697018e-05,
+      "loss": 0.1107,
+      "step": 4470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1152050583288513e-05,
+      "loss": 0.1194,
+      "step": 4471
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11486011106877e-05,
+      "loss": 0.1096,
+      "step": 4472
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114515149958369e-05,
+      "loss": 0.1311,
+      "step": 4473
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114170175039246e-05,
+      "loss": 0.1089,
+      "step": 4474
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138251863529984e-05,
+      "loss": 0.1082,
+      "step": 4475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134801839412266e-05,
+      "loss": 0.1008,
+      "step": 4476
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1131351678455331e-05,
+      "loss": 0.1138,
+      "step": 4477
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1127901381075212e-05,
+      "loss": 0.1028,
+      "step": 4478
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1124450947687958e-05,
+      "loss": 0.1096,
+      "step": 4479
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1121000378709634e-05,
+      "loss": 0.1207,
+      "step": 4480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117549674556333e-05,
+      "loss": 0.1191,
+      "step": 4481
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111409883564415e-05,
+      "loss": 0.1133,
+      "step": 4482
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11106478623892e-05,
+      "loss": 0.1161,
+      "step": 4483
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1107196755207622e-05,
+      "loss": 0.1149,
+      "step": 4484
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110374551451556e-05,
+      "loss": 0.1108,
+      "step": 4485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1100294140729182e-05,
+      "loss": 0.1178,
+      "step": 4486
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1096842634264666e-05,
+      "loss": 0.1068,
+      "step": 4487
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1093390995538213e-05,
+      "loss": 0.1221,
+      "step": 4488
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1089939224966035e-05,
+      "loss": 0.1101,
+      "step": 4489
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1086487322964361e-05,
+      "loss": 0.1094,
+      "step": 4490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1083035289949433e-05,
+      "loss": 0.1062,
+      "step": 4491
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1079583126337519e-05,
+      "loss": 0.1101,
+      "step": 4492
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1076130832544888e-05,
+      "loss": 0.12,
+      "step": 4493
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1072678408987837e-05,
+      "loss": 0.1209,
+      "step": 4494
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1069225856082673e-05,
+      "loss": 0.1109,
+      "step": 4495
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106577317424572e-05,
+      "loss": 0.1242,
+      "step": 4496
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1062320363893313e-05,
+      "loss": 0.1174,
+      "step": 4497
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058867425441808e-05,
+      "loss": 0.1106,
+      "step": 4498
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105541435930758e-05,
+      "loss": 0.1036,
+      "step": 4499
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105196116590701e-05,
+      "loss": 0.1056,
+      "step": 4500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1048507845656497e-05,
+      "loss": 0.1133,
+      "step": 4501
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045054398972462e-05,
+      "loss": 0.1041,
+      "step": 4502
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1041600826271334e-05,
+      "loss": 0.1175,
+      "step": 4503
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1038147127969559e-05,
+      "loss": 0.1234,
+      "step": 4504
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034693304483596e-05,
+      "loss": 0.1043,
+      "step": 4505
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1031239356229928e-05,
+      "loss": 0.1099,
+      "step": 4506
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102778528362504e-05,
+      "loss": 0.1201,
+      "step": 4507
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1024331087085443e-05,
+      "loss": 0.117,
+      "step": 4508
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1020876767027657e-05,
+      "loss": 0.0951,
+      "step": 4509
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101742232386822e-05,
+      "loss": 0.1233,
+      "step": 4510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101396775802368e-05,
+      "loss": 0.1094,
+      "step": 4511
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101051306991061e-05,
+      "loss": 0.1169,
+      "step": 4512
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1007058259945584e-05,
+      "loss": 0.1046,
+      "step": 4513
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.10036033285452e-05,
+      "loss": 0.1022,
+      "step": 4514
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.100014827612607e-05,
+      "loss": 0.1063,
+      "step": 4515
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0996693103104816e-05,
+      "loss": 0.1131,
+      "step": 4516
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0993237809898082e-05,
+      "loss": 0.1014,
+      "step": 4517
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0989782396922517e-05,
+      "loss": 0.1042,
+      "step": 4518
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098632686459479e-05,
+      "loss": 0.1115,
+      "step": 4519
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0982871213331587e-05,
+      "loss": 0.1188,
+      "step": 4520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0979415443549604e-05,
+      "loss": 0.1024,
+      "step": 4521
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0975959555665547e-05,
+      "loss": 0.1131,
+      "step": 4522
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0972503550096147e-05,
+      "loss": 0.1226,
+      "step": 4523
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0969047427258148e-05,
+      "loss": 0.1257,
+      "step": 4524
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0965591187568295e-05,
+      "loss": 0.123,
+      "step": 4525
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0962134831443358e-05,
+      "loss": 0.1214,
+      "step": 4526
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0958678359300126e-05,
+      "loss": 0.1174,
+      "step": 4527
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0955221771555387e-05,
+      "loss": 0.1086,
+      "step": 4528
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0951765068625952e-05,
+      "loss": 0.1085,
+      "step": 4529
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0948308250928647e-05,
+      "loss": 0.1251,
+      "step": 4530
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944851318880314e-05,
+      "loss": 0.1021,
+      "step": 4531
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0941394272897799e-05,
+      "loss": 0.1163,
+      "step": 4532
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937937113397964e-05,
+      "loss": 0.114,
+      "step": 4533
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0934479840797696e-05,
+      "loss": 0.1121,
+      "step": 4534
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0931022455513886e-05,
+      "loss": 0.1172,
+      "step": 4535
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0927564957963435e-05,
+      "loss": 0.1212,
+      "step": 4536
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0924107348563267e-05,
+      "loss": 0.1182,
+      "step": 4537
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0920649627730316e-05,
+      "loss": 0.1096,
+      "step": 4538
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0917191795881524e-05,
+      "loss": 0.1145,
+      "step": 4539
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0913733853433854e-05,
+      "loss": 0.1139,
+      "step": 4540
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091027580080428e-05,
+      "loss": 0.1158,
+      "step": 4541
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090681763840979e-05,
+      "loss": 0.1059,
+      "step": 4542
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0903359366667379e-05,
+      "loss": 0.1183,
+      "step": 4543
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0899900985994063e-05,
+      "loss": 0.1078,
+      "step": 4544
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0896442496806868e-05,
+      "loss": 0.1138,
+      "step": 4545
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0892983899522836e-05,
+      "loss": 0.1198,
+      "step": 4546
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0889525194559012e-05,
+      "loss": 0.1029,
+      "step": 4547
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0886066382332469e-05,
+      "loss": 0.1195,
+      "step": 4548
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0882607463260283e-05,
+      "loss": 0.1236,
+      "step": 4549
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0879148437759543e-05,
+      "loss": 0.1146,
+      "step": 4550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0875689306247354e-05,
+      "loss": 0.1156,
+      "step": 4551
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0872230069140834e-05,
+      "loss": 0.1291,
+      "step": 4552
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0868770726857112e-05,
+      "loss": 0.1222,
+      "step": 4553
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0865311279813329e-05,
+      "loss": 0.106,
+      "step": 4554
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.086185172842664e-05,
+      "loss": 0.1248,
+      "step": 4555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0858392073114216e-05,
+      "loss": 0.1125,
+      "step": 4556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0854932314293231e-05,
+      "loss": 0.1147,
+      "step": 4557
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0851472452380881e-05,
+      "loss": 0.1154,
+      "step": 4558
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0848012487794368e-05,
+      "loss": 0.116,
+      "step": 4559
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0844552420950916e-05,
+      "loss": 0.1191,
+      "step": 4560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0841092252267747e-05,
+      "loss": 0.1056,
+      "step": 4561
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0837631982162105e-05,
+      "loss": 0.1094,
+      "step": 4562
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834171611051245e-05,
+      "loss": 0.1071,
+      "step": 4563
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0830711139352434e-05,
+      "loss": 0.1151,
+      "step": 4564
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0827250567482949e-05,
+      "loss": 0.1105,
+      "step": 4565
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0823789895860081e-05,
+      "loss": 0.1149,
+      "step": 4566
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0820329124901132e-05,
+      "loss": 0.1246,
+      "step": 4567
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0816868255023418e-05,
+      "loss": 0.1182,
+      "step": 4568
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0813407286644262e-05,
+      "loss": 0.1148,
+      "step": 4569
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809946220181004e-05,
+      "loss": 0.1085,
+      "step": 4570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0806485056050997e-05,
+      "loss": 0.123,
+      "step": 4571
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.08030237946716e-05,
+      "loss": 0.1119,
+      "step": 4572
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799562436460182e-05,
+      "loss": 0.1117,
+      "step": 4573
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0796100981834136e-05,
+      "loss": 0.1045,
+      "step": 4574
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0792639431210859e-05,
+      "loss": 0.1199,
+      "step": 4575
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0789177785007754e-05,
+      "loss": 0.1155,
+      "step": 4576
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0785716043642242e-05,
+      "loss": 0.1118,
+      "step": 4577
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0782254207531758e-05,
+      "loss": 0.1181,
+      "step": 4578
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0778792277093741e-05,
+      "loss": 0.1106,
+      "step": 4579
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0775330252745648e-05,
+      "loss": 0.1216,
+      "step": 4580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0771868134904944e-05,
+      "loss": 0.0985,
+      "step": 4581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0768405923989106e-05,
+      "loss": 0.1156,
+      "step": 4582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076494362041562e-05,
+      "loss": 0.1192,
+      "step": 4583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761481224601987e-05,
+      "loss": 0.1215,
+      "step": 4584
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0758018736965719e-05,
+      "loss": 0.1089,
+      "step": 4585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0754556157924334e-05,
+      "loss": 0.1152,
+      "step": 4586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0751093487895366e-05,
+      "loss": 0.1166,
+      "step": 4587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0747630727296357e-05,
+      "loss": 0.1132,
+      "step": 4588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0744167876544867e-05,
+      "loss": 0.1029,
+      "step": 4589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0740704936058453e-05,
+      "loss": 0.1182,
+      "step": 4590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0737241906254696e-05,
+      "loss": 0.1159,
+      "step": 4591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0733778787551184e-05,
+      "loss": 0.0983,
+      "step": 4592
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730315580365514e-05,
+      "loss": 0.1147,
+      "step": 4593
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0726852285115287e-05,
+      "loss": 0.121,
+      "step": 4594
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723388902218133e-05,
+      "loss": 0.1136,
+      "step": 4595
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0719925432091671e-05,
+      "loss": 0.1256,
+      "step": 4596
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071646187515355e-05,
+      "loss": 0.1263,
+      "step": 4597
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0712998231821415e-05,
+      "loss": 0.1017,
+      "step": 4598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0709534502512929e-05,
+      "loss": 0.1138,
+      "step": 4599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0706070687645762e-05,
+      "loss": 0.1214,
+      "step": 4600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702606787637596e-05,
+      "loss": 0.1054,
+      "step": 4601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069914280290612e-05,
+      "loss": 0.128,
+      "step": 4602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0695678733869042e-05,
+      "loss": 0.1156,
+      "step": 4603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069221458094407e-05,
+      "loss": 0.1098,
+      "step": 4604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0688750344548923e-05,
+      "loss": 0.1212,
+      "step": 4605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0685286025101343e-05,
+      "loss": 0.1089,
+      "step": 4606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681821623019064e-05,
+      "loss": 0.1102,
+      "step": 4607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067835713871984e-05,
+      "loss": 0.1251,
+      "step": 4608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674892572621433e-05,
+      "loss": 0.1125,
+      "step": 4609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0671427925141623e-05,
+      "loss": 0.1193,
+      "step": 4610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066796319669818e-05,
+      "loss": 0.105,
+      "step": 4611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.06644983877089e-05,
+      "loss": 0.1035,
+      "step": 4612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0661033498591584e-05,
+      "loss": 0.1099,
+      "step": 4613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0657568529764048e-05,
+      "loss": 0.1171,
+      "step": 4614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0654103481644106e-05,
+      "loss": 0.1085,
+      "step": 4615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650638354649593e-05,
+      "loss": 0.129,
+      "step": 4616
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0647173149198348e-05,
+      "loss": 0.1132,
+      "step": 4617
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0643707865708216e-05,
+      "loss": 0.1187,
+      "step": 4618
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0640242504597058e-05,
+      "loss": 0.1202,
+      "step": 4619
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0636777066282744e-05,
+      "loss": 0.1181,
+      "step": 4620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.063331155118315e-05,
+      "loss": 0.1172,
+      "step": 4621
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062984595971616e-05,
+      "loss": 0.1107,
+      "step": 4622
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0626380292299673e-05,
+      "loss": 0.1002,
+      "step": 4623
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0622914549351598e-05,
+      "loss": 0.099,
+      "step": 4624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0619448731289837e-05,
+      "loss": 0.1193,
+      "step": 4625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0615982838532322e-05,
+      "loss": 0.1071,
+      "step": 4626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061251687149698e-05,
+      "loss": 0.1201,
+      "step": 4627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0609050830601761e-05,
+      "loss": 0.1018,
+      "step": 4628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0605584716264606e-05,
+      "loss": 0.1165,
+      "step": 4629
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0602118528903473e-05,
+      "loss": 0.1109,
+      "step": 4630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598652268936336e-05,
+      "loss": 0.0967,
+      "step": 4631
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0595185936781169e-05,
+      "loss": 0.1251,
+      "step": 4632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0591719532855958e-05,
+      "loss": 0.1149,
+      "step": 4633
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0588253057578692e-05,
+      "loss": 0.1198,
+      "step": 4634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0584786511367379e-05,
+      "loss": 0.1228,
+      "step": 4635
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0581319894640026e-05,
+      "loss": 0.1051,
+      "step": 4636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0577853207814652e-05,
+      "loss": 0.1217,
+      "step": 4637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574386451309288e-05,
+      "loss": 0.1074,
+      "step": 4638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0570919625541971e-05,
+      "loss": 0.1236,
+      "step": 4639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0567452730930743e-05,
+      "loss": 0.093,
+      "step": 4640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0563985767893654e-05,
+      "loss": 0.1155,
+      "step": 4641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0560518736848768e-05,
+      "loss": 0.1281,
+      "step": 4642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0557051638214159e-05,
+      "loss": 0.1187,
+      "step": 4643
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0553584472407896e-05,
+      "loss": 0.1154,
+      "step": 4644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055011723984807e-05,
+      "loss": 0.1104,
+      "step": 4645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0546649940952776e-05,
+      "loss": 0.1183,
+      "step": 4646
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.054318257614011e-05,
+      "loss": 0.1198,
+      "step": 4647
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0539715145828182e-05,
+      "loss": 0.1114,
+      "step": 4648
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0536247650435114e-05,
+      "loss": 0.1051,
+      "step": 4649
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0532780090379031e-05,
+      "loss": 0.1321,
+      "step": 4650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0529312466078064e-05,
+      "loss": 0.11,
+      "step": 4651
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0525844777950352e-05,
+      "loss": 0.1052,
+      "step": 4652
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0522377026414048e-05,
+      "loss": 0.1173,
+      "step": 4653
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0518909211887302e-05,
+      "loss": 0.1075,
+      "step": 4654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0515441334788285e-05,
+      "loss": 0.1205,
+      "step": 4655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511973395535162e-05,
+      "loss": 0.109,
+      "step": 4656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050850539454612e-05,
+      "loss": 0.1197,
+      "step": 4657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0505037332239336e-05,
+      "loss": 0.1099,
+      "step": 4658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0501569209033006e-05,
+      "loss": 0.1145,
+      "step": 4659
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0498101025345338e-05,
+      "loss": 0.1131,
+      "step": 4660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0494632781594531e-05,
+      "loss": 0.1027,
+      "step": 4661
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491164478198807e-05,
+      "loss": 0.1204,
+      "step": 4662
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0487696115576384e-05,
+      "loss": 0.1036,
+      "step": 4663
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0484227694145497e-05,
+      "loss": 0.1147,
+      "step": 4664
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0480759214324378e-05,
+      "loss": 0.1088,
+      "step": 4665
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0477290676531276e-05,
+      "loss": 0.1186,
+      "step": 4666
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0473822081184435e-05,
+      "loss": 0.1143,
+      "step": 4667
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.047035342870212e-05,
+      "loss": 0.1183,
+      "step": 4668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0466884719502592e-05,
+      "loss": 0.099,
+      "step": 4669
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0463415954004123e-05,
+      "loss": 0.1203,
+      "step": 4670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0459947132624992e-05,
+      "loss": 0.1226,
+      "step": 4671
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045647825578349e-05,
+      "loss": 0.1037,
+      "step": 4672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0453009323897895e-05,
+      "loss": 0.1163,
+      "step": 4673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0449540337386516e-05,
+      "loss": 0.0989,
+      "step": 4674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044607129666766e-05,
+      "loss": 0.1073,
+      "step": 4675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044260220215963e-05,
+      "loss": 0.0988,
+      "step": 4676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043913305428075e-05,
+      "loss": 0.1098,
+      "step": 4677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0435663853449343e-05,
+      "loss": 0.1094,
+      "step": 4678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043219460008374e-05,
+      "loss": 0.1215,
+      "step": 4679
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042872529460228e-05,
+      "loss": 0.1125,
+      "step": 4680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0425255937423303e-05,
+      "loss": 0.0972,
+      "step": 4681
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0421786528965163e-05,
+      "loss": 0.114,
+      "step": 4682
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0418317069646211e-05,
+      "loss": 0.1149,
+      "step": 4683
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0414847559884813e-05,
+      "loss": 0.1138,
+      "step": 4684
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041137800009934e-05,
+      "loss": 0.115,
+      "step": 4685
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0407908390708156e-05,
+      "loss": 0.1142,
+      "step": 4686
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0404438732129652e-05,
+      "loss": 0.1094,
+      "step": 4687
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0400969024782208e-05,
+      "loss": 0.104,
+      "step": 4688
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039749926908422e-05,
+      "loss": 0.1183,
+      "step": 4689
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0394029465454081e-05,
+      "loss": 0.1289,
+      "step": 4690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0390559614310196e-05,
+      "loss": 0.1089,
+      "step": 4691
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038708971607098e-05,
+      "loss": 0.1023,
+      "step": 4692
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038361977115484e-05,
+      "loss": 0.1044,
+      "step": 4693
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380149779980203e-05,
+      "loss": 0.1255,
+      "step": 4694
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037667974296549e-05,
+      "loss": 0.1058,
+      "step": 4695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0373209660529137e-05,
+      "loss": 0.1211,
+      "step": 4696
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.036973953308958e-05,
+      "loss": 0.1224,
+      "step": 4697
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366269361065263e-05,
+      "loss": 0.1115,
+      "step": 4698
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0362799144874629e-05,
+      "loss": 0.1041,
+      "step": 4699
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359328884936138e-05,
+      "loss": 0.1175,
+      "step": 4700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0355858581668244e-05,
+      "loss": 0.1085,
+      "step": 4701
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035238823548941e-05,
+      "loss": 0.1221,
+      "step": 4702
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0348917846818113e-05,
+      "loss": 0.1007,
+      "step": 4703
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345447416072815e-05,
+      "loss": 0.1148,
+      "step": 4704
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0341976943672005e-05,
+      "loss": 0.1235,
+      "step": 4705
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338506430034166e-05,
+      "loss": 0.1057,
+      "step": 4706
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335035875577784e-05,
+      "loss": 0.1074,
+      "step": 4707
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0331565280721354e-05,
+      "loss": 0.1083,
+      "step": 4708
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0328094645883377e-05,
+      "loss": 0.1122,
+      "step": 4709
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0324623971482356e-05,
+      "loss": 0.1077,
+      "step": 4710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0321153257936798e-05,
+      "loss": 0.1087,
+      "step": 4711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0317682505665216e-05,
+      "loss": 0.0962,
+      "step": 4712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0314211715086131e-05,
+      "loss": 0.1055,
+      "step": 4713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0310740886618067e-05,
+      "loss": 0.1115,
+      "step": 4714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0307270020679545e-05,
+      "loss": 0.113,
+      "step": 4715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0303799117689102e-05,
+      "loss": 0.1104,
+      "step": 4716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030032817806527e-05,
+      "loss": 0.1119,
+      "step": 4717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0296857202226596e-05,
+      "loss": 0.1072,
+      "step": 4718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0293386190591619e-05,
+      "loss": 0.1129,
+      "step": 4719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.028991514357889e-05,
+      "loss": 0.0984,
+      "step": 4720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0286444061606963e-05,
+      "loss": 0.1145,
+      "step": 4721
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0282972945094397e-05,
+      "loss": 0.1271,
+      "step": 4722
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279501794459752e-05,
+      "loss": 0.1046,
+      "step": 4723
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0276030610121593e-05,
+      "loss": 0.1201,
+      "step": 4724
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272559392498496e-05,
+      "loss": 0.1101,
+      "step": 4725
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0269088142009028e-05,
+      "loss": 0.0983,
+      "step": 4726
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026561685907177e-05,
+      "loss": 0.1096,
+      "step": 4727
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262145544105304e-05,
+      "loss": 0.1019,
+      "step": 4728
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0258674197528218e-05,
+      "loss": 0.1134,
+      "step": 4729
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255202819759098e-05,
+      "loss": 0.1186,
+      "step": 4730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0251731411216539e-05,
+      "loss": 0.1194,
+      "step": 4731
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0248259972319142e-05,
+      "loss": 0.115,
+      "step": 4732
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.02447885034855e-05,
+      "loss": 0.1147,
+      "step": 4733
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0241317005134224e-05,
+      "loss": 0.1085,
+      "step": 4734
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0237845477683916e-05,
+      "loss": 0.1137,
+      "step": 4735
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0234373921553198e-05,
+      "loss": 0.1186,
+      "step": 4736
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0230902337160673e-05,
+      "loss": 0.1114,
+      "step": 4737
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227430724924966e-05,
+      "loss": 0.1169,
+      "step": 4738
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0223959085264697e-05,
+      "loss": 0.1292,
+      "step": 4739
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0220487418598491e-05,
+      "loss": 0.0966,
+      "step": 4740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0217015725344977e-05,
+      "loss": 0.1165,
+      "step": 4741
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0213544005922785e-05,
+      "loss": 0.1123,
+      "step": 4742
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0210072260750554e-05,
+      "loss": 0.1236,
+      "step": 4743
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0206600490246917e-05,
+      "loss": 0.1104,
+      "step": 4744
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0203128694830512e-05,
+      "loss": 0.1239,
+      "step": 4745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199656874919992e-05,
+      "loss": 0.1115,
+      "step": 4746
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0196185030933999e-05,
+      "loss": 0.1195,
+      "step": 4747
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0192713163291181e-05,
+      "loss": 0.1207,
+      "step": 4748
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0189241272410191e-05,
+      "loss": 0.1191,
+      "step": 4749
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.018576935870969e-05,
+      "loss": 0.1041,
+      "step": 4750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0182297422608326e-05,
+      "loss": 0.0979,
+      "step": 4751
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0178825464524769e-05,
+      "loss": 0.1043,
+      "step": 4752
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0175353484877677e-05,
+      "loss": 0.1105,
+      "step": 4753
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0171881484085719e-05,
+      "loss": 0.1246,
+      "step": 4754
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0168409462567561e-05,
+      "loss": 0.1151,
+      "step": 4755
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0164937420741876e-05,
+      "loss": 0.1072,
+      "step": 4756
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0161465359027337e-05,
+      "loss": 0.1181,
+      "step": 4757
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0157993277842622e-05,
+      "loss": 0.1054,
+      "step": 4758
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0154521177606405e-05,
+      "loss": 0.1119,
+      "step": 4759
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0151049058737368e-05,
+      "loss": 0.1132,
+      "step": 4760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0147576921654199e-05,
+      "loss": 0.1036,
+      "step": 4761
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144104766775574e-05,
+      "loss": 0.1226,
+      "step": 4762
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0140632594520183e-05,
+      "loss": 0.1098,
+      "step": 4763
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0137160405306723e-05,
+      "loss": 0.1219,
+      "step": 4764
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133688199553878e-05,
+      "loss": 0.1274,
+      "step": 4765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013021597768034e-05,
+      "loss": 0.1121,
+      "step": 4766
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0126743740104807e-05,
+      "loss": 0.1051,
+      "step": 4767
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.012327148724598e-05,
+      "loss": 0.0895,
+      "step": 4768
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011979921952255e-05,
+      "loss": 0.1129,
+      "step": 4769
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0116326937353225e-05,
+      "loss": 0.1186,
+      "step": 4770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0112854641156704e-05,
+      "loss": 0.1185,
+      "step": 4771
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0109382331351689e-05,
+      "loss": 0.1153,
+      "step": 4772
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.010591000835689e-05,
+      "loss": 0.1187,
+      "step": 4773
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0102437672591012e-05,
+      "loss": 0.111,
+      "step": 4774
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098965324472767e-05,
+      "loss": 0.1126,
+      "step": 4775
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0095492964420861e-05,
+      "loss": 0.1006,
+      "step": 4776
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0092020592854009e-05,
+      "loss": 0.1082,
+      "step": 4777
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0088548210190922e-05,
+      "loss": 0.1059,
+      "step": 4778
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008507581685032e-05,
+      "loss": 0.1245,
+      "step": 4779
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008160341325091e-05,
+      "loss": 0.1177,
+      "step": 4780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0078130999811417e-05,
+      "loss": 0.1175,
+      "step": 4781
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.007465857695056e-05,
+      "loss": 0.1093,
+      "step": 4782
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071186145087052e-05,
+      "loss": 0.1215,
+      "step": 4783
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0067713704639615e-05,
+      "loss": 0.1234,
+      "step": 4784
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0064241256026972e-05,
+      "loss": 0.1113,
+      "step": 4785
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0060768799667849e-05,
+      "loss": 0.1254,
+      "step": 4786
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0057296335980962e-05,
+      "loss": 0.112,
+      "step": 4787
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053823865385042e-05,
+      "loss": 0.1119,
+      "step": 4788
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050351388298811e-05,
+      "loss": 0.1212,
+      "step": 4789
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0046878905141e-05,
+      "loss": 0.1267,
+      "step": 4790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0043406416330328e-05,
+      "loss": 0.1089,
+      "step": 4791
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039933922285525e-05,
+      "loss": 0.1115,
+      "step": 4792
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0036461423425326e-05,
+      "loss": 0.1112,
+      "step": 4793
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0032988920168448e-05,
+      "loss": 0.1034,
+      "step": 4794
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029516412933627e-05,
+      "loss": 0.1186,
+      "step": 4795
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026043902139592e-05,
+      "loss": 0.1159,
+      "step": 4796
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0022571388205076e-05,
+      "loss": 0.1028,
+      "step": 4797
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019098871548803e-05,
+      "loss": 0.1184,
+      "step": 4798
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015626352589506e-05,
+      "loss": 0.1049,
+      "step": 4799
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001215383174592e-05,
+      "loss": 0.1158,
+      "step": 4800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0008681309436775e-05,
+      "loss": 0.1073,
+      "step": 4801
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0005208786080796e-05,
+      "loss": 0.1152,
+      "step": 4802
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0001736262096722e-05,
+      "loss": 0.1114,
+      "step": 4803
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.998263737903282e-06,
+      "loss": 0.1163,
+      "step": 4804
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.994791213919208e-06,
+      "loss": 0.1119,
+      "step": 4805
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.99131869056323e-06,
+      "loss": 0.1013,
+      "step": 4806
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987846168254081e-06,
+      "loss": 0.1063,
+      "step": 4807
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984373647410495e-06,
+      "loss": 0.1122,
+      "step": 4808
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9809011284512e-06,
+      "loss": 0.1098,
+      "step": 4809
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977428611794927e-06,
+      "loss": 0.1213,
+      "step": 4810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97395609786041e-06,
+      "loss": 0.103,
+      "step": 4811
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.970483587066375e-06,
+      "loss": 0.1003,
+      "step": 4812
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.967011079831556e-06,
+      "loss": 0.1106,
+      "step": 4813
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.963538576574678e-06,
+      "loss": 0.1138,
+      "step": 4814
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960066077714477e-06,
+      "loss": 0.106,
+      "step": 4815
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.956593583669675e-06,
+      "loss": 0.1013,
+      "step": 4816
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.953121094859003e-06,
+      "loss": 0.1116,
+      "step": 4817
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94964861170119e-06,
+      "loss": 0.118,
+      "step": 4818
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946176134614961e-06,
+      "loss": 0.1173,
+      "step": 4819
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94270366401904e-06,
+      "loss": 0.0927,
+      "step": 4820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.939231200332155e-06,
+      "loss": 0.1025,
+      "step": 4821
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.935758743973031e-06,
+      "loss": 0.1318,
+      "step": 4822
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93228629536039e-06,
+      "loss": 0.1112,
+      "step": 4823
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.928813854912953e-06,
+      "loss": 0.1127,
+      "step": 4824
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.925341423049444e-06,
+      "loss": 0.1163,
+      "step": 4825
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.921869000188586e-06,
+      "loss": 0.1205,
+      "step": 4826
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918396586749091e-06,
+      "loss": 0.1028,
+      "step": 4827
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.914924183149683e-06,
+      "loss": 0.1088,
+      "step": 4828
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91145178980908e-06,
+      "loss": 0.1158,
+      "step": 4829
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.907979407145995e-06,
+      "loss": 0.1142,
+      "step": 4830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.904507035579142e-06,
+      "loss": 0.1163,
+      "step": 4831
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901034675527236e-06,
+      "loss": 0.112,
+      "step": 4832
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.897562327408991e-06,
+      "loss": 0.1128,
+      "step": 4833
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.894089991643114e-06,
+      "loss": 0.1055,
+      "step": 4834
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.890617668648313e-06,
+      "loss": 0.1082,
+      "step": 4835
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8871453588433e-06,
+      "loss": 0.1007,
+      "step": 4836
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88367306264678e-06,
+      "loss": 0.1049,
+      "step": 4837
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.880200780477451e-06,
+      "loss": 0.1046,
+      "step": 4838
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.876728512754022e-06,
+      "loss": 0.1088,
+      "step": 4839
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.873256259895196e-06,
+      "loss": 0.1025,
+      "step": 4840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.869784022319663e-06,
+      "loss": 0.1206,
+      "step": 4841
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.866311800446128e-06,
+      "loss": 0.1084,
+      "step": 4842
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86283959469328e-06,
+      "loss": 0.1127,
+      "step": 4843
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.859367405479819e-06,
+      "loss": 0.1194,
+      "step": 4844
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855895233224431e-06,
+      "loss": 0.1222,
+      "step": 4845
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.852423078345806e-06,
+      "loss": 0.1118,
+      "step": 4846
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.848950941262637e-06,
+      "loss": 0.1049,
+      "step": 4847
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8454788223936e-06,
+      "loss": 0.1212,
+      "step": 4848
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.842006722157383e-06,
+      "loss": 0.1023,
+      "step": 4849
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.838534640972666e-06,
+      "loss": 0.113,
+      "step": 4850
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.835062579258129e-06,
+      "loss": 0.1032,
+      "step": 4851
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.831590537432444e-06,
+      "loss": 0.1096,
+      "step": 4852
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.828118515914286e-06,
+      "loss": 0.1002,
+      "step": 4853
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.824646515122326e-06,
+      "loss": 0.105,
+      "step": 4854
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.821174535475236e-06,
+      "loss": 0.1137,
+      "step": 4855
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.817702577391679e-06,
+      "loss": 0.1146,
+      "step": 4856
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.814230641290315e-06,
+      "loss": 0.1133,
+      "step": 4857
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.810758727589814e-06,
+      "loss": 0.1175,
+      "step": 4858
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807286836708825e-06,
+      "loss": 0.1044,
+      "step": 4859
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.803814969066006e-06,
+      "loss": 0.1193,
+      "step": 4860
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800343125080011e-06,
+      "loss": 0.1199,
+      "step": 4861
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.796871305169493e-06,
+      "loss": 0.1059,
+      "step": 4862
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79339950975309e-06,
+      "loss": 0.1041,
+      "step": 4863
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.789927739249451e-06,
+      "loss": 0.1185,
+      "step": 4864
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786455994077217e-06,
+      "loss": 0.114,
+      "step": 4865
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.782984274655028e-06,
+      "loss": 0.1208,
+      "step": 4866
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.779512581401514e-06,
+      "loss": 0.1044,
+      "step": 4867
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.776040914735307e-06,
+      "loss": 0.1078,
+      "step": 4868
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77256927507504e-06,
+      "loss": 0.1182,
+      "step": 4869
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769097662839332e-06,
+      "loss": 0.1108,
+      "step": 4870
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.765626078446807e-06,
+      "loss": 0.1156,
+      "step": 4871
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.762154522316086e-06,
+      "loss": 0.1143,
+      "step": 4872
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.758682994865783e-06,
+      "loss": 0.1209,
+      "step": 4873
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.755211496514504e-06,
+      "loss": 0.1046,
+      "step": 4874
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.751740027680863e-06,
+      "loss": 0.0986,
+      "step": 4875
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.748268588783466e-06,
+      "loss": 0.1226,
+      "step": 4876
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.744797180240907e-06,
+      "loss": 0.1045,
+      "step": 4877
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.741325802471787e-06,
+      "loss": 0.107,
+      "step": 4878
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.737854455894699e-06,
+      "loss": 0.1144,
+      "step": 4879
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734383140928237e-06,
+      "loss": 0.104,
+      "step": 4880
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.730911857990977e-06,
+      "loss": 0.1034,
+      "step": 4881
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72744060750151e-06,
+      "loss": 0.1079,
+      "step": 4882
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72396938987841e-06,
+      "loss": 0.1,
+      "step": 4883
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.720498205540253e-06,
+      "loss": 0.1172,
+      "step": 4884
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717027054905608e-06,
+      "loss": 0.1146,
+      "step": 4885
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.713555938393042e-06,
+      "loss": 0.112,
+      "step": 4886
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.710084856421117e-06,
+      "loss": 0.1109,
+      "step": 4887
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.706613809408382e-06,
+      "loss": 0.1133,
+      "step": 4888
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.703142797773408e-06,
+      "loss": 0.1137,
+      "step": 4889
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69967182193473e-06,
+      "loss": 0.0982,
+      "step": 4890
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.6962008823109e-06,
+      "loss": 0.104,
+      "step": 4891
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.692729979320456e-06,
+      "loss": 0.1156,
+      "step": 4892
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.689259113381936e-06,
+      "loss": 0.1041,
+      "step": 4893
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685788284913869e-06,
+      "loss": 0.1147,
+      "step": 4894
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.682317494334784e-06,
+      "loss": 0.1071,
+      "step": 4895
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.678846742063202e-06,
+      "loss": 0.1007,
+      "step": 4896
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.675376028517646e-06,
+      "loss": 0.1065,
+      "step": 4897
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.671905354116625e-06,
+      "loss": 0.1012,
+      "step": 4898
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.668434719278646e-06,
+      "loss": 0.1135,
+      "step": 4899
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664964124422218e-06,
+      "loss": 0.1043,
+      "step": 4900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.661493569965836e-06,
+      "loss": 0.0992,
+      "step": 4901
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.658023056327993e-06,
+      "loss": 0.1172,
+      "step": 4902
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654552583927185e-06,
+      "loss": 0.1046,
+      "step": 4903
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65108215318189e-06,
+      "loss": 0.1104,
+      "step": 4904
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64761176451059e-06,
+      "loss": 0.1251,
+      "step": 4905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.644141418331758e-06,
+      "loss": 0.1035,
+      "step": 4906
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640671115063866e-06,
+      "loss": 0.0981,
+      "step": 4907
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.637200855125373e-06,
+      "loss": 0.1105,
+      "step": 4908
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.633730638934739e-06,
+      "loss": 0.1092,
+      "step": 4909
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63026046691042e-06,
+      "loss": 0.1011,
+      "step": 4910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.626790339470864e-06,
+      "loss": 0.1059,
+      "step": 4911
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62332025703451e-06,
+      "loss": 0.1193,
+      "step": 4912
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619850220019799e-06,
+      "loss": 0.1124,
+      "step": 4913
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61638022884516e-06,
+      "loss": 0.1181,
+      "step": 4914
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.612910283929023e-06,
+      "loss": 0.102,
+      "step": 4915
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.609440385689804e-06,
+      "loss": 0.1069,
+      "step": 4916
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60597053454592e-06,
+      "loss": 0.1003,
+      "step": 4917
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.602500730915784e-06,
+      "loss": 0.1135,
+      "step": 4918
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.599030975217793e-06,
+      "loss": 0.1071,
+      "step": 4919
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.59556126787035e-06,
+      "loss": 0.1216,
+      "step": 4920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.592091609291844e-06,
+      "loss": 0.1173,
+      "step": 4921
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.588621999900664e-06,
+      "loss": 0.1145,
+      "step": 4922
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.585152440115187e-06,
+      "loss": 0.1112,
+      "step": 4923
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58168293035379e-06,
+      "loss": 0.1106,
+      "step": 4924
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.578213471034839e-06,
+      "loss": 0.1096,
+      "step": 4925
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.574744062576699e-06,
+      "loss": 0.1133,
+      "step": 4926
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.571274705397722e-06,
+      "loss": 0.1075,
+      "step": 4927
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56780539991626e-06,
+      "loss": 0.1099,
+      "step": 4928
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.564336146550658e-06,
+      "loss": 0.117,
+      "step": 4929
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.560866945719252e-06,
+      "loss": 0.111,
+      "step": 4930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.557397797840372e-06,
+      "loss": 0.1138,
+      "step": 4931
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.553928703332342e-06,
+      "loss": 0.1166,
+      "step": 4932
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.550459662613486e-06,
+      "loss": 0.1112,
+      "step": 4933
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.546990676102106e-06,
+      "loss": 0.1086,
+      "step": 4934
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.543521744216515e-06,
+      "loss": 0.1141,
+      "step": 4935
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54005286737501e-06,
+      "loss": 0.1126,
+      "step": 4936
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536584045995879e-06,
+      "loss": 0.1224,
+      "step": 4937
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.53311528049741e-06,
+      "loss": 0.1088,
+      "step": 4938
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.529646571297882e-06,
+      "loss": 0.1046,
+      "step": 4939
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.526177918815568e-06,
+      "loss": 0.1195,
+      "step": 4940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.522709323468729e-06,
+      "loss": 0.1098,
+      "step": 4941
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.519240785675623e-06,
+      "loss": 0.11,
+      "step": 4942
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.515772305854505e-06,
+      "loss": 0.1212,
+      "step": 4943
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.512303884423618e-06,
+      "loss": 0.1151,
+      "step": 4944
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.508835521801197e-06,
+      "loss": 0.1218,
+      "step": 4945
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50536721840547e-06,
+      "loss": 0.1135,
+      "step": 4946
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.501898974654667e-06,
+      "loss": 0.1094,
+      "step": 4947
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.498430790966995e-06,
+      "loss": 0.114,
+      "step": 4948
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.494962667760666e-06,
+      "loss": 0.1052,
+      "step": 4949
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.491494605453883e-06,
+      "loss": 0.1064,
+      "step": 4950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48802660446484e-06,
+      "loss": 0.1162,
+      "step": 4951
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.484558665211719e-06,
+      "loss": 0.1034,
+      "step": 4952
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.4810907881127e-06,
+      "loss": 0.1147,
+      "step": 4953
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.477622973585955e-06,
+      "loss": 0.112,
+      "step": 4954
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.474155222049652e-06,
+      "loss": 0.1126,
+      "step": 4955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47068753392194e-06,
+      "loss": 0.1045,
+      "step": 4956
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46721990962097e-06,
+      "loss": 0.1168,
+      "step": 4957
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.463752349564888e-06,
+      "loss": 0.1106,
+      "step": 4958
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.46028485417182e-06,
+      "loss": 0.1055,
+      "step": 4959
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.456817423859893e-06,
+      "loss": 0.111,
+      "step": 4960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.453350059047227e-06,
+      "loss": 0.1077,
+      "step": 4961
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.449882760151932e-06,
+      "loss": 0.1114,
+      "step": 4962
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.446415527592106e-06,
+      "loss": 0.1069,
+      "step": 4963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.442948361785844e-06,
+      "loss": 0.1206,
+      "step": 4964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.439481263151234e-06,
+      "loss": 0.1135,
+      "step": 4965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.436014232106349e-06,
+      "loss": 0.1082,
+      "step": 4966
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43254726906926e-06,
+      "loss": 0.1021,
+      "step": 4967
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42908037445803e-06,
+      "loss": 0.1078,
+      "step": 4968
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425613548690713e-06,
+      "loss": 0.1146,
+      "step": 4969
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42214679218535e-06,
+      "loss": 0.1206,
+      "step": 4970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418680105359977e-06,
+      "loss": 0.1086,
+      "step": 4971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415213488632624e-06,
+      "loss": 0.1137,
+      "step": 4972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.411746942421311e-06,
+      "loss": 0.1166,
+      "step": 4973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.408280467144047e-06,
+      "loss": 0.1036,
+      "step": 4974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404814063218833e-06,
+      "loss": 0.115,
+      "step": 4975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401347731063667e-06,
+      "loss": 0.1061,
+      "step": 4976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39788147109653e-06,
+      "loss": 0.1081,
+      "step": 4977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.394415283735399e-06,
+      "loss": 0.0998,
+      "step": 4978
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.390949169398242e-06,
+      "loss": 0.1184,
+      "step": 4979
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38748312850302e-06,
+      "loss": 0.0995,
+      "step": 4980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.384017161467682e-06,
+      "loss": 0.1046,
+      "step": 4981
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.380551268710166e-06,
+      "loss": 0.11,
+      "step": 4982
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.377085450648407e-06,
+      "loss": 0.1047,
+      "step": 4983
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.373619707700328e-06,
+      "loss": 0.1061,
+      "step": 4984
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370154040283841e-06,
+      "loss": 0.1153,
+      "step": 4985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366688448816852e-06,
+      "loss": 0.1146,
+      "step": 4986
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.363222933717258e-06,
+      "loss": 0.1122,
+      "step": 4987
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359757495402944e-06,
+      "loss": 0.1204,
+      "step": 4988
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.356292134291787e-06,
+      "loss": 0.107,
+      "step": 4989
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.352826850801655e-06,
+      "loss": 0.1101,
+      "step": 4990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.34936164535041e-06,
+      "loss": 0.1054,
+      "step": 4991
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.345896518355895e-06,
+      "loss": 0.1,
+      "step": 4992
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.342431470235953e-06,
+      "loss": 0.1158,
+      "step": 4993
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338966501408417e-06,
+      "loss": 0.1055,
+      "step": 4994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.335501612291103e-06,
+      "loss": 0.1103,
+      "step": 4995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.332036803301824e-06,
+      "loss": 0.1218,
+      "step": 4996
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32857207485838e-06,
+      "loss": 0.1095,
+      "step": 4997
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.325107427378568e-06,
+      "loss": 0.1073,
+      "step": 4998
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.321642861280162e-06,
+      "loss": 0.111,
+      "step": 4999
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.318178376980939e-06,
+      "loss": 0.1209,
+      "step": 5000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314713974898659e-06,
+      "loss": 0.1103,
+      "step": 5001
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.311249655451078e-06,
+      "loss": 0.1107,
+      "step": 5002
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307785419055935e-06,
+      "loss": 0.1063,
+      "step": 5003
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.304321266130961e-06,
+      "loss": 0.1016,
+      "step": 5004
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300857197093882e-06,
+      "loss": 0.1149,
+      "step": 5005
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29739321236241e-06,
+      "loss": 0.1104,
+      "step": 5006
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293929312354242e-06,
+      "loss": 0.1175,
+      "step": 5007
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.290465497487073e-06,
+      "loss": 0.1085,
+      "step": 5008
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.287001768178588e-06,
+      "loss": 0.1047,
+      "step": 5009
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.283538124846454e-06,
+      "loss": 0.1095,
+      "step": 5010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28007456790833e-06,
+      "loss": 0.1059,
+      "step": 5011
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276611097781872e-06,
+      "loss": 0.0952,
+      "step": 5012
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.273147714884718e-06,
+      "loss": 0.1204,
+      "step": 5013
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269684419634493e-06,
+      "loss": 0.1157,
+      "step": 5014
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26622121244882e-06,
+      "loss": 0.1197,
+      "step": 5015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262758093745308e-06,
+      "loss": 0.105,
+      "step": 5016
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25929506394155e-06,
+      "loss": 0.1127,
+      "step": 5017
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.255832123455138e-06,
+      "loss": 0.1236,
+      "step": 5018
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252369272703646e-06,
+      "loss": 0.102,
+      "step": 5019
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.248906512104641e-06,
+      "loss": 0.1067,
+      "step": 5020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.245443842075672e-06,
+      "loss": 0.1245,
+      "step": 5021
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.241981263034286e-06,
+      "loss": 0.1182,
+      "step": 5022
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238518775398018e-06,
+      "loss": 0.1031,
+      "step": 5023
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.235056379584385e-06,
+      "loss": 0.1123,
+      "step": 5024
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231594076010899e-06,
+      "loss": 0.1079,
+      "step": 5025
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22813186509506e-06,
+      "loss": 0.1183,
+      "step": 5026
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.224669747254357e-06,
+      "loss": 0.1071,
+      "step": 5027
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.221207722906264e-06,
+      "loss": 0.1168,
+      "step": 5028
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.217745792468247e-06,
+      "loss": 0.1085,
+      "step": 5029
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.214283956357761e-06,
+      "loss": 0.1219,
+      "step": 5030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210822214992253e-06,
+      "loss": 0.1096,
+      "step": 5031
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207360568789146e-06,
+      "loss": 0.1031,
+      "step": 5032
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.203899018165866e-06,
+      "loss": 0.11,
+      "step": 5033
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200437563539823e-06,
+      "loss": 0.1205,
+      "step": 5034
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196976205328408e-06,
+      "loss": 0.1091,
+      "step": 5035
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.193514943949008e-06,
+      "loss": 0.1021,
+      "step": 5036
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190053779819e-06,
+      "loss": 0.1196,
+      "step": 5037
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.186592713355743e-06,
+      "loss": 0.1138,
+      "step": 5038
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183131744976589e-06,
+      "loss": 0.1098,
+      "step": 5039
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.179670875098871e-06,
+      "loss": 0.1144,
+      "step": 5040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176210104139924e-06,
+      "loss": 0.1163,
+      "step": 5041
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.172749432517054e-06,
+      "loss": 0.1056,
+      "step": 5042
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16928886064757e-06,
+      "loss": 0.1078,
+      "step": 5043
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165828388948757e-06,
+      "loss": 0.1135,
+      "step": 5044
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.1623680178379e-06,
+      "loss": 0.1103,
+      "step": 5045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.158907747732258e-06,
+      "loss": 0.1119,
+      "step": 5046
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15544757904909e-06,
+      "loss": 0.1139,
+      "step": 5047
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.151987512205635e-06,
+      "loss": 0.1072,
+      "step": 5048
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.148527547619124e-06,
+      "loss": 0.1151,
+      "step": 5049
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145067685706774e-06,
+      "loss": 0.1128,
+      "step": 5050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.141607926885789e-06,
+      "loss": 0.0973,
+      "step": 5051
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.138148271573365e-06,
+      "loss": 0.1159,
+      "step": 5052
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.134688720186676e-06,
+      "loss": 0.1119,
+      "step": 5053
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.131229273142893e-06,
+      "loss": 0.1185,
+      "step": 5054
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.127769930859168e-06,
+      "loss": 0.1045,
+      "step": 5055
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.124310693752648e-06,
+      "loss": 0.1131,
+      "step": 5056
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120851562240459e-06,
+      "loss": 0.1057,
+      "step": 5057
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.11739253673972e-06,
+      "loss": 0.1111,
+      "step": 5058
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.113933617667533e-06,
+      "loss": 0.1055,
+      "step": 5059
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110474805440988e-06,
+      "loss": 0.0947,
+      "step": 5060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.107016100477166e-06,
+      "loss": 0.1168,
+      "step": 5061
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103557503193134e-06,
+      "loss": 0.1025,
+      "step": 5062
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.100099014005939e-06,
+      "loss": 0.1046,
+      "step": 5063
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.096640633332623e-06,
+      "loss": 0.1039,
+      "step": 5064
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.093182361590214e-06,
+      "loss": 0.1089,
+      "step": 5065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.089724199195721e-06,
+      "loss": 0.1029,
+      "step": 5066
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.086266146566146e-06,
+      "loss": 0.1006,
+      "step": 5067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.082808204118476e-06,
+      "loss": 0.1159,
+      "step": 5068
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.079350372269687e-06,
+      "loss": 0.0993,
+      "step": 5069
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.075892651436735e-06,
+      "loss": 0.111,
+      "step": 5070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.072435042036565e-06,
+      "loss": 0.108,
+      "step": 5071
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.068977544486115e-06,
+      "loss": 0.1031,
+      "step": 5072
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.065520159202305e-06,
+      "loss": 0.1178,
+      "step": 5073
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062062886602036e-06,
+      "loss": 0.1126,
+      "step": 5074
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.058605727102203e-06,
+      "loss": 0.1113,
+      "step": 5075
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055148681119688e-06,
+      "loss": 0.1091,
+      "step": 5076
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.051691749071351e-06,
+      "loss": 0.0959,
+      "step": 5077
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.04823493137405e-06,
+      "loss": 0.1067,
+      "step": 5078
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044778228444615e-06,
+      "loss": 0.1065,
+      "step": 5079
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.041321640699879e-06,
+      "loss": 0.107,
+      "step": 5080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.037865168556642e-06,
+      "loss": 0.1073,
+      "step": 5081
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.034408812431707e-06,
+      "loss": 0.1124,
+      "step": 5082
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.030952572741855e-06,
+      "loss": 0.1152,
+      "step": 5083
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.027496449903852e-06,
+      "loss": 0.1152,
+      "step": 5084
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.024040444334453e-06,
+      "loss": 0.0977,
+      "step": 5085
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.020584556450398e-06,
+      "loss": 0.1047,
+      "step": 5086
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.017128786668415e-06,
+      "loss": 0.1085,
+      "step": 5087
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01367313540521e-06,
+      "loss": 0.1087,
+      "step": 5088
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.010217603077484e-06,
+      "loss": 0.109,
+      "step": 5089
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.00676219010192e-06,
+      "loss": 0.1064,
+      "step": 5090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.003306896895184e-06,
+      "loss": 0.1067,
+      "step": 5091
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99985172387393e-06,
+      "loss": 0.107,
+      "step": 5092
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.9963966714548e-06,
+      "loss": 0.0985,
+      "step": 5093
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.992941740054418e-06,
+      "loss": 0.1115,
+      "step": 5094
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.989486930089392e-06,
+      "loss": 0.1039,
+      "step": 5095
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98603224197632e-06,
+      "loss": 0.1085,
+      "step": 5096
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.982577676131781e-06,
+      "loss": 0.1084,
+      "step": 5097
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.979123232972346e-06,
+      "loss": 0.119,
+      "step": 5098
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97566891291456e-06,
+      "loss": 0.1042,
+      "step": 5099
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.972214716374963e-06,
+      "loss": 0.1138,
+      "step": 5100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.968760643770075e-06,
+      "loss": 0.1117,
+      "step": 5101
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.965306695516407e-06,
+      "loss": 0.1081,
+      "step": 5102
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.961852872030445e-06,
+      "loss": 0.1046,
+      "step": 5103
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.958399173728668e-06,
+      "loss": 0.1155,
+      "step": 5104
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.95494560102754e-06,
+      "loss": 0.1237,
+      "step": 5105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951492154343505e-06,
+      "loss": 0.1149,
+      "step": 5106
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.948038834092993e-06,
+      "loss": 0.1147,
+      "step": 5107
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.944585640692421e-06,
+      "loss": 0.1172,
+      "step": 5108
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941132574558193e-06,
+      "loss": 0.1064,
+      "step": 5109
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.937679636106692e-06,
+      "loss": 0.1317,
+      "step": 5110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.934226825754285e-06,
+      "loss": 0.1075,
+      "step": 5111
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93077414391733e-06,
+      "loss": 0.1088,
+      "step": 5112
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.927321591012164e-06,
+      "loss": 0.1153,
+      "step": 5113
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.923869167455113e-06,
+      "loss": 0.1169,
+      "step": 5114
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.920416873662483e-06,
+      "loss": 0.1064,
+      "step": 5115
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.916964710050568e-06,
+      "loss": 0.1191,
+      "step": 5116
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.913512677035644e-06,
+      "loss": 0.1107,
+      "step": 5117
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910060775033968e-06,
+      "loss": 0.1074,
+      "step": 5118
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.906609004461788e-06,
+      "loss": 0.1029,
+      "step": 5119
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.903157365735337e-06,
+      "loss": 0.1146,
+      "step": 5120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.899705859270823e-06,
+      "loss": 0.1111,
+      "step": 5121
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896254485484443e-06,
+      "loss": 0.1106,
+      "step": 5122
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.892803244792382e-06,
+      "loss": 0.1167,
+      "step": 5123
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8893521376108e-06,
+      "loss": 0.1161,
+      "step": 5124
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.885901164355854e-06,
+      "loss": 0.0943,
+      "step": 5125
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882450325443668e-06,
+      "loss": 0.107,
+      "step": 5126
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.878999621290369e-06,
+      "loss": 0.1117,
+      "step": 5127
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.875549052312047e-06,
+      "loss": 0.1136,
+      "step": 5128
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.872098618924791e-06,
+      "loss": 0.1134,
+      "step": 5129
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.868648321544672e-06,
+      "loss": 0.1023,
+      "step": 5130
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865198160587735e-06,
+      "loss": 0.1164,
+      "step": 5131
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861748136470021e-06,
+      "loss": 0.1184,
+      "step": 5132
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.858298249607544e-06,
+      "loss": 0.1075,
+      "step": 5133
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.854848500416311e-06,
+      "loss": 0.1117,
+      "step": 5134
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.851398889312301e-06,
+      "loss": 0.1092,
+      "step": 5135
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.847949416711488e-06,
+      "loss": 0.1108,
+      "step": 5136
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844500083029822e-06,
+      "loss": 0.1182,
+      "step": 5137
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.841050888683238e-06,
+      "loss": 0.1187,
+      "step": 5138
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.837601834087656e-06,
+      "loss": 0.1199,
+      "step": 5139
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834152919658976e-06,
+      "loss": 0.1092,
+      "step": 5140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.830704145813086e-06,
+      "loss": 0.1057,
+      "step": 5141
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.827255512965846e-06,
+      "loss": 0.1075,
+      "step": 5142
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.823807021533113e-06,
+      "loss": 0.112,
+      "step": 5143
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.820358671930721e-06,
+      "loss": 0.1162,
+      "step": 5144
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.816910464574486e-06,
+      "loss": 0.1089,
+      "step": 5145
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.813462399880206e-06,
+      "loss": 0.1218,
+      "step": 5146
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.810014478263662e-06,
+      "loss": 0.0995,
+      "step": 5147
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806566700140624e-06,
+      "loss": 0.0965,
+      "step": 5148
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.803119065926837e-06,
+      "loss": 0.1002,
+      "step": 5149
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79967157603803e-06,
+      "loss": 0.1183,
+      "step": 5150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.796224230889915e-06,
+      "loss": 0.0986,
+      "step": 5151
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.792777030898194e-06,
+      "loss": 0.1016,
+      "step": 5152
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789329976478535e-06,
+      "loss": 0.1042,
+      "step": 5153
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.785883068046607e-06,
+      "loss": 0.1125,
+      "step": 5154
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.78243630601805e-06,
+      "loss": 0.0941,
+      "step": 5155
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77898969080849e-06,
+      "loss": 0.1052,
+      "step": 5156
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.775543222833532e-06,
+      "loss": 0.1219,
+      "step": 5157
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.772096902508765e-06,
+      "loss": 0.1071,
+      "step": 5158
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.768650730249768e-06,
+      "loss": 0.0951,
+      "step": 5159
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765204706472086e-06,
+      "loss": 0.1048,
+      "step": 5160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761758831591262e-06,
+      "loss": 0.1185,
+      "step": 5161
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75831310602281e-06,
+      "loss": 0.1091,
+      "step": 5162
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754867530182234e-06,
+      "loss": 0.1057,
+      "step": 5163
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.751422104485013e-06,
+      "loss": 0.1017,
+      "step": 5164
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74797682934661e-06,
+      "loss": 0.1084,
+      "step": 5165
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744531705182475e-06,
+      "loss": 0.1113,
+      "step": 5166
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.741086732408036e-06,
+      "loss": 0.1027,
+      "step": 5167
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.737641911438696e-06,
+      "loss": 0.1056,
+      "step": 5168
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.734197242689851e-06,
+      "loss": 0.095,
+      "step": 5169
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.730752726576874e-06,
+      "loss": 0.1118,
+      "step": 5170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.727308363515117e-06,
+      "loss": 0.1091,
+      "step": 5171
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.723864153919918e-06,
+      "loss": 0.1111,
+      "step": 5172
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.720420098206592e-06,
+      "loss": 0.1095,
+      "step": 5173
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716976196790444e-06,
+      "loss": 0.1195,
+      "step": 5174
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.713532450086745e-06,
+      "loss": 0.1084,
+      "step": 5175
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710088858510764e-06,
+      "loss": 0.1173,
+      "step": 5176
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.706645422477739e-06,
+      "loss": 0.1059,
+      "step": 5177
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.703202142402898e-06,
+      "loss": 0.1111,
+      "step": 5178
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.699759018701444e-06,
+      "loss": 0.1138,
+      "step": 5179
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.696316051788564e-06,
+      "loss": 0.114,
+      "step": 5180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692873242079428e-06,
+      "loss": 0.1098,
+      "step": 5181
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68943058998918e-06,
+      "loss": 0.1128,
+      "step": 5182
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.685988095932951e-06,
+      "loss": 0.121,
+      "step": 5183
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.682545760325854e-06,
+      "loss": 0.1031,
+      "step": 5184
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67910358358298e-06,
+      "loss": 0.0994,
+      "step": 5185
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.675661566119397e-06,
+      "loss": 0.1132,
+      "step": 5186
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.672219708350162e-06,
+      "loss": 0.1078,
+      "step": 5187
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66877801069031e-06,
+      "loss": 0.1247,
+      "step": 5188
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.665336473554854e-06,
+      "loss": 0.103,
+      "step": 5189
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661895097358786e-06,
+      "loss": 0.1113,
+      "step": 5190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.658453882517084e-06,
+      "loss": 0.1245,
+      "step": 5191
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.655012829444707e-06,
+      "loss": 0.1027,
+      "step": 5192
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65157193855659e-06,
+      "loss": 0.1135,
+      "step": 5193
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.648131210267644e-06,
+      "loss": 0.1121,
+      "step": 5194
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644690644992777e-06,
+      "loss": 0.0985,
+      "step": 5195
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.641250243146863e-06,
+      "loss": 0.1171,
+      "step": 5196
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.637810005144757e-06,
+      "loss": 0.1136,
+      "step": 5197
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.634369931401302e-06,
+      "loss": 0.1061,
+      "step": 5198
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.630930022331315e-06,
+      "loss": 0.0952,
+      "step": 5199
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.627490278349593e-06,
+      "loss": 0.1132,
+      "step": 5200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.624050699870916e-06,
+      "loss": 0.1177,
+      "step": 5201
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.620611287310044e-06,
+      "loss": 0.112,
+      "step": 5202
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.617172041081717e-06,
+      "loss": 0.1135,
+      "step": 5203
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613732961600651e-06,
+      "loss": 0.1032,
+      "step": 5204
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.610294049281544e-06,
+      "loss": 0.1148,
+      "step": 5205
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.606855304539079e-06,
+      "loss": 0.1104,
+      "step": 5206
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.603416727787912e-06,
+      "loss": 0.104,
+      "step": 5207
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.599978319442681e-06,
+      "loss": 0.1134,
+      "step": 5208
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.596540079918003e-06,
+      "loss": 0.1176,
+      "step": 5209
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59310200962848e-06,
+      "loss": 0.1141,
+      "step": 5210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.589664108988683e-06,
+      "loss": 0.1128,
+      "step": 5211
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586226378413172e-06,
+      "loss": 0.1109,
+      "step": 5212
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.582788818316481e-06,
+      "loss": 0.1331,
+      "step": 5213
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.57935142911313e-06,
+      "loss": 0.1193,
+      "step": 5214
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.575914211217612e-06,
+      "loss": 0.1135,
+      "step": 5215
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.572477165044395e-06,
+      "loss": 0.106,
+      "step": 5216
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.569040291007944e-06,
+      "loss": 0.1119,
+      "step": 5217
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565603589522683e-06,
+      "loss": 0.1089,
+      "step": 5218
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.562167061003025e-06,
+      "loss": 0.124,
+      "step": 5219
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.558730705863367e-06,
+      "loss": 0.1054,
+      "step": 5220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.555294524518074e-06,
+      "loss": 0.1162,
+      "step": 5221
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.551858517381496e-06,
+      "loss": 0.1141,
+      "step": 5222
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54842268486796e-06,
+      "loss": 0.1065,
+      "step": 5223
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.544987027391778e-06,
+      "loss": 0.1136,
+      "step": 5224
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.541551545367232e-06,
+      "loss": 0.1214,
+      "step": 5225
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.538116239208588e-06,
+      "loss": 0.1126,
+      "step": 5226
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.534681109330094e-06,
+      "loss": 0.1028,
+      "step": 5227
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.531246156145968e-06,
+      "loss": 0.0992,
+      "step": 5228
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52781138007041e-06,
+      "loss": 0.1082,
+      "step": 5229
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.524376781517605e-06,
+      "loss": 0.1062,
+      "step": 5230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.520942360901704e-06,
+      "loss": 0.1242,
+      "step": 5231
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.517508118636852e-06,
+      "loss": 0.1086,
+      "step": 5232
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.514074055137158e-06,
+      "loss": 0.1111,
+      "step": 5233
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.510640170816722e-06,
+      "loss": 0.1175,
+      "step": 5234
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.507206466089612e-06,
+      "loss": 0.1128,
+      "step": 5235
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.503772941369878e-06,
+      "loss": 0.1086,
+      "step": 5236
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.500339597071552e-06,
+      "loss": 0.096,
+      "step": 5237
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.496906433608642e-06,
+      "loss": 0.1106,
+      "step": 5238
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.493473451395129e-06,
+      "loss": 0.1088,
+      "step": 5239
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.490040650844978e-06,
+      "loss": 0.1013,
+      "step": 5240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486608032372135e-06,
+      "loss": 0.0941,
+      "step": 5241
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.483175596390515e-06,
+      "loss": 0.1181,
+      "step": 5242
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.479743343314016e-06,
+      "loss": 0.0973,
+      "step": 5243
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.476311273556516e-06,
+      "loss": 0.0946,
+      "step": 5244
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.472879387531867e-06,
+      "loss": 0.1033,
+      "step": 5245
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.469447685653902e-06,
+      "loss": 0.1026,
+      "step": 5246
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.466016168336425e-06,
+      "loss": 0.1237,
+      "step": 5247
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46258483599323e-06,
+      "loss": 0.1104,
+      "step": 5248
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.459153689038077e-06,
+      "loss": 0.1098,
+      "step": 5249
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.45572272788471e-06,
+      "loss": 0.1108,
+      "step": 5250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.452291952946847e-06,
+      "loss": 0.1017,
+      "step": 5251
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44886136463819e-06,
+      "loss": 0.1129,
+      "step": 5252
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44543096337241e-06,
+      "loss": 0.1155,
+      "step": 5253
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.442000749563157e-06,
+      "loss": 0.1115,
+      "step": 5254
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.438570723624066e-06,
+      "loss": 0.1075,
+      "step": 5255
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.435140885968744e-06,
+      "loss": 0.1112,
+      "step": 5256
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.431711237010768e-06,
+      "loss": 0.1248,
+      "step": 5257
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42828177716371e-06,
+      "loss": 0.1119,
+      "step": 5258
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.424852506841102e-06,
+      "loss": 0.1154,
+      "step": 5259
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42142342645646e-06,
+      "loss": 0.1001,
+      "step": 5260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41799453642328e-06,
+      "loss": 0.1099,
+      "step": 5261
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41456583715503e-06,
+      "loss": 0.1212,
+      "step": 5262
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41113732906516e-06,
+      "loss": 0.1028,
+      "step": 5263
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40770901256709e-06,
+      "loss": 0.0889,
+      "step": 5264
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.404280888074223e-06,
+      "loss": 0.1091,
+      "step": 5265
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.400852955999938e-06,
+      "loss": 0.0984,
+      "step": 5266
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397425216757588e-06,
+      "loss": 0.125,
+      "step": 5267
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.393997670760505e-06,
+      "loss": 0.0973,
+      "step": 5268
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.390570318421996e-06,
+      "loss": 0.1112,
+      "step": 5269
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.387143160155347e-06,
+      "loss": 0.1073,
+      "step": 5270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.383716196373817e-06,
+      "loss": 0.1135,
+      "step": 5271
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.380289427490647e-06,
+      "loss": 0.1065,
+      "step": 5272
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37686285391905e-06,
+      "loss": 0.0988,
+      "step": 5273
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.373436476072217e-06,
+      "loss": 0.0884,
+      "step": 5274
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370010294363313e-06,
+      "loss": 0.0969,
+      "step": 5275
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366584309205483e-06,
+      "loss": 0.1103,
+      "step": 5276
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.363158521011851e-06,
+      "loss": 0.1121,
+      "step": 5277
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359732930195506e-06,
+      "loss": 0.1156,
+      "step": 5278
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356307537169524e-06,
+      "loss": 0.1094,
+      "step": 5279
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.352882342346951e-06,
+      "loss": 0.1078,
+      "step": 5280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.349457346140818e-06,
+      "loss": 0.1192,
+      "step": 5281
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.346032548964118e-06,
+      "loss": 0.1104,
+      "step": 5282
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.34260795122983e-06,
+      "loss": 0.1093,
+      "step": 5283
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.339183553350908e-06,
+      "loss": 0.124,
+      "step": 5284
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.335759355740282e-06,
+      "loss": 0.1006,
+      "step": 5285
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33233535881085e-06,
+      "loss": 0.1149,
+      "step": 5286
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328911562975498e-06,
+      "loss": 0.1212,
+      "step": 5287
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.325487968647081e-06,
+      "loss": 0.1023,
+      "step": 5288
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.322064576238426e-06,
+      "loss": 0.0997,
+      "step": 5289
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.318641386162345e-06,
+      "loss": 0.1123,
+      "step": 5290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31521839883162e-06,
+      "loss": 0.1081,
+      "step": 5291
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311795614659008e-06,
+      "loss": 0.1172,
+      "step": 5292
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308373034057243e-06,
+      "loss": 0.1156,
+      "step": 5293
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.304950657439034e-06,
+      "loss": 0.1384,
+      "step": 5294
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.301528485217067e-06,
+      "loss": 0.0965,
+      "step": 5295
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.298106517804003e-06,
+      "loss": 0.1115,
+      "step": 5296
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.294684755612475e-06,
+      "loss": 0.1131,
+      "step": 5297
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.291263199055092e-06,
+      "loss": 0.1069,
+      "step": 5298
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287841848544446e-06,
+      "loss": 0.1187,
+      "step": 5299
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.284420704493092e-06,
+      "loss": 0.1076,
+      "step": 5300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.280999767313567e-06,
+      "loss": 0.1075,
+      "step": 5301
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.277579037418385e-06,
+      "loss": 0.1073,
+      "step": 5302
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.274158515220031e-06,
+      "loss": 0.1044,
+      "step": 5303
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270738201130964e-06,
+      "loss": 0.1093,
+      "step": 5304
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.267318095563623e-06,
+      "loss": 0.1127,
+      "step": 5305
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263898198930416e-06,
+      "loss": 0.1116,
+      "step": 5306
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.260478511643731e-06,
+      "loss": 0.1159,
+      "step": 5307
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.257059034115926e-06,
+      "loss": 0.1159,
+      "step": 5308
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.253639766759338e-06,
+      "loss": 0.1145,
+      "step": 5309
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.250220709986278e-06,
+      "loss": 0.1098,
+      "step": 5310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.246801864209027e-06,
+      "loss": 0.1017,
+      "step": 5311
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.243383229839846e-06,
+      "loss": 0.1051,
+      "step": 5312
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239964807290968e-06,
+      "loss": 0.1149,
+      "step": 5313
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.236546596974602e-06,
+      "loss": 0.1026,
+      "step": 5314
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.233128599302928e-06,
+      "loss": 0.1098,
+      "step": 5315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.229710814688104e-06,
+      "loss": 0.1145,
+      "step": 5316
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22629324354226e-06,
+      "loss": 0.1066,
+      "step": 5317
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222875886277504e-06,
+      "loss": 0.097,
+      "step": 5318
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21945874330591e-06,
+      "loss": 0.1113,
+      "step": 5319
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.216041815039538e-06,
+      "loss": 0.108,
+      "step": 5320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.212625101890415e-06,
+      "loss": 0.1063,
+      "step": 5321
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209208604270536e-06,
+      "loss": 0.1069,
+      "step": 5322
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.205792322591883e-06,
+      "loss": 0.1094,
+      "step": 5323
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.202376257266404e-06,
+      "loss": 0.1021,
+      "step": 5324
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198960408706024e-06,
+      "loss": 0.1097,
+      "step": 5325
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.195544777322639e-06,
+      "loss": 0.112,
+      "step": 5326
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19212936352812e-06,
+      "loss": 0.098,
+      "step": 5327
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.188714167734313e-06,
+      "loss": 0.1155,
+      "step": 5328
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.185299190353035e-06,
+      "loss": 0.1086,
+      "step": 5329
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.181884431796082e-06,
+      "loss": 0.1116,
+      "step": 5330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178469892475216e-06,
+      "loss": 0.1128,
+      "step": 5331
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.175055572802183e-06,
+      "loss": 0.1005,
+      "step": 5332
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.171641473188687e-06,
+      "loss": 0.1069,
+      "step": 5333
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16822759404642e-06,
+      "loss": 0.1057,
+      "step": 5334
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.164813935787044e-06,
+      "loss": 0.1111,
+      "step": 5335
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161400498822188e-06,
+      "loss": 0.1114,
+      "step": 5336
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15798728356346e-06,
+      "loss": 0.0938,
+      "step": 5337
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15457429042244e-06,
+      "loss": 0.1059,
+      "step": 5338
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.151161519810685e-06,
+      "loss": 0.0933,
+      "step": 5339
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147748972139715e-06,
+      "loss": 0.105,
+      "step": 5340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.144336647821033e-06,
+      "loss": 0.1091,
+      "step": 5341
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.140924547266111e-06,
+      "loss": 0.107,
+      "step": 5342
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.137512670886397e-06,
+      "loss": 0.099,
+      "step": 5343
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.134101019093305e-06,
+      "loss": 0.1113,
+      "step": 5344
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.130689592298229e-06,
+      "loss": 0.1133,
+      "step": 5345
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.127278390912535e-06,
+      "loss": 0.1055,
+      "step": 5346
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.123867415347556e-06,
+      "loss": 0.1169,
+      "step": 5347
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.120456666014603e-06,
+      "loss": 0.1009,
+      "step": 5348
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11704614332496e-06,
+      "loss": 0.1052,
+      "step": 5349
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.113635847689884e-06,
+      "loss": 0.1076,
+      "step": 5350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.1102257795206e-06,
+      "loss": 0.0968,
+      "step": 5351
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.106815939228309e-06,
+      "loss": 0.0972,
+      "step": 5352
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.103406327224185e-06,
+      "loss": 0.1078,
+      "step": 5353
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.099996943919372e-06,
+      "loss": 0.1138,
+      "step": 5354
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096587789724988e-06,
+      "loss": 0.1051,
+      "step": 5355
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.093178865052125e-06,
+      "loss": 0.1015,
+      "step": 5356
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.089770170311845e-06,
+      "loss": 0.1221,
+      "step": 5357
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.086361705915182e-06,
+      "loss": 0.1134,
+      "step": 5358
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.082953472273144e-06,
+      "loss": 0.0916,
+      "step": 5359
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.079545469796708e-06,
+      "loss": 0.1033,
+      "step": 5360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07613769889683e-06,
+      "loss": 0.1016,
+      "step": 5361
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.072730159984429e-06,
+      "loss": 0.1115,
+      "step": 5362
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.069322853470402e-06,
+      "loss": 0.1141,
+      "step": 5363
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.065915779765619e-06,
+      "loss": 0.1117,
+      "step": 5364
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.062508939280915e-06,
+      "loss": 0.11,
+      "step": 5365
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.059102332427102e-06,
+      "loss": 0.1103,
+      "step": 5366
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.055695959614967e-06,
+      "loss": 0.1111,
+      "step": 5367
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.052289821255262e-06,
+      "loss": 0.1149,
+      "step": 5368
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.048883917758713e-06,
+      "loss": 0.0983,
+      "step": 5369
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.045478249536019e-06,
+      "loss": 0.1018,
+      "step": 5370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04207281699785e-06,
+      "loss": 0.1192,
+      "step": 5371
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.038667620554851e-06,
+      "loss": 0.1102,
+      "step": 5372
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.035262660617628e-06,
+      "loss": 0.0998,
+      "step": 5373
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.03185793759677e-06,
+      "loss": 0.1132,
+      "step": 5374
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.028453451902836e-06,
+      "loss": 0.1036,
+      "step": 5375
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.025049203946346e-06,
+      "loss": 0.1047,
+      "step": 5376
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.021645194137801e-06,
+      "loss": 0.0921,
+      "step": 5377
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.018241422887673e-06,
+      "loss": 0.1014,
+      "step": 5378
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.014837890606407e-06,
+      "loss": 0.1044,
+      "step": 5379
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.011434597704405e-06,
+      "loss": 0.1034,
+      "step": 5380
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.008031544592056e-06,
+      "loss": 0.0961,
+      "step": 5381
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00462873167972e-06,
+      "loss": 0.1032,
+      "step": 5382
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.001226159377713e-06,
+      "loss": 0.11,
+      "step": 5383
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.997823828096336e-06,
+      "loss": 0.0955,
+      "step": 5384
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.994421738245856e-06,
+      "loss": 0.1067,
+      "step": 5385
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991019890236517e-06,
+      "loss": 0.1062,
+      "step": 5386
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98761828447852e-06,
+      "loss": 0.1021,
+      "step": 5387
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.984216921382046e-06,
+      "loss": 0.1123,
+      "step": 5388
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.980815801357249e-06,
+      "loss": 0.1196,
+      "step": 5389
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.977414924814249e-06,
+      "loss": 0.0976,
+      "step": 5390
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.97401429216314e-06,
+      "loss": 0.1101,
+      "step": 5391
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.970613903813984e-06,
+      "loss": 0.1109,
+      "step": 5392
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.967213760176813e-06,
+      "loss": 0.1167,
+      "step": 5393
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.963813861661632e-06,
+      "loss": 0.1137,
+      "step": 5394
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.960414208678416e-06,
+      "loss": 0.1115,
+      "step": 5395
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.957014801637104e-06,
+      "loss": 0.1087,
+      "step": 5396
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.953615640947618e-06,
+      "loss": 0.1162,
+      "step": 5397
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.95021672701984e-06,
+      "loss": 0.1108,
+      "step": 5398
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.946818060263625e-06,
+      "loss": 0.1051,
+      "step": 5399
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.943419641088802e-06,
+      "loss": 0.112,
+      "step": 5400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940021469905159e-06,
+      "loss": 0.1131,
+      "step": 5401
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.936623547122471e-06,
+      "loss": 0.1197,
+      "step": 5402
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93322587315047e-06,
+      "loss": 0.097,
+      "step": 5403
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92982844839886e-06,
+      "loss": 0.113,
+      "step": 5404
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92643127327732e-06,
+      "loss": 0.1046,
+      "step": 5405
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.923034348195497e-06,
+      "loss": 0.1131,
+      "step": 5406
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919637673563003e-06,
+      "loss": 0.1081,
+      "step": 5407
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.916241249789423e-06,
+      "loss": 0.1076,
+      "step": 5408
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.912845077284314e-06,
+      "loss": 0.1045,
+      "step": 5409
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.909449156457202e-06,
+      "loss": 0.1175,
+      "step": 5410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90605348771758e-06,
+      "loss": 0.0934,
+      "step": 5411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.902658071474912e-06,
+      "loss": 0.1133,
+      "step": 5412
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899262908138633e-06,
+      "loss": 0.12,
+      "step": 5413
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.895867998118148e-06,
+      "loss": 0.1129,
+      "step": 5414
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.892473341822823e-06,
+      "loss": 0.1093,
+      "step": 5415
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.889078939662006e-06,
+      "loss": 0.1174,
+      "step": 5416
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885684792045006e-06,
+      "loss": 0.1227,
+      "step": 5417
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882290899381105e-06,
+      "loss": 0.1159,
+      "step": 5418
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87889726207955e-06,
+      "loss": 0.1074,
+      "step": 5419
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.875503880549563e-06,
+      "loss": 0.1106,
+      "step": 5420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.872110755200335e-06,
+      "loss": 0.1009,
+      "step": 5421
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.868717886441017e-06,
+      "loss": 0.1093,
+      "step": 5422
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.865325274680739e-06,
+      "loss": 0.1235,
+      "step": 5423
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.861932920328598e-06,
+      "loss": 0.1176,
+      "step": 5424
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.858540823793655e-06,
+      "loss": 0.1101,
+      "step": 5425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.855148985484946e-06,
+      "loss": 0.1159,
+      "step": 5426
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.851757405811471e-06,
+      "loss": 0.1121,
+      "step": 5427
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.848366085182207e-06,
+      "loss": 0.1032,
+      "step": 5428
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.844975024006085e-06,
+      "loss": 0.1034,
+      "step": 5429
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.841584222692017e-06,
+      "loss": 0.101,
+      "step": 5430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.838193681648883e-06,
+      "loss": 0.1108,
+      "step": 5431
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.834803401285528e-06,
+      "loss": 0.1089,
+      "step": 5432
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.831413382010763e-06,
+      "loss": 0.1066,
+      "step": 5433
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.828023624233373e-06,
+      "loss": 0.1022,
+      "step": 5434
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.824634128362111e-06,
+      "loss": 0.1043,
+      "step": 5435
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.821244894805695e-06,
+      "loss": 0.1058,
+      "step": 5436
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.817855923972813e-06,
+      "loss": 0.1094,
+      "step": 5437
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.814467216272122e-06,
+      "loss": 0.1157,
+      "step": 5438
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.811078772112248e-06,
+      "loss": 0.1107,
+      "step": 5439
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807690591901783e-06,
+      "loss": 0.106,
+      "step": 5440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.804302676049285e-06,
+      "loss": 0.1079,
+      "step": 5441
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800915024963289e-06,
+      "loss": 0.0975,
+      "step": 5442
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.797527639052288e-06,
+      "loss": 0.1039,
+      "step": 5443
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.794140518724747e-06,
+      "loss": 0.107,
+      "step": 5444
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790753664389103e-06,
+      "loss": 0.0985,
+      "step": 5445
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.787367076453755e-06,
+      "loss": 0.1087,
+      "step": 5446
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.783980755327074e-06,
+      "loss": 0.1021,
+      "step": 5447
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780594701417394e-06,
+      "loss": 0.0967,
+      "step": 5448
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.777208915133022e-06,
+      "loss": 0.1151,
+      "step": 5449
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77382339688223e-06,
+      "loss": 0.1142,
+      "step": 5450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.770438147073256e-06,
+      "loss": 0.1113,
+      "step": 5451
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.767053166114312e-06,
+      "loss": 0.1114,
+      "step": 5452
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76366845441357e-06,
+      "loss": 0.1143,
+      "step": 5453
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.760284012379174e-06,
+      "loss": 0.1116,
+      "step": 5454
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.756899840419234e-06,
+      "loss": 0.107,
+      "step": 5455
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.753515938941826e-06,
+      "loss": 0.1057,
+      "step": 5456
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.750132308355002e-06,
+      "loss": 0.1083,
+      "step": 5457
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.746748949066768e-06,
+      "loss": 0.1229,
+      "step": 5458
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.743365861485103e-06,
+      "loss": 0.1058,
+      "step": 5459
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73998304601796e-06,
+      "loss": 0.0871,
+      "step": 5460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73660050307325e-06,
+      "loss": 0.1082,
+      "step": 5461
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.733218233058854e-06,
+      "loss": 0.0998,
+      "step": 5462
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.729836236382618e-06,
+      "loss": 0.1062,
+      "step": 5463
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.726454513452366e-06,
+      "loss": 0.109,
+      "step": 5464
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72307306467587e-06,
+      "loss": 0.1071,
+      "step": 5465
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719691890460884e-06,
+      "loss": 0.1217,
+      "step": 5466
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716310991215126e-06,
+      "loss": 0.1052,
+      "step": 5467
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.712930367346279e-06,
+      "loss": 0.1044,
+      "step": 5468
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.709550019261991e-06,
+      "loss": 0.1059,
+      "step": 5469
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70616994736988e-06,
+      "loss": 0.1088,
+      "step": 5470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702790152077528e-06,
+      "loss": 0.1098,
+      "step": 5471
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.699410633792486e-06,
+      "loss": 0.1069,
+      "step": 5472
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.696031392922269e-06,
+      "loss": 0.1058,
+      "step": 5473
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.692652429874363e-06,
+      "loss": 0.108,
+      "step": 5474
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.689273745056219e-06,
+      "loss": 0.1064,
+      "step": 5475
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.685895338875248e-06,
+      "loss": 0.1084,
+      "step": 5476
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.682517211738834e-06,
+      "loss": 0.1093,
+      "step": 5477
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67913936405433e-06,
+      "loss": 0.1026,
+      "step": 5478
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.675761796229048e-06,
+      "loss": 0.106,
+      "step": 5479
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.672384508670267e-06,
+      "loss": 0.1053,
+      "step": 5480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66900750178524e-06,
+      "loss": 0.1031,
+      "step": 5481
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66563077598118e-06,
+      "loss": 0.1062,
+      "step": 5482
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.662254331665261e-06,
+      "loss": 0.1189,
+      "step": 5483
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.658878169244634e-06,
+      "loss": 0.1081,
+      "step": 5484
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65550228912641e-06,
+      "loss": 0.1037,
+      "step": 5485
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65212669171767e-06,
+      "loss": 0.1076,
+      "step": 5486
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.648751377425452e-06,
+      "loss": 0.1059,
+      "step": 5487
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.645376346656769e-06,
+      "loss": 0.1167,
+      "step": 5488
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642001599818596e-06,
+      "loss": 0.0956,
+      "step": 5489
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.638627137317877e-06,
+      "loss": 0.1106,
+      "step": 5490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635252959561513e-06,
+      "loss": 0.1103,
+      "step": 5491
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631879066956382e-06,
+      "loss": 0.0993,
+      "step": 5492
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6285054599093215e-06,
+      "loss": 0.1086,
+      "step": 5493
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.625132138827132e-06,
+      "loss": 0.1017,
+      "step": 5494
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.621759104116586e-06,
+      "loss": 0.1006,
+      "step": 5495
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6183863561844175e-06,
+      "loss": 0.1014,
+      "step": 5496
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615013895437327e-06,
+      "loss": 0.1143,
+      "step": 5497
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6116417222819794e-06,
+      "loss": 0.1139,
+      "step": 5498
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.608269837125006e-06,
+      "loss": 0.1085,
+      "step": 5499
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6048982403730056e-06,
+      "loss": 0.1015,
+      "step": 5500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.601526932432535e-06,
+      "loss": 0.0997,
+      "step": 5501
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.598155913710121e-06,
+      "loss": 0.1065,
+      "step": 5502
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.59478518461226e-06,
+      "loss": 0.1027,
+      "step": 5503
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5914147455454065e-06,
+      "loss": 0.1025,
+      "step": 5504
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.588044596915981e-06,
+      "loss": 0.1121,
+      "step": 5505
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.584674739130371e-06,
+      "loss": 0.1039,
+      "step": 5506
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.581305172594928e-06,
+      "loss": 0.1176,
+      "step": 5507
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5779358977159715e-06,
+      "loss": 0.112,
+      "step": 5508
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.574566914899779e-06,
+      "loss": 0.0982,
+      "step": 5509
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.571198224552598e-06,
+      "loss": 0.117,
+      "step": 5510
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5678298270806414e-06,
+      "loss": 0.1076,
+      "step": 5511
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564461722890082e-06,
+      "loss": 0.1056,
+      "step": 5512
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561093912387062e-06,
+      "loss": 0.1075,
+      "step": 5513
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.557726395977685e-06,
+      "loss": 0.1129,
+      "step": 5514
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.554359174068022e-06,
+      "loss": 0.1115,
+      "step": 5515
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5509922470641066e-06,
+      "loss": 0.1062,
+      "step": 5516
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.547625615371934e-06,
+      "loss": 0.108,
+      "step": 5517
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54425927939747e-06,
+      "loss": 0.1093,
+      "step": 5518
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.540893239546645e-06,
+      "loss": 0.1059,
+      "step": 5519
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.537527496225343e-06,
+      "loss": 0.1045,
+      "step": 5520
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.534162049839423e-06,
+      "loss": 0.0965,
+      "step": 5521
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.530796900794707e-06,
+      "loss": 0.1079,
+      "step": 5522
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.527432049496975e-06,
+      "loss": 0.1012,
+      "step": 5523
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.524067496351978e-06,
+      "loss": 0.0925,
+      "step": 5524
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.520703241765426e-06,
+      "loss": 0.1029,
+      "step": 5525
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.517339286142997e-06,
+      "loss": 0.1193,
+      "step": 5526
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.51397562989033e-06,
+      "loss": 0.1168,
+      "step": 5527
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510612273413029e-06,
+      "loss": 0.1206,
+      "step": 5528
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5072492171166636e-06,
+      "loss": 0.1014,
+      "step": 5529
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503886461406761e-06,
+      "loss": 0.1051,
+      "step": 5530
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5005240066888195e-06,
+      "loss": 0.0996,
+      "step": 5531
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497161853368298e-06,
+      "loss": 0.1109,
+      "step": 5532
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4938000018506214e-06,
+      "loss": 0.1026,
+      "step": 5533
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.490438452541172e-06,
+      "loss": 0.1045,
+      "step": 5534
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.487077205845301e-06,
+      "loss": 0.1027,
+      "step": 5535
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4837162621683215e-06,
+      "loss": 0.1072,
+      "step": 5536
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.480355621915514e-06,
+      "loss": 0.1028,
+      "step": 5537
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.476995285492115e-06,
+      "loss": 0.1101,
+      "step": 5538
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4736352533033265e-06,
+      "loss": 0.1072,
+      "step": 5539
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4702755257543204e-06,
+      "loss": 0.1047,
+      "step": 5540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.466916103250224e-06,
+      "loss": 0.0981,
+      "step": 5541
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46355698619613e-06,
+      "loss": 0.1029,
+      "step": 5542
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4601981749970955e-06,
+      "loss": 0.1108,
+      "step": 5543
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4568396700581425e-06,
+      "loss": 0.1099,
+      "step": 5544
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.453481471784252e-06,
+      "loss": 0.1149,
+      "step": 5545
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.450123580580369e-06,
+      "loss": 0.1148,
+      "step": 5546
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.446765996851401e-06,
+      "loss": 0.1126,
+      "step": 5547
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4434087210022255e-06,
+      "loss": 0.1078,
+      "step": 5548
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.440051753437671e-06,
+      "loss": 0.1089,
+      "step": 5549
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.436695094562538e-06,
+      "loss": 0.0999,
+      "step": 5550
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.433338744781588e-06,
+      "loss": 0.1102,
+      "step": 5551
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42998270449954e-06,
+      "loss": 0.0996,
+      "step": 5552
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.426626974121081e-06,
+      "loss": 0.1071,
+      "step": 5553
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.423271554050859e-06,
+      "loss": 0.0946,
+      "step": 5554
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.419916444693489e-06,
+      "loss": 0.1158,
+      "step": 5555
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416561646453537e-06,
+      "loss": 0.1124,
+      "step": 5556
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.413207159735543e-06,
+      "loss": 0.1092,
+      "step": 5557
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4098529849440035e-06,
+      "loss": 0.1002,
+      "step": 5558
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4064991224833815e-06,
+      "loss": 0.12,
+      "step": 5559
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.403145572758101e-06,
+      "loss": 0.1071,
+      "step": 5560
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.399792336172543e-06,
+      "loss": 0.1055,
+      "step": 5561
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.396439413131058e-06,
+      "loss": 0.1054,
+      "step": 5562
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.393086804037954e-06,
+      "loss": 0.1121,
+      "step": 5563
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.389734509297506e-06,
+      "loss": 0.1136,
+      "step": 5564
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.386382529313942e-06,
+      "loss": 0.0967,
+      "step": 5565
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383030864491464e-06,
+      "loss": 0.1088,
+      "step": 5566
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.379679515234226e-06,
+      "loss": 0.1097,
+      "step": 5567
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.376328481946352e-06,
+      "loss": 0.1159,
+      "step": 5568
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.372977765031919e-06,
+      "loss": 0.1114,
+      "step": 5569
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.369627364894974e-06,
+      "loss": 0.1159,
+      "step": 5570
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3662772819395215e-06,
+      "loss": 0.1082,
+      "step": 5571
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.362927516569527e-06,
+      "loss": 0.1149,
+      "step": 5572
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.359578069188922e-06,
+      "loss": 0.1166,
+      "step": 5573
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.356228940201595e-06,
+      "loss": 0.1001,
+      "step": 5574
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.352880130011402e-06,
+      "loss": 0.1144,
+      "step": 5575
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3495316390221505e-06,
+      "loss": 0.103,
+      "step": 5576
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34618346763762e-06,
+      "loss": 0.1056,
+      "step": 5577
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.342835616261547e-06,
+      "loss": 0.1111,
+      "step": 5578
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.33948808529763e-06,
+      "loss": 0.0992,
+      "step": 5579
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.336140875149527e-06,
+      "loss": 0.0968,
+      "step": 5580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.332793986220857e-06,
+      "loss": 0.115,
+      "step": 5581
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3294474189152084e-06,
+      "loss": 0.1044,
+      "step": 5582
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3261011736361175e-06,
+      "loss": 0.1052,
+      "step": 5583
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.322755250787092e-06,
+      "loss": 0.1136,
+      "step": 5584
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3194096507715986e-06,
+      "loss": 0.109,
+      "step": 5585
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3160643739930635e-06,
+      "loss": 0.1122,
+      "step": 5586
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3127194208548706e-06,
+      "loss": 0.1073,
+      "step": 5587
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309374791760373e-06,
+      "loss": 0.1093,
+      "step": 5588
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.306030487112878e-06,
+      "loss": 0.1209,
+      "step": 5589
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.302686507315657e-06,
+      "loss": 0.1217,
+      "step": 5590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.29934285277194e-06,
+      "loss": 0.1141,
+      "step": 5591
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295999523884921e-06,
+      "loss": 0.0971,
+      "step": 5592
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.292656521057753e-06,
+      "loss": 0.1128,
+      "step": 5593
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.289313844693546e-06,
+      "loss": 0.1192,
+      "step": 5594
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.285971495195377e-06,
+      "loss": 0.1129,
+      "step": 5595
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2826294729662805e-06,
+      "loss": 0.1036,
+      "step": 5596
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.279287778409252e-06,
+      "loss": 0.1101,
+      "step": 5597
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.275946411927245e-06,
+      "loss": 0.1015,
+      "step": 5598
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.272605373923177e-06,
+      "loss": 0.1108,
+      "step": 5599
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.269264664799927e-06,
+      "loss": 0.1104,
+      "step": 5600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.265924284960328e-06,
+      "loss": 0.0957,
+      "step": 5601
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.262584234807177e-06,
+      "loss": 0.1085,
+      "step": 5602
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2592445147432335e-06,
+      "loss": 0.1013,
+      "step": 5603
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.255905125171218e-06,
+      "loss": 0.1147,
+      "step": 5604
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.252566066493803e-06,
+      "loss": 0.1015,
+      "step": 5605
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.249227339113627e-06,
+      "loss": 0.1037,
+      "step": 5606
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.245888943433289e-06,
+      "loss": 0.1136,
+      "step": 5607
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.24255087985535e-06,
+      "loss": 0.1076,
+      "step": 5608
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.239213148782324e-06,
+      "loss": 0.1065,
+      "step": 5609
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.235875750616687e-06,
+      "loss": 0.1193,
+      "step": 5610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.232538685760885e-06,
+      "loss": 0.107,
+      "step": 5611
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.229201954617306e-06,
+      "loss": 0.1123,
+      "step": 5612
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.225865557588312e-06,
+      "loss": 0.1106,
+      "step": 5613
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.222529495076218e-06,
+      "loss": 0.1136,
+      "step": 5614
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.219193767483304e-06,
+      "loss": 0.108,
+      "step": 5615
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2158583752118016e-06,
+      "loss": 0.104,
+      "step": 5616
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.212523318663909e-06,
+      "loss": 0.1134,
+      "step": 5617
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209188598241782e-06,
+      "loss": 0.1054,
+      "step": 5618
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.205854214347534e-06,
+      "loss": 0.1011,
+      "step": 5619
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.202520167383238e-06,
+      "loss": 0.1065,
+      "step": 5620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199186457750931e-06,
+      "loss": 0.1082,
+      "step": 5621
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1958530858526055e-06,
+      "loss": 0.102,
+      "step": 5622
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.19252005209021e-06,
+      "loss": 0.1111,
+      "step": 5623
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1891873568656575e-06,
+      "loss": 0.101,
+      "step": 5624
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.185855000580819e-06,
+      "loss": 0.1164,
+      "step": 5625
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.182522983637527e-06,
+      "loss": 0.109,
+      "step": 5626
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1791913064375665e-06,
+      "loss": 0.1044,
+      "step": 5627
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1758599693826845e-06,
+      "loss": 0.1013,
+      "step": 5628
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.172528972874593e-06,
+      "loss": 0.1282,
+      "step": 5629
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.16919831731495e-06,
+      "loss": 0.1044,
+      "step": 5630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.165868003105385e-06,
+      "loss": 0.1106,
+      "step": 5631
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.162538030647481e-06,
+      "loss": 0.1009,
+      "step": 5632
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.159208400342783e-06,
+      "loss": 0.1016,
+      "step": 5633
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.155879112592785e-06,
+      "loss": 0.1127,
+      "step": 5634
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.152550167798952e-06,
+      "loss": 0.1053,
+      "step": 5635
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.149221566362699e-06,
+      "loss": 0.1218,
+      "step": 5636
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1458933086854075e-06,
+      "loss": 0.1096,
+      "step": 5637
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.14256539516841e-06,
+      "loss": 0.1143,
+      "step": 5638
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.139237826212998e-06,
+      "loss": 0.1087,
+      "step": 5639
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.135910602220428e-06,
+      "loss": 0.1109,
+      "step": 5640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.13258372359191e-06,
+      "loss": 0.1031,
+      "step": 5641
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.12925719072861e-06,
+      "loss": 0.1107,
+      "step": 5642
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.125931004031659e-06,
+      "loss": 0.1101,
+      "step": 5643
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.122605163902142e-06,
+      "loss": 0.1055,
+      "step": 5644
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.119279670741101e-06,
+      "loss": 0.1037,
+      "step": 5645
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.11595452494954e-06,
+      "loss": 0.1097,
+      "step": 5646
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.112629726928421e-06,
+      "loss": 0.1039,
+      "step": 5647
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1093052770786566e-06,
+      "loss": 0.1033,
+      "step": 5648
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1059811758011265e-06,
+      "loss": 0.0923,
+      "step": 5649
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102657423496666e-06,
+      "loss": 0.0989,
+      "step": 5650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.099334020566066e-06,
+      "loss": 0.1064,
+      "step": 5651
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.096010967410077e-06,
+      "loss": 0.1034,
+      "step": 5652
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.092688264429405e-06,
+      "loss": 0.1045,
+      "step": 5653
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0893659120247175e-06,
+      "loss": 0.1055,
+      "step": 5654
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.086043910596638e-06,
+      "loss": 0.1197,
+      "step": 5655
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0827222605457465e-06,
+      "loss": 0.1037,
+      "step": 5656
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079400962272582e-06,
+      "loss": 0.1207,
+      "step": 5657
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0760800161776425e-06,
+      "loss": 0.1172,
+      "step": 5658
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.072759422661378e-06,
+      "loss": 0.0989,
+      "step": 5659
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.069439182124203e-06,
+      "loss": 0.1138,
+      "step": 5660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.066119294966482e-06,
+      "loss": 0.1033,
+      "step": 5661
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0627997615885505e-06,
+      "loss": 0.1138,
+      "step": 5662
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0594805823906795e-06,
+      "loss": 0.1066,
+      "step": 5663
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056161757773118e-06,
+      "loss": 0.1054,
+      "step": 5664
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.052843288136063e-06,
+      "loss": 0.1133,
+      "step": 5665
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0495251738796655e-06,
+      "loss": 0.105,
+      "step": 5666
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046207415404043e-06,
+      "loss": 0.1221,
+      "step": 5667
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0428900131092605e-06,
+      "loss": 0.11,
+      "step": 5668
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0395729673953494e-06,
+      "loss": 0.1011,
+      "step": 5669
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036256278662288e-06,
+      "loss": 0.1142,
+      "step": 5670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.032939947310018e-06,
+      "loss": 0.1164,
+      "step": 5671
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0296239737384374e-06,
+      "loss": 0.1135,
+      "step": 5672
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026308358347405e-06,
+      "loss": 0.114,
+      "step": 5673
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022993101536724e-06,
+      "loss": 0.1129,
+      "step": 5674
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.019678203706164e-06,
+      "loss": 0.1021,
+      "step": 5675
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.016363665255454e-06,
+      "loss": 0.0945,
+      "step": 5676
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.013049486584271e-06,
+      "loss": 0.1137,
+      "step": 5677
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.009735668092252e-06,
+      "loss": 0.1036,
+      "step": 5678
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.006422210178996e-06,
+      "loss": 0.1079,
+      "step": 5679
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003109113244051e-06,
+      "loss": 0.0949,
+      "step": 5680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.999796377686922e-06,
+      "loss": 0.0958,
+      "step": 5681
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9964840039070735e-06,
+      "loss": 0.098,
+      "step": 5682
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.993171992303928e-06,
+      "loss": 0.0965,
+      "step": 5683
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.989860343276864e-06,
+      "loss": 0.1034,
+      "step": 5684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9865490572252095e-06,
+      "loss": 0.1007,
+      "step": 5685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983238134548254e-06,
+      "loss": 0.108,
+      "step": 5686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.979927575645245e-06,
+      "loss": 0.1032,
+      "step": 5687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.97661738091538e-06,
+      "loss": 0.1029,
+      "step": 5688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9733075507578195e-06,
+      "loss": 0.1041,
+      "step": 5689
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.969998085571677e-06,
+      "loss": 0.1132,
+      "step": 5690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966688985756021e-06,
+      "loss": 0.1046,
+      "step": 5691
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.963380251709875e-06,
+      "loss": 0.1147,
+      "step": 5692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.960071883832222e-06,
+      "loss": 0.1166,
+      "step": 5693
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.956763882522002e-06,
+      "loss": 0.1036,
+      "step": 5694
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.953456248178101e-06,
+      "loss": 0.1025,
+      "step": 5695
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.950148981199372e-06,
+      "loss": 0.1,
+      "step": 5696
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.94684208198462e-06,
+      "loss": 0.1073,
+      "step": 5697
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.943535550932603e-06,
+      "loss": 0.103,
+      "step": 5698
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940229388442037e-06,
+      "loss": 0.1056,
+      "step": 5699
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.936923594911592e-06,
+      "loss": 0.1167,
+      "step": 5700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.933618170739896e-06,
+      "loss": 0.1074,
+      "step": 5701
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9303131163255335e-06,
+      "loss": 0.1007,
+      "step": 5702
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.927008432067036e-06,
+      "loss": 0.0994,
+      "step": 5703
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9237041183629005e-06,
+      "loss": 0.11,
+      "step": 5704
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.920400175611577e-06,
+      "loss": 0.1061,
+      "step": 5705
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.917096604211463e-06,
+      "loss": 0.1172,
+      "step": 5706
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.913793404560921e-06,
+      "loss": 0.1048,
+      "step": 5707
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.910490577058264e-06,
+      "loss": 0.1078,
+      "step": 5708
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.907188122101763e-06,
+      "loss": 0.1141,
+      "step": 5709
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.903886040089639e-06,
+      "loss": 0.0967,
+      "step": 5710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900584331420073e-06,
+      "loss": 0.1161,
+      "step": 5711
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897282996491197e-06,
+      "loss": 0.0965,
+      "step": 5712
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.893982035701103e-06,
+      "loss": 0.1069,
+      "step": 5713
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.890681449447831e-06,
+      "loss": 0.1054,
+      "step": 5714
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.887381238129384e-06,
+      "loss": 0.1005,
+      "step": 5715
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.884081402143714e-06,
+      "loss": 0.1015,
+      "step": 5716
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8807819418887255e-06,
+      "loss": 0.1186,
+      "step": 5717
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.877482857762285e-06,
+      "loss": 0.1083,
+      "step": 5718
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874184150162211e-06,
+      "loss": 0.1041,
+      "step": 5719
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.870885819486275e-06,
+      "loss": 0.104,
+      "step": 5720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8675878661322014e-06,
+      "loss": 0.1096,
+      "step": 5721
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.864290290497672e-06,
+      "loss": 0.0975,
+      "step": 5722
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.860993092980326e-06,
+      "loss": 0.1026,
+      "step": 5723
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8576962739777495e-06,
+      "loss": 0.1108,
+      "step": 5724
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.854399833887484e-06,
+      "loss": 0.1019,
+      "step": 5725
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851103773107037e-06,
+      "loss": 0.113,
+      "step": 5726
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.847808092033856e-06,
+      "loss": 0.0923,
+      "step": 5727
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.84451279106535e-06,
+      "loss": 0.1071,
+      "step": 5728
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.841217870598881e-06,
+      "loss": 0.1001,
+      "step": 5729
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837923331031761e-06,
+      "loss": 0.1048,
+      "step": 5730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.834629172761262e-06,
+      "loss": 0.1144,
+      "step": 5731
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.831335396184606e-06,
+      "loss": 0.1137,
+      "step": 5732
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.828042001698975e-06,
+      "loss": 0.106,
+      "step": 5733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.824748989701494e-06,
+      "loss": 0.113,
+      "step": 5734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.821456360589251e-06,
+      "loss": 0.0929,
+      "step": 5735
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.81816411475929e-06,
+      "loss": 0.1134,
+      "step": 5736
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.814872252608598e-06,
+      "loss": 0.1046,
+      "step": 5737
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.811580774534121e-06,
+      "loss": 0.1124,
+      "step": 5738
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808289680932764e-06,
+      "loss": 0.1086,
+      "step": 5739
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.80499897220138e-06,
+      "loss": 0.1093,
+      "step": 5740
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8017086487367735e-06,
+      "loss": 0.1118,
+      "step": 5741
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.798418710935707e-06,
+      "loss": 0.1069,
+      "step": 5742
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.795129159194896e-06,
+      "loss": 0.0974,
+      "step": 5743
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7918399939110115e-06,
+      "loss": 0.1123,
+      "step": 5744
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7885512154806694e-06,
+      "loss": 0.123,
+      "step": 5745
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.785262824300447e-06,
+      "loss": 0.108,
+      "step": 5746
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.781974820766874e-06,
+      "loss": 0.1018,
+      "step": 5747
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.778687205276428e-06,
+      "loss": 0.1127,
+      "step": 5748
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.775399978225547e-06,
+      "loss": 0.0978,
+      "step": 5749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772113140010619e-06,
+      "loss": 0.1052,
+      "step": 5750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.768826691027983e-06,
+      "loss": 0.1099,
+      "step": 5751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.765540631673934e-06,
+      "loss": 0.1096,
+      "step": 5752
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7622549623447165e-06,
+      "loss": 0.1239,
+      "step": 5753
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.758969683436536e-06,
+      "loss": 0.1108,
+      "step": 5754
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.75568479534554e-06,
+      "loss": 0.1073,
+      "step": 5755
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.752400298467835e-06,
+      "loss": 0.1121,
+      "step": 5756
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7491161931994806e-06,
+      "loss": 0.11,
+      "step": 5757
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.745832479936492e-06,
+      "loss": 0.1074,
+      "step": 5758
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.742549159074827e-06,
+      "loss": 0.1112,
+      "step": 5759
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.739266231010404e-06,
+      "loss": 0.0979,
+      "step": 5760
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.735983696139094e-06,
+      "loss": 0.1179,
+      "step": 5761
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.73270155485672e-06,
+      "loss": 0.1051,
+      "step": 5762
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.729419807559053e-06,
+      "loss": 0.111,
+      "step": 5763
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.726138454641822e-06,
+      "loss": 0.1068,
+      "step": 5764
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.722857496500708e-06,
+      "loss": 0.105,
+      "step": 5765
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.719576933531339e-06,
+      "loss": 0.1037,
+      "step": 5766
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.716296766129302e-06,
+      "loss": 0.1042,
+      "step": 5767
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7130169946901315e-06,
+      "loss": 0.0928,
+      "step": 5768
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.709737619609321e-06,
+      "loss": 0.1008,
+      "step": 5769
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.706458641282305e-06,
+      "loss": 0.1025,
+      "step": 5770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70318006010448e-06,
+      "loss": 0.1079,
+      "step": 5771
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699901876471191e-06,
+      "loss": 0.1175,
+      "step": 5772
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.696624090777738e-06,
+      "loss": 0.1103,
+      "step": 5773
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693346703419365e-06,
+      "loss": 0.1094,
+      "step": 5774
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.690069714791275e-06,
+      "loss": 0.0957,
+      "step": 5775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.686793125288624e-06,
+      "loss": 0.1017,
+      "step": 5776
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.683516935306515e-06,
+      "loss": 0.1102,
+      "step": 5777
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.680241145240003e-06,
+      "loss": 0.1149,
+      "step": 5778
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6769657554841e-06,
+      "loss": 0.1161,
+      "step": 5779
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673690766433765e-06,
+      "loss": 0.1073,
+      "step": 5780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.67041617848391e-06,
+      "loss": 0.1142,
+      "step": 5781
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667141992029397e-06,
+      "loss": 0.1079,
+      "step": 5782
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.663868207465045e-06,
+      "loss": 0.1055,
+      "step": 5783
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6605948251856175e-06,
+      "loss": 0.1183,
+      "step": 5784
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657321845585836e-06,
+      "loss": 0.1121,
+      "step": 5785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.654049269060366e-06,
+      "loss": 0.1007,
+      "step": 5786
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.650777096003834e-06,
+      "loss": 0.0973,
+      "step": 5787
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.647505326810808e-06,
+      "loss": 0.103,
+      "step": 5788
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.644233961875813e-06,
+      "loss": 0.1051,
+      "step": 5789
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.640963001593326e-06,
+      "loss": 0.1138,
+      "step": 5790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6376924463577705e-06,
+      "loss": 0.1053,
+      "step": 5791
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.634422296563527e-06,
+      "loss": 0.1109,
+      "step": 5792
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.63115255260492e-06,
+      "loss": 0.1078,
+      "step": 5793
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.627883214876234e-06,
+      "loss": 0.1104,
+      "step": 5794
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.624614283771694e-06,
+      "loss": 0.104,
+      "step": 5795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.621345759685486e-06,
+      "loss": 0.1056,
+      "step": 5796
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6180776430117396e-06,
+      "loss": 0.1008,
+      "step": 5797
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.614809934144542e-06,
+      "loss": 0.1115,
+      "step": 5798
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.611542633477924e-06,
+      "loss": 0.1215,
+      "step": 5799
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.608275741405872e-06,
+      "loss": 0.1068,
+      "step": 5800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.605009258322319e-06,
+      "loss": 0.101,
+      "step": 5801
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.601743184621158e-06,
+      "loss": 0.1,
+      "step": 5802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.598477520696218e-06,
+      "loss": 0.102,
+      "step": 5803
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.595212266941291e-06,
+      "loss": 0.101,
+      "step": 5804
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.591947423750117e-06,
+      "loss": 0.114,
+      "step": 5805
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.58868299151638e-06,
+      "loss": 0.1051,
+      "step": 5806
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585418970633721e-06,
+      "loss": 0.1046,
+      "step": 5807
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.582155361495729e-06,
+      "loss": 0.1008,
+      "step": 5808
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.578892164495947e-06,
+      "loss": 0.1103,
+      "step": 5809
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.575629380027861e-06,
+      "loss": 0.0943,
+      "step": 5810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5723670084849125e-06,
+      "loss": 0.1069,
+      "step": 5811
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5691050502604955e-06,
+      "loss": 0.105,
+      "step": 5812
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.565843505747947e-06,
+      "loss": 0.1049,
+      "step": 5813
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.562582375340556e-06,
+      "loss": 0.1121,
+      "step": 5814
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5593216594315675e-06,
+      "loss": 0.1075,
+      "step": 5815
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556061358414174e-06,
+      "loss": 0.104,
+      "step": 5816
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5528014726815104e-06,
+      "loss": 0.1059,
+      "step": 5817
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.549542002626671e-06,
+      "loss": 0.1113,
+      "step": 5818
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.546282948642697e-06,
+      "loss": 0.1215,
+      "step": 5819
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54302431112258e-06,
+      "loss": 0.0918,
+      "step": 5820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.539766090459256e-06,
+      "loss": 0.1109,
+      "step": 5821
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536508287045618e-06,
+      "loss": 0.106,
+      "step": 5822
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533250901274508e-06,
+      "loss": 0.1034,
+      "step": 5823
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5299939335387105e-06,
+      "loss": 0.1023,
+      "step": 5824
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.526737384230967e-06,
+      "loss": 0.109,
+      "step": 5825
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.523481253743966e-06,
+      "loss": 0.1084,
+      "step": 5826
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.520225542470347e-06,
+      "loss": 0.1077,
+      "step": 5827
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.516970250802693e-06,
+      "loss": 0.108,
+      "step": 5828
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513715379133545e-06,
+      "loss": 0.107,
+      "step": 5829
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.510460927855387e-06,
+      "loss": 0.1157,
+      "step": 5830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507206897360658e-06,
+      "loss": 0.1077,
+      "step": 5831
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.50395328804174e-06,
+      "loss": 0.1062,
+      "step": 5832
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.500700100290965e-06,
+      "loss": 0.1047,
+      "step": 5833
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.497447334500622e-06,
+      "loss": 0.1207,
+      "step": 5834
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.494194991062937e-06,
+      "loss": 0.1109,
+      "step": 5835
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.490943070370095e-06,
+      "loss": 0.1032,
+      "step": 5836
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487691572814226e-06,
+      "loss": 0.1013,
+      "step": 5837
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.484440498787409e-06,
+      "loss": 0.1067,
+      "step": 5838
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.48118984868167e-06,
+      "loss": 0.0948,
+      "step": 5839
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.477939622888991e-06,
+      "loss": 0.0915,
+      "step": 5840
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474689821801295e-06,
+      "loss": 0.1145,
+      "step": 5841
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.471440445810458e-06,
+      "loss": 0.108,
+      "step": 5842
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4681914953083e-06,
+      "loss": 0.1049,
+      "step": 5843
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464942970686598e-06,
+      "loss": 0.1102,
+      "step": 5844
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.461694872337072e-06,
+      "loss": 0.106,
+      "step": 5845
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.458447200651388e-06,
+      "loss": 0.1147,
+      "step": 5846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.455199956021167e-06,
+      "loss": 0.0962,
+      "step": 5847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451953138837975e-06,
+      "loss": 0.1004,
+      "step": 5848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448706749493329e-06,
+      "loss": 0.1089,
+      "step": 5849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445460788378688e-06,
+      "loss": 0.1045,
+      "step": 5850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.442215255885467e-06,
+      "loss": 0.1029,
+      "step": 5851
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438970152405026e-06,
+      "loss": 0.1034,
+      "step": 5852
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.435725478328671e-06,
+      "loss": 0.1037,
+      "step": 5853
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.43248123404766e-06,
+      "loss": 0.114,
+      "step": 5854
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.429237419953199e-06,
+      "loss": 0.108,
+      "step": 5855
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4259940364364404e-06,
+      "loss": 0.1027,
+      "step": 5856
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.422751083888484e-06,
+      "loss": 0.1086,
+      "step": 5857
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.419508562700378e-06,
+      "loss": 0.1003,
+      "step": 5858
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4162664732631195e-06,
+      "loss": 0.1138,
+      "step": 5859
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.413024815967659e-06,
+      "loss": 0.1107,
+      "step": 5860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4097835912048815e-06,
+      "loss": 0.0987,
+      "step": 5861
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4065427993656314e-06,
+      "loss": 0.1019,
+      "step": 5862
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.403302440840698e-06,
+      "loss": 0.1022,
+      "step": 5863
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.400062516020814e-06,
+      "loss": 0.1062,
+      "step": 5864
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.396823025296666e-06,
+      "loss": 0.1204,
+      "step": 5865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3935839690588856e-06,
+      "loss": 0.1051,
+      "step": 5866
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.390345347698052e-06,
+      "loss": 0.1015,
+      "step": 5867
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.387107161604689e-06,
+      "loss": 0.101,
+      "step": 5868
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3838694111692745e-06,
+      "loss": 0.1114,
+      "step": 5869
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.380632096782229e-06,
+      "loss": 0.1015,
+      "step": 5870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37739521883392e-06,
+      "loss": 0.102,
+      "step": 5871
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.374158777714666e-06,
+      "loss": 0.1107,
+      "step": 5872
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370922773814729e-06,
+      "loss": 0.1082,
+      "step": 5873
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.367687207524324e-06,
+      "loss": 0.0945,
+      "step": 5874
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.364452079233605e-06,
+      "loss": 0.1089,
+      "step": 5875
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361217389332678e-06,
+      "loss": 0.1031,
+      "step": 5876
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.357983138211596e-06,
+      "loss": 0.1212,
+      "step": 5877
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3547493262603634e-06,
+      "loss": 0.1127,
+      "step": 5878
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.351515953868918e-06,
+      "loss": 0.1081,
+      "step": 5879
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348283021427162e-06,
+      "loss": 0.1119,
+      "step": 5880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.345050529324932e-06,
+      "loss": 0.0903,
+      "step": 5881
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.341818477952014e-06,
+      "loss": 0.1037,
+      "step": 5882
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.338586867698146e-06,
+      "loss": 0.1041,
+      "step": 5883
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.335355698953007e-06,
+      "loss": 0.1076,
+      "step": 5884
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332124972106228e-06,
+      "loss": 0.1089,
+      "step": 5885
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.32889468754738e-06,
+      "loss": 0.1033,
+      "step": 5886
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.325664845665986e-06,
+      "loss": 0.1019,
+      "step": 5887
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3224354468515145e-06,
+      "loss": 0.1052,
+      "step": 5888
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.319206491493382e-06,
+      "loss": 0.0993,
+      "step": 5889
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3159779799809455e-06,
+      "loss": 0.1077,
+      "step": 5890
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.312749912703513e-06,
+      "loss": 0.1184,
+      "step": 5891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.309522290050344e-06,
+      "loss": 0.1096,
+      "step": 5892
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.306295112410627e-06,
+      "loss": 0.1039,
+      "step": 5893
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.303068380173525e-06,
+      "loss": 0.1033,
+      "step": 5894
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.299842093728119e-06,
+      "loss": 0.1136,
+      "step": 5895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.296616253463452e-06,
+      "loss": 0.1122,
+      "step": 5896
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.293390859768511e-06,
+      "loss": 0.1082,
+      "step": 5897
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.290165913032229e-06,
+      "loss": 0.1113,
+      "step": 5898
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.286941413643477e-06,
+      "loss": 0.0998,
+      "step": 5899
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.283717361991084e-06,
+      "loss": 0.0996,
+      "step": 5900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.280493758463821e-06,
+      "loss": 0.1041,
+      "step": 5901
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.277270603450398e-06,
+      "loss": 0.1021,
+      "step": 5902
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2740478973394815e-06,
+      "loss": 0.1077,
+      "step": 5903
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270825640519678e-06,
+      "loss": 0.1135,
+      "step": 5904
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267603833379542e-06,
+      "loss": 0.102,
+      "step": 5905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.264382476307569e-06,
+      "loss": 0.1032,
+      "step": 5906
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2611615696922065e-06,
+      "loss": 0.1111,
+      "step": 5907
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.257941113921846e-06,
+      "loss": 0.1154,
+      "step": 5908
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.254721109384823e-06,
+      "loss": 0.1089,
+      "step": 5909
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.251501556469417e-06,
+      "loss": 0.0967,
+      "step": 5910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.248282455563859e-06,
+      "loss": 0.1122,
+      "step": 5911
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245063807056321e-06,
+      "loss": 0.096,
+      "step": 5912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241845611334918e-06,
+      "loss": 0.1059,
+      "step": 5913
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.238627868787717e-06,
+      "loss": 0.1166,
+      "step": 5914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.235410579802725e-06,
+      "loss": 0.1108,
+      "step": 5915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.232193744767899e-06,
+      "loss": 0.1082,
+      "step": 5916
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.228977364071136e-06,
+      "loss": 0.1019,
+      "step": 5917
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.225761438100282e-06,
+      "loss": 0.1148,
+      "step": 5918
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.222545967243127e-06,
+      "loss": 0.0972,
+      "step": 5919
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.219330951887408e-06,
+      "loss": 0.1013,
+      "step": 5920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2161163924208016e-06,
+      "loss": 0.1105,
+      "step": 5921
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.212902289230933e-06,
+      "loss": 0.117,
+      "step": 5922
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.209688642705377e-06,
+      "loss": 0.1066,
+      "step": 5923
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.206475453231644e-06,
+      "loss": 0.1142,
+      "step": 5924
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.203262721197196e-06,
+      "loss": 0.1064,
+      "step": 5925
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.200050446989436e-06,
+      "loss": 0.1007,
+      "step": 5926
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19683863099572e-06,
+      "loss": 0.1036,
+      "step": 5927
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1936272736033335e-06,
+      "loss": 0.1063,
+      "step": 5928
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19041637519952e-06,
+      "loss": 0.1079,
+      "step": 5929
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.187205936171467e-06,
+      "loss": 0.1047,
+      "step": 5930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1839959569062945e-06,
+      "loss": 0.1002,
+      "step": 5931
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.180786437791082e-06,
+      "loss": 0.106,
+      "step": 5932
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.177577379212843e-06,
+      "loss": 0.104,
+      "step": 5933
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.174368781558544e-06,
+      "loss": 0.1038,
+      "step": 5934
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.171160645215086e-06,
+      "loss": 0.1163,
+      "step": 5935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167952970569322e-06,
+      "loss": 0.1117,
+      "step": 5936
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.164745758008048e-06,
+      "loss": 0.1099,
+      "step": 5937
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.161539007918005e-06,
+      "loss": 0.0979,
+      "step": 5938
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.15833272068587e-06,
+      "loss": 0.105,
+      "step": 5939
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155126896698276e-06,
+      "loss": 0.1036,
+      "step": 5940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151921536341796e-06,
+      "loss": 0.1011,
+      "step": 5941
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.14871664000294e-06,
+      "loss": 0.1096,
+      "step": 5942
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1455122080681735e-06,
+      "loss": 0.1095,
+      "step": 5943
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.142308240923898e-06,
+      "loss": 0.1119,
+      "step": 5944
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.139104738956463e-06,
+      "loss": 0.1084,
+      "step": 5945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.135901702552158e-06,
+      "loss": 0.1072,
+      "step": 5946
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1326991320972195e-06,
+      "loss": 0.12,
+      "step": 5947
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129497027977829e-06,
+      "loss": 0.1125,
+      "step": 5948
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126295390580109e-06,
+      "loss": 0.1125,
+      "step": 5949
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1230942202901245e-06,
+      "loss": 0.115,
+      "step": 5950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.119893517493887e-06,
+      "loss": 0.1018,
+      "step": 5951
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116693282577355e-06,
+      "loss": 0.1098,
+      "step": 5952
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.113493515926419e-06,
+      "loss": 0.093,
+      "step": 5953
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.110294217926925e-06,
+      "loss": 0.0963,
+      "step": 5954
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.107095388964658e-06,
+      "loss": 0.0981,
+      "step": 5955
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.103897029425347e-06,
+      "loss": 0.095,
+      "step": 5956
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1006991396946604e-06,
+      "loss": 0.1139,
+      "step": 5957
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.097501720158215e-06,
+      "loss": 0.1068,
+      "step": 5958
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094304771201571e-06,
+      "loss": 0.1049,
+      "step": 5959
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.091108293210228e-06,
+      "loss": 0.102,
+      "step": 5960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.08791228656963e-06,
+      "loss": 0.1093,
+      "step": 5961
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.084716751665167e-06,
+      "loss": 0.1004,
+      "step": 5962
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.081521688882173e-06,
+      "loss": 0.0995,
+      "step": 5963
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.078327098605914e-06,
+      "loss": 0.117,
+      "step": 5964
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.075132981221614e-06,
+      "loss": 0.0919,
+      "step": 5965
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.071939337114431e-06,
+      "loss": 0.1114,
+      "step": 5966
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.06874616666947e-06,
+      "loss": 0.1117,
+      "step": 5967
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065553470271775e-06,
+      "loss": 0.1067,
+      "step": 5968
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0623612483063335e-06,
+      "loss": 0.1112,
+      "step": 5969
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.059169501158082e-06,
+      "loss": 0.113,
+      "step": 5970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.055978229211889e-06,
+      "loss": 0.1127,
+      "step": 5971
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.052787432852576e-06,
+      "loss": 0.1038,
+      "step": 5972
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0495971124649e-06,
+      "loss": 0.1088,
+      "step": 5973
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.046407268433567e-06,
+      "loss": 0.099,
+      "step": 5974
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.043217901143219e-06,
+      "loss": 0.1116,
+      "step": 5975
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.040029010978443e-06,
+      "loss": 0.1006,
+      "step": 5976
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.036840598323772e-06,
+      "loss": 0.1091,
+      "step": 5977
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.033652663563675e-06,
+      "loss": 0.1061,
+      "step": 5978
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.030465207082568e-06,
+      "loss": 0.111,
+      "step": 5979
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027278229264807e-06,
+      "loss": 0.1172,
+      "step": 5980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.024091730494694e-06,
+      "loss": 0.1076,
+      "step": 5981
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0209057111564685e-06,
+      "loss": 0.1058,
+      "step": 5982
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.017720171634314e-06,
+      "loss": 0.1026,
+      "step": 5983
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014535112312356e-06,
+      "loss": 0.1129,
+      "step": 5984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.011350533574667e-06,
+      "loss": 0.1024,
+      "step": 5985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.00816643580525e-06,
+      "loss": 0.1133,
+      "step": 5986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.004982819388061e-06,
+      "loss": 0.1044,
+      "step": 5987
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.001799684706997e-06,
+      "loss": 0.1033,
+      "step": 5988
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.998617032145886e-06,
+      "loss": 0.1047,
+      "step": 5989
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995434862088511e-06,
+      "loss": 0.1109,
+      "step": 5990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.992253174918592e-06,
+      "loss": 0.1122,
+      "step": 5991
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9890719710197884e-06,
+      "loss": 0.113,
+      "step": 5992
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.985891250775704e-06,
+      "loss": 0.1,
+      "step": 5993
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.982711014569882e-06,
+      "loss": 0.1111,
+      "step": 5994
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979531262785811e-06,
+      "loss": 0.0965,
+      "step": 5995
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9763519958069205e-06,
+      "loss": 0.1102,
+      "step": 5996
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.973173214016574e-06,
+      "loss": 0.0966,
+      "step": 5997
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.969994917798088e-06,
+      "loss": 0.1052,
+      "step": 5998
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.966817107534715e-06,
+      "loss": 0.1076,
+      "step": 5999
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.963639783609643e-06,
+      "loss": 0.101,
+      "step": 6000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9604629464060125e-06,
+      "loss": 0.1065,
+      "step": 6001
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.957286596306899e-06,
+      "loss": 0.103,
+      "step": 6002
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.954110733695322e-06,
+      "loss": 0.1158,
+      "step": 6003
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.950935358954235e-06,
+      "loss": 0.0959,
+      "step": 6004
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947760472466541e-06,
+      "loss": 0.1304,
+      "step": 6005
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944586074615086e-06,
+      "loss": 0.0905,
+      "step": 6006
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941412165782645e-06,
+      "loss": 0.1117,
+      "step": 6007
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9382387463519455e-06,
+      "loss": 0.0947,
+      "step": 6008
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.935065816705649e-06,
+      "loss": 0.1014,
+      "step": 6009
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931893377226368e-06,
+      "loss": 0.1102,
+      "step": 6010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.92872142829664e-06,
+      "loss": 0.1089,
+      "step": 6011
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9255499702989536e-06,
+      "loss": 0.1001,
+      "step": 6012
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.922379003615742e-06,
+      "loss": 0.1048,
+      "step": 6013
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.91920852862937e-06,
+      "loss": 0.0995,
+      "step": 6014
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916038545722146e-06,
+      "loss": 0.0963,
+      "step": 6015
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.91286905527632e-06,
+      "loss": 0.1034,
+      "step": 6016
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.909700057674088e-06,
+      "loss": 0.1015,
+      "step": 6017
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906531553297574e-06,
+      "loss": 0.1072,
+      "step": 6018
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.903363542528854e-06,
+      "loss": 0.0977,
+      "step": 6019
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.900196025749938e-06,
+      "loss": 0.0945,
+      "step": 6020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.897029003342782e-06,
+      "loss": 0.1045,
+      "step": 6021
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.893862475689274e-06,
+      "loss": 0.1063,
+      "step": 6022
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.890696443171251e-06,
+      "loss": 0.0992,
+      "step": 6023
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887530906170484e-06,
+      "loss": 0.0988,
+      "step": 6024
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.884365865068693e-06,
+      "loss": 0.1033,
+      "step": 6025
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.881201320247524e-06,
+      "loss": 0.1105,
+      "step": 6026
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.878037272088575e-06,
+      "loss": 0.1078,
+      "step": 6027
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.874873720973382e-06,
+      "loss": 0.108,
+      "step": 6028
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8717106672834146e-06,
+      "loss": 0.111,
+      "step": 6029
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868548111400092e-06,
+      "loss": 0.1032,
+      "step": 6030
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.865386053704764e-06,
+      "loss": 0.1012,
+      "step": 6031
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.862224494578733e-06,
+      "loss": 0.0899,
+      "step": 6032
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.859063434403225e-06,
+      "loss": 0.0958,
+      "step": 6033
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.855902873559416e-06,
+      "loss": 0.0961,
+      "step": 6034
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852742812428425e-06,
+      "loss": 0.1174,
+      "step": 6035
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.849583251391297e-06,
+      "loss": 0.1115,
+      "step": 6036
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.84642419082903e-06,
+      "loss": 0.0924,
+      "step": 6037
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843265631122558e-06,
+      "loss": 0.1061,
+      "step": 6038
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.840107572652753e-06,
+      "loss": 0.0978,
+      "step": 6039
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.836950015800422e-06,
+      "loss": 0.1161,
+      "step": 6040
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.833792960946326e-06,
+      "loss": 0.0989,
+      "step": 6041
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830636408471151e-06,
+      "loss": 0.1051,
+      "step": 6042
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827480358755522e-06,
+      "loss": 0.1083,
+      "step": 6043
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.824324812180019e-06,
+      "loss": 0.096,
+      "step": 6044
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.821169769125146e-06,
+      "loss": 0.1007,
+      "step": 6045
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.818015229971349e-06,
+      "loss": 0.1017,
+      "step": 6046
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8148611950990215e-06,
+      "loss": 0.1132,
+      "step": 6047
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.811707664888489e-06,
+      "loss": 0.1066,
+      "step": 6048
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.80855463972001e-06,
+      "loss": 0.0979,
+      "step": 6049
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.805402119973802e-06,
+      "loss": 0.1118,
+      "step": 6050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.802250106029997e-06,
+      "loss": 0.1059,
+      "step": 6051
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7990985982686885e-06,
+      "loss": 0.1152,
+      "step": 6052
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.795947597069895e-06,
+      "loss": 0.1152,
+      "step": 6053
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.792797102813574e-06,
+      "loss": 0.1023,
+      "step": 6054
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7896471158796305e-06,
+      "loss": 0.0928,
+      "step": 6055
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.786497636647903e-06,
+      "loss": 0.1044,
+      "step": 6056
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.78334866549816e-06,
+      "loss": 0.0977,
+      "step": 6057
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7802002028101335e-06,
+      "loss": 0.108,
+      "step": 6058
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7770522489634675e-06,
+      "loss": 0.1097,
+      "step": 6059
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.773904804337754e-06,
+      "loss": 0.0993,
+      "step": 6060
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.770757869312531e-06,
+      "loss": 0.1097,
+      "step": 6061
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7676114442672695e-06,
+      "loss": 0.1112,
+      "step": 6062
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.764465529581375e-06,
+      "loss": 0.1127,
+      "step": 6063
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7613201256342e-06,
+      "loss": 0.1122,
+      "step": 6064
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.758175232805028e-06,
+      "loss": 0.1092,
+      "step": 6065
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.75503085147308e-06,
+      "loss": 0.0956,
+      "step": 6066
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.751886982017525e-06,
+      "loss": 0.1109,
+      "step": 6067
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748743624817463e-06,
+      "loss": 0.103,
+      "step": 6068
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.745600780251927e-06,
+      "loss": 0.1091,
+      "step": 6069
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.742458448699904e-06,
+      "loss": 0.0894,
+      "step": 6070
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.739316630540299e-06,
+      "loss": 0.0994,
+      "step": 6071
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736175326151976e-06,
+      "loss": 0.0996,
+      "step": 6072
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.733034535913724e-06,
+      "loss": 0.1114,
+      "step": 6073
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.729894260204266e-06,
+      "loss": 0.1013,
+      "step": 6074
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.72675449940228e-06,
+      "loss": 0.1247,
+      "step": 6075
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.723615253886366e-06,
+      "loss": 0.1102,
+      "step": 6076
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.720476524035064e-06,
+      "loss": 0.1065,
+      "step": 6077
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.717338310226864e-06,
+      "loss": 0.1057,
+      "step": 6078
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.714200612840182e-06,
+      "loss": 0.1211,
+      "step": 6079
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.711063432253368e-06,
+      "loss": 0.1012,
+      "step": 6080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.707926768844727e-06,
+      "loss": 0.1069,
+      "step": 6081
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.704790622992482e-06,
+      "loss": 0.1062,
+      "step": 6082
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.701654995074811e-06,
+      "loss": 0.1089,
+      "step": 6083
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698519885469818e-06,
+      "loss": 0.0936,
+      "step": 6084
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695385294555542e-06,
+      "loss": 0.094,
+      "step": 6085
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692251222709977e-06,
+      "loss": 0.1074,
+      "step": 6086
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689117670311034e-06,
+      "loss": 0.103,
+      "step": 6087
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.685984637736569e-06,
+      "loss": 0.1008,
+      "step": 6088
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.682852125364384e-06,
+      "loss": 0.1121,
+      "step": 6089
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6797201335722064e-06,
+      "loss": 0.1176,
+      "step": 6090
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6765886627377e-06,
+      "loss": 0.1117,
+      "step": 6091
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67345771323848e-06,
+      "loss": 0.1096,
+      "step": 6092
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.670327285452082e-06,
+      "loss": 0.1112,
+      "step": 6093
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6671973797559935e-06,
+      "loss": 0.1122,
+      "step": 6094
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.664067996527627e-06,
+      "loss": 0.1117,
+      "step": 6095
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6609391361443335e-06,
+      "loss": 0.1079,
+      "step": 6096
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657810798983413e-06,
+      "loss": 0.0984,
+      "step": 6097
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65468298542209e-06,
+      "loss": 0.0989,
+      "step": 6098
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651555695837523e-06,
+      "loss": 0.1039,
+      "step": 6099
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648428930606825e-06,
+      "loss": 0.1024,
+      "step": 6100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645302690107029e-06,
+      "loss": 0.1033,
+      "step": 6101
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6421769747151055e-06,
+      "loss": 0.0985,
+      "step": 6102
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639051784807977e-06,
+      "loss": 0.0993,
+      "step": 6103
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.635927120762481e-06,
+      "loss": 0.1016,
+      "step": 6104
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632802982955414e-06,
+      "loss": 0.102,
+      "step": 6105
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.62967937176349e-06,
+      "loss": 0.1005,
+      "step": 6106
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6265562875633675e-06,
+      "loss": 0.1122,
+      "step": 6107
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.623433730731646e-06,
+      "loss": 0.1097,
+      "step": 6108
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.620311701644856e-06,
+      "loss": 0.1089,
+      "step": 6109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.617190200679456e-06,
+      "loss": 0.0987,
+      "step": 6110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.614069228211862e-06,
+      "loss": 0.1063,
+      "step": 6111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.610948784618409e-06,
+      "loss": 0.1006,
+      "step": 6112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.607828870275369e-06,
+      "loss": 0.0893,
+      "step": 6113
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604709485558963e-06,
+      "loss": 0.1062,
+      "step": 6114
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.601590630845335e-06,
+      "loss": 0.1033,
+      "step": 6115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598472306510567e-06,
+      "loss": 0.1053,
+      "step": 6116
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.595354512930686e-06,
+      "loss": 0.0977,
+      "step": 6117
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592237250481641e-06,
+      "loss": 0.0992,
+      "step": 6118
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.589120519539334e-06,
+      "loss": 0.1019,
+      "step": 6119
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.586004320479589e-06,
+      "loss": 0.1024,
+      "step": 6120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5828886536781665e-06,
+      "loss": 0.1043,
+      "step": 6121
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579773519510774e-06,
+      "loss": 0.1161,
+      "step": 6122
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576658918353045e-06,
+      "loss": 0.1111,
+      "step": 6123
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573544850580548e-06,
+      "loss": 0.0963,
+      "step": 6124
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570431316568795e-06,
+      "loss": 0.1139,
+      "step": 6125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567318316693229e-06,
+      "loss": 0.1005,
+      "step": 6126
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.564205851329221e-06,
+      "loss": 0.109,
+      "step": 6127
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.561093920852097e-06,
+      "loss": 0.1022,
+      "step": 6128
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5579825256370975e-06,
+      "loss": 0.1105,
+      "step": 6129
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.554871666059414e-06,
+      "loss": 0.1047,
+      "step": 6130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.551761342494166e-06,
+      "loss": 0.1101,
+      "step": 6131
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548651555316404e-06,
+      "loss": 0.0975,
+      "step": 6132
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.545542304901126e-06,
+      "loss": 0.099,
+      "step": 6133
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.542433591623256e-06,
+      "loss": 0.1042,
+      "step": 6134
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.539325415857652e-06,
+      "loss": 0.0987,
+      "step": 6135
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.53621777797912e-06,
+      "loss": 0.1025,
+      "step": 6136
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5331106783623865e-06,
+      "loss": 0.1048,
+      "step": 6137
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.530004117382115e-06,
+      "loss": 0.0967,
+      "step": 6138
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.526898095412916e-06,
+      "loss": 0.0986,
+      "step": 6139
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.523792612829319e-06,
+      "loss": 0.1062,
+      "step": 6140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.520687670005805e-06,
+      "loss": 0.1058,
+      "step": 6141
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.517583267316775e-06,
+      "loss": 0.1093,
+      "step": 6142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514479405136569e-06,
+      "loss": 0.0994,
+      "step": 6143
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.511376083839471e-06,
+      "loss": 0.0922,
+      "step": 6144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.508273303799689e-06,
+      "loss": 0.1115,
+      "step": 6145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.505171065391366e-06,
+      "loss": 0.107,
+      "step": 6146
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502069368988588e-06,
+      "loss": 0.0995,
+      "step": 6147
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.498968214965371e-06,
+      "loss": 0.1031,
+      "step": 6148
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.495867603695659e-06,
+      "loss": 0.0978,
+      "step": 6149
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.492767535553345e-06,
+      "loss": 0.1121,
+      "step": 6150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489668010912242e-06,
+      "loss": 0.1153,
+      "step": 6151
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.486569030146107e-06,
+      "loss": 0.1007,
+      "step": 6152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48347059362863e-06,
+      "loss": 0.0936,
+      "step": 6153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.480372701733427e-06,
+      "loss": 0.1073,
+      "step": 6154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.477275354834064e-06,
+      "loss": 0.1057,
+      "step": 6155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.474178553304027e-06,
+      "loss": 0.1018,
+      "step": 6156
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471082297516738e-06,
+      "loss": 0.1038,
+      "step": 6157
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.467986587845564e-06,
+      "loss": 0.1146,
+      "step": 6158
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.464891424663795e-06,
+      "loss": 0.1084,
+      "step": 6159
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.461796808344657e-06,
+      "loss": 0.1058,
+      "step": 6160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.458702739261317e-06,
+      "loss": 0.1083,
+      "step": 6161
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455609217786868e-06,
+      "loss": 0.099,
+      "step": 6162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.452516244294337e-06,
+      "loss": 0.0929,
+      "step": 6163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.449423819156694e-06,
+      "loss": 0.1039,
+      "step": 6164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4463319427468305e-06,
+      "loss": 0.1081,
+      "step": 6165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.443240615437586e-06,
+      "loss": 0.1014,
+      "step": 6166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.44014983760172e-06,
+      "loss": 0.0888,
+      "step": 6167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.43705960961193e-06,
+      "loss": 0.0956,
+      "step": 6168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.433969931840854e-06,
+      "loss": 0.1057,
+      "step": 6169
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.430880804661059e-06,
+      "loss": 0.1094,
+      "step": 6170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4277922284450355e-06,
+      "loss": 0.1096,
+      "step": 6171
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4247042035652285e-06,
+      "loss": 0.1059,
+      "step": 6172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.421616730394e-06,
+      "loss": 0.1085,
+      "step": 6173
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418529809303648e-06,
+      "loss": 0.1121,
+      "step": 6174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.415443440666413e-06,
+      "loss": 0.0986,
+      "step": 6175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.412357624854456e-06,
+      "loss": 0.1012,
+      "step": 6176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4092723622398835e-06,
+      "loss": 0.1015,
+      "step": 6177
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.406187653194728e-06,
+      "loss": 0.1105,
+      "step": 6178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40310349809095e-06,
+      "loss": 0.1102,
+      "step": 6179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400019897300462e-06,
+      "loss": 0.1118,
+      "step": 6180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.396936851195089e-06,
+      "loss": 0.1047,
+      "step": 6181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.393854360146597e-06,
+      "loss": 0.1058,
+      "step": 6182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.390772424526693e-06,
+      "loss": 0.1105,
+      "step": 6183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387691044707005e-06,
+      "loss": 0.1069,
+      "step": 6184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3846102210590964e-06,
+      "loss": 0.1008,
+      "step": 6185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381529953954474e-06,
+      "loss": 0.1005,
+      "step": 6186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.378450243764559e-06,
+      "loss": 0.1039,
+      "step": 6187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.375371090860727e-06,
+      "loss": 0.1162,
+      "step": 6188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372292495614269e-06,
+      "loss": 0.0923,
+      "step": 6189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369214458396412e-06,
+      "loss": 0.1044,
+      "step": 6190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.366136979578329e-06,
+      "loss": 0.0991,
+      "step": 6191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.363060059531107e-06,
+      "loss": 0.1121,
+      "step": 6192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.359983698625772e-06,
+      "loss": 0.0967,
+      "step": 6193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3569078972332945e-06,
+      "loss": 0.1058,
+      "step": 6194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.353832655724562e-06,
+      "loss": 0.1001,
+      "step": 6195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.350757974470396e-06,
+      "loss": 0.0975,
+      "step": 6196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.347683853841563e-06,
+      "loss": 0.1121,
+      "step": 6197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.344610294208745e-06,
+      "loss": 0.1078,
+      "step": 6198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.341537295942575e-06,
+      "loss": 0.104,
+      "step": 6199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3384648594136014e-06,
+      "loss": 0.1039,
+      "step": 6200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.335392984992308e-06,
+      "loss": 0.1113,
+      "step": 6201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.332321673049124e-06,
+      "loss": 0.1063,
+      "step": 6202
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.329250923954397e-06,
+      "loss": 0.1017,
+      "step": 6203
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.326180738078406e-06,
+      "loss": 0.1081,
+      "step": 6204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.323111115791374e-06,
+      "loss": 0.0983,
+      "step": 6205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3200420574634484e-06,
+      "loss": 0.0937,
+      "step": 6206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.316973563464703e-06,
+      "loss": 0.1089,
+      "step": 6207
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.313905634165156e-06,
+      "loss": 0.1007,
+      "step": 6208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.310838269934748e-06,
+      "loss": 0.0978,
+      "step": 6209
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.307771471143359e-06,
+      "loss": 0.1104,
+      "step": 6210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.304705238160795e-06,
+      "loss": 0.1014,
+      "step": 6211
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30163957135679e-06,
+      "loss": 0.0977,
+      "step": 6212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298574471101022e-06,
+      "loss": 0.1145,
+      "step": 6213
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.295509937763092e-06,
+      "loss": 0.0916,
+      "step": 6214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2924459717125296e-06,
+      "loss": 0.0969,
+      "step": 6215
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2893825733188096e-06,
+      "loss": 0.0911,
+      "step": 6216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286319742951325e-06,
+      "loss": 0.1132,
+      "step": 6217
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2832574809794e-06,
+      "loss": 0.1006,
+      "step": 6218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.280195787772304e-06,
+      "loss": 0.0784,
+      "step": 6219
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.277134663699225e-06,
+      "loss": 0.0743,
+      "step": 6220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274074109129281e-06,
+      "loss": 0.064,
+      "step": 6221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.271014124431537e-06,
+      "loss": 0.0707,
+      "step": 6222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.26795470997497e-06,
+      "loss": 0.0826,
+      "step": 6223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2648958661285044e-06,
+      "loss": 0.0821,
+      "step": 6224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.261837593260986e-06,
+      "loss": 0.0684,
+      "step": 6225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.258779891741188e-06,
+      "loss": 0.0769,
+      "step": 6226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25572276193783e-06,
+      "loss": 0.0679,
+      "step": 6227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.252666204219548e-06,
+      "loss": 0.0727,
+      "step": 6228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2496102189549146e-06,
+      "loss": 0.0637,
+      "step": 6229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246554806512442e-06,
+      "loss": 0.071,
+      "step": 6230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.243499967260557e-06,
+      "loss": 0.0678,
+      "step": 6231
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2404457015676225e-06,
+      "loss": 0.0695,
+      "step": 6232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237392009801944e-06,
+      "loss": 0.0773,
+      "step": 6233
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.234338892331743e-06,
+      "loss": 0.0692,
+      "step": 6234
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.231286349525173e-06,
+      "loss": 0.0722,
+      "step": 6235
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2282343817503326e-06,
+      "loss": 0.0704,
+      "step": 6236
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.225182989375235e-06,
+      "loss": 0.0678,
+      "step": 6237
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.222132172767827e-06,
+      "loss": 0.0709,
+      "step": 6238
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.219081932295996e-06,
+      "loss": 0.0728,
+      "step": 6239
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.216032268327546e-06,
+      "loss": 0.069,
+      "step": 6240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.212983181230227e-06,
+      "loss": 0.0668,
+      "step": 6241
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209934671371706e-06,
+      "loss": 0.079,
+      "step": 6242
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.206886739119581e-06,
+      "loss": 0.0687,
+      "step": 6243
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.203839384841394e-06,
+      "loss": 0.0683,
+      "step": 6244
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.200792608904601e-06,
+      "loss": 0.0697,
+      "step": 6245
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.197746411676595e-06,
+      "loss": 0.0691,
+      "step": 6246
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.194700793524705e-06,
+      "loss": 0.0685,
+      "step": 6247
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.19165575481618e-06,
+      "loss": 0.0717,
+      "step": 6248
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1886112959182e-06,
+      "loss": 0.0741,
+      "step": 6249
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.18556741719789e-06,
+      "loss": 0.0612,
+      "step": 6250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.182524119022285e-06,
+      "loss": 0.073,
+      "step": 6251
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.179481401758357e-06,
+      "loss": 0.0717,
+      "step": 6252
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.176439265773019e-06,
+      "loss": 0.0723,
+      "step": 6253
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.173397711433095e-06,
+      "loss": 0.0776,
+      "step": 6254
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170356739105357e-06,
+      "loss": 0.0652,
+      "step": 6255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.167316349156495e-06,
+      "loss": 0.0614,
+      "step": 6256
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.164276541953126e-06,
+      "loss": 0.0749,
+      "step": 6257
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.161237317861812e-06,
+      "loss": 0.0694,
+      "step": 6258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158198677249033e-06,
+      "loss": 0.0669,
+      "step": 6259
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.155160620481197e-06,
+      "loss": 0.0808,
+      "step": 6260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.15212314792465e-06,
+      "loss": 0.0717,
+      "step": 6261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.149086259945664e-06,
+      "loss": 0.0733,
+      "step": 6262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.146049956910434e-06,
+      "loss": 0.0781,
+      "step": 6263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.143014239185098e-06,
+      "loss": 0.0686,
+      "step": 6264
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.139979107135708e-06,
+      "loss": 0.0677,
+      "step": 6265
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.136944561128261e-06,
+      "loss": 0.0721,
+      "step": 6266
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.133910601528671e-06,
+      "loss": 0.0695,
+      "step": 6267
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.130877228702782e-06,
+      "loss": 0.0676,
+      "step": 6268
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.127844443016381e-06,
+      "loss": 0.0682,
+      "step": 6269
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1248122448351664e-06,
+      "loss": 0.0677,
+      "step": 6270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121780634524772e-06,
+      "loss": 0.0723,
+      "step": 6271
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.11874961245077e-06,
+      "loss": 0.0646,
+      "step": 6272
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.11571917897865e-06,
+      "loss": 0.0654,
+      "step": 6273
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.112689334473829e-06,
+      "loss": 0.0706,
+      "step": 6274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.109660079301668e-06,
+      "loss": 0.0635,
+      "step": 6275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.10663141382744e-06,
+      "loss": 0.0683,
+      "step": 6276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1036033384163605e-06,
+      "loss": 0.0682,
+      "step": 6277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100575853433565e-06,
+      "loss": 0.0681,
+      "step": 6278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.097548959244116e-06,
+      "loss": 0.0675,
+      "step": 6279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094522656213017e-06,
+      "loss": 0.0735,
+      "step": 6280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.09149694470519e-06,
+      "loss": 0.064,
+      "step": 6281
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.088471825085481e-06,
+      "loss": 0.0716,
+      "step": 6282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.085447297718683e-06,
+      "loss": 0.0671,
+      "step": 6283
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.082423362969501e-06,
+      "loss": 0.0795,
+      "step": 6284
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.07940002120257e-06,
+      "loss": 0.063,
+      "step": 6285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.076377272782464e-06,
+      "loss": 0.0646,
+      "step": 6286
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.073355118073673e-06,
+      "loss": 0.0661,
+      "step": 6287
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0703335574406274e-06,
+      "loss": 0.07,
+      "step": 6288
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0673125912476786e-06,
+      "loss": 0.0698,
+      "step": 6289
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0642922198591e-06,
+      "loss": 0.0669,
+      "step": 6290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.061272443639111e-06,
+      "loss": 0.0671,
+      "step": 6291
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.058253262951844e-06,
+      "loss": 0.0708,
+      "step": 6292
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0552346781613624e-06,
+      "loss": 0.0716,
+      "step": 6293
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.052216689631666e-06,
+      "loss": 0.0654,
+      "step": 6294
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.049199297726673e-06,
+      "loss": 0.0704,
+      "step": 6295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0461825028102305e-06,
+      "loss": 0.0698,
+      "step": 6296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0431663052461235e-06,
+      "loss": 0.0758,
+      "step": 6297
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.040150705398049e-06,
+      "loss": 0.0691,
+      "step": 6298
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03713570362965e-06,
+      "loss": 0.069,
+      "step": 6299
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034121300304484e-06,
+      "loss": 0.0717,
+      "step": 6300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.031107495786037e-06,
+      "loss": 0.0722,
+      "step": 6301
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0280942904377335e-06,
+      "loss": 0.0672,
+      "step": 6302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.025081684622915e-06,
+      "loss": 0.0716,
+      "step": 6303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.022069678704851e-06,
+      "loss": 0.0685,
+      "step": 6304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.019058273046748e-06,
+      "loss": 0.0736,
+      "step": 6305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.016047468011733e-06,
+      "loss": 0.0755,
+      "step": 6306
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.013037263962854e-06,
+      "loss": 0.0667,
+      "step": 6307
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.010027661263106e-06,
+      "loss": 0.073,
+      "step": 6308
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.007018660275392e-06,
+      "loss": 0.0732,
+      "step": 6309
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0040102613625484e-06,
+      "loss": 0.0709,
+      "step": 6310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.001002464887348e-06,
+      "loss": 0.0721,
+      "step": 6311
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.997995271212477e-06,
+      "loss": 0.0732,
+      "step": 6312
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994988680700561e-06,
+      "loss": 0.0736,
+      "step": 6313
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.991982693714146e-06,
+      "loss": 0.0737,
+      "step": 6314
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9889773106157e-06,
+      "loss": 0.0676,
+      "step": 6315
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.985972531767635e-06,
+      "loss": 0.0654,
+      "step": 6316
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982968357532276e-06,
+      "loss": 0.0678,
+      "step": 6317
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.979964788271874e-06,
+      "loss": 0.0676,
+      "step": 6318
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9769618243486205e-06,
+      "loss": 0.0658,
+      "step": 6319
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.973959466124622e-06,
+      "loss": 0.0763,
+      "step": 6320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9709577139619115e-06,
+      "loss": 0.0619,
+      "step": 6321
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.967956568222459e-06,
+      "loss": 0.0715,
+      "step": 6322
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964956029268151e-06,
+      "loss": 0.0755,
+      "step": 6323
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.961956097460811e-06,
+      "loss": 0.0687,
+      "step": 6324
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.958956773162179e-06,
+      "loss": 0.0686,
+      "step": 6325
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9559580567339235e-06,
+      "loss": 0.0702,
+      "step": 6326
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9529599485376514e-06,
+      "loss": 0.0716,
+      "step": 6327
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.949962448934881e-06,
+      "loss": 0.0738,
+      "step": 6328
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.94696555828706e-06,
+      "loss": 0.0688,
+      "step": 6329
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943969276955575e-06,
+      "loss": 0.0729,
+      "step": 6330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.940973605301726e-06,
+      "loss": 0.0656,
+      "step": 6331
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.93797854368674e-06,
+      "loss": 0.0687,
+      "step": 6332
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.934984092471782e-06,
+      "loss": 0.0747,
+      "step": 6333
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.931990252017928e-06,
+      "loss": 0.0677,
+      "step": 6334
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.928997022686195e-06,
+      "loss": 0.0671,
+      "step": 6335
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.926004404837517e-06,
+      "loss": 0.0679,
+      "step": 6336
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.923012398832751e-06,
+      "loss": 0.067,
+      "step": 6337
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920021005032693e-06,
+      "loss": 0.0695,
+      "step": 6338
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.917030223798057e-06,
+      "loss": 0.0711,
+      "step": 6339
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9140400554894775e-06,
+      "loss": 0.0648,
+      "step": 6340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91105050046753e-06,
+      "loss": 0.0664,
+      "step": 6341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.908061559092705e-06,
+      "loss": 0.0736,
+      "step": 6342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.905073231725418e-06,
+      "loss": 0.0698,
+      "step": 6343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902085518726019e-06,
+      "loss": 0.0829,
+      "step": 6344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.899098420454775e-06,
+      "loss": 0.0632,
+      "step": 6345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896111937271888e-06,
+      "loss": 0.0685,
+      "step": 6346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.893126069537479e-06,
+      "loss": 0.0742,
+      "step": 6347
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.890140817611592e-06,
+      "loss": 0.0691,
+      "step": 6348
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.887156181854208e-06,
+      "loss": 0.0694,
+      "step": 6349
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.884172162625225e-06,
+      "loss": 0.0692,
+      "step": 6350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881188760284462e-06,
+      "loss": 0.0644,
+      "step": 6351
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878205975191682e-06,
+      "loss": 0.0675,
+      "step": 6352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.875223807706555e-06,
+      "loss": 0.0769,
+      "step": 6353
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872242258188681e-06,
+      "loss": 0.0756,
+      "step": 6354
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8692613269975945e-06,
+      "loss": 0.0713,
+      "step": 6355
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866281014492742e-06,
+      "loss": 0.0766,
+      "step": 6356
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.86330132103351e-06,
+      "loss": 0.0664,
+      "step": 6357
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8603222469791975e-06,
+      "loss": 0.0681,
+      "step": 6358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.857343792689032e-06,
+      "loss": 0.0716,
+      "step": 6359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.854365958522174e-06,
+      "loss": 0.0635,
+      "step": 6360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.851388744837701e-06,
+      "loss": 0.0743,
+      "step": 6361
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.848412151994613e-06,
+      "loss": 0.0756,
+      "step": 6362
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845436180351848e-06,
+      "loss": 0.0746,
+      "step": 6363
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.842460830268259e-06,
+      "loss": 0.0713,
+      "step": 6364
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.83948610210262e-06,
+      "loss": 0.0716,
+      "step": 6365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836511996213646e-06,
+      "loss": 0.0677,
+      "step": 6366
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.833538512959963e-06,
+      "loss": 0.0718,
+      "step": 6367
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.830565652700122e-06,
+      "loss": 0.066,
+      "step": 6368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.827593415792612e-06,
+      "loss": 0.0679,
+      "step": 6369
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.824621802595828e-06,
+      "loss": 0.0733,
+      "step": 6370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.821650813468109e-06,
+      "loss": 0.0748,
+      "step": 6371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.818680448767705e-06,
+      "loss": 0.0729,
+      "step": 6372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.815710708852792e-06,
+      "loss": 0.0682,
+      "step": 6373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812741594081482e-06,
+      "loss": 0.0724,
+      "step": 6374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.809773104811797e-06,
+      "loss": 0.0674,
+      "step": 6375
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.806805241401689e-06,
+      "loss": 0.0679,
+      "step": 6376
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8038380042090425e-06,
+      "loss": 0.0722,
+      "step": 6377
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.800871393591655e-06,
+      "loss": 0.0676,
+      "step": 6378
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.79790540990725e-06,
+      "loss": 0.07,
+      "step": 6379
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794940053513485e-06,
+      "loss": 0.0691,
+      "step": 6380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7919753247679276e-06,
+      "loss": 0.0689,
+      "step": 6381
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789011224028087e-06,
+      "loss": 0.0722,
+      "step": 6382
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7860477516513816e-06,
+      "loss": 0.0743,
+      "step": 6383
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.783084907995156e-06,
+      "loss": 0.0727,
+      "step": 6384
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.78012269341669e-06,
+      "loss": 0.0693,
+      "step": 6385
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.777161108273176e-06,
+      "loss": 0.0634,
+      "step": 6386
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774200152921731e-06,
+      "loss": 0.0636,
+      "step": 6387
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.771239827719406e-06,
+      "loss": 0.0698,
+      "step": 6388
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7682801330231684e-06,
+      "loss": 0.0798,
+      "step": 6389
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.765321069189904e-06,
+      "loss": 0.0625,
+      "step": 6390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.76236263657644e-06,
+      "loss": 0.0722,
+      "step": 6391
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.759404835539506e-06,
+      "loss": 0.0714,
+      "step": 6392
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.756447666435776e-06,
+      "loss": 0.0674,
+      "step": 6393
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.753491129621833e-06,
+      "loss": 0.0694,
+      "step": 6394
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.750535225454186e-06,
+      "loss": 0.0742,
+      "step": 6395
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.747579954289273e-06,
+      "loss": 0.0625,
+      "step": 6396
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.744625316483458e-06,
+      "loss": 0.0707,
+      "step": 6397
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.741671312393019e-06,
+      "loss": 0.07,
+      "step": 6398
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.738717942374159e-06,
+      "loss": 0.0706,
+      "step": 6399
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7357652067830166e-06,
+      "loss": 0.0677,
+      "step": 6400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.732813105975636e-06,
+      "loss": 0.0728,
+      "step": 6401
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7298616403080015e-06,
+      "loss": 0.0698,
+      "step": 6402
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.72691081013601e-06,
+      "loss": 0.0736,
+      "step": 6403
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.723960615815483e-06,
+      "loss": 0.0676,
+      "step": 6404
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7210110577021725e-06,
+      "loss": 0.0696,
+      "step": 6405
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7180621361517464e-06,
+      "loss": 0.0747,
+      "step": 6406
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.715113851519794e-06,
+      "loss": 0.073,
+      "step": 6407
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.71216620416184e-06,
+      "loss": 0.0709,
+      "step": 6408
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709219194433319e-06,
+      "loss": 0.0701,
+      "step": 6409
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.706272822689592e-06,
+      "loss": 0.0737,
+      "step": 6410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703327089285951e-06,
+      "loss": 0.0671,
+      "step": 6411
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7003819945776e-06,
+      "loss": 0.0754,
+      "step": 6412
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697437538919675e-06,
+      "loss": 0.0683,
+      "step": 6413
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6944937226672295e-06,
+      "loss": 0.0674,
+      "step": 6414
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.691550546175237e-06,
+      "loss": 0.0689,
+      "step": 6415
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6886080097986056e-06,
+      "loss": 0.0698,
+      "step": 6416
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685666113892155e-06,
+      "loss": 0.0718,
+      "step": 6417
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6827248588106286e-06,
+      "loss": 0.0699,
+      "step": 6418
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.679784244908702e-06,
+      "loss": 0.0693,
+      "step": 6419
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.676844272540964e-06,
+      "loss": 0.0663,
+      "step": 6420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.673904942061923e-06,
+      "loss": 0.0686,
+      "step": 6421
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.670966253826027e-06,
+      "loss": 0.0716,
+      "step": 6422
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.668028208187625e-06,
+      "loss": 0.0763,
+      "step": 6423
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.665090805501008e-06,
+      "loss": 0.0705,
+      "step": 6424
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6621540461203765e-06,
+      "loss": 0.0686,
+      "step": 6425
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659217930399853e-06,
+      "loss": 0.0697,
+      "step": 6426
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.656282458693495e-06,
+      "loss": 0.0668,
+      "step": 6427
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65334763135527e-06,
+      "loss": 0.0655,
+      "step": 6428
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6504134487390685e-06,
+      "loss": 0.0763,
+      "step": 6429
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6474799111987136e-06,
+      "loss": 0.0697,
+      "step": 6430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64454701908794e-06,
+      "loss": 0.0648,
+      "step": 6431
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641614772760406e-06,
+      "loss": 0.0725,
+      "step": 6432
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6386831725697005e-06,
+      "loss": 0.0755,
+      "step": 6433
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.635752218869321e-06,
+      "loss": 0.0703,
+      "step": 6434
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6328219120127e-06,
+      "loss": 0.0716,
+      "step": 6435
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.629892252353186e-06,
+      "loss": 0.063,
+      "step": 6436
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.626963240244044e-06,
+      "loss": 0.0709,
+      "step": 6437
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.624034876038476e-06,
+      "loss": 0.0728,
+      "step": 6438
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621107160089589e-06,
+      "loss": 0.0634,
+      "step": 6439
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.618180092750421e-06,
+      "loss": 0.0645,
+      "step": 6440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.615253674373933e-06,
+      "loss": 0.0701,
+      "step": 6441
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.612327905313003e-06,
+      "loss": 0.0705,
+      "step": 6442
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.60940278592043e-06,
+      "loss": 0.0695,
+      "step": 6443
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.606478316548942e-06,
+      "loss": 0.0733,
+      "step": 6444
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.603554497551184e-06,
+      "loss": 0.0658,
+      "step": 6445
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.600631329279716e-06,
+      "loss": 0.0678,
+      "step": 6446
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597708812087033e-06,
+      "loss": 0.0715,
+      "step": 6447
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.594786946325539e-06,
+      "loss": 0.0655,
+      "step": 6448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.591865732347571e-06,
+      "loss": 0.0694,
+      "step": 6449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.58894517050538e-06,
+      "loss": 0.0663,
+      "step": 6450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5860252611511325e-06,
+      "loss": 0.0692,
+      "step": 6451
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.583106004636934e-06,
+      "loss": 0.0696,
+      "step": 6452
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.580187401314795e-06,
+      "loss": 0.0645,
+      "step": 6453
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.577269451536651e-06,
+      "loss": 0.0731,
+      "step": 6454
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.574352155654368e-06,
+      "loss": 0.0734,
+      "step": 6455
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5714355140197205e-06,
+      "loss": 0.0691,
+      "step": 6456
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5685195269844076e-06,
+      "loss": 0.0631,
+      "step": 6457
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.565604194900059e-06,
+      "loss": 0.0679,
+      "step": 6458
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5626895181182075e-06,
+      "loss": 0.0701,
+      "step": 6459
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559775496990329e-06,
+      "loss": 0.0679,
+      "step": 6460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.556862131867801e-06,
+      "loss": 0.0659,
+      "step": 6461
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.553949423101928e-06,
+      "loss": 0.0776,
+      "step": 6462
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.551037371043943e-06,
+      "loss": 0.069,
+      "step": 6463
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54812597604499e-06,
+      "loss": 0.0769,
+      "step": 6464
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5452152384561346e-06,
+      "loss": 0.0661,
+      "step": 6465
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.542305158628373e-06,
+      "loss": 0.0702,
+      "step": 6466
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.539395736912609e-06,
+      "loss": 0.073,
+      "step": 6467
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.536486973659674e-06,
+      "loss": 0.067,
+      "step": 6468
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.533578869220322e-06,
+      "loss": 0.076,
+      "step": 6469
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.53067142394522e-06,
+      "loss": 0.0687,
+      "step": 6470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.527764638184965e-06,
+      "loss": 0.0724,
+      "step": 6471
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524858512290068e-06,
+      "loss": 0.0736,
+      "step": 6472
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5219530466109575e-06,
+      "loss": 0.0712,
+      "step": 6473
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.519048241497994e-06,
+      "loss": 0.0731,
+      "step": 6474
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.516144097301448e-06,
+      "loss": 0.0737,
+      "step": 6475
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513240614371509e-06,
+      "loss": 0.0638,
+      "step": 6476
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5103377930583005e-06,
+      "loss": 0.0713,
+      "step": 6477
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.507435633711851e-06,
+      "loss": 0.0604,
+      "step": 6478
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.504534136682113e-06,
+      "loss": 0.0672,
+      "step": 6479
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.501633302318968e-06,
+      "loss": 0.0812,
+      "step": 6480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.498733130972205e-06,
+      "loss": 0.0632,
+      "step": 6481
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.495833622991545e-06,
+      "loss": 0.0648,
+      "step": 6482
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.492934778726618e-06,
+      "loss": 0.0684,
+      "step": 6483
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4900365985269775e-06,
+      "loss": 0.0737,
+      "step": 6484
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.487139082742106e-06,
+      "loss": 0.0616,
+      "step": 6485
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4842422317213915e-06,
+      "loss": 0.069,
+      "step": 6486
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.481346045814149e-06,
+      "loss": 0.0751,
+      "step": 6487
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.478450525369618e-06,
+      "loss": 0.0713,
+      "step": 6488
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.475555670736948e-06,
+      "loss": 0.068,
+      "step": 6489
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4726614822652105e-06,
+      "loss": 0.0641,
+      "step": 6490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469767960303406e-06,
+      "loss": 0.0666,
+      "step": 6491
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.466875105200441e-06,
+      "loss": 0.0697,
+      "step": 6492
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.463982917305155e-06,
+      "loss": 0.0767,
+      "step": 6493
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.461091396966297e-06,
+      "loss": 0.0609,
+      "step": 6494
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.458200544532535e-06,
+      "loss": 0.0692,
+      "step": 6495
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455310360352467e-06,
+      "loss": 0.0621,
+      "step": 6496
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452420844774601e-06,
+      "loss": 0.0654,
+      "step": 6497
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4495319981473614e-06,
+      "loss": 0.0624,
+      "step": 6498
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.446643820819108e-06,
+      "loss": 0.0762,
+      "step": 6499
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.443756313138104e-06,
+      "loss": 0.073,
+      "step": 6500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440869475452533e-06,
+      "loss": 0.0711,
+      "step": 6501
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.43798330811051e-06,
+      "loss": 0.0632,
+      "step": 6502
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.435097811460058e-06,
+      "loss": 0.0755,
+      "step": 6503
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.432212985849119e-06,
+      "loss": 0.0639,
+      "step": 6504
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429328831625565e-06,
+      "loss": 0.075,
+      "step": 6505
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.426445349137172e-06,
+      "loss": 0.075,
+      "step": 6506
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423562538731648e-06,
+      "loss": 0.0671,
+      "step": 6507
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.420680400756614e-06,
+      "loss": 0.0699,
+      "step": 6508
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.417798935559604e-06,
+      "loss": 0.0685,
+      "step": 6509
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.414918143488086e-06,
+      "loss": 0.0672,
+      "step": 6510
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4120380248894356e-06,
+      "loss": 0.065,
+      "step": 6511
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.409158580110944e-06,
+      "loss": 0.0626,
+      "step": 6512
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.406279809499834e-06,
+      "loss": 0.0682,
+      "step": 6513
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.403401713403238e-06,
+      "loss": 0.0694,
+      "step": 6514
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4005242921682056e-06,
+      "loss": 0.0705,
+      "step": 6515
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397647546141713e-06,
+      "loss": 0.0674,
+      "step": 6516
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394771475670646e-06,
+      "loss": 0.072,
+      "step": 6517
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.39189608110182e-06,
+      "loss": 0.0746,
+      "step": 6518
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.389021362781958e-06,
+      "loss": 0.0628,
+      "step": 6519
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386147321057702e-06,
+      "loss": 0.0734,
+      "step": 6520
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3832739562756245e-06,
+      "loss": 0.0604,
+      "step": 6521
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380401268782203e-06,
+      "loss": 0.0699,
+      "step": 6522
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.377529258923835e-06,
+      "loss": 0.0719,
+      "step": 6523
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.374657927046848e-06,
+      "loss": 0.0711,
+      "step": 6524
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.371787273497475e-06,
+      "loss": 0.0713,
+      "step": 6525
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3689172986218675e-06,
+      "loss": 0.0667,
+      "step": 6526
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.366048002766108e-06,
+      "loss": 0.0717,
+      "step": 6527
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.363179386276178e-06,
+      "loss": 0.0707,
+      "step": 6528
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.360311449497999e-06,
+      "loss": 0.0704,
+      "step": 6529
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.357444192777393e-06,
+      "loss": 0.0723,
+      "step": 6530
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.354577616460101e-06,
+      "loss": 0.0684,
+      "step": 6531
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.351711720891797e-06,
+      "loss": 0.0701,
+      "step": 6532
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.348846506418057e-06,
+      "loss": 0.0754,
+      "step": 6533
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.345981973384379e-06,
+      "loss": 0.0735,
+      "step": 6534
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.343118122136186e-06,
+      "loss": 0.069,
+      "step": 6535
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.340254953018811e-06,
+      "loss": 0.0686,
+      "step": 6536
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.337392466377503e-06,
+      "loss": 0.0725,
+      "step": 6537
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.334530662557439e-06,
+      "loss": 0.0665,
+      "step": 6538
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331669541903703e-06,
+      "loss": 0.0712,
+      "step": 6539
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.328809104761305e-06,
+      "loss": 0.0661,
+      "step": 6540
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.325949351475166e-06,
+      "loss": 0.0658,
+      "step": 6541
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.323090282390126e-06,
+      "loss": 0.0685,
+      "step": 6542
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.320231897850947e-06,
+      "loss": 0.0782,
+      "step": 6543
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3173741982023055e-06,
+      "loss": 0.0636,
+      "step": 6544
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3145171837887885e-06,
+      "loss": 0.0696,
+      "step": 6545
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3116608549549155e-06,
+      "loss": 0.069,
+      "step": 6546
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3088052120451104e-06,
+      "loss": 0.0667,
+      "step": 6547
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.305950255403716e-06,
+      "loss": 0.0692,
+      "step": 6548
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.303095985375001e-06,
+      "loss": 0.073,
+      "step": 6549
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.300242402303139e-06,
+      "loss": 0.0667,
+      "step": 6550
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297389506532234e-06,
+      "loss": 0.062,
+      "step": 6551
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2945372984062985e-06,
+      "loss": 0.0702,
+      "step": 6552
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.291685778269258e-06,
+      "loss": 0.0755,
+      "step": 6553
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.288834946464968e-06,
+      "loss": 0.0708,
+      "step": 6554
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.285984803337192e-06,
+      "loss": 0.0677,
+      "step": 6555
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.283135349229608e-06,
+      "loss": 0.0672,
+      "step": 6556
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2802865844858225e-06,
+      "loss": 0.0656,
+      "step": 6557
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.277438509449347e-06,
+      "loss": 0.0726,
+      "step": 6558
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.274591124463613e-06,
+      "loss": 0.0668,
+      "step": 6559
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.271744429871976e-06,
+      "loss": 0.0674,
+      "step": 6560
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.268898426017698e-06,
+      "loss": 0.067,
+      "step": 6561
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266053113243961e-06,
+      "loss": 0.0741,
+      "step": 6562
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.263208491893867e-06,
+      "loss": 0.0704,
+      "step": 6563
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.260364562310435e-06,
+      "loss": 0.0703,
+      "step": 6564
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2575213248365934e-06,
+      "loss": 0.0716,
+      "step": 6565
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2546787798151975e-06,
+      "loss": 0.068,
+      "step": 6566
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.251836927589012e-06,
+      "loss": 0.0732,
+      "step": 6567
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.248995768500712e-06,
+      "loss": 0.0642,
+      "step": 6568
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.246155302892907e-06,
+      "loss": 0.0659,
+      "step": 6569
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243315531108103e-06,
+      "loss": 0.068,
+      "step": 6570
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.240476453488741e-06,
+      "loss": 0.0754,
+      "step": 6571
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2376380703771636e-06,
+      "loss": 0.0686,
+      "step": 6572
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234800382115632e-06,
+      "loss": 0.072,
+      "step": 6573
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.231963389046335e-06,
+      "loss": 0.0686,
+      "step": 6574
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.229127091511366e-06,
+      "loss": 0.0684,
+      "step": 6575
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2262914898527305e-06,
+      "loss": 0.0674,
+      "step": 6576
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.223456584412369e-06,
+      "loss": 0.0709,
+      "step": 6577
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.22062237553212e-06,
+      "loss": 0.0682,
+      "step": 6578
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.217788863553742e-06,
+      "loss": 0.0673,
+      "step": 6579
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.214956048818919e-06,
+      "loss": 0.0713,
+      "step": 6580
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.212123931669236e-06,
+      "loss": 0.081,
+      "step": 6581
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.209292512446211e-06,
+      "loss": 0.0693,
+      "step": 6582
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.206461791491261e-06,
+      "loss": 0.0711,
+      "step": 6583
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.203631769145726e-06,
+      "loss": 0.0642,
+      "step": 6584
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2008024457508676e-06,
+      "loss": 0.0676,
+      "step": 6585
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.197973821647856e-06,
+      "loss": 0.0661,
+      "step": 6586
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.195145897177773e-06,
+      "loss": 0.07,
+      "step": 6587
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.192318672681631e-06,
+      "loss": 0.0694,
+      "step": 6588
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189492148500343e-06,
+      "loss": 0.0661,
+      "step": 6589
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.18666632497474e-06,
+      "loss": 0.0768,
+      "step": 6590
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183841202445581e-06,
+      "loss": 0.0696,
+      "step": 6591
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.181016781253526e-06,
+      "loss": 0.0648,
+      "step": 6592
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.178193061739152e-06,
+      "loss": 0.0786,
+      "step": 6593
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.175370044242962e-06,
+      "loss": 0.0765,
+      "step": 6594
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.172547729105362e-06,
+      "loss": 0.0682,
+      "step": 6595
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.169726116666685e-06,
+      "loss": 0.0729,
+      "step": 6596
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1669052072671675e-06,
+      "loss": 0.0658,
+      "step": 6597
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.164085001246967e-06,
+      "loss": 0.0651,
+      "step": 6598
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161265498946159e-06,
+      "loss": 0.0723,
+      "step": 6599
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.15844670070473e-06,
+      "loss": 0.0703,
+      "step": 6600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.155628606862578e-06,
+      "loss": 0.0761,
+      "step": 6601
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.152811217759529e-06,
+      "loss": 0.061,
+      "step": 6602
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.149994533735312e-06,
+      "loss": 0.0655,
+      "step": 6603
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.14717855512957e-06,
+      "loss": 0.0682,
+      "step": 6604
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144363282281873e-06,
+      "loss": 0.0672,
+      "step": 6605
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.141548715531693e-06,
+      "loss": 0.0731,
+      "step": 6606
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.138734855218427e-06,
+      "loss": 0.0728,
+      "step": 6607
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.13592170168138e-06,
+      "loss": 0.0725,
+      "step": 6608
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.13310925525977e-06,
+      "loss": 0.0682,
+      "step": 6609
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.130297516292743e-06,
+      "loss": 0.0707,
+      "step": 6610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127486485119344e-06,
+      "loss": 0.0683,
+      "step": 6611
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.124676162078537e-06,
+      "loss": 0.0749,
+      "step": 6612
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.121866547509209e-06,
+      "loss": 0.0725,
+      "step": 6613
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.119057641750153e-06,
+      "loss": 0.0755,
+      "step": 6614
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.116249445140074e-06,
+      "loss": 0.0635,
+      "step": 6615
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1134419580176024e-06,
+      "loss": 0.0691,
+      "step": 6616
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110635180721272e-06,
+      "loss": 0.0731,
+      "step": 6617
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1078291135895425e-06,
+      "loss": 0.0644,
+      "step": 6618
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.105023756960777e-06,
+      "loss": 0.0741,
+      "step": 6619
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1022191111732545e-06,
+      "loss": 0.0728,
+      "step": 6620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.099415176565177e-06,
+      "loss": 0.077,
+      "step": 6621
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.096611953474653e-06,
+      "loss": 0.0706,
+      "step": 6622
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0938094422397025e-06,
+      "loss": 0.0766,
+      "step": 6623
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091007643198271e-06,
+      "loss": 0.0618,
+      "step": 6624
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.088206556688209e-06,
+      "loss": 0.0724,
+      "step": 6625
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.085406183047278e-06,
+      "loss": 0.0732,
+      "step": 6626
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082606522613168e-06,
+      "loss": 0.0664,
+      "step": 6627
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.079807575723466e-06,
+      "loss": 0.0692,
+      "step": 6628
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.077009342715687e-06,
+      "loss": 0.0694,
+      "step": 6629
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.074211823927252e-06,
+      "loss": 0.0677,
+      "step": 6630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071415019695493e-06,
+      "loss": 0.0698,
+      "step": 6631
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0686189303576674e-06,
+      "loss": 0.0711,
+      "step": 6632
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.065823556250937e-06,
+      "loss": 0.0728,
+      "step": 6633
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.063028897712377e-06,
+      "loss": 0.07,
+      "step": 6634
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.060234955078984e-06,
+      "loss": 0.0746,
+      "step": 6635
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0574417286876595e-06,
+      "loss": 0.0719,
+      "step": 6636
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.054649218875223e-06,
+      "loss": 0.0685,
+      "step": 6637
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0518574259784105e-06,
+      "loss": 0.0628,
+      "step": 6638
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.049066350333862e-06,
+      "loss": 0.0586,
+      "step": 6639
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.046275992278146e-06,
+      "loss": 0.0716,
+      "step": 6640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043486352147729e-06,
+      "loss": 0.0695,
+      "step": 6641
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0406974302789994e-06,
+      "loss": 0.0722,
+      "step": 6642
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.037909227008259e-06,
+      "loss": 0.0704,
+      "step": 6643
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.035121742671719e-06,
+      "loss": 0.0698,
+      "step": 6644
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0323349776055045e-06,
+      "loss": 0.0637,
+      "step": 6645
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.029548932145659e-06,
+      "loss": 0.0659,
+      "step": 6646
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026763606628137e-06,
+      "loss": 0.0687,
+      "step": 6647
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.023979001388797e-06,
+      "loss": 0.0659,
+      "step": 6648
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.021195116763428e-06,
+      "loss": 0.0673,
+      "step": 6649
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.018411953087719e-06,
+      "loss": 0.0692,
+      "step": 6650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.01562951069727e-06,
+      "loss": 0.0641,
+      "step": 6651
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.012847789927608e-06,
+      "loss": 0.0714,
+      "step": 6652
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.010066791114159e-06,
+      "loss": 0.0687,
+      "step": 6653
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.007286514592273e-06,
+      "loss": 0.0741,
+      "step": 6654
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.004506960697206e-06,
+      "loss": 0.067,
+      "step": 6655
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.001728129764122e-06,
+      "loss": 0.065,
+      "step": 6656
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.998950022128113e-06,
+      "loss": 0.0693,
+      "step": 6657
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.996172638124172e-06,
+      "loss": 0.0679,
+      "step": 6658
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.993395978087202e-06,
+      "loss": 0.0625,
+      "step": 6659
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.990620042352034e-06,
+      "loss": 0.0675,
+      "step": 6660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987844831253398e-06,
+      "loss": 0.0695,
+      "step": 6661
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.985070345125936e-06,
+      "loss": 0.0684,
+      "step": 6662
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.982296584304215e-06,
+      "loss": 0.0685,
+      "step": 6663
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.979523549122699e-06,
+      "loss": 0.0663,
+      "step": 6664
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.976751239915781e-06,
+      "loss": 0.0817,
+      "step": 6665
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9739796570177544e-06,
+      "loss": 0.0723,
+      "step": 6666
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.971208800762823e-06,
+      "loss": 0.0654,
+      "step": 6667
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.968438671485119e-06,
+      "loss": 0.0733,
+      "step": 6668
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.965669269518667e-06,
+      "loss": 0.0706,
+      "step": 6669
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962900595197415e-06,
+      "loss": 0.0665,
+      "step": 6670
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.960132648855226e-06,
+      "loss": 0.0641,
+      "step": 6671
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.957365430825868e-06,
+      "loss": 0.0669,
+      "step": 6672
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.95459894144302e-06,
+      "loss": 0.0711,
+      "step": 6673
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.951833181040284e-06,
+      "loss": 0.0697,
+      "step": 6674
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.949068149951161e-06,
+      "loss": 0.0734,
+      "step": 6675
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9463038485090755e-06,
+      "loss": 0.0695,
+      "step": 6676
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9435402770473575e-06,
+      "loss": 0.0655,
+      "step": 6677
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.940777435899245e-06,
+      "loss": 0.0677,
+      "step": 6678
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9380153253978995e-06,
+      "loss": 0.0608,
+      "step": 6679
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935253945876386e-06,
+      "loss": 0.0734,
+      "step": 6680
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93249329766768e-06,
+      "loss": 0.078,
+      "step": 6681
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929733381104678e-06,
+      "loss": 0.0758,
+      "step": 6682
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926974196520179e-06,
+      "loss": 0.0741,
+      "step": 6683
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.924215744246896e-06,
+      "loss": 0.0654,
+      "step": 6684
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.921458024617457e-06,
+      "loss": 0.0763,
+      "step": 6685
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9187010379643975e-06,
+      "loss": 0.0705,
+      "step": 6686
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915944784620171e-06,
+      "loss": 0.0744,
+      "step": 6687
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913189264917137e-06,
+      "loss": 0.0678,
+      "step": 6688
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.910434479187561e-06,
+      "loss": 0.0753,
+      "step": 6689
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907680427763636e-06,
+      "loss": 0.0723,
+      "step": 6690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.904927110977453e-06,
+      "loss": 0.0712,
+      "step": 6691
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902174529161016e-06,
+      "loss": 0.0687,
+      "step": 6692
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.899422682646249e-06,
+      "loss": 0.0743,
+      "step": 6693
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.896671571764977e-06,
+      "loss": 0.0724,
+      "step": 6694
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89392119684894e-06,
+      "loss": 0.0669,
+      "step": 6695
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.891171558229794e-06,
+      "loss": 0.0756,
+      "step": 6696
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8884226562390995e-06,
+      "loss": 0.0747,
+      "step": 6697
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8856744912083276e-06,
+      "loss": 0.0704,
+      "step": 6698
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882927063468871e-06,
+      "loss": 0.0732,
+      "step": 6699
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.880180373352018e-06,
+      "loss": 0.0643,
+      "step": 6700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.877434421188985e-06,
+      "loss": 0.0749,
+      "step": 6701
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8746892073108855e-06,
+      "loss": 0.073,
+      "step": 6702
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.871944732048746e-06,
+      "loss": 0.0681,
+      "step": 6703
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.869200995733513e-06,
+      "loss": 0.0681,
+      "step": 6704
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.866457998696036e-06,
+      "loss": 0.0754,
+      "step": 6705
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.863715741267073e-06,
+      "loss": 0.0656,
+      "step": 6706
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.860974223777304e-06,
+      "loss": 0.0704,
+      "step": 6707
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85823344655731e-06,
+      "loss": 0.0758,
+      "step": 6708
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.855493409937581e-06,
+      "loss": 0.0707,
+      "step": 6709
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85275411424853e-06,
+      "loss": 0.0673,
+      "step": 6710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.850015559820465e-06,
+      "loss": 0.0694,
+      "step": 6711
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8472777469836195e-06,
+      "loss": 0.0715,
+      "step": 6712
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.844540676068129e-06,
+      "loss": 0.0703,
+      "step": 6713
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8418043474040356e-06,
+      "loss": 0.0762,
+      "step": 6714
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839068761321305e-06,
+      "loss": 0.0683,
+      "step": 6715
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8363339181498025e-06,
+      "loss": 0.0676,
+      "step": 6716
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8335998182193045e-06,
+      "loss": 0.0699,
+      "step": 6717
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.830866461859505e-06,
+      "loss": 0.0704,
+      "step": 6718
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828133849400002e-06,
+      "loss": 0.0728,
+      "step": 6719
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.825401981170303e-06,
+      "loss": 0.0741,
+      "step": 6720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822670857499832e-06,
+      "loss": 0.0752,
+      "step": 6721
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.819940478717914e-06,
+      "loss": 0.0675,
+      "step": 6722
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817210845153798e-06,
+      "loss": 0.0642,
+      "step": 6723
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.81448195713663e-06,
+      "loss": 0.0686,
+      "step": 6724
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.811753814995468e-06,
+      "loss": 0.0664,
+      "step": 6725
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8090264190592886e-06,
+      "loss": 0.0686,
+      "step": 6726
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8062997696569717e-06,
+      "loss": 0.0671,
+      "step": 6727
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.803573867117302e-06,
+      "loss": 0.0723,
+      "step": 6728
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8008487117689897e-06,
+      "loss": 0.0707,
+      "step": 6729
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.79812430394064e-06,
+      "loss": 0.0668,
+      "step": 6730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.795400643960769e-06,
+      "loss": 0.0631,
+      "step": 6731
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.792677732157819e-06,
+      "loss": 0.0733,
+      "step": 6732
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.789955568860124e-06,
+      "loss": 0.0671,
+      "step": 6733
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.787234154395931e-06,
+      "loss": 0.0681,
+      "step": 6734
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7845134890934066e-06,
+      "loss": 0.0594,
+      "step": 6735
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7817935732806167e-06,
+      "loss": 0.0671,
+      "step": 6736
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7790744072855356e-06,
+      "loss": 0.0678,
+      "step": 6737
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7763559914360603e-06,
+      "loss": 0.0702,
+      "step": 6738
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.773638326059984e-06,
+      "loss": 0.0646,
+      "step": 6739
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7709214114850124e-06,
+      "loss": 0.0701,
+      "step": 6740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.768205248038769e-06,
+      "loss": 0.0764,
+      "step": 6741
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7654898360487723e-06,
+      "loss": 0.0731,
+      "step": 6742
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.762775175842467e-06,
+      "loss": 0.0711,
+      "step": 6743
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7600612677471936e-06,
+      "loss": 0.0698,
+      "step": 6744
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7573481120902035e-06,
+      "loss": 0.0663,
+      "step": 6745
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.754635709198667e-06,
+      "loss": 0.078,
+      "step": 6746
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7519240593996553e-06,
+      "loss": 0.0731,
+      "step": 6747
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7492131630201446e-06,
+      "loss": 0.0663,
+      "step": 6748
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7465030203870356e-06,
+      "loss": 0.0713,
+      "step": 6749
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7437936318271238e-06,
+      "loss": 0.0658,
+      "step": 6750
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.741084997667116e-06,
+      "loss": 0.0726,
+      "step": 6751
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7383771182336382e-06,
+      "loss": 0.0675,
+      "step": 6752
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7356699938532103e-06,
+      "loss": 0.0674,
+      "step": 6753
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.732963624852275e-06,
+      "loss": 0.0725,
+      "step": 6754
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.730258011557176e-06,
+      "loss": 0.0756,
+      "step": 6755
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7275531542941633e-06,
+      "loss": 0.0656,
+      "step": 6756
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7248490533894076e-06,
+      "loss": 0.0739,
+      "step": 6757
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7221457091689763e-06,
+      "loss": 0.0663,
+      "step": 6758
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7194431219588468e-06,
+      "loss": 0.0666,
+      "step": 6759
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7167412920849165e-06,
+      "loss": 0.0712,
+      "step": 6760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7140402198729795e-06,
+      "loss": 0.0707,
+      "step": 6761
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.711339905648739e-06,
+      "loss": 0.0681,
+      "step": 6762
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7086403497378166e-06,
+      "loss": 0.0707,
+      "step": 6763
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7059415524657304e-06,
+      "loss": 0.0666,
+      "step": 6764
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.703243514157919e-06,
+      "loss": 0.0666,
+      "step": 6765
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7005462351397216e-06,
+      "loss": 0.0731,
+      "step": 6766
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6978497157363812e-06,
+      "loss": 0.0722,
+      "step": 6767
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6951539562730644e-06,
+      "loss": 0.0737,
+      "step": 6768
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6924589570748327e-06,
+      "loss": 0.0722,
+      "step": 6769
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6897647184666583e-06,
+      "loss": 0.0694,
+      "step": 6770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.687071240773431e-06,
+      "loss": 0.0751,
+      "step": 6771
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.684378524319937e-06,
+      "loss": 0.0717,
+      "step": 6772
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6816865694308722e-06,
+      "loss": 0.0666,
+      "step": 6773
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.678995376430852e-06,
+      "loss": 0.0759,
+      "step": 6774
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.676304945644383e-06,
+      "loss": 0.0633,
+      "step": 6775
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.673615277395898e-06,
+      "loss": 0.0662,
+      "step": 6776
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.670926372009724e-06,
+      "loss": 0.0624,
+      "step": 6777
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.668238229810097e-06,
+      "loss": 0.067,
+      "step": 6778
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6655508511211725e-06,
+      "loss": 0.0657,
+      "step": 6779
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.662864236267001e-06,
+      "loss": 0.0696,
+      "step": 6780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6601783855715435e-06,
+      "loss": 0.073,
+      "step": 6781
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6574932993586774e-06,
+      "loss": 0.0705,
+      "step": 6782
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.654808977952179e-06,
+      "loss": 0.0717,
+      "step": 6783
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.652125421675732e-06,
+      "loss": 0.0713,
+      "step": 6784
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.649442630852937e-06,
+      "loss": 0.0756,
+      "step": 6785
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6467606058072937e-06,
+      "loss": 0.0655,
+      "step": 6786
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.644079346862208e-06,
+      "loss": 0.0683,
+      "step": 6787
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641398854341005e-06,
+      "loss": 0.0752,
+      "step": 6788
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.638719128566902e-06,
+      "loss": 0.0719,
+      "step": 6789
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6360401698630397e-06,
+      "loss": 0.0664,
+      "step": 6790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.633361978552454e-06,
+      "loss": 0.0643,
+      "step": 6791
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6306845549580903e-06,
+      "loss": 0.0629,
+      "step": 6792
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.628007899402809e-06,
+      "loss": 0.0558,
+      "step": 6793
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6253320122093704e-06,
+      "loss": 0.0663,
+      "step": 6794
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.62265689370044e-06,
+      "loss": 0.0707,
+      "step": 6795
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6199825441986024e-06,
+      "loss": 0.0693,
+      "step": 6796
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.617308964026338e-06,
+      "loss": 0.0711,
+      "step": 6797
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.614636153506036e-06,
+      "loss": 0.0692,
+      "step": 6798
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6119641129600015e-06,
+      "loss": 0.0669,
+      "step": 6799
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6092928427104323e-06,
+      "loss": 0.0697,
+      "step": 6800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.606622343079451e-06,
+      "loss": 0.0733,
+      "step": 6801
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6039526143890723e-06,
+      "loss": 0.0712,
+      "step": 6802
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.601283656961222e-06,
+      "loss": 0.0659,
+      "step": 6803
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.598615471117738e-06,
+      "loss": 0.067,
+      "step": 6804
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5959480571803607e-06,
+      "loss": 0.0671,
+      "step": 6805
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.593281415470734e-06,
+      "loss": 0.0566,
+      "step": 6806
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5906155463104196e-06,
+      "loss": 0.078,
+      "step": 6807
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5879504500208762e-06,
+      "loss": 0.0705,
+      "step": 6808
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5852861269234697e-06,
+      "loss": 0.0702,
+      "step": 6809
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5826225773394796e-06,
+      "loss": 0.0638,
+      "step": 6810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.579959801590084e-06,
+      "loss": 0.0628,
+      "step": 6811
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5772977999963778e-06,
+      "loss": 0.069,
+      "step": 6812
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.574636572879352e-06,
+      "loss": 0.0767,
+      "step": 6813
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5719761205599058e-06,
+      "loss": 0.0723,
+      "step": 6814
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5693164433588556e-06,
+      "loss": 0.076,
+      "step": 6815
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.566657541596912e-06,
+      "loss": 0.0686,
+      "step": 6816
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.563999415594693e-06,
+      "loss": 0.0688,
+      "step": 6817
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.561342065672735e-06,
+      "loss": 0.0682,
+      "step": 6818
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.558685492151468e-06,
+      "loss": 0.0651,
+      "step": 6819
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5560296953512296e-06,
+      "loss": 0.0722,
+      "step": 6820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.553374675592275e-06,
+      "loss": 0.0747,
+      "step": 6821
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5507204331947486e-06,
+      "loss": 0.0646,
+      "step": 6822
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5480669684787193e-06,
+      "loss": 0.0643,
+      "step": 6823
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.545414281764148e-06,
+      "loss": 0.0729,
+      "step": 6824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5427623733709047e-06,
+      "loss": 0.072,
+      "step": 6825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5401112436187745e-06,
+      "loss": 0.0714,
+      "step": 6826
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.537460892827439e-06,
+      "loss": 0.0762,
+      "step": 6827
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5348113213164823e-06,
+      "loss": 0.0675,
+      "step": 6828
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5321625294054117e-06,
+      "loss": 0.0699,
+      "step": 6829
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5295145174136247e-06,
+      "loss": 0.0659,
+      "step": 6830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.526867285660426e-06,
+      "loss": 0.0632,
+      "step": 6831
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5242208344650364e-06,
+      "loss": 0.0675,
+      "step": 6832
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.521575164146571e-06,
+      "loss": 0.0719,
+      "step": 6833
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5189302750240618e-06,
+      "loss": 0.0687,
+      "step": 6834
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5162861674164383e-06,
+      "loss": 0.0748,
+      "step": 6835
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.513642841642533e-06,
+      "loss": 0.0776,
+      "step": 6836
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.511000298021098e-06,
+      "loss": 0.0726,
+      "step": 6837
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5083585368707775e-06,
+      "loss": 0.0691,
+      "step": 6838
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.505717558510123e-06,
+      "loss": 0.0723,
+      "step": 6839
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5030773632576044e-06,
+      "loss": 0.0731,
+      "step": 6840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.50043795143158e-06,
+      "loss": 0.0635,
+      "step": 6841
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4977993233503214e-06,
+      "loss": 0.0684,
+      "step": 6842
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.49516147933201e-06,
+      "loss": 0.0715,
+      "step": 6843
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.492524419694726e-06,
+      "loss": 0.0707,
+      "step": 6844
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.489888144756454e-06,
+      "loss": 0.0682,
+      "step": 6845
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4872526548350928e-06,
+      "loss": 0.0689,
+      "step": 6846
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4846179502484358e-06,
+      "loss": 0.0706,
+      "step": 6847
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.481984031314193e-06,
+      "loss": 0.0725,
+      "step": 6848
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.479350898349969e-06,
+      "loss": 0.0665,
+      "step": 6849
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.476718551673276e-06,
+      "loss": 0.0662,
+      "step": 6850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4740869916015398e-06,
+      "loss": 0.0645,
+      "step": 6851
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.471456218452083e-06,
+      "loss": 0.066,
+      "step": 6852
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4688262325421295e-06,
+      "loss": 0.0636,
+      "step": 6853
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.466197034188823e-06,
+      "loss": 0.0666,
+      "step": 6854
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.463568623709199e-06,
+      "loss": 0.0673,
+      "step": 6855
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4609410014202005e-06,
+      "loss": 0.067,
+      "step": 6856
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.458314167638682e-06,
+      "loss": 0.0676,
+      "step": 6857
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4556881226813932e-06,
+      "loss": 0.0671,
+      "step": 6858
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.453062866864999e-06,
+      "loss": 0.0639,
+      "step": 6859
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.450438400506061e-06,
+      "loss": 0.0702,
+      "step": 6860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.447814723921046e-06,
+      "loss": 0.0751,
+      "step": 6861
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4451918374263337e-06,
+      "loss": 0.0705,
+      "step": 6862
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.442569741338201e-06,
+      "loss": 0.0731,
+      "step": 6863
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.439948435972826e-06,
+      "loss": 0.0634,
+      "step": 6864
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4373279216463053e-06,
+      "loss": 0.0691,
+      "step": 6865
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.434708198674627e-06,
+      "loss": 0.0686,
+      "step": 6866
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.432089267373685e-06,
+      "loss": 0.0693,
+      "step": 6867
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4294711280592896e-06,
+      "loss": 0.0687,
+      "step": 6868
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.426853781047138e-06,
+      "loss": 0.0676,
+      "step": 6869
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.42423722665285e-06,
+      "loss": 0.0635,
+      "step": 6870
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4216214651919366e-06,
+      "loss": 0.0681,
+      "step": 6871
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.419006496979813e-06,
+      "loss": 0.0689,
+      "step": 6872
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.416392322331812e-06,
+      "loss": 0.069,
+      "step": 6873
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.413778941563156e-06,
+      "loss": 0.0605,
+      "step": 6874
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4111663549889764e-06,
+      "loss": 0.072,
+      "step": 6875
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4085545629243154e-06,
+      "loss": 0.072,
+      "step": 6876
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.405943565684111e-06,
+      "loss": 0.0668,
+      "step": 6877
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4033333635832056e-06,
+      "loss": 0.069,
+      "step": 6878
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.400723956936354e-06,
+      "loss": 0.0647,
+      "step": 6879
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3981153460582027e-06,
+      "loss": 0.0692,
+      "step": 6880
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.395507531263318e-06,
+      "loss": 0.0642,
+      "step": 6881
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.392900512866155e-06,
+      "loss": 0.0694,
+      "step": 6882
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3902942911810775e-06,
+      "loss": 0.0641,
+      "step": 6883
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.387688866522361e-06,
+      "loss": 0.066,
+      "step": 6884
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.385084239204175e-06,
+      "loss": 0.0632,
+      "step": 6885
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.382480409540594e-06,
+      "loss": 0.0734,
+      "step": 6886
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.379877377845604e-06,
+      "loss": 0.0714,
+      "step": 6887
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3772751444330874e-06,
+      "loss": 0.0748,
+      "step": 6888
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3746737096168304e-06,
+      "loss": 0.0659,
+      "step": 6889
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.372073073710529e-06,
+      "loss": 0.0704,
+      "step": 6890
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3694732370277737e-06,
+      "loss": 0.0739,
+      "step": 6891
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.366874199882071e-06,
+      "loss": 0.0649,
+      "step": 6892
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3642759625868193e-06,
+      "loss": 0.0782,
+      "step": 6893
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.361678525455323e-06,
+      "loss": 0.0687,
+      "step": 6894
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.359081888800798e-06,
+      "loss": 0.0691,
+      "step": 6895
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.356486052936354e-06,
+      "loss": 0.0755,
+      "step": 6896
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3538910181750063e-06,
+      "loss": 0.0688,
+      "step": 6897
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.351296784829676e-06,
+      "loss": 0.0705,
+      "step": 6898
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.348703353213192e-06,
+      "loss": 0.0616,
+      "step": 6899
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.346110723638274e-06,
+      "loss": 0.068,
+      "step": 6900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.343518896417559e-06,
+      "loss": 0.0717,
+      "step": 6901
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3409278718635772e-06,
+      "loss": 0.0647,
+      "step": 6902
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3383376502887623e-06,
+      "loss": 0.0681,
+      "step": 6903
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.33574823200546e-06,
+      "loss": 0.0644,
+      "step": 6904
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3331596173259117e-06,
+      "loss": 0.0625,
+      "step": 6905
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3305718065622583e-06,
+      "loss": 0.0678,
+      "step": 6906
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.327984800026556e-06,
+      "loss": 0.0747,
+      "step": 6907
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3253985980307547e-06,
+      "loss": 0.0691,
+      "step": 6908
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3228132008867052e-06,
+      "loss": 0.0639,
+      "step": 6909
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3202286089061732e-06,
+      "loss": 0.074,
+      "step": 6910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.317644822400813e-06,
+      "loss": 0.0718,
+      "step": 6911
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3150618416821946e-06,
+      "loss": 0.0623,
+      "step": 6912
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312479667061782e-06,
+      "loss": 0.0678,
+      "step": 6913
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.309898298850941e-06,
+      "loss": 0.0665,
+      "step": 6914
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.307317737360951e-06,
+      "loss": 0.0739,
+      "step": 6915
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3047379829029835e-06,
+      "loss": 0.0626,
+      "step": 6916
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.302159035788113e-06,
+      "loss": 0.0683,
+      "step": 6917
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2995808963273256e-06,
+      "loss": 0.0725,
+      "step": 6918
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2970035648315023e-06,
+      "loss": 0.0641,
+      "step": 6919
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2944270416114256e-06,
+      "loss": 0.0729,
+      "step": 6920
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2918513269777894e-06,
+      "loss": 0.0638,
+      "step": 6921
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2892764212411777e-06,
+      "loss": 0.0738,
+      "step": 6922
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2867023247120898e-06,
+      "loss": 0.0704,
+      "step": 6923
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2841290377009196e-06,
+      "loss": 0.0646,
+      "step": 6924
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.281556560517959e-06,
+      "loss": 0.0695,
+      "step": 6925
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.278984893473417e-06,
+      "loss": 0.0708,
+      "step": 6926
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.276414036877392e-06,
+      "loss": 0.0663,
+      "step": 6927
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2738439910398854e-06,
+      "loss": 0.0688,
+      "step": 6928
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2712747562708115e-06,
+      "loss": 0.0651,
+      "step": 6929
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.268706332879975e-06,
+      "loss": 0.078,
+      "step": 6930
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2661387211770846e-06,
+      "loss": 0.0693,
+      "step": 6931
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.263571921471761e-06,
+      "loss": 0.0687,
+      "step": 6932
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2610059340735156e-06,
+      "loss": 0.0713,
+      "step": 6933
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2584407592917645e-06,
+      "loss": 0.0679,
+      "step": 6934
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.255876397435831e-06,
+      "loss": 0.0713,
+      "step": 6935
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2533128488149324e-06,
+      "loss": 0.0674,
+      "step": 6936
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.250750113738198e-06,
+      "loss": 0.0669,
+      "step": 6937
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.248188192514651e-06,
+      "loss": 0.0668,
+      "step": 6938
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2456270854532137e-06,
+      "loss": 0.072,
+      "step": 6939
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2430667928627235e-06,
+      "loss": 0.0723,
+      "step": 6940
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.240507315051907e-06,
+      "loss": 0.0628,
+      "step": 6941
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2379486523293936e-06,
+      "loss": 0.0674,
+      "step": 6942
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2353908050037252e-06,
+      "loss": 0.0648,
+      "step": 6943
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.232833773383334e-06,
+      "loss": 0.0731,
+      "step": 6944
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2302775577765535e-06,
+      "loss": 0.0669,
+      "step": 6945
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.227722158491632e-06,
+      "loss": 0.0669,
+      "step": 6946
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2251675758367007e-06,
+      "loss": 0.067,
+      "step": 6947
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2226138101198113e-06,
+      "loss": 0.0647,
+      "step": 6948
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2200608616489036e-06,
+      "loss": 0.0658,
+      "step": 6949
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2175087307318185e-06,
+      "loss": 0.0608,
+      "step": 6950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.214957417676311e-06,
+      "loss": 0.0678,
+      "step": 6951
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2124069227900258e-06,
+      "loss": 0.0707,
+      "step": 6952
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2098572463805076e-06,
+      "loss": 0.07,
+      "step": 6953
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2073083887552147e-06,
+      "loss": 0.0719,
+      "step": 6954
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2047603502214963e-06,
+      "loss": 0.07,
+      "step": 6955
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2022131310866022e-06,
+      "loss": 0.0705,
+      "step": 6956
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1996667316576925e-06,
+      "loss": 0.0756,
+      "step": 6957
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1971211522418177e-06,
+      "loss": 0.0709,
+      "step": 6958
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1945763931459405e-06,
+      "loss": 0.0712,
+      "step": 6959
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1920324546769156e-06,
+      "loss": 0.0623,
+      "step": 6960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.189489337141497e-06,
+      "loss": 0.07,
+      "step": 6961
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1869470408463544e-06,
+      "loss": 0.0723,
+      "step": 6962
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1844055660980432e-06,
+      "loss": 0.0704,
+      "step": 6963
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.181864913203022e-06,
+      "loss": 0.0702,
+      "step": 6964
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1793250824676604e-06,
+      "loss": 0.0677,
+      "step": 6965
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.176786074198219e-06,
+      "loss": 0.078,
+      "step": 6966
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.174247888700859e-06,
+      "loss": 0.0721,
+      "step": 6967
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1717105262816507e-06,
+      "loss": 0.0694,
+      "step": 6968
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1691739872465555e-06,
+      "loss": 0.074,
+      "step": 6969
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1666382719014454e-06,
+      "loss": 0.0671,
+      "step": 6970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1641033805520837e-06,
+      "loss": 0.0663,
+      "step": 6971
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.161569313504137e-06,
+      "loss": 0.0709,
+      "step": 6972
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1590360710631796e-06,
+      "loss": 0.0749,
+      "step": 6973
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.156503653534677e-06,
+      "loss": 0.0716,
+      "step": 6974
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1539720612239953e-06,
+      "loss": 0.0711,
+      "step": 6975
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1514412944364114e-06,
+      "loss": 0.0695,
+      "step": 6976
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1489113534770943e-06,
+      "loss": 0.0723,
+      "step": 6977
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.146382238651109e-06,
+      "loss": 0.0688,
+      "step": 6978
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1438539502634346e-06,
+      "loss": 0.0655,
+      "step": 6979
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.141326488618938e-06,
+      "loss": 0.0687,
+      "step": 6980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.138799854022395e-06,
+      "loss": 0.0689,
+      "step": 6981
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.136274046778476e-06,
+      "loss": 0.0721,
+      "step": 6982
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1337490671917504e-06,
+      "loss": 0.0677,
+      "step": 6983
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1312249155666973e-06,
+      "loss": 0.0612,
+      "step": 6984
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1287015922076858e-06,
+      "loss": 0.0615,
+      "step": 6985
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1261790974189867e-06,
+      "loss": 0.068,
+      "step": 6986
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1236574315047797e-06,
+      "loss": 0.0698,
+      "step": 6987
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.121136594769133e-06,
+      "loss": 0.0711,
+      "step": 6988
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1186165875160178e-06,
+      "loss": 0.0705,
+      "step": 6989
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1160974100493145e-06,
+      "loss": 0.0669,
+      "step": 6990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1135790626727913e-06,
+      "loss": 0.0702,
+      "step": 6991
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1110615456901196e-06,
+      "loss": 0.0653,
+      "step": 6992
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1085448594048773e-06,
+      "loss": 0.0742,
+      "step": 6993
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.106029004120531e-06,
+      "loss": 0.0673,
+      "step": 6994
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.10351398014046e-06,
+      "loss": 0.0644,
+      "step": 6995
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1009997877679343e-06,
+      "loss": 0.0655,
+      "step": 6996
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.098486427306121e-06,
+      "loss": 0.0739,
+      "step": 6997
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.095973899058099e-06,
+      "loss": 0.0767,
+      "step": 6998
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0934622033268356e-06,
+      "loss": 0.0693,
+      "step": 6999
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0909513404151993e-06,
+      "loss": 0.068,
+      "step": 7000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.088441310625968e-06,
+      "loss": 0.0654,
+      "step": 7001
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.085932114261806e-06,
+      "loss": 0.0681,
+      "step": 7002
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0834237516252817e-06,
+      "loss": 0.0689,
+      "step": 7003
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0809162230188695e-06,
+      "loss": 0.067,
+      "step": 7004
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0784095287449323e-06,
+      "loss": 0.0694,
+      "step": 7005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.075903669105744e-06,
+      "loss": 0.0766,
+      "step": 7006
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.073398644403468e-06,
+      "loss": 0.0709,
+      "step": 7007
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0708944549401677e-06,
+      "loss": 0.0632,
+      "step": 7008
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0683911010178156e-06,
+      "loss": 0.0669,
+      "step": 7009
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0658885829382745e-06,
+      "loss": 0.0697,
+      "step": 7010
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0633869010033047e-06,
+      "loss": 0.0762,
+      "step": 7011
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0608860555145748e-06,
+      "loss": 0.0674,
+      "step": 7012
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.058386046773645e-06,
+      "loss": 0.0722,
+      "step": 7013
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0558868750819748e-06,
+      "loss": 0.0672,
+      "step": 7014
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0533885407409304e-06,
+      "loss": 0.0733,
+      "step": 7015
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0508910440517647e-06,
+      "loss": 0.0695,
+      "step": 7016
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.048394385315644e-06,
+      "loss": 0.0646,
+      "step": 7017
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0458985648336212e-06,
+      "loss": 0.0666,
+      "step": 7018
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.043403582906651e-06,
+      "loss": 0.067,
+      "step": 7019
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0409094398355955e-06,
+      "loss": 0.0677,
+      "step": 7020
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.038416135921205e-06,
+      "loss": 0.0717,
+      "step": 7021
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0359236714641305e-06,
+      "loss": 0.0679,
+      "step": 7022
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0334320467649292e-06,
+      "loss": 0.0745,
+      "step": 7023
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0309412621240484e-06,
+      "loss": 0.0709,
+      "step": 7024
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.028451317841835e-06,
+      "loss": 0.0647,
+      "step": 7025
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.025962214218544e-06,
+      "loss": 0.0703,
+      "step": 7026
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0234739515543153e-06,
+      "loss": 0.066,
+      "step": 7027
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0209865301492e-06,
+      "loss": 0.0743,
+      "step": 7028
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0184999503031387e-06,
+      "loss": 0.0644,
+      "step": 7029
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0160142123159707e-06,
+      "loss": 0.0647,
+      "step": 7030
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0135293164874447e-06,
+      "loss": 0.0706,
+      "step": 7031
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.011045263117194e-06,
+      "loss": 0.064,
+      "step": 7032
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0085620525047553e-06,
+      "loss": 0.0669,
+      "step": 7033
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.00607968494957e-06,
+      "loss": 0.0662,
+      "step": 7034
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.00359816075097e-06,
+      "loss": 0.0646,
+      "step": 7035
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0011174802081844e-06,
+      "loss": 0.0652,
+      "step": 7036
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.998637643620351e-06,
+      "loss": 0.0712,
+      "step": 7037
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9961586512864947e-06,
+      "loss": 0.0694,
+      "step": 7038
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.993680503505542e-06,
+      "loss": 0.0749,
+      "step": 7039
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9912032005763225e-06,
+      "loss": 0.0681,
+      "step": 7040
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.988726742797554e-06,
+      "loss": 0.0671,
+      "step": 7041
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9862511304678655e-06,
+      "loss": 0.0721,
+      "step": 7042
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.983776363885773e-06,
+      "loss": 0.0698,
+      "step": 7043
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.981302443349692e-06,
+      "loss": 0.07,
+      "step": 7044
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9788293691579427e-06,
+      "loss": 0.0734,
+      "step": 7045
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.976357141608738e-06,
+      "loss": 0.064,
+      "step": 7046
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9738857610001846e-06,
+      "loss": 0.0691,
+      "step": 7047
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.971415227630301e-06,
+      "loss": 0.0705,
+      "step": 7048
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.968945541796988e-06,
+      "loss": 0.0627,
+      "step": 7049
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.96647670379805e-06,
+      "loss": 0.0659,
+      "step": 7050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9640087139311968e-06,
+      "loss": 0.0716,
+      "step": 7051
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.96154157249402e-06,
+      "loss": 0.0767,
+      "step": 7052
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.959075279784026e-06,
+      "loss": 0.0699,
+      "step": 7053
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9566098360986085e-06,
+      "loss": 0.0667,
+      "step": 7054
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.954145241735057e-06,
+      "loss": 0.0674,
+      "step": 7055
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9516814969905684e-06,
+      "loss": 0.0664,
+      "step": 7056
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9492186021622293e-06,
+      "loss": 0.0602,
+      "step": 7057
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9467565575470225e-06,
+      "loss": 0.0693,
+      "step": 7058
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.944295363441839e-06,
+      "loss": 0.0589,
+      "step": 7059
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9418350201434555e-06,
+      "loss": 0.071,
+      "step": 7060
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.939375527948548e-06,
+      "loss": 0.0637,
+      "step": 7061
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.936916887153699e-06,
+      "loss": 0.0632,
+      "step": 7062
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9344590980553757e-06,
+      "loss": 0.0698,
+      "step": 7063
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9320021609499536e-06,
+      "loss": 0.0674,
+      "step": 7064
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9295460761337003e-06,
+      "loss": 0.0702,
+      "step": 7065
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9270908439027747e-06,
+      "loss": 0.0683,
+      "step": 7066
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9246364645532443e-06,
+      "loss": 0.0699,
+      "step": 7067
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.922182938381072e-06,
+      "loss": 0.0666,
+      "step": 7068
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9197302656821056e-06,
+      "loss": 0.0654,
+      "step": 7069
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9172784467521076e-06,
+      "loss": 0.0656,
+      "step": 7070
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9148274818867237e-06,
+      "loss": 0.0717,
+      "step": 7071
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9123773713815007e-06,
+      "loss": 0.0555,
+      "step": 7072
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.909928115531887e-06,
+      "loss": 0.0725,
+      "step": 7073
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9074797146332236e-06,
+      "loss": 0.0684,
+      "step": 7074
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9050321689807447e-06,
+      "loss": 0.0665,
+      "step": 7075
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9025854788695916e-06,
+      "loss": 0.073,
+      "step": 7076
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.900139644594794e-06,
+      "loss": 0.0654,
+      "step": 7077
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8976946664512773e-06,
+      "loss": 0.0745,
+      "step": 7078
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.895250544733875e-06,
+      "loss": 0.0729,
+      "step": 7079
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8928072797373052e-06,
+      "loss": 0.0719,
+      "step": 7080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.890364871756185e-06,
+      "loss": 0.0699,
+      "step": 7081
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8879233210850355e-06,
+      "loss": 0.0717,
+      "step": 7082
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8854826280182635e-06,
+      "loss": 0.0663,
+      "step": 7083
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8830427928501847e-06,
+      "loss": 0.0718,
+      "step": 7084
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8806038158750007e-06,
+      "loss": 0.0692,
+      "step": 7085
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.878165697386812e-06,
+      "loss": 0.0683,
+      "step": 7086
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8757284376796224e-06,
+      "loss": 0.0664,
+      "step": 7087
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.873292037047324e-06,
+      "loss": 0.068,
+      "step": 7088
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.870856495783706e-06,
+      "loss": 0.0682,
+      "step": 7089
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8684218141824616e-06,
+      "loss": 0.0711,
+      "step": 7090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8659879925371716e-06,
+      "loss": 0.0688,
+      "step": 7091
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.863555031141315e-06,
+      "loss": 0.0738,
+      "step": 7092
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8611229302882737e-06,
+      "loss": 0.0681,
+      "step": 7093
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.858691690271315e-06,
+      "loss": 0.0688,
+      "step": 7094
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8562613113836136e-06,
+      "loss": 0.0667,
+      "step": 7095
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8538317939182325e-06,
+      "loss": 0.0687,
+      "step": 7096
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8514031381681293e-06,
+      "loss": 0.0732,
+      "step": 7097
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8489753444261703e-06,
+      "loss": 0.0592,
+      "step": 7098
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8465484129851028e-06,
+      "loss": 0.0663,
+      "step": 7099
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8441223441375753e-06,
+      "loss": 0.0722,
+      "step": 7100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.841697138176138e-06,
+      "loss": 0.0683,
+      "step": 7101
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.839272795393232e-06,
+      "loss": 0.0723,
+      "step": 7102
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8368493160811892e-06,
+      "loss": 0.0671,
+      "step": 7103
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.834426700532251e-06,
+      "loss": 0.0643,
+      "step": 7104
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8320049490385402e-06,
+      "loss": 0.0662,
+      "step": 7105
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.829584061892088e-06,
+      "loss": 0.0694,
+      "step": 7106
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8271640393848108e-06,
+      "loss": 0.0666,
+      "step": 7107
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8247448818085233e-06,
+      "loss": 0.0706,
+      "step": 7108
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8223265894549434e-06,
+      "loss": 0.0756,
+      "step": 7109
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.819909162615677e-06,
+      "loss": 0.0734,
+      "step": 7110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8174926015822245e-06,
+      "loss": 0.0692,
+      "step": 7111
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8150769066459895e-06,
+      "loss": 0.0683,
+      "step": 7112
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.812662078098266e-06,
+      "loss": 0.0717,
+      "step": 7113
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8102481162302386e-06,
+      "loss": 0.0622,
+      "step": 7114
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.807835021333002e-06,
+      "loss": 0.0713,
+      "step": 7115
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8054227936975298e-06,
+      "loss": 0.067,
+      "step": 7116
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.803011433614704e-06,
+      "loss": 0.0727,
+      "step": 7117
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8006009413752956e-06,
+      "loss": 0.068,
+      "step": 7118
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.798191317269967e-06,
+      "loss": 0.0729,
+      "step": 7119
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7957825615892873e-06,
+      "loss": 0.0666,
+      "step": 7120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.793374674623713e-06,
+      "loss": 0.0673,
+      "step": 7121
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7909676566635936e-06,
+      "loss": 0.0687,
+      "step": 7122
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.788561507999182e-06,
+      "loss": 0.0684,
+      "step": 7123
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.786156228920621e-06,
+      "loss": 0.0671,
+      "step": 7124
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.783751819717945e-06,
+      "loss": 0.0685,
+      "step": 7125
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7813482806810943e-06,
+      "loss": 0.0711,
+      "step": 7126
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.778945612099895e-06,
+      "loss": 0.0703,
+      "step": 7127
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7765438142640675e-06,
+      "loss": 0.0627,
+      "step": 7128
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.774142887463237e-06,
+      "loss": 0.066,
+      "step": 7129
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.771742831986912e-06,
+      "loss": 0.0718,
+      "step": 7130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.769343648124506e-06,
+      "loss": 0.0653,
+      "step": 7131
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.766945336165322e-06,
+      "loss": 0.0701,
+      "step": 7132
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.764547896398554e-06,
+      "loss": 0.0649,
+      "step": 7133
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7621513291133016e-06,
+      "loss": 0.0691,
+      "step": 7134
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.75975563459855e-06,
+      "loss": 0.07,
+      "step": 7135
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7573608131431796e-06,
+      "loss": 0.0674,
+      "step": 7136
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.754966865035974e-06,
+      "loss": 0.0701,
+      "step": 7137
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7525737905656026e-06,
+      "loss": 0.0699,
+      "step": 7138
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7501815900206298e-06,
+      "loss": 0.072,
+      "step": 7139
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7477902636895226e-06,
+      "loss": 0.0677,
+      "step": 7140
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.745399811860632e-06,
+      "loss": 0.0677,
+      "step": 7141
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7430102348222144e-06,
+      "loss": 0.0651,
+      "step": 7142
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.740621532862412e-06,
+      "loss": 0.0732,
+      "step": 7143
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.738233706269261e-06,
+      "loss": 0.069,
+      "step": 7144
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7358467553307033e-06,
+      "loss": 0.07,
+      "step": 7145
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7334606803345643e-06,
+      "loss": 0.0722,
+      "step": 7146
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7310754815685627e-06,
+      "loss": 0.0619,
+      "step": 7147
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7286911593203238e-06,
+      "loss": 0.0651,
+      "step": 7148
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.726307713877354e-06,
+      "loss": 0.0703,
+      "step": 7149
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7239251455270577e-06,
+      "loss": 0.0663,
+      "step": 7150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7215434545567408e-06,
+      "loss": 0.0671,
+      "step": 7151
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.719162641253592e-06,
+      "loss": 0.0584,
+      "step": 7152
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.716782705904706e-06,
+      "loss": 0.0695,
+      "step": 7153
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7144036487970617e-06,
+      "loss": 0.0735,
+      "step": 7154
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.712025470217533e-06,
+      "loss": 0.0766,
+      "step": 7155
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.709648170452898e-06,
+      "loss": 0.0631,
+      "step": 7156
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.707271749789816e-06,
+      "loss": 0.0674,
+      "step": 7157
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7048962085148456e-06,
+      "loss": 0.0633,
+      "step": 7158
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.702521546914445e-06,
+      "loss": 0.065,
+      "step": 7159
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7001477652749576e-06,
+      "loss": 0.0645,
+      "step": 7160
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.697774863882621e-06,
+      "loss": 0.0676,
+      "step": 7161
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6954028430235745e-06,
+      "loss": 0.0724,
+      "step": 7162
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6930317029838426e-06,
+      "loss": 0.0622,
+      "step": 7163
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6906614440493526e-06,
+      "loss": 0.0725,
+      "step": 7164
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6882920665059175e-06,
+      "loss": 0.0672,
+      "step": 7165
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6859235706392428e-06,
+      "loss": 0.0704,
+      "step": 7166
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.68355595673494e-06,
+      "loss": 0.0689,
+      "step": 7167
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.681189225078501e-06,
+      "loss": 0.0658,
+      "step": 7168
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.678823375955314e-06,
+      "loss": 0.0731,
+      "step": 7169
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6764584096506686e-06,
+      "loss": 0.0695,
+      "step": 7170
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.674094326449741e-06,
+      "loss": 0.0758,
+      "step": 7171
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.671731126637598e-06,
+      "loss": 0.0732,
+      "step": 7172
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.669368810499211e-06,
+      "loss": 0.0696,
+      "step": 7173
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6670073783194317e-06,
+      "loss": 0.0638,
+      "step": 7174
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6646468303830174e-06,
+      "loss": 0.0675,
+      "step": 7175
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.662287166974611e-06,
+      "loss": 0.0676,
+      "step": 7176
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6599283883787463e-06,
+      "loss": 0.062,
+      "step": 7177
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.657570494879862e-06,
+      "loss": 0.0689,
+      "step": 7178
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6552134867622793e-06,
+      "loss": 0.0686,
+      "step": 7179
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.652857364310213e-06,
+      "loss": 0.0635,
+      "step": 7180
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6505021278077802e-06,
+      "loss": 0.0719,
+      "step": 7181
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.648147777538984e-06,
+      "loss": 0.0664,
+      "step": 7182
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6457943137877175e-06,
+      "loss": 0.0736,
+      "step": 7183
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6434417368377776e-06,
+      "loss": 0.0704,
+      "step": 7184
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.641090046972845e-06,
+      "loss": 0.0719,
+      "step": 7185
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6387392444764936e-06,
+      "loss": 0.0661,
+      "step": 7186
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6363893296321987e-06,
+      "loss": 0.0722,
+      "step": 7187
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6340403027233174e-06,
+      "loss": 0.0665,
+      "step": 7188
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.631692164033112e-06,
+      "loss": 0.0657,
+      "step": 7189
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.629344913844727e-06,
+      "loss": 0.0688,
+      "step": 7190
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6269985524412002e-06,
+      "loss": 0.0676,
+      "step": 7191
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.624653080105474e-06,
+      "loss": 0.0711,
+      "step": 7192
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6223084971203694e-06,
+      "loss": 0.0643,
+      "step": 7193
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6199648037686056e-06,
+      "loss": 0.0671,
+      "step": 7194
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6176220003328e-06,
+      "loss": 0.0707,
+      "step": 7195
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6152800870954564e-06,
+      "loss": 0.0724,
+      "step": 7196
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.612939064338967e-06,
+      "loss": 0.0711,
+      "step": 7197
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6105989323456305e-06,
+      "loss": 0.064,
+      "step": 7198
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6082596913976223e-06,
+      "loss": 0.0683,
+      "step": 7199
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.605921341777026e-06,
+      "loss": 0.0676,
+      "step": 7200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6035838837658046e-06,
+      "loss": 0.065,
+      "step": 7201
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6012473176458162e-06,
+      "loss": 0.068,
+      "step": 7202
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5989116436988206e-06,
+      "loss": 0.0781,
+      "step": 7203
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.59657686220646e-06,
+      "loss": 0.0726,
+      "step": 7204
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5942429734502683e-06,
+      "loss": 0.0788,
+      "step": 7205
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.591909977711684e-06,
+      "loss": 0.0718,
+      "step": 7206
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5895778752720247e-06,
+      "loss": 0.0616,
+      "step": 7207
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5872466664125017e-06,
+      "loss": 0.0761,
+      "step": 7208
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5849163514142306e-06,
+      "loss": 0.0676,
+      "step": 7209
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5825869305582018e-06,
+      "loss": 0.0652,
+      "step": 7210
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5802584041253143e-06,
+      "loss": 0.0712,
+      "step": 7211
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.577930772396349e-06,
+      "loss": 0.0667,
+      "step": 7212
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5756040356519785e-06,
+      "loss": 0.0686,
+      "step": 7213
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5732781941727747e-06,
+      "loss": 0.0747,
+      "step": 7214
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5709532482391973e-06,
+      "loss": 0.067,
+      "step": 7215
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5686291981315926e-06,
+      "loss": 0.073,
+      "step": 7216
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5663060441302125e-06,
+      "loss": 0.0723,
+      "step": 7217
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5639837865151885e-06,
+      "loss": 0.0653,
+      "step": 7218
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5616624255665455e-06,
+      "loss": 0.0653,
+      "step": 7219
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.559341961564209e-06,
+      "loss": 0.0651,
+      "step": 7220
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5570223947879845e-06,
+      "loss": 0.0642,
+      "step": 7221
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.554703725517582e-06,
+      "loss": 0.0713,
+      "step": 7222
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5523859540325925e-06,
+      "loss": 0.0717,
+      "step": 7223
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5500690806124993e-06,
+      "loss": 0.0692,
+      "step": 7224
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.547753105536689e-06,
+      "loss": 0.0699,
+      "step": 7225
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.545438029084425e-06,
+      "loss": 0.0755,
+      "step": 7226
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.54312385153487e-06,
+      "loss": 0.0682,
+      "step": 7227
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5408105731670806e-06,
+      "loss": 0.0698,
+      "step": 7228
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5384981942599997e-06,
+      "loss": 0.0675,
+      "step": 7229
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5361867150924614e-06,
+      "loss": 0.0614,
+      "step": 7230
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5338761359431986e-06,
+      "loss": 0.0681,
+      "step": 7231
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5315664570908284e-06,
+      "loss": 0.0651,
+      "step": 7232
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5292576788138556e-06,
+      "loss": 0.0759,
+      "step": 7233
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5269498013906935e-06,
+      "loss": 0.0653,
+      "step": 7234
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.524642825099631e-06,
+      "loss": 0.0641,
+      "step": 7235
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.52233675021885e-06,
+      "loss": 0.0701,
+      "step": 7236
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5200315770264318e-06,
+      "loss": 0.064,
+      "step": 7237
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5177273058003414e-06,
+      "loss": 0.0691,
+      "step": 7238
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5154239368184364e-06,
+      "loss": 0.0673,
+      "step": 7239
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5131214703584693e-06,
+      "loss": 0.0624,
+      "step": 7240
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5108199066980786e-06,
+      "loss": 0.0679,
+      "step": 7241
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5085192461148e-06,
+      "loss": 0.0614,
+      "step": 7242
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.506219488886056e-06,
+      "loss": 0.0658,
+      "step": 7243
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5039206352891566e-06,
+      "loss": 0.0645,
+      "step": 7244
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.501622685601315e-06,
+      "loss": 0.0713,
+      "step": 7245
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4993256400996223e-06,
+      "loss": 0.0631,
+      "step": 7246
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4970294990610642e-06,
+      "loss": 0.0684,
+      "step": 7247
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.494734262762527e-06,
+      "loss": 0.0688,
+      "step": 7248
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4924399314807735e-06,
+      "loss": 0.0648,
+      "step": 7249
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4901465054924634e-06,
+      "loss": 0.0665,
+      "step": 7250
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4878539850741533e-06,
+      "loss": 0.0759,
+      "step": 7251
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.485562370502279e-06,
+      "loss": 0.0617,
+      "step": 7252
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4832716620531783e-06,
+      "loss": 0.0717,
+      "step": 7253
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.480981860003072e-06,
+      "loss": 0.0672,
+      "step": 7254
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478692964628071e-06,
+      "loss": 0.0695,
+      "step": 7255
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4764049762041874e-06,
+      "loss": 0.0711,
+      "step": 7256
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.474117895007312e-06,
+      "loss": 0.0715,
+      "step": 7257
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.471831721313227e-06,
+      "loss": 0.0707,
+      "step": 7258
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.469546455397618e-06,
+      "loss": 0.0664,
+      "step": 7259
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.467262097536046e-06,
+      "loss": 0.0743,
+      "step": 7260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4649786480039672e-06,
+      "loss": 0.0644,
+      "step": 7261
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.462696107076734e-06,
+      "loss": 0.0687,
+      "step": 7262
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4604144750295812e-06,
+      "loss": 0.0658,
+      "step": 7263
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4581337521376427e-06,
+      "loss": 0.0731,
+      "step": 7264
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.455853938675934e-06,
+      "loss": 0.0759,
+      "step": 7265
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.453575034919362e-06,
+      "loss": 0.0692,
+      "step": 7266
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.451297041142733e-06,
+      "loss": 0.0629,
+      "step": 7267
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.449019957620734e-06,
+      "loss": 0.0662,
+      "step": 7268
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.446743784627943e-06,
+      "loss": 0.066,
+      "step": 7269
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4444685224388355e-06,
+      "loss": 0.0763,
+      "step": 7270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.44219417132777e-06,
+      "loss": 0.0665,
+      "step": 7271
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.439920731568993e-06,
+      "loss": 0.0711,
+      "step": 7272
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4376482034366533e-06,
+      "loss": 0.0696,
+      "step": 7273
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4353765872047786e-06,
+      "loss": 0.071,
+      "step": 7274
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4331058831472866e-06,
+      "loss": 0.0724,
+      "step": 7275
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.430836091537995e-06,
+      "loss": 0.0723,
+      "step": 7276
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4285672126505975e-06,
+      "loss": 0.0666,
+      "step": 7277
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4262992467586933e-06,
+      "loss": 0.0692,
+      "step": 7278
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424032194135758e-06,
+      "loss": 0.064,
+      "step": 7279
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4217660550551614e-06,
+      "loss": 0.0692,
+      "step": 7280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.419500829790169e-06,
+      "loss": 0.0678,
+      "step": 7281
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.417236518613929e-06,
+      "loss": 0.0674,
+      "step": 7282
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4149731217994775e-06,
+      "loss": 0.073,
+      "step": 7283
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.412710639619752e-06,
+      "loss": 0.0656,
+      "step": 7284
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.410449072347568e-06,
+      "loss": 0.0722,
+      "step": 7285
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.408188420255633e-06,
+      "loss": 0.0699,
+      "step": 7286
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4059286836165507e-06,
+      "loss": 0.0702,
+      "step": 7287
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4036698627028045e-06,
+      "loss": 0.072,
+      "step": 7288
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.401411957786779e-06,
+      "loss": 0.0668,
+      "step": 7289
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3991549691407378e-06,
+      "loss": 0.0741,
+      "step": 7290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3968988970368366e-06,
+      "loss": 0.0728,
+      "step": 7291
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3946437417471268e-06,
+      "loss": 0.0653,
+      "step": 7292
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3923895035435418e-06,
+      "loss": 0.0692,
+      "step": 7293
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.390136182697905e-06,
+      "loss": 0.0674,
+      "step": 7294
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3878837794819378e-06,
+      "loss": 0.0632,
+      "step": 7295
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3856322941672396e-06,
+      "loss": 0.0686,
+      "step": 7296
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3833817270253024e-06,
+      "loss": 0.0708,
+      "step": 7297
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.381132078327515e-06,
+      "loss": 0.0633,
+      "step": 7298
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3788833483451434e-06,
+      "loss": 0.0686,
+      "step": 7299
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3766355373493557e-06,
+      "loss": 0.0643,
+      "step": 7300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3743886456111977e-06,
+      "loss": 0.0599,
+      "step": 7301
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3721426734016086e-06,
+      "loss": 0.0662,
+      "step": 7302
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3698976209914205e-06,
+      "loss": 0.0691,
+      "step": 7303
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3676534886513515e-06,
+      "loss": 0.0704,
+      "step": 7304
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3654102766520026e-06,
+      "loss": 0.0715,
+      "step": 7305
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3631679852638778e-06,
+      "loss": 0.0744,
+      "step": 7306
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3609266147573575e-06,
+      "loss": 0.068,
+      "step": 7307
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3586861654027147e-06,
+      "loss": 0.0787,
+      "step": 7308
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3564466374701165e-06,
+      "loss": 0.0679,
+      "step": 7309
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.35420803122961e-06,
+      "loss": 0.0668,
+      "step": 7310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.35197034695114e-06,
+      "loss": 0.0685,
+      "step": 7311
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3497335849045355e-06,
+      "loss": 0.0607,
+      "step": 7312
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3474977453595093e-06,
+      "loss": 0.0739,
+      "step": 7313
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.345262828585676e-06,
+      "loss": 0.0685,
+      "step": 7314
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3430288348525285e-06,
+      "loss": 0.0657,
+      "step": 7315
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.340795764429448e-06,
+      "loss": 0.0664,
+      "step": 7316
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.338563617585712e-06,
+      "loss": 0.0727,
+      "step": 7317
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.33633239459048e-06,
+      "loss": 0.0691,
+      "step": 7318
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3341020957128014e-06,
+      "loss": 0.0707,
+      "step": 7319
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3318727212216187e-06,
+      "loss": 0.0651,
+      "step": 7320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.329644271385757e-06,
+      "loss": 0.0685,
+      "step": 7321
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3274167464739292e-06,
+      "loss": 0.0711,
+      "step": 7322
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.325190146754747e-06,
+      "loss": 0.0737,
+      "step": 7323
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.322964472496695e-06,
+      "loss": 0.068,
+      "step": 7324
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.320739723968162e-06,
+      "loss": 0.0681,
+      "step": 7325
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.318515901437415e-06,
+      "loss": 0.0709,
+      "step": 7326
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.316293005172607e-06,
+      "loss": 0.0687,
+      "step": 7327
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3140710354417917e-06,
+      "loss": 0.0706,
+      "step": 7328
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3118499925129e-06,
+      "loss": 0.0694,
+      "step": 7329
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.309629876653753e-06,
+      "loss": 0.0682,
+      "step": 7330
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3074106881320655e-06,
+      "loss": 0.0615,
+      "step": 7331
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3051924272154358e-06,
+      "loss": 0.064,
+      "step": 7332
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.302975094171347e-06,
+      "loss": 0.0724,
+      "step": 7333
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.300758689267181e-06,
+      "loss": 0.0656,
+      "step": 7334
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2985432127701945e-06,
+      "loss": 0.0722,
+      "step": 7335
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.296328664947547e-06,
+      "loss": 0.0643,
+      "step": 7336
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.294115046066272e-06,
+      "loss": 0.0659,
+      "step": 7337
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2919023563932974e-06,
+      "loss": 0.0666,
+      "step": 7338
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.289690596195442e-06,
+      "loss": 0.0693,
+      "step": 7339
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2874797657394076e-06,
+      "loss": 0.0681,
+      "step": 7340
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2852698652917827e-06,
+      "loss": 0.0659,
+      "step": 7341
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2830608951190525e-06,
+      "loss": 0.0663,
+      "step": 7342
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.28085285548758e-06,
+      "loss": 0.065,
+      "step": 7343
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.278645746663617e-06,
+      "loss": 0.0749,
+      "step": 7344
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2764395689133123e-06,
+      "loss": 0.0683,
+      "step": 7345
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.274234322502692e-06,
+      "loss": 0.0721,
+      "step": 7346
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2720300076976774e-06,
+      "loss": 0.0668,
+      "step": 7347
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2698266247640723e-06,
+      "loss": 0.0698,
+      "step": 7348
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2676241739675675e-06,
+      "loss": 0.0664,
+      "step": 7349
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.265422655573748e-06,
+      "loss": 0.0663,
+      "step": 7350
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2632220698480813e-06,
+      "loss": 0.0705,
+      "step": 7351
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.26102241705592e-06,
+      "loss": 0.0676,
+      "step": 7352
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2588236974625124e-06,
+      "loss": 0.0657,
+      "step": 7353
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.256625911332987e-06,
+      "loss": 0.0697,
+      "step": 7354
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.25442905893236e-06,
+      "loss": 0.0624,
+      "step": 7355
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2522331405255426e-06,
+      "loss": 0.0638,
+      "step": 7356
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2500381563773222e-06,
+      "loss": 0.0713,
+      "step": 7357
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.247844106752384e-06,
+      "loss": 0.0694,
+      "step": 7358
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2456509919152934e-06,
+      "loss": 0.0692,
+      "step": 7359
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2434588121305044e-06,
+      "loss": 0.0692,
+      "step": 7360
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.241267567662363e-06,
+      "loss": 0.0698,
+      "step": 7361
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2390772587750963e-06,
+      "loss": 0.0663,
+      "step": 7362
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.236887885732819e-06,
+      "loss": 0.0687,
+      "step": 7363
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.23469944879954e-06,
+      "loss": 0.0754,
+      "step": 7364
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2325119482391466e-06,
+      "loss": 0.0748,
+      "step": 7365
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.230325384315416e-06,
+      "loss": 0.0679,
+      "step": 7366
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2281397572920184e-06,
+      "loss": 0.0705,
+      "step": 7367
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2259550674325002e-06,
+      "loss": 0.0732,
+      "step": 7368
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.223771315000305e-06,
+      "loss": 0.0654,
+      "step": 7369
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.221588500258758e-06,
+      "loss": 0.0735,
+      "step": 7370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2194066234710688e-06,
+      "loss": 0.0645,
+      "step": 7371
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2172256849003415e-06,
+      "loss": 0.0625,
+      "step": 7372
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2150456848095624e-06,
+      "loss": 0.066,
+      "step": 7373
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2128666234616014e-06,
+      "loss": 0.068,
+      "step": 7374
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.210688501119225e-06,
+      "loss": 0.0696,
+      "step": 7375
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2085113180450756e-06,
+      "loss": 0.07,
+      "step": 7376
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2063350745016865e-06,
+      "loss": 0.069,
+      "step": 7377
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.204159770751484e-06,
+      "loss": 0.0704,
+      "step": 7378
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2019854070567715e-06,
+      "loss": 0.0671,
+      "step": 7379
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1998119836797414e-06,
+      "loss": 0.0634,
+      "step": 7380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1976395008824792e-06,
+      "loss": 0.0674,
+      "step": 7381
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1954679589269454e-06,
+      "loss": 0.0675,
+      "step": 7382
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1932973580750016e-06,
+      "loss": 0.0659,
+      "step": 7383
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1911276985883846e-06,
+      "loss": 0.0711,
+      "step": 7384
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.188958980728717e-06,
+      "loss": 0.0731,
+      "step": 7385
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.186791204757519e-06,
+      "loss": 0.0694,
+      "step": 7386
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.184624370936187e-06,
+      "loss": 0.0753,
+      "step": 7387
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.182458479526004e-06,
+      "loss": 0.0673,
+      "step": 7388
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.180293530788148e-06,
+      "loss": 0.0635,
+      "step": 7389
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1781295249836754e-06,
+      "loss": 0.0657,
+      "step": 7390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1759664623735277e-06,
+      "loss": 0.0661,
+      "step": 7391
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1738043432185417e-06,
+      "loss": 0.0664,
+      "step": 7392
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1716431677794305e-06,
+      "loss": 0.0666,
+      "step": 7393
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1694829363168023e-06,
+      "loss": 0.0807,
+      "step": 7394
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1673236490911442e-06,
+      "loss": 0.064,
+      "step": 7395
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1651653063628297e-06,
+      "loss": 0.0668,
+      "step": 7396
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1630079083921273e-06,
+      "loss": 0.072,
+      "step": 7397
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.160851455439181e-06,
+      "loss": 0.0686,
+      "step": 7398
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1586959477640233e-06,
+      "loss": 0.0665,
+      "step": 7399
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.156541385626579e-06,
+      "loss": 0.0679,
+      "step": 7400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.15438776928665e-06,
+      "loss": 0.0637,
+      "step": 7401
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1522350990039297e-06,
+      "loss": 0.0702,
+      "step": 7402
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.150083375038001e-06,
+      "loss": 0.0694,
+      "step": 7403
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1479325976483247e-06,
+      "loss": 0.0671,
+      "step": 7404
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1457827670942457e-06,
+      "loss": 0.0679,
+      "step": 7405
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1436338836350078e-06,
+      "loss": 0.0673,
+      "step": 7406
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1414859475297277e-06,
+      "loss": 0.0683,
+      "step": 7407
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1393389590374126e-06,
+      "loss": 0.0732,
+      "step": 7408
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1371929184169583e-06,
+      "loss": 0.0708,
+      "step": 7409
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.135047825927141e-06,
+      "loss": 0.0677,
+      "step": 7410
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1329036818266246e-06,
+      "loss": 0.07,
+      "step": 7411
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1307604863739617e-06,
+      "loss": 0.0682,
+      "step": 7412
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1286182398275847e-06,
+      "loss": 0.0642,
+      "step": 7413
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1264769424458187e-06,
+      "loss": 0.0679,
+      "step": 7414
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1243365944868687e-06,
+      "loss": 0.0631,
+      "step": 7415
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.122197196208824e-06,
+      "loss": 0.0633,
+      "step": 7416
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1200587478696667e-06,
+      "loss": 0.0671,
+      "step": 7417
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1179212497272582e-06,
+      "loss": 0.0669,
+      "step": 7418
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.115784702039345e-06,
+      "loss": 0.064,
+      "step": 7419
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1136491050635653e-06,
+      "loss": 0.0642,
+      "step": 7420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1115144590574365e-06,
+      "loss": 0.0674,
+      "step": 7421
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10938076427836e-06,
+      "loss": 0.0665,
+      "step": 7422
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1072480209836323e-06,
+      "loss": 0.0646,
+      "step": 7423
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1051162294304226e-06,
+      "loss": 0.0687,
+      "step": 7424
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1029853898757955e-06,
+      "loss": 0.0662,
+      "step": 7425
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1008555025766954e-06,
+      "loss": 0.0721,
+      "step": 7426
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0987265677899504e-06,
+      "loss": 0.0743,
+      "step": 7427
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0965985857722817e-06,
+      "loss": 0.0673,
+      "step": 7428
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0944715567802878e-06,
+      "loss": 0.0678,
+      "step": 7429
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0923454810704515e-06,
+      "loss": 0.0629,
+      "step": 7430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.09022035889915e-06,
+      "loss": 0.0669,
+      "step": 7431
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.088096190522636e-06,
+      "loss": 0.0736,
+      "step": 7432
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0859729761970493e-06,
+      "loss": 0.0672,
+      "step": 7433
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0838507161784216e-06,
+      "loss": 0.071,
+      "step": 7434
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0817294107226572e-06,
+      "loss": 0.0669,
+      "step": 7435
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.079609060085559e-06,
+      "loss": 0.0686,
+      "step": 7436
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.077489664522805e-06,
+      "loss": 0.0646,
+      "step": 7437
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0753712242899558e-06,
+      "loss": 0.0701,
+      "step": 7438
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0732537396424712e-06,
+      "loss": 0.0647,
+      "step": 7439
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.071137210835681e-06,
+      "loss": 0.0656,
+      "step": 7440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.069021638124803e-06,
+      "loss": 0.078,
+      "step": 7441
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.066907021764948e-06,
+      "loss": 0.0656,
+      "step": 7442
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.064793362011103e-06,
+      "loss": 0.068,
+      "step": 7443
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.062680659118137e-06,
+      "loss": 0.0658,
+      "step": 7444
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.060568913340817e-06,
+      "loss": 0.0704,
+      "step": 7445
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0584581249337797e-06,
+      "loss": 0.0633,
+      "step": 7446
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0563482941515577e-06,
+      "loss": 0.0695,
+      "step": 7447
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0542394212485617e-06,
+      "loss": 0.0717,
+      "step": 7448
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.052131506479086e-06,
+      "loss": 0.0699,
+      "step": 7449
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.050024550097316e-06,
+      "loss": 0.0679,
+      "step": 7450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.047918552357315e-06,
+      "loss": 0.0618,
+      "step": 7451
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045813513513032e-06,
+      "loss": 0.0621,
+      "step": 7452
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.043709433818305e-06,
+      "loss": 0.0651,
+      "step": 7453
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0416063135268503e-06,
+      "loss": 0.0652,
+      "step": 7454
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.03950415289227e-06,
+      "loss": 0.0685,
+      "step": 7455
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.037402952168056e-06,
+      "loss": 0.0646,
+      "step": 7456
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.035302711607574e-06,
+      "loss": 0.0718,
+      "step": 7457
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0332034314640857e-06,
+      "loss": 0.0614,
+      "step": 7458
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.03110511199073e-06,
+      "loss": 0.0737,
+      "step": 7459
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.029007753440526e-06,
+      "loss": 0.0693,
+      "step": 7460
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.026911356066389e-06,
+      "loss": 0.0677,
+      "step": 7461
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.024815920121108e-06,
+      "loss": 0.061,
+      "step": 7462
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0227214458573586e-06,
+      "loss": 0.0726,
+      "step": 7463
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0206279335277034e-06,
+      "loss": 0.0734,
+      "step": 7464
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0185353833845876e-06,
+      "loss": 0.071,
+      "step": 7465
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.016443795680336e-06,
+      "loss": 0.0731,
+      "step": 7466
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.014353170667166e-06,
+      "loss": 0.0664,
+      "step": 7467
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.012263508597172e-06,
+      "loss": 0.072,
+      "step": 7468
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0101748097223317e-06,
+      "loss": 0.0728,
+      "step": 7469
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0080870742945136e-06,
+      "loss": 0.0675,
+      "step": 7470
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0060003025654605e-06,
+      "loss": 0.0682,
+      "step": 7471
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0039144947868117e-06,
+      "loss": 0.0711,
+      "step": 7472
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.001829651210078e-06,
+      "loss": 0.0661,
+      "step": 7473
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9997457720866554e-06,
+      "loss": 0.0675,
+      "step": 7474
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9976628576678336e-06,
+      "loss": 0.0589,
+      "step": 7475
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.995580908204776e-06,
+      "loss": 0.0635,
+      "step": 7476
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9934999239485307e-06,
+      "loss": 0.0684,
+      "step": 7477
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9914199051500373e-06,
+      "loss": 0.0742,
+      "step": 7478
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.98934085206011e-06,
+      "loss": 0.0678,
+      "step": 7479
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.987262764929447e-06,
+      "loss": 0.0668,
+      "step": 7480
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985185644008638e-06,
+      "loss": 0.0625,
+      "step": 7481
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9831094895481473e-06,
+      "loss": 0.0671,
+      "step": 7482
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.981034301798329e-06,
+      "loss": 0.0679,
+      "step": 7483
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9789600810094177e-06,
+      "loss": 0.07,
+      "step": 7484
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.976886827431529e-06,
+      "loss": 0.0709,
+      "step": 7485
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974814541314669e-06,
+      "loss": 0.0642,
+      "step": 7486
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9727432229087195e-06,
+      "loss": 0.0663,
+      "step": 7487
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9706728724634485e-06,
+      "loss": 0.068,
+      "step": 7488
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.968603490228511e-06,
+      "loss": 0.0684,
+      "step": 7489
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9665350764534395e-06,
+      "loss": 0.06,
+      "step": 7490
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9644676313876497e-06,
+      "loss": 0.0681,
+      "step": 7491
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9624011552804476e-06,
+      "loss": 0.0702,
+      "step": 7492
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9603356483810143e-06,
+      "loss": 0.0678,
+      "step": 7493
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.958271110938421e-06,
+      "loss": 0.0674,
+      "step": 7494
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.956207543201616e-06,
+      "loss": 0.074,
+      "step": 7495
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9541449454194307e-06,
+      "loss": 0.0711,
+      "step": 7496
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9520833178405865e-06,
+      "loss": 0.068,
+      "step": 7497
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9500226607136807e-06,
+      "loss": 0.0633,
+      "step": 7498
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9479629742871954e-06,
+      "loss": 0.0723,
+      "step": 7499
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.945904258809499e-06,
+      "loss": 0.069,
+      "step": 7500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9438465145288377e-06,
+      "loss": 0.0683,
+      "step": 7501
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9417897416933428e-06,
+      "loss": 0.0652,
+      "step": 7502
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9397339405510306e-06,
+      "loss": 0.0724,
+      "step": 7503
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.937679111349796e-06,
+      "loss": 0.0668,
+      "step": 7504
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9356252543374233e-06,
+      "loss": 0.068,
+      "step": 7505
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.933572369761573e-06,
+      "loss": 0.0632,
+      "step": 7506
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9315204578697856e-06,
+      "loss": 0.0641,
+      "step": 7507
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.929469518909498e-06,
+      "loss": 0.067,
+      "step": 7508
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.927419553128016e-06,
+      "loss": 0.0672,
+      "step": 7509
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9253705607725328e-06,
+      "loss": 0.0702,
+      "step": 7510
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.923322542090127e-06,
+      "loss": 0.069,
+      "step": 7511
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.921275497327757e-06,
+      "loss": 0.072,
+      "step": 7512
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919229426732261e-06,
+      "loss": 0.0662,
+      "step": 7513
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.917184330550368e-06,
+      "loss": 0.0683,
+      "step": 7514
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9151402090286786e-06,
+      "loss": 0.0666,
+      "step": 7515
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.913097062413688e-06,
+      "loss": 0.0643,
+      "step": 7516
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9110548909517633e-06,
+      "loss": 0.07,
+      "step": 7517
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.909013694889158e-06,
+      "loss": 0.0677,
+      "step": 7518
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9069734744720115e-06,
+      "loss": 0.0656,
+      "step": 7519
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9049342299463392e-06,
+      "loss": 0.067,
+      "step": 7520
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.902895961558041e-06,
+      "loss": 0.0679,
+      "step": 7521
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.900858669552903e-06,
+      "loss": 0.0718,
+      "step": 7522
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8988223541765904e-06,
+      "loss": 0.0647,
+      "step": 7523
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8967870156746471e-06,
+      "loss": 0.0664,
+      "step": 7524
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8947526542925066e-06,
+      "loss": 0.0743,
+      "step": 7525
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8927192702754804e-06,
+      "loss": 0.07,
+      "step": 7526
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8906868638687592e-06,
+      "loss": 0.0669,
+      "step": 7527
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8886554353174235e-06,
+      "loss": 0.0672,
+      "step": 7528
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8866249848664275e-06,
+      "loss": 0.0675,
+      "step": 7529
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8845955127606175e-06,
+      "loss": 0.071,
+      "step": 7530
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8825670192447109e-06,
+      "loss": 0.0646,
+      "step": 7531
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8805395045633101e-06,
+      "loss": 0.07,
+      "step": 7532
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8785129689609082e-06,
+      "loss": 0.0702,
+      "step": 7533
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.87648741268187e-06,
+      "loss": 0.0696,
+      "step": 7534
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.874462835970443e-06,
+      "loss": 0.0734,
+      "step": 7535
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8724392390707635e-06,
+      "loss": 0.0545,
+      "step": 7536
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8704166222268438e-06,
+      "loss": 0.0635,
+      "step": 7537
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8683949856825778e-06,
+      "loss": 0.0653,
+      "step": 7538
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8663743296817472e-06,
+      "loss": 0.068,
+      "step": 7539
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8643546544680058e-06,
+      "loss": 0.0667,
+      "step": 7540
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8623359602849001e-06,
+      "loss": 0.0715,
+      "step": 7541
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8603182473758497e-06,
+      "loss": 0.0673,
+      "step": 7542
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8583015159841578e-06,
+      "loss": 0.065,
+      "step": 7543
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8562857663530143e-06,
+      "loss": 0.0724,
+      "step": 7544
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8542709987254859e-06,
+      "loss": 0.0719,
+      "step": 7545
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.852257213344517e-06,
+      "loss": 0.0667,
+      "step": 7546
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.850244410452945e-06,
+      "loss": 0.0687,
+      "step": 7547
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8482325902934796e-06,
+      "loss": 0.0696,
+      "step": 7548
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8462217531087112e-06,
+      "loss": 0.0665,
+      "step": 7549
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8442118991411207e-06,
+      "loss": 0.0772,
+      "step": 7550
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8422030286330605e-06,
+      "loss": 0.0705,
+      "step": 7551
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8401951418267715e-06,
+      "loss": 0.0663,
+      "step": 7552
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8381882389643734e-06,
+      "loss": 0.0686,
+      "step": 7553
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8361823202878636e-06,
+      "loss": 0.066,
+      "step": 7554
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8341773860391288e-06,
+      "loss": 0.0726,
+      "step": 7555
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8321734364599298e-06,
+      "loss": 0.0681,
+      "step": 7556
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.83017047179191e-06,
+      "loss": 0.068,
+      "step": 7557
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8281684922765996e-06,
+      "loss": 0.0718,
+      "step": 7558
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8261674981554035e-06,
+      "loss": 0.0667,
+      "step": 7559
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.824167489669606e-06,
+      "loss": 0.0684,
+      "step": 7560
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8221684670603845e-06,
+      "loss": 0.0712,
+      "step": 7561
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8201704305687829e-06,
+      "loss": 0.0722,
+      "step": 7562
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8181733804357381e-06,
+      "loss": 0.0694,
+      "step": 7563
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8161773169020602e-06,
+      "loss": 0.0698,
+      "step": 7564
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8141822402084407e-06,
+      "loss": 0.0746,
+      "step": 7565
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8121881505954597e-06,
+      "loss": 0.0682,
+      "step": 7566
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8101950483035701e-06,
+      "loss": 0.0678,
+      "step": 7567
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8082029335731066e-06,
+      "loss": 0.0696,
+      "step": 7568
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.806211806644289e-06,
+      "loss": 0.0612,
+      "step": 7569
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8042216677572178e-06,
+      "loss": 0.0628,
+      "step": 7570
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8022325171518683e-06,
+      "loss": 0.0696,
+      "step": 7571
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.800244355068106e-06,
+      "loss": 0.0688,
+      "step": 7572
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7982571817456685e-06,
+      "loss": 0.0684,
+      "step": 7573
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7962709974241744e-06,
+      "loss": 0.0749,
+      "step": 7574
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7942858023431331e-06,
+      "loss": 0.0644,
+      "step": 7575
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7923015967419245e-06,
+      "loss": 0.0701,
+      "step": 7576
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7903183808598102e-06,
+      "loss": 0.069,
+      "step": 7577
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.788336154935939e-06,
+      "loss": 0.0687,
+      "step": 7578
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7863549192093342e-06,
+      "loss": 0.0666,
+      "step": 7579
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7843746739188994e-06,
+      "loss": 0.0664,
+      "step": 7580
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7823954193034255e-06,
+      "loss": 0.0693,
+      "step": 7581
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7804171556015749e-06,
+      "loss": 0.0669,
+      "step": 7582
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7784398830519002e-06,
+      "loss": 0.0683,
+      "step": 7583
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7764636018928249e-06,
+      "loss": 0.0648,
+      "step": 7584
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7744883123626577e-06,
+      "loss": 0.069,
+      "step": 7585
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7725140146995901e-06,
+      "loss": 0.0671,
+      "step": 7586
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7705407091416904e-06,
+      "loss": 0.0641,
+      "step": 7587
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.768568395926904e-06,
+      "loss": 0.0704,
+      "step": 7588
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7665970752930673e-06,
+      "loss": 0.0694,
+      "step": 7589
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7646267474778877e-06,
+      "loss": 0.0692,
+      "step": 7590
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7626574127189522e-06,
+      "loss": 0.068,
+      "step": 7591
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7606890712537382e-06,
+      "loss": 0.0686,
+      "step": 7592
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7587217233195907e-06,
+      "loss": 0.0716,
+      "step": 7593
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7567553691537454e-06,
+      "loss": 0.068,
+      "step": 7594
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7547900089933112e-06,
+      "loss": 0.0648,
+      "step": 7595
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7528256430752787e-06,
+      "loss": 0.0704,
+      "step": 7596
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.750862271636522e-06,
+      "loss": 0.0644,
+      "step": 7597
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7488998949137926e-06,
+      "loss": 0.0716,
+      "step": 7598
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7469385131437178e-06,
+      "loss": 0.0657,
+      "step": 7599
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7449781265628162e-06,
+      "loss": 0.0765,
+      "step": 7600
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7430187354074758e-06,
+      "loss": 0.0691,
+      "step": 7601
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7410603399139648e-06,
+      "loss": 0.0686,
+      "step": 7602
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7391029403184423e-06,
+      "loss": 0.0661,
+      "step": 7603
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.737146536856933e-06,
+      "loss": 0.0677,
+      "step": 7604
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7351911297653544e-06,
+      "loss": 0.073,
+      "step": 7605
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.733236719279494e-06,
+      "loss": 0.0652,
+      "step": 7606
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.731283305635022e-06,
+      "loss": 0.0697,
+      "step": 7607
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.729330889067493e-06,
+      "loss": 0.0742,
+      "step": 7608
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7273794698123359e-06,
+      "loss": 0.0674,
+      "step": 7609
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7254290481048586e-06,
+      "loss": 0.0726,
+      "step": 7610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7234796241802555e-06,
+      "loss": 0.0633,
+      "step": 7611
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7215311982735949e-06,
+      "loss": 0.0678,
+      "step": 7612
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7195837706198238e-06,
+      "loss": 0.066,
+      "step": 7613
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.717637341453775e-06,
+      "loss": 0.0704,
+      "step": 7614
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7156919110101554e-06,
+      "loss": 0.0684,
+      "step": 7615
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7137474795235508e-06,
+      "loss": 0.0671,
+      "step": 7616
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7118040472284347e-06,
+      "loss": 0.0639,
+      "step": 7617
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7098616143591483e-06,
+      "loss": 0.0666,
+      "step": 7618
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7079201811499247e-06,
+      "loss": 0.0678,
+      "step": 7619
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7059797478348661e-06,
+      "loss": 0.0664,
+      "step": 7620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7040403146479557e-06,
+      "loss": 0.0666,
+      "step": 7621
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.702101881823065e-06,
+      "loss": 0.0615,
+      "step": 7622
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7001644495939352e-06,
+      "loss": 0.0694,
+      "step": 7623
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.698228018194188e-06,
+      "loss": 0.0685,
+      "step": 7624
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.69629258785733e-06,
+      "loss": 0.0707,
+      "step": 7625
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6943581588167436e-06,
+      "loss": 0.0679,
+      "step": 7626
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6924247313056852e-06,
+      "loss": 0.0639,
+      "step": 7627
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.690492305557303e-06,
+      "loss": 0.0709,
+      "step": 7628
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6885608818046117e-06,
+      "loss": 0.0663,
+      "step": 7629
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6866304602805138e-06,
+      "loss": 0.0705,
+      "step": 7630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6847010412177866e-06,
+      "loss": 0.0679,
+      "step": 7631
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6827726248490862e-06,
+      "loss": 0.0726,
+      "step": 7632
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6808452114069518e-06,
+      "loss": 0.0691,
+      "step": 7633
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6789188011237977e-06,
+      "loss": 0.0575,
+      "step": 7634
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6769933942319172e-06,
+      "loss": 0.0711,
+      "step": 7635
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6750689909634877e-06,
+      "loss": 0.0682,
+      "step": 7636
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6731455915505602e-06,
+      "loss": 0.0637,
+      "step": 7637
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6712231962250624e-06,
+      "loss": 0.0667,
+      "step": 7638
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6693018052188114e-06,
+      "loss": 0.0705,
+      "step": 7639
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6673814187634918e-06,
+      "loss": 0.0703,
+      "step": 7640
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.665462037090676e-06,
+      "loss": 0.0587,
+      "step": 7641
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.663543660431809e-06,
+      "loss": 0.0675,
+      "step": 7642
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6616262890182145e-06,
+      "loss": 0.0742,
+      "step": 7643
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.659709923081102e-06,
+      "loss": 0.0676,
+      "step": 7644
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6577945628515523e-06,
+      "loss": 0.0656,
+      "step": 7645
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6558802085605252e-06,
+      "loss": 0.0656,
+      "step": 7646
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6539668604388681e-06,
+      "loss": 0.0713,
+      "step": 7647
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6520545187172965e-06,
+      "loss": 0.0729,
+      "step": 7648
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.650143183626407e-06,
+      "loss": 0.0608,
+      "step": 7649
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6482328553966808e-06,
+      "loss": 0.0691,
+      "step": 7650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6463235342584694e-06,
+      "loss": 0.0727,
+      "step": 7651
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6444152204420106e-06,
+      "loss": 0.0615,
+      "step": 7652
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6425079141774147e-06,
+      "loss": 0.0738,
+      "step": 7653
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6406016156946715e-06,
+      "loss": 0.0663,
+      "step": 7654
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6386963252236544e-06,
+      "loss": 0.0661,
+      "step": 7655
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6367920429941087e-06,
+      "loss": 0.0651,
+      "step": 7656
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6348887692356596e-06,
+      "loss": 0.0676,
+      "step": 7657
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.632986504177816e-06,
+      "loss": 0.0673,
+      "step": 7658
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6310852480499584e-06,
+      "loss": 0.068,
+      "step": 7659
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.629185001081347e-06,
+      "loss": 0.0715,
+      "step": 7660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6272857635011252e-06,
+      "loss": 0.0684,
+      "step": 7661
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6253875355383097e-06,
+      "loss": 0.0655,
+      "step": 7662
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6234903174217942e-06,
+      "loss": 0.0706,
+      "step": 7663
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6215941093803578e-06,
+      "loss": 0.065,
+      "step": 7664
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6196989116426487e-06,
+      "loss": 0.0667,
+      "step": 7665
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.617804724437202e-06,
+      "loss": 0.069,
+      "step": 7666
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6159115479924259e-06,
+      "loss": 0.0636,
+      "step": 7667
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6140193825366046e-06,
+      "loss": 0.062,
+      "step": 7668
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6121282282979077e-06,
+      "loss": 0.0687,
+      "step": 7669
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6102380855043764e-06,
+      "loss": 0.0688,
+      "step": 7670
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6083489543839304e-06,
+      "loss": 0.0607,
+      "step": 7671
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.606460835164373e-06,
+      "loss": 0.0687,
+      "step": 7672
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6045737280733798e-06,
+      "loss": 0.0646,
+      "step": 7673
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6026876333385045e-06,
+      "loss": 0.0678,
+      "step": 7674
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6008025511871827e-06,
+      "loss": 0.0618,
+      "step": 7675
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.598918481846724e-06,
+      "loss": 0.0703,
+      "step": 7676
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5970354255443199e-06,
+      "loss": 0.0672,
+      "step": 7677
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5951533825070365e-06,
+      "loss": 0.0643,
+      "step": 7678
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5932723529618156e-06,
+      "loss": 0.0612,
+      "step": 7679
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5913923371354833e-06,
+      "loss": 0.0626,
+      "step": 7680
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5895133352547388e-06,
+      "loss": 0.0713,
+      "step": 7681
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.587635347546158e-06,
+      "loss": 0.0693,
+      "step": 7682
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5857583742362003e-06,
+      "loss": 0.0677,
+      "step": 7683
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5838824155511978e-06,
+      "loss": 0.0624,
+      "step": 7684
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5820074717173583e-06,
+      "loss": 0.0662,
+      "step": 7685
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5801335429607755e-06,
+      "loss": 0.065,
+      "step": 7686
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.578260629507411e-06,
+      "loss": 0.0729,
+      "step": 7687
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.576388731583114e-06,
+      "loss": 0.0654,
+      "step": 7688
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5745178494136026e-06,
+      "loss": 0.0638,
+      "step": 7689
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5726479832244733e-06,
+      "loss": 0.0634,
+      "step": 7690
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5707791332412082e-06,
+      "loss": 0.0593,
+      "step": 7691
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5689112996891576e-06,
+      "loss": 0.0618,
+      "step": 7692
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5670444827935515e-06,
+      "loss": 0.0744,
+      "step": 7693
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5651786827795024e-06,
+      "loss": 0.0733,
+      "step": 7694
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5633138998719955e-06,
+      "loss": 0.0626,
+      "step": 7695
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.56145013429589e-06,
+      "loss": 0.0599,
+      "step": 7696
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5595873862759325e-06,
+      "loss": 0.0647,
+      "step": 7697
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.557725656036736e-06,
+      "loss": 0.07,
+      "step": 7698
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5558649438028007e-06,
+      "loss": 0.068,
+      "step": 7699
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.554005249798497e-06,
+      "loss": 0.0669,
+      "step": 7700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5521465742480713e-06,
+      "loss": 0.0668,
+      "step": 7701
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5502889173756563e-06,
+      "loss": 0.0693,
+      "step": 7702
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5484322794052542e-06,
+      "loss": 0.063,
+      "step": 7703
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5465766605607424e-06,
+      "loss": 0.067,
+      "step": 7704
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5447220610658863e-06,
+      "loss": 0.0664,
+      "step": 7705
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5428684811443162e-06,
+      "loss": 0.0722,
+      "step": 7706
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5410159210195442e-06,
+      "loss": 0.067,
+      "step": 7707
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5391643809149648e-06,
+      "loss": 0.0613,
+      "step": 7708
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5373138610538385e-06,
+      "loss": 0.0669,
+      "step": 7709
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5354643616593135e-06,
+      "loss": 0.0702,
+      "step": 7710
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.533615882954409e-06,
+      "loss": 0.0644,
+      "step": 7711
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5317684251620203e-06,
+      "loss": 0.065,
+      "step": 7712
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5299219885049255e-06,
+      "loss": 0.0633,
+      "step": 7713
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5280765732057734e-06,
+      "loss": 0.0702,
+      "step": 7714
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5262321794870904e-06,
+      "loss": 0.0681,
+      "step": 7715
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5243888075712843e-06,
+      "loss": 0.0629,
+      "step": 7716
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.522546457680637e-06,
+      "loss": 0.0646,
+      "step": 7717
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5207051300373021e-06,
+      "loss": 0.0686,
+      "step": 7718
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5188648248633208e-06,
+      "loss": 0.063,
+      "step": 7719
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5170255423806023e-06,
+      "loss": 0.065,
+      "step": 7720
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5151872828109326e-06,
+      "loss": 0.0638,
+      "step": 7721
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.513350046375982e-06,
+      "loss": 0.0713,
+      "step": 7722
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5115138332972867e-06,
+      "loss": 0.0615,
+      "step": 7723
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5096786437962697e-06,
+      "loss": 0.071,
+      "step": 7724
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5078444780942248e-06,
+      "loss": 0.0605,
+      "step": 7725
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5060113364123197e-06,
+      "loss": 0.0678,
+      "step": 7726
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5041792189716088e-06,
+      "loss": 0.0697,
+      "step": 7727
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.502348125993013e-06,
+      "loss": 0.0661,
+      "step": 7728
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5005180576973299e-06,
+      "loss": 0.0667,
+      "step": 7729
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4986890143052435e-06,
+      "loss": 0.0713,
+      "step": 7730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4968609960373048e-06,
+      "loss": 0.0678,
+      "step": 7731
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4950340031139421e-06,
+      "loss": 0.0684,
+      "step": 7732
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4932080357554646e-06,
+      "loss": 0.0706,
+      "step": 7733
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.491383094182053e-06,
+      "loss": 0.0698,
+      "step": 7734
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4895591786137699e-06,
+      "loss": 0.0681,
+      "step": 7735
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4877362892705461e-06,
+      "loss": 0.071,
+      "step": 7736
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.485914426372198e-06,
+      "loss": 0.0646,
+      "step": 7737
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4840935901384102e-06,
+      "loss": 0.0711,
+      "step": 7738
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4822737807887488e-06,
+      "loss": 0.0656,
+      "step": 7739
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.480454998542652e-06,
+      "loss": 0.068,
+      "step": 7740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4786372436194397e-06,
+      "loss": 0.0663,
+      "step": 7741
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.476820516238303e-06,
+      "loss": 0.0773,
+      "step": 7742
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4750048166183073e-06,
+      "loss": 0.0674,
+      "step": 7743
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4731901449784024e-06,
+      "loss": 0.0663,
+      "step": 7744
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.471376501537407e-06,
+      "loss": 0.0655,
+      "step": 7745
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.469563886514016e-06,
+      "loss": 0.0643,
+      "step": 7746
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4677523001268056e-06,
+      "loss": 0.0651,
+      "step": 7747
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4659417425942235e-06,
+      "loss": 0.0715,
+      "step": 7748
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4641322141345915e-06,
+      "loss": 0.0798,
+      "step": 7749
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.462323714966114e-06,
+      "loss": 0.0744,
+      "step": 7750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4605162453068667e-06,
+      "loss": 0.0656,
+      "step": 7751
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4587098053747994e-06,
+      "loss": 0.0618,
+      "step": 7752
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4569043953877437e-06,
+      "loss": 0.0662,
+      "step": 7753
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.455100015563401e-06,
+      "loss": 0.0749,
+      "step": 7754
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4532966661193536e-06,
+      "loss": 0.0692,
+      "step": 7755
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4514943472730558e-06,
+      "loss": 0.0635,
+      "step": 7756
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4496930592418368e-06,
+      "loss": 0.0647,
+      "step": 7757
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4478928022429073e-06,
+      "loss": 0.0694,
+      "step": 7758
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4460935764933482e-06,
+      "loss": 0.0646,
+      "step": 7759
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4442953822101157e-06,
+      "loss": 0.0714,
+      "step": 7760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.442498219610048e-06,
+      "loss": 0.0631,
+      "step": 7761
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.440702088909851e-06,
+      "loss": 0.0733,
+      "step": 7762
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4389069903261099e-06,
+      "loss": 0.0674,
+      "step": 7763
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.437112924075289e-06,
+      "loss": 0.0633,
+      "step": 7764
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4353198903737188e-06,
+      "loss": 0.0686,
+      "step": 7765
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.433527889437616e-06,
+      "loss": 0.0665,
+      "step": 7766
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4317369214830656e-06,
+      "loss": 0.0681,
+      "step": 7767
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4299469867260284e-06,
+      "loss": 0.0671,
+      "step": 7768
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4281580853823463e-06,
+      "loss": 0.0685,
+      "step": 7769
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4263702176677297e-06,
+      "loss": 0.0616,
+      "step": 7770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4245833837977662e-06,
+      "loss": 0.0724,
+      "step": 7771
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4227975839879237e-06,
+      "loss": 0.0638,
+      "step": 7772
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.42101281845354e-06,
+      "loss": 0.07,
+      "step": 7773
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4192290874098257e-06,
+      "loss": 0.0724,
+      "step": 7774
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4174463910718771e-06,
+      "loss": 0.0665,
+      "step": 7775
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4156647296546533e-06,
+      "loss": 0.0681,
+      "step": 7776
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.413884103373001e-06,
+      "loss": 0.0711,
+      "step": 7777
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.412104512441631e-06,
+      "loss": 0.0651,
+      "step": 7778
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4103259570751326e-06,
+      "loss": 0.063,
+      "step": 7779
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4085484374879776e-06,
+      "loss": 0.0675,
+      "step": 7780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4067719538945024e-06,
+      "loss": 0.067,
+      "step": 7781
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.404996506508921e-06,
+      "loss": 0.076,
+      "step": 7782
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4032220955453312e-06,
+      "loss": 0.0664,
+      "step": 7783
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.401448721217693e-06,
+      "loss": 0.0745,
+      "step": 7784
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3996763837398487e-06,
+      "loss": 0.0712,
+      "step": 7785
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.397905083325517e-06,
+      "loss": 0.0639,
+      "step": 7786
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3961348201882842e-06,
+      "loss": 0.0677,
+      "step": 7787
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3943655945416213e-06,
+      "loss": 0.064,
+      "step": 7788
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3925974065988658e-06,
+      "loss": 0.0654,
+      "step": 7789
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3908302565732322e-06,
+      "loss": 0.0747,
+      "step": 7790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3890641446778152e-06,
+      "loss": 0.0625,
+      "step": 7791
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3872990711255775e-06,
+      "loss": 0.0652,
+      "step": 7792
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3855350361293563e-06,
+      "loss": 0.0646,
+      "step": 7793
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3837720399018727e-06,
+      "loss": 0.0646,
+      "step": 7794
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3820100826557126e-06,
+      "loss": 0.0709,
+      "step": 7795
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3802491646033377e-06,
+      "loss": 0.0628,
+      "step": 7796
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.378489285957093e-06,
+      "loss": 0.0654,
+      "step": 7797
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3767304469291865e-06,
+      "loss": 0.073,
+      "step": 7798
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3749726477317104e-06,
+      "loss": 0.0647,
+      "step": 7799
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3732158885766277e-06,
+      "loss": 0.0646,
+      "step": 7800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3714601696757713e-06,
+      "loss": 0.0704,
+      "step": 7801
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.369705491240858e-06,
+      "loss": 0.0735,
+      "step": 7802
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3679518534834735e-06,
+      "loss": 0.0609,
+      "step": 7803
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3661992566150761e-06,
+      "loss": 0.0688,
+      "step": 7804
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3644477008470048e-06,
+      "loss": 0.0655,
+      "step": 7805
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.362697186390468e-06,
+      "loss": 0.0673,
+      "step": 7806
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3609477134565486e-06,
+      "loss": 0.0645,
+      "step": 7807
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3591992822562095e-06,
+      "loss": 0.0669,
+      "step": 7808
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3574518930002812e-06,
+      "loss": 0.0708,
+      "step": 7809
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3557055458994695e-06,
+      "loss": 0.0684,
+      "step": 7810
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3539602411643616e-06,
+      "loss": 0.064,
+      "step": 7811
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3522159790054079e-06,
+      "loss": 0.0669,
+      "step": 7812
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3504727596329437e-06,
+      "loss": 0.067,
+      "step": 7813
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3487305832571717e-06,
+      "loss": 0.0646,
+      "step": 7814
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3469894500881698e-06,
+      "loss": 0.0674,
+      "step": 7815
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.345249360335894e-06,
+      "loss": 0.0689,
+      "step": 7816
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3435103142101701e-06,
+      "loss": 0.0687,
+      "step": 7817
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3417723119206982e-06,
+      "loss": 0.0653,
+      "step": 7818
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3400353536770573e-06,
+      "loss": 0.0674,
+      "step": 7819
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3382994396886951e-06,
+      "loss": 0.0642,
+      "step": 7820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3365645701649343e-06,
+      "loss": 0.0608,
+      "step": 7821
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3348307453149755e-06,
+      "loss": 0.0637,
+      "step": 7822
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3330979653478882e-06,
+      "loss": 0.0698,
+      "step": 7823
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3313662304726215e-06,
+      "loss": 0.0665,
+      "step": 7824
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3296355408979944e-06,
+      "loss": 0.0671,
+      "step": 7825
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3279058968326963e-06,
+      "loss": 0.0659,
+      "step": 7826
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3261772984853017e-06,
+      "loss": 0.0687,
+      "step": 7827
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3244497460642492e-06,
+      "loss": 0.0695,
+      "step": 7828
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.322723239777851e-06,
+      "loss": 0.0701,
+      "step": 7829
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3209977798343044e-06,
+      "loss": 0.067,
+      "step": 7830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3192733664416669e-06,
+      "loss": 0.0694,
+      "step": 7831
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3175499998078745e-06,
+      "loss": 0.0684,
+      "step": 7832
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3158276801407432e-06,
+      "loss": 0.0644,
+      "step": 7833
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3141064076479526e-06,
+      "loss": 0.0692,
+      "step": 7834
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3123861825370655e-06,
+      "loss": 0.0642,
+      "step": 7835
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3106670050155114e-06,
+      "loss": 0.0609,
+      "step": 7836
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3089488752905944e-06,
+      "loss": 0.073,
+      "step": 7837
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.307231793569499e-06,
+      "loss": 0.0705,
+      "step": 7838
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.305515760059274e-06,
+      "loss": 0.0625,
+      "step": 7839
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3038007749668457e-06,
+      "loss": 0.0673,
+      "step": 7840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.302086838499017e-06,
+      "loss": 0.0704,
+      "step": 7841
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3003739508624613e-06,
+      "loss": 0.067,
+      "step": 7842
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2986621122637222e-06,
+      "loss": 0.0728,
+      "step": 7843
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2969513229092245e-06,
+      "loss": 0.0704,
+      "step": 7844
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.29524158300526e-06,
+      "loss": 0.0773,
+      "step": 7845
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2935328927579993e-06,
+      "loss": 0.0721,
+      "step": 7846
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.291825252373482e-06,
+      "loss": 0.0651,
+      "step": 7847
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2901186620576189e-06,
+      "loss": 0.0665,
+      "step": 7848
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2884131220162044e-06,
+      "loss": 0.0647,
+      "step": 7849
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2867086324548972e-06,
+      "loss": 0.0703,
+      "step": 7850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2850051935792285e-06,
+      "loss": 0.0626,
+      "step": 7851
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2833028055946105e-06,
+      "loss": 0.071,
+      "step": 7852
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2816014687063238e-06,
+      "loss": 0.0704,
+      "step": 7853
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2799011831195197e-06,
+      "loss": 0.0638,
+      "step": 7854
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2782019490392294e-06,
+      "loss": 0.0693,
+      "step": 7855
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2765037666703538e-06,
+      "loss": 0.0626,
+      "step": 7856
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2748066362176625e-06,
+      "loss": 0.0633,
+      "step": 7857
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2731105578858082e-06,
+      "loss": 0.0661,
+      "step": 7858
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2714155318793065e-06,
+      "loss": 0.0632,
+      "step": 7859
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2697215584025547e-06,
+      "loss": 0.0717,
+      "step": 7860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2680286376598183e-06,
+      "loss": 0.0655,
+      "step": 7861
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2663367698552343e-06,
+      "loss": 0.068,
+      "step": 7862
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.264645955192819e-06,
+      "loss": 0.0637,
+      "step": 7863
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2629561938764557e-06,
+      "loss": 0.0667,
+      "step": 7864
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2612674861099028e-06,
+      "loss": 0.065,
+      "step": 7865
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2595798320967933e-06,
+      "loss": 0.0716,
+      "step": 7866
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2578932320406322e-06,
+      "loss": 0.0681,
+      "step": 7867
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2562076861447925e-06,
+      "loss": 0.0655,
+      "step": 7868
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2545231946125303e-06,
+      "loss": 0.0714,
+      "step": 7869
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2528397576469642e-06,
+      "loss": 0.0708,
+      "step": 7870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2511573754510942e-06,
+      "loss": 0.0681,
+      "step": 7871
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2494760482277869e-06,
+      "loss": 0.0747,
+      "step": 7872
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.247795776179781e-06,
+      "loss": 0.0654,
+      "step": 7873
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2461165595096958e-06,
+      "loss": 0.0679,
+      "step": 7874
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.244438398420017e-06,
+      "loss": 0.0677,
+      "step": 7875
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2427612931131016e-06,
+      "loss": 0.0771,
+      "step": 7876
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2410852437911857e-06,
+      "loss": 0.0686,
+      "step": 7877
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2394102506563744e-06,
+      "loss": 0.0656,
+      "step": 7878
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2377363139106402e-06,
+      "loss": 0.0672,
+      "step": 7879
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2360634337558396e-06,
+      "loss": 0.0677,
+      "step": 7880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2343916103936914e-06,
+      "loss": 0.0707,
+      "step": 7881
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2327208440257954e-06,
+      "loss": 0.0656,
+      "step": 7882
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2310511348536181e-06,
+      "loss": 0.0651,
+      "step": 7883
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2293824830784962e-06,
+      "loss": 0.0642,
+      "step": 7884
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2277148889016476e-06,
+      "loss": 0.0686,
+      "step": 7885
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2260483525241574e-06,
+      "loss": 0.0652,
+      "step": 7886
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.22438287414698e-06,
+      "loss": 0.0603,
+      "step": 7887
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.222718453970949e-06,
+      "loss": 0.0615,
+      "step": 7888
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2210550921967679e-06,
+      "loss": 0.0728,
+      "step": 7889
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.219392789025009e-06,
+      "loss": 0.0638,
+      "step": 7890
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2177315446561222e-06,
+      "loss": 0.0636,
+      "step": 7891
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.216071359290425e-06,
+      "loss": 0.0668,
+      "step": 7892
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.214412233128114e-06,
+      "loss": 0.0705,
+      "step": 7893
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2127541663692499e-06,
+      "loss": 0.0739,
+      "step": 7894
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2110971592137688e-06,
+      "loss": 0.0657,
+      "step": 7895
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.209441211861484e-06,
+      "loss": 0.0668,
+      "step": 7896
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2077863245120725e-06,
+      "loss": 0.0644,
+      "step": 7897
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.206132497365088e-06,
+      "loss": 0.0676,
+      "step": 7898
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2044797306199585e-06,
+      "loss": 0.0708,
+      "step": 7899
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2028280244759804e-06,
+      "loss": 0.0619,
+      "step": 7900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2011773791323223e-06,
+      "loss": 0.0656,
+      "step": 7901
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.199527794788028e-06,
+      "loss": 0.0592,
+      "step": 7902
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1978792716420074e-06,
+      "loss": 0.0656,
+      "step": 7903
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1962318098930503e-06,
+      "loss": 0.0651,
+      "step": 7904
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1945854097398158e-06,
+      "loss": 0.0721,
+      "step": 7905
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1929400713808316e-06,
+      "loss": 0.068,
+      "step": 7906
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1912957950144976e-06,
+      "loss": 0.07,
+      "step": 7907
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1896525808390913e-06,
+      "loss": 0.068,
+      "step": 7908
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1880104290527572e-06,
+      "loss": 0.0655,
+      "step": 7909
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1863693398535115e-06,
+      "loss": 0.0644,
+      "step": 7910
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1847293134392457e-06,
+      "loss": 0.0603,
+      "step": 7911
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1830903500077174e-06,
+      "loss": 0.064,
+      "step": 7912
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.181452449756566e-06,
+      "loss": 0.0662,
+      "step": 7913
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1798156128832927e-06,
+      "loss": 0.0783,
+      "step": 7914
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1781798395852716e-06,
+      "loss": 0.0618,
+      "step": 7915
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1765451300597574e-06,
+      "loss": 0.0686,
+      "step": 7916
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1749114845038668e-06,
+      "loss": 0.0701,
+      "step": 7917
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1732789031145886e-06,
+      "loss": 0.0745,
+      "step": 7918
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1716473860887933e-06,
+      "loss": 0.0655,
+      "step": 7919
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1700169336232125e-06,
+      "loss": 0.0651,
+      "step": 7920
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.16838754591445e-06,
+      "loss": 0.059,
+      "step": 7921
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1667592231589908e-06,
+      "loss": 0.0652,
+      "step": 7922
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1651319655531789e-06,
+      "loss": 0.0674,
+      "step": 7923
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.163505773293241e-06,
+      "loss": 0.0621,
+      "step": 7924
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1618806465752674e-06,
+      "loss": 0.0645,
+      "step": 7925
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1602565855952207e-06,
+      "loss": 0.0643,
+      "step": 7926
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.158633590548943e-06,
+      "loss": 0.0685,
+      "step": 7927
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.157011661632138e-06,
+      "loss": 0.0594,
+      "step": 7928
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1553907990403835e-06,
+      "loss": 0.0678,
+      "step": 7929
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.153771002969134e-06,
+      "loss": 0.0664,
+      "step": 7930
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1521522736137081e-06,
+      "loss": 0.0663,
+      "step": 7931
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1505346111692994e-06,
+      "loss": 0.0634,
+      "step": 7932
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1489180158309755e-06,
+      "loss": 0.0703,
+      "step": 7933
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1473024877936667e-06,
+      "loss": 0.0773,
+      "step": 7934
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1456880272521865e-06,
+      "loss": 0.0667,
+      "step": 7935
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.144074634401211e-06,
+      "loss": 0.0608,
+      "step": 7936
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.142462309435286e-06,
+      "loss": 0.0686,
+      "step": 7937
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1408510525488393e-06,
+      "loss": 0.0635,
+      "step": 7938
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1392408639361596e-06,
+      "loss": 0.0674,
+      "step": 7939
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1376317437914074e-06,
+      "loss": 0.0662,
+      "step": 7940
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1360236923086232e-06,
+      "loss": 0.0675,
+      "step": 7941
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1344167096817093e-06,
+      "loss": 0.0601,
+      "step": 7942
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.132810796104441e-06,
+      "loss": 0.0665,
+      "step": 7943
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.131205951770471e-06,
+      "loss": 0.0702,
+      "step": 7944
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1296021768733145e-06,
+      "loss": 0.0721,
+      "step": 7945
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1279994716063603e-06,
+      "loss": 0.0736,
+      "step": 7946
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1263978361628735e-06,
+      "loss": 0.0679,
+      "step": 7947
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.124797270735981e-06,
+      "loss": 0.0623,
+      "step": 7948
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1231977755186929e-06,
+      "loss": 0.0719,
+      "step": 7949
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.121599350703877e-06,
+      "loss": 0.0746,
+      "step": 7950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.120001996484279e-06,
+      "loss": 0.0645,
+      "step": 7951
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1184057130525184e-06,
+      "loss": 0.0634,
+      "step": 7952
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1168105006010777e-06,
+      "loss": 0.0682,
+      "step": 7953
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1152163593223154e-06,
+      "loss": 0.0671,
+      "step": 7954
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.113623289408462e-06,
+      "loss": 0.0591,
+      "step": 7955
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.112031291051614e-06,
+      "loss": 0.0678,
+      "step": 7956
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1104403644437412e-06,
+      "loss": 0.0721,
+      "step": 7957
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1088505097766867e-06,
+      "loss": 0.0651,
+      "step": 7958
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1072617272421592e-06,
+      "loss": 0.0649,
+      "step": 7959
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1056740170317438e-06,
+      "loss": 0.0654,
+      "step": 7960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1040873793368923e-06,
+      "loss": 0.0653,
+      "step": 7961
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1025018143489253e-06,
+      "loss": 0.066,
+      "step": 7962
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1009173222590407e-06,
+      "loss": 0.0685,
+      "step": 7963
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0993339032583028e-06,
+      "loss": 0.0694,
+      "step": 7964
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.097751557537643e-06,
+      "loss": 0.0714,
+      "step": 7965
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0961702852878731e-06,
+      "loss": 0.064,
+      "step": 7966
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0945900866996673e-06,
+      "loss": 0.0656,
+      "step": 7967
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0930109619635699e-06,
+      "loss": 0.063,
+      "step": 7968
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0914329112700018e-06,
+      "loss": 0.0665,
+      "step": 7969
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0898559348092485e-06,
+      "loss": 0.0681,
+      "step": 7970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0882800327714727e-06,
+      "loss": 0.0675,
+      "step": 7971
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0867052053466997e-06,
+      "loss": 0.0614,
+      "step": 7972
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.085131452724828e-06,
+      "loss": 0.0615,
+      "step": 7973
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0835587750956323e-06,
+      "loss": 0.0712,
+      "step": 7974
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0819871726487485e-06,
+      "loss": 0.0671,
+      "step": 7975
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0804166455736865e-06,
+      "loss": 0.0735,
+      "step": 7976
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.078847194059831e-06,
+      "loss": 0.0726,
+      "step": 7977
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0772788182964311e-06,
+      "loss": 0.0679,
+      "step": 7978
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0757115184726052e-06,
+      "loss": 0.0672,
+      "step": 7979
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0741452947773511e-06,
+      "loss": 0.0604,
+      "step": 7980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0725801473995245e-06,
+      "loss": 0.0719,
+      "step": 7981
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0710160765278621e-06,
+      "loss": 0.065,
+      "step": 7982
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0694530823509653e-06,
+      "loss": 0.0708,
+      "step": 7983
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0678911650573043e-06,
+      "loss": 0.0675,
+      "step": 7984
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0663303248352242e-06,
+      "loss": 0.0687,
+      "step": 7985
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0647705618729375e-06,
+      "loss": 0.0672,
+      "step": 7986
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0632118763585242e-06,
+      "loss": 0.0712,
+      "step": 7987
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0616542684799414e-06,
+      "loss": 0.0714,
+      "step": 7988
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0600977384250099e-06,
+      "loss": 0.0637,
+      "step": 7989
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0585422863814198e-06,
+      "loss": 0.0621,
+      "step": 7990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.05698791253674e-06,
+      "loss": 0.0677,
+      "step": 7991
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.055434617078399e-06,
+      "loss": 0.0666,
+      "step": 7992
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0538824001937032e-06,
+      "loss": 0.0731,
+      "step": 7993
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0523312620698234e-06,
+      "loss": 0.0671,
+      "step": 7994
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0507812028938003e-06,
+      "loss": 0.0612,
+      "step": 7995
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.049232222852551e-06,
+      "loss": 0.0716,
+      "step": 7996
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0476843221328559e-06,
+      "loss": 0.0642,
+      "step": 7997
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.046137500921366e-06,
+      "loss": 0.0607,
+      "step": 7998
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0445917594046073e-06,
+      "loss": 0.0675,
+      "step": 7999
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.04304709776897e-06,
+      "loss": 0.0658,
+      "step": 8000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0415035162007124e-06,
+      "loss": 0.0714,
+      "step": 8001
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.039961014885973e-06,
+      "loss": 0.0705,
+      "step": 8002
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0384195940107489e-06,
+      "loss": 0.0641,
+      "step": 8003
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0368792537609096e-06,
+      "loss": 0.0687,
+      "step": 8004
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0353399943221998e-06,
+      "loss": 0.071,
+      "step": 8005
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0338018158802266e-06,
+      "loss": 0.0647,
+      "step": 8006
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0322647186204736e-06,
+      "loss": 0.0604,
+      "step": 8007
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0307287027282885e-06,
+      "loss": 0.0647,
+      "step": 8008
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0291937683888886e-06,
+      "loss": 0.0673,
+      "step": 8009
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0276599157873678e-06,
+      "loss": 0.0757,
+      "step": 8010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0261271451086808e-06,
+      "loss": 0.0689,
+      "step": 8011
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0245954565376547e-06,
+      "loss": 0.0651,
+      "step": 8012
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0230648502589902e-06,
+      "loss": 0.0696,
+      "step": 8013
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0215353264572537e-06,
+      "loss": 0.0667,
+      "step": 8014
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0200068853168787e-06,
+      "loss": 0.0698,
+      "step": 8015
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0184795270221747e-06,
+      "loss": 0.0686,
+      "step": 8016
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0169532517573145e-06,
+      "loss": 0.0701,
+      "step": 8017
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0154280597063449e-06,
+      "loss": 0.0639,
+      "step": 8018
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.01390395105318e-06,
+      "loss": 0.0665,
+      "step": 8019
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0123809259815998e-06,
+      "loss": 0.0619,
+      "step": 8020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0108589846752614e-06,
+      "loss": 0.0677,
+      "step": 8021
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.009338127317686e-06,
+      "loss": 0.0708,
+      "step": 8022
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0078183540922616e-06,
+      "loss": 0.0691,
+      "step": 8023
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0062996651822543e-06,
+      "loss": 0.0693,
+      "step": 8024
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0047820607707903e-06,
+      "loss": 0.0719,
+      "step": 8025
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0032655410408687e-06,
+      "loss": 0.0717,
+      "step": 8026
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.00175010617536e-06,
+      "loss": 0.0648,
+      "step": 8027
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0002357563569997e-06,
+      "loss": 0.0698,
+      "step": 8028
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.987224917683968e-07,
+      "loss": 0.061,
+      "step": 8029
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.972103125920264e-07,
+      "loss": 0.072,
+      "step": 8030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95699219010231e-07,
+      "loss": 0.065,
+      "step": 8031
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.941892112052288e-07,
+      "loss": 0.0689,
+      "step": 8032
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.926802893591003e-07,
+      "loss": 0.0687,
+      "step": 8033
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.91172453653797e-07,
+      "loss": 0.0617,
+      "step": 8034
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.896657042711422e-07,
+      "loss": 0.0699,
+      "step": 8035
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.88160041392826e-07,
+      "loss": 0.0713,
+      "step": 8036
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.866554652004056e-07,
+      "loss": 0.0642,
+      "step": 8037
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.851519758753114e-07,
+      "loss": 0.0639,
+      "step": 8038
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.836495735988382e-07,
+      "loss": 0.0692,
+      "step": 8039
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.821482585521548e-07,
+      "loss": 0.0656,
+      "step": 8040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.806480309162958e-07,
+      "loss": 0.0628,
+      "step": 8041
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.791488908721614e-07,
+      "loss": 0.0665,
+      "step": 8042
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77650838600529e-07,
+      "loss": 0.0702,
+      "step": 8043
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.761538742820377e-07,
+      "loss": 0.0689,
+      "step": 8044
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.74657998097196e-07,
+      "loss": 0.0666,
+      "step": 8045
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.731632102263878e-07,
+      "loss": 0.0676,
+      "step": 8046
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.716695108498575e-07,
+      "loss": 0.0637,
+      "step": 8047
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.701769001477201e-07,
+      "loss": 0.0611,
+      "step": 8048
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.686853782999662e-07,
+      "loss": 0.0652,
+      "step": 8049
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.671949454864449e-07,
+      "loss": 0.0717,
+      "step": 8050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.657056018868827e-07,
+      "loss": 0.0678,
+      "step": 8051
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.642173476808681e-07,
+      "loss": 0.0681,
+      "step": 8052
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.627301830478618e-07,
+      "loss": 0.0688,
+      "step": 8053
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.61244108167194e-07,
+      "loss": 0.0675,
+      "step": 8054
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.597591232180614e-07,
+      "loss": 0.0693,
+      "step": 8055
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.58275228379526e-07,
+      "loss": 0.0658,
+      "step": 8056
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.567924238305281e-07,
+      "loss": 0.0648,
+      "step": 8057
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.55310709749867e-07,
+      "loss": 0.0666,
+      "step": 8058
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.538300863162143e-07,
+      "loss": 0.0655,
+      "step": 8059
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.523505537081113e-07,
+      "loss": 0.0642,
+      "step": 8060
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.508721121039643e-07,
+      "loss": 0.0602,
+      "step": 8061
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.49394761682051e-07,
+      "loss": 0.0678,
+      "step": 8062
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.479185026205173e-07,
+      "loss": 0.0676,
+      "step": 8063
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.464433350973757e-07,
+      "loss": 0.0646,
+      "step": 8064
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.44969259290509e-07,
+      "loss": 0.0655,
+      "step": 8065
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.434962753776677e-07,
+      "loss": 0.0673,
+      "step": 8066
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.420243835364684e-07,
+      "loss": 0.0697,
+      "step": 8067
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.405535839444013e-07,
+      "loss": 0.0661,
+      "step": 8068
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.390838767788202e-07,
+      "loss": 0.0669,
+      "step": 8069
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.376152622169454e-07,
+      "loss": 0.0669,
+      "step": 8070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.361477404358732e-07,
+      "loss": 0.0714,
+      "step": 8071
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.34681311612563e-07,
+      "loss": 0.0627,
+      "step": 8072
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.332159759238401e-07,
+      "loss": 0.0649,
+      "step": 8073
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.317517335464044e-07,
+      "loss": 0.0688,
+      "step": 8074
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.302885846568199e-07,
+      "loss": 0.0699,
+      "step": 8075
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.288265294315157e-07,
+      "loss": 0.0711,
+      "step": 8076
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.273655680467974e-07,
+      "loss": 0.0713,
+      "step": 8077
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.259057006788308e-07,
+      "loss": 0.0704,
+      "step": 8078
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.244469275036527e-07,
+      "loss": 0.0666,
+      "step": 8079
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.229892486971714e-07,
+      "loss": 0.0732,
+      "step": 8080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.215326644351563e-07,
+      "loss": 0.068,
+      "step": 8081
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.200771748932513e-07,
+      "loss": 0.0667,
+      "step": 8082
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.186227802469638e-07,
+      "loss": 0.0637,
+      "step": 8083
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.171694806716703e-07,
+      "loss": 0.0681,
+      "step": 8084
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.157172763426192e-07,
+      "loss": 0.0668,
+      "step": 8085
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.142661674349207e-07,
+      "loss": 0.0622,
+      "step": 8086
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.128161541235536e-07,
+      "loss": 0.0691,
+      "step": 8087
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.113672365833726e-07,
+      "loss": 0.0631,
+      "step": 8088
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.0991941498909e-07,
+      "loss": 0.0673,
+      "step": 8089
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.084726895152885e-07,
+      "loss": 0.0712,
+      "step": 8090
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.070270603364262e-07,
+      "loss": 0.0676,
+      "step": 8091
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.055825276268193e-07,
+      "loss": 0.0672,
+      "step": 8092
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.041390915606551e-07,
+      "loss": 0.0675,
+      "step": 8093
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.026967523119922e-07,
+      "loss": 0.0669,
+      "step": 8094
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.012555100547515e-07,
+      "loss": 0.0673,
+      "step": 8095
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.998153649627262e-07,
+      "loss": 0.0631,
+      "step": 8096
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.983763172095739e-07,
+      "loss": 0.0615,
+      "step": 8097
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.969383669688181e-07,
+      "loss": 0.0706,
+      "step": 8098
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.955015144138591e-07,
+      "loss": 0.0704,
+      "step": 8099
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.940657597179559e-07,
+      "loss": 0.0655,
+      "step": 8100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.926311030542345e-07,
+      "loss": 0.0666,
+      "step": 8101
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.911975445956978e-07,
+      "loss": 0.0701,
+      "step": 8102
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.897650845152061e-07,
+      "loss": 0.0624,
+      "step": 8103
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.883337229854916e-07,
+      "loss": 0.063,
+      "step": 8104
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.869034601791581e-07,
+      "loss": 0.0638,
+      "step": 8105
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.854742962686669e-07,
+      "loss": 0.068,
+      "step": 8106
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.840462314263575e-07,
+      "loss": 0.0663,
+      "step": 8107
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.826192658244303e-07,
+      "loss": 0.0677,
+      "step": 8108
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.811933996349531e-07,
+      "loss": 0.061,
+      "step": 8109
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.79768633029866e-07,
+      "loss": 0.0691,
+      "step": 8110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.783449661809729e-07,
+      "loss": 0.0677,
+      "step": 8111
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.769223992599418e-07,
+      "loss": 0.0721,
+      "step": 8112
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.755009324383168e-07,
+      "loss": 0.0689,
+      "step": 8113
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.740805658875018e-07,
+      "loss": 0.0674,
+      "step": 8114
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.726612997787697e-07,
+      "loss": 0.059,
+      "step": 8115
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.712431342832639e-07,
+      "loss": 0.0624,
+      "step": 8116
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.698260695719917e-07,
+      "loss": 0.0627,
+      "step": 8117
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.684101058158301e-07,
+      "loss": 0.0664,
+      "step": 8118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.669952431855199e-07,
+      "loss": 0.0718,
+      "step": 8119
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.655814818516717e-07,
+      "loss": 0.0684,
+      "step": 8120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.641688219847644e-07,
+      "loss": 0.0657,
+      "step": 8121
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.627572637551418e-07,
+      "loss": 0.0642,
+      "step": 8122
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.613468073330145e-07,
+      "loss": 0.0659,
+      "step": 8123
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.59937452888463e-07,
+      "loss": 0.0659,
+      "step": 8124
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.585292005914314e-07,
+      "loss": 0.0673,
+      "step": 8125
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.571220506117328e-07,
+      "loss": 0.064,
+      "step": 8126
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.557160031190503e-07,
+      "loss": 0.0699,
+      "step": 8127
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.543110582829272e-07,
+      "loss": 0.0643,
+      "step": 8128
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.529072162727802e-07,
+      "loss": 0.0604,
+      "step": 8129
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.515044772578907e-07,
+      "loss": 0.0605,
+      "step": 8130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.501028414074042e-07,
+      "loss": 0.0662,
+      "step": 8131
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.48702308890339e-07,
+      "loss": 0.0675,
+      "step": 8132
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.473028798755767e-07,
+      "loss": 0.062,
+      "step": 8133
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.459045545318633e-07,
+      "loss": 0.0593,
+      "step": 8134
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.445073330278197e-07,
+      "loss": 0.0668,
+      "step": 8135
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.431112155319254e-07,
+      "loss": 0.0779,
+      "step": 8136
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.417162022125302e-07,
+      "loss": 0.0728,
+      "step": 8137
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.403222932378518e-07,
+      "loss": 0.0698,
+      "step": 8138
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.389294887759724e-07,
+      "loss": 0.0719,
+      "step": 8139
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.375377889948455e-07,
+      "loss": 0.0651,
+      "step": 8140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.361471940622856e-07,
+      "loss": 0.0683,
+      "step": 8141
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.347577041459754e-07,
+      "loss": 0.0658,
+      "step": 8142
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.333693194134685e-07,
+      "loss": 0.0683,
+      "step": 8143
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.31982040032181e-07,
+      "loss": 0.0588,
+      "step": 8144
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.305958661693947e-07,
+      "loss": 0.0661,
+      "step": 8145
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.292107979922637e-07,
+      "loss": 0.0634,
+      "step": 8146
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.278268356678044e-07,
+      "loss": 0.0676,
+      "step": 8147
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.264439793628987e-07,
+      "loss": 0.0677,
+      "step": 8148
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.250622292443e-07,
+      "loss": 0.0621,
+      "step": 8149
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.236815854786251e-07,
+      "loss": 0.0635,
+      "step": 8150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.223020482323552e-07,
+      "loss": 0.0633,
+      "step": 8151
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.209236176718449e-07,
+      "loss": 0.0656,
+      "step": 8152
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.195462939633081e-07,
+      "loss": 0.0725,
+      "step": 8153
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.181700772728307e-07,
+      "loss": 0.0634,
+      "step": 8154
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.167949677663611e-07,
+      "loss": 0.0707,
+      "step": 8155
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.154209656097145e-07,
+      "loss": 0.0641,
+      "step": 8156
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.140480709685783e-07,
+      "loss": 0.0643,
+      "step": 8157
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.126762840084989e-07,
+      "loss": 0.071,
+      "step": 8158
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.113056048948908e-07,
+      "loss": 0.0667,
+      "step": 8159
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.099360337930395e-07,
+      "loss": 0.0729,
+      "step": 8160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.085675708680929e-07,
+      "loss": 0.065,
+      "step": 8161
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.072002162850635e-07,
+      "loss": 0.0622,
+      "step": 8162
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.058339702088358e-07,
+      "loss": 0.0699,
+      "step": 8163
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.04468832804155e-07,
+      "loss": 0.065,
+      "step": 8164
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.031048042356393e-07,
+      "loss": 0.0678,
+      "step": 8165
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.01741884667765e-07,
+      "loss": 0.0683,
+      "step": 8166
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.003800742648793e-07,
+      "loss": 0.0636,
+      "step": 8167
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.990193731911966e-07,
+      "loss": 0.0643,
+      "step": 8168
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.976597816107967e-07,
+      "loss": 0.0695,
+      "step": 8169
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.963012996876207e-07,
+      "loss": 0.0695,
+      "step": 8170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.949439275854842e-07,
+      "loss": 0.0671,
+      "step": 8171
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.935876654680641e-07,
+      "loss": 0.07,
+      "step": 8172
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.922325134989028e-07,
+      "loss": 0.0679,
+      "step": 8173
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.90878471841412e-07,
+      "loss": 0.0735,
+      "step": 8174
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.895255406588653e-07,
+      "loss": 0.0641,
+      "step": 8175
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.88173720114408e-07,
+      "loss": 0.0655,
+      "step": 8176
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.868230103710473e-07,
+      "loss": 0.0661,
+      "step": 8177
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.854734115916563e-07,
+      "loss": 0.0672,
+      "step": 8178
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.84124923938977e-07,
+      "loss": 0.064,
+      "step": 8179
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.827775475756149e-07,
+      "loss": 0.0689,
+      "step": 8180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.814312826640402e-07,
+      "loss": 0.0624,
+      "step": 8181
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.800861293665962e-07,
+      "loss": 0.0642,
+      "step": 8182
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.787420878454832e-07,
+      "loss": 0.0684,
+      "step": 8183
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.773991582627716e-07,
+      "loss": 0.0638,
+      "step": 8184
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.760573407804006e-07,
+      "loss": 0.0655,
+      "step": 8185
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.747166355601688e-07,
+      "loss": 0.0679,
+      "step": 8186
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.733770427637466e-07,
+      "loss": 0.0668,
+      "step": 8187
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.720385625526672e-07,
+      "loss": 0.0683,
+      "step": 8188
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.70701195088327e-07,
+      "loss": 0.0671,
+      "step": 8189
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.693649405319969e-07,
+      "loss": 0.0673,
+      "step": 8190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.680297990448049e-07,
+      "loss": 0.0685,
+      "step": 8191
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.666957707877476e-07,
+      "loss": 0.0654,
+      "step": 8192
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.653628559216896e-07,
+      "loss": 0.0635,
+      "step": 8193
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.640310546073593e-07,
+      "loss": 0.0661,
+      "step": 8194
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.62700367005349e-07,
+      "loss": 0.0656,
+      "step": 8195
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.613707932761205e-07,
+      "loss": 0.0671,
+      "step": 8196
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.600423335800001e-07,
+      "loss": 0.0732,
+      "step": 8197
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.58714988077175e-07,
+      "loss": 0.0679,
+      "step": 8198
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.573887569277071e-07,
+      "loss": 0.0688,
+      "step": 8199
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.560636402915145e-07,
+      "loss": 0.0612,
+      "step": 8200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.547396383283901e-07,
+      "loss": 0.0665,
+      "step": 8201
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.534167511979862e-07,
+      "loss": 0.0605,
+      "step": 8202
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.520949790598187e-07,
+      "loss": 0.0655,
+      "step": 8203
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.507743220732766e-07,
+      "loss": 0.0705,
+      "step": 8204
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.494547803976093e-07,
+      "loss": 0.0697,
+      "step": 8205
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.481363541919306e-07,
+      "loss": 0.0684,
+      "step": 8206
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.468190436152245e-07,
+      "loss": 0.0719,
+      "step": 8207
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.455028488263371e-07,
+      "loss": 0.065,
+      "step": 8208
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.441877699839794e-07,
+      "loss": 0.0712,
+      "step": 8209
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.428738072467323e-07,
+      "loss": 0.0642,
+      "step": 8210
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.415609607730345e-07,
+      "loss": 0.0694,
+      "step": 8211
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.402492307211995e-07,
+      "loss": 0.0697,
+      "step": 8212
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.389386172493996e-07,
+      "loss": 0.0675,
+      "step": 8213
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.376291205156705e-07,
+      "loss": 0.0695,
+      "step": 8214
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.363207406779227e-07,
+      "loss": 0.0679,
+      "step": 8215
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.350134778939233e-07,
+      "loss": 0.0725,
+      "step": 8216
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.337073323213051e-07,
+      "loss": 0.0682,
+      "step": 8217
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.324023041175732e-07,
+      "loss": 0.0696,
+      "step": 8218
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.310983934400917e-07,
+      "loss": 0.0635,
+      "step": 8219
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.297956004460904e-07,
+      "loss": 0.071,
+      "step": 8220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.28493925292667e-07,
+      "loss": 0.0638,
+      "step": 8221
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.271933681367827e-07,
+      "loss": 0.0668,
+      "step": 8222
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.25893929135264e-07,
+      "loss": 0.065,
+      "step": 8223
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.245956084448035e-07,
+      "loss": 0.0639,
+      "step": 8224
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.232984062219572e-07,
+      "loss": 0.0642,
+      "step": 8225
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.220023226231488e-07,
+      "loss": 0.0629,
+      "step": 8226
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.207073578046642e-07,
+      "loss": 0.0676,
+      "step": 8227
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.194135119226542e-07,
+      "loss": 0.0638,
+      "step": 8228
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.181207851331396e-07,
+      "loss": 0.0675,
+      "step": 8229
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.168291775920011e-07,
+      "loss": 0.0641,
+      "step": 8230
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.155386894549854e-07,
+      "loss": 0.0642,
+      "step": 8231
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.14249320877708e-07,
+      "loss": 0.0677,
+      "step": 8232
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.129610720156422e-07,
+      "loss": 0.0676,
+      "step": 8233
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.116739430241349e-07,
+      "loss": 0.067,
+      "step": 8234
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.103879340583908e-07,
+      "loss": 0.0636,
+      "step": 8235
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.091030452734826e-07,
+      "loss": 0.0647,
+      "step": 8236
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.078192768243486e-07,
+      "loss": 0.0634,
+      "step": 8237
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.065366288657916e-07,
+      "loss": 0.0634,
+      "step": 8238
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.052551015524756e-07,
+      "loss": 0.069,
+      "step": 8239
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.039746950389337e-07,
+      "loss": 0.0635,
+      "step": 8240
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.026954094795657e-07,
+      "loss": 0.0662,
+      "step": 8241
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.014172450286305e-07,
+      "loss": 0.0641,
+      "step": 8242
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.001402018402558e-07,
+      "loss": 0.0616,
+      "step": 8243
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.988642800684342e-07,
+      "loss": 0.0649,
+      "step": 8244
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.975894798670168e-07,
+      "loss": 0.066,
+      "step": 8245
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.963158013897298e-07,
+      "loss": 0.0644,
+      "step": 8246
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.950432447901568e-07,
+      "loss": 0.068,
+      "step": 8247
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.937718102217461e-07,
+      "loss": 0.065,
+      "step": 8248
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.925014978378153e-07,
+      "loss": 0.0636,
+      "step": 8249
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.91232307791544e-07,
+      "loss": 0.0629,
+      "step": 8250
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.899642402359741e-07,
+      "loss": 0.0653,
+      "step": 8251
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.886972953240178e-07,
+      "loss": 0.0673,
+      "step": 8252
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.874314732084453e-07,
+      "loss": 0.066,
+      "step": 8253
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.861667740418976e-07,
+      "loss": 0.0658,
+      "step": 8254
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.849031979768772e-07,
+      "loss": 0.0637,
+      "step": 8255
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.83640745165749e-07,
+      "loss": 0.0621,
+      "step": 8256
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.823794157607477e-07,
+      "loss": 0.0682,
+      "step": 8257
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.811192099139697e-07,
+      "loss": 0.0635,
+      "step": 8258
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.798601277773731e-07,
+      "loss": 0.0665,
+      "step": 8259
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.786021695027867e-07,
+      "loss": 0.0662,
+      "step": 8260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.77345335241898e-07,
+      "loss": 0.0655,
+      "step": 8261
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.760896251462623e-07,
+      "loss": 0.0649,
+      "step": 8262
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.748350393672986e-07,
+      "loss": 0.0593,
+      "step": 8263
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.735815780562893e-07,
+      "loss": 0.0676,
+      "step": 8264
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.723292413643845e-07,
+      "loss": 0.0704,
+      "step": 8265
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.710780294425956e-07,
+      "loss": 0.0624,
+      "step": 8266
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.698279424417953e-07,
+      "loss": 0.0711,
+      "step": 8267
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.685789805127286e-07,
+      "loss": 0.0677,
+      "step": 8268
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.673311438060004e-07,
+      "loss": 0.0628,
+      "step": 8269
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.660844324720761e-07,
+      "loss": 0.0715,
+      "step": 8270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.648388466612943e-07,
+      "loss": 0.0657,
+      "step": 8271
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.635943865238514e-07,
+      "loss": 0.0613,
+      "step": 8272
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.623510522098075e-07,
+      "loss": 0.0609,
+      "step": 8273
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.611088438690927e-07,
+      "loss": 0.0652,
+      "step": 8274
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.598677616514948e-07,
+      "loss": 0.0613,
+      "step": 8275
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.586278057066708e-07,
+      "loss": 0.0669,
+      "step": 8276
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.57388976184139e-07,
+      "loss": 0.0715,
+      "step": 8277
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.56151273233282e-07,
+      "loss": 0.0667,
+      "step": 8278
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.549146970033482e-07,
+      "loss": 0.067,
+      "step": 8279
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.536792476434506e-07,
+      "loss": 0.0679,
+      "step": 8280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.5244492530256e-07,
+      "loss": 0.07,
+      "step": 8281
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.512117301295229e-07,
+      "loss": 0.0657,
+      "step": 8282
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.499796622730382e-07,
+      "loss": 0.0648,
+      "step": 8283
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.487487218816746e-07,
+      "loss": 0.0659,
+      "step": 8284
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.475189091038669e-07,
+      "loss": 0.0734,
+      "step": 8285
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.462902240879099e-07,
+      "loss": 0.0741,
+      "step": 8286
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.450626669819604e-07,
+      "loss": 0.0631,
+      "step": 8287
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.438362379340468e-07,
+      "loss": 0.0716,
+      "step": 8288
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.426109370920541e-07,
+      "loss": 0.0687,
+      "step": 8289
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.413867646037375e-07,
+      "loss": 0.0648,
+      "step": 8290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.401637206167099e-07,
+      "loss": 0.0602,
+      "step": 8291
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.389418052784512e-07,
+      "loss": 0.0628,
+      "step": 8292
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.377210187363081e-07,
+      "loss": 0.0704,
+      "step": 8293
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.36501361137486e-07,
+      "loss": 0.0665,
+      "step": 8294
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.35282832629055e-07,
+      "loss": 0.0627,
+      "step": 8295
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.340654333579544e-07,
+      "loss": 0.068,
+      "step": 8296
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.328491634709799e-07,
+      "loss": 0.0813,
+      "step": 8297
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.316340231147955e-07,
+      "loss": 0.0674,
+      "step": 8298
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.304200124359284e-07,
+      "loss": 0.0611,
+      "step": 8299
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.29207131580768e-07,
+      "loss": 0.0715,
+      "step": 8300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.279953806955708e-07,
+      "loss": 0.0707,
+      "step": 8301
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.267847599264531e-07,
+      "loss": 0.071,
+      "step": 8302
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.255752694193961e-07,
+      "loss": 0.0711,
+      "step": 8303
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.243669093202487e-07,
+      "loss": 0.0653,
+      "step": 8304
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.231596797747163e-07,
+      "loss": 0.0705,
+      "step": 8305
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.219535809283728e-07,
+      "loss": 0.0676,
+      "step": 8306
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.20748612926656e-07,
+      "loss": 0.0666,
+      "step": 8307
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.195447759148654e-07,
+      "loss": 0.0655,
+      "step": 8308
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.183420700381637e-07,
+      "loss": 0.0703,
+      "step": 8309
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.171404954415794e-07,
+      "loss": 0.0652,
+      "step": 8310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.15940052270002e-07,
+      "loss": 0.059,
+      "step": 8311
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.147407406681893e-07,
+      "loss": 0.0679,
+      "step": 8312
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.135425607807576e-07,
+      "loss": 0.071,
+      "step": 8313
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.12345512752186e-07,
+      "loss": 0.0673,
+      "step": 8314
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.111495967268233e-07,
+      "loss": 0.0643,
+      "step": 8315
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.099548128488774e-07,
+      "loss": 0.0643,
+      "step": 8316
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.087611612624167e-07,
+      "loss": 0.0662,
+      "step": 8317
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.075686421113825e-07,
+      "loss": 0.0646,
+      "step": 8318
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.063772555395708e-07,
+      "loss": 0.0662,
+      "step": 8319
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.051870016906436e-07,
+      "loss": 0.0665,
+      "step": 8320
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.039978807081281e-07,
+      "loss": 0.07,
+      "step": 8321
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.028098927354109e-07,
+      "loss": 0.074,
+      "step": 8322
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.016230379157495e-07,
+      "loss": 0.0625,
+      "step": 8323
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.004373163922561e-07,
+      "loss": 0.0685,
+      "step": 8324
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.992527283079097e-07,
+      "loss": 0.0696,
+      "step": 8325
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.98069273805556e-07,
+      "loss": 0.0712,
+      "step": 8326
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.968869530278987e-07,
+      "loss": 0.0692,
+      "step": 8327
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.95705766117507e-07,
+      "loss": 0.0635,
+      "step": 8328
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.945257132168148e-07,
+      "loss": 0.0654,
+      "step": 8329
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.93346794468117e-07,
+      "loss": 0.07,
+      "step": 8330
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.921690100135713e-07,
+      "loss": 0.0703,
+      "step": 8331
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.909923599952028e-07,
+      "loss": 0.0656,
+      "step": 8332
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.898168445548946e-07,
+      "loss": 0.0639,
+      "step": 8333
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.886424638343979e-07,
+      "loss": 0.0671,
+      "step": 8334
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.874692179753216e-07,
+      "loss": 0.0692,
+      "step": 8335
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.862971071191414e-07,
+      "loss": 0.0628,
+      "step": 8336
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.851261314071977e-07,
+      "loss": 0.0639,
+      "step": 8337
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.839562909806884e-07,
+      "loss": 0.0714,
+      "step": 8338
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.827875859806797e-07,
+      "loss": 0.0676,
+      "step": 8339
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.81620016548099e-07,
+      "loss": 0.065,
+      "step": 8340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.804535828237367e-07,
+      "loss": 0.0628,
+      "step": 8341
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.792882849482451e-07,
+      "loss": 0.0682,
+      "step": 8342
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.781241230621426e-07,
+      "loss": 0.0622,
+      "step": 8343
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.769610973058082e-07,
+      "loss": 0.0674,
+      "step": 8344
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.757992078194819e-07,
+      "loss": 0.0653,
+      "step": 8345
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.746384547432738e-07,
+      "loss": 0.0678,
+      "step": 8346
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.734788382171485e-07,
+      "loss": 0.069,
+      "step": 8347
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.723203583809412e-07,
+      "loss": 0.0642,
+      "step": 8348
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.711630153743441e-07,
+      "loss": 0.0694,
+      "step": 8349
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.700068093369126e-07,
+      "loss": 0.0665,
+      "step": 8350
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.688517404080706e-07,
+      "loss": 0.0646,
+      "step": 8351
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.676978087271001e-07,
+      "loss": 0.0624,
+      "step": 8352
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.665450144331452e-07,
+      "loss": 0.0669,
+      "step": 8353
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.653933576652171e-07,
+      "loss": 0.0651,
+      "step": 8354
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.642428385621868e-07,
+      "loss": 0.0637,
+      "step": 8355
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.630934572627878e-07,
+      "loss": 0.0712,
+      "step": 8356
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.619452139056181e-07,
+      "loss": 0.0719,
+      "step": 8357
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.60798108629137e-07,
+      "loss": 0.072,
+      "step": 8358
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.596521415716694e-07,
+      "loss": 0.0672,
+      "step": 8359
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.585073128714003e-07,
+      "loss": 0.0639,
+      "step": 8360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.573636226663748e-07,
+      "loss": 0.0671,
+      "step": 8361
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.562210710945082e-07,
+      "loss": 0.0709,
+      "step": 8362
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.550796582935725e-07,
+      "loss": 0.0691,
+      "step": 8363
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.53939384401202e-07,
+      "loss": 0.0694,
+      "step": 8364
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.52800249554899e-07,
+      "loss": 0.0695,
+      "step": 8365
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.516622538920246e-07,
+      "loss": 0.0676,
+      "step": 8366
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.505253975498004e-07,
+      "loss": 0.0622,
+      "step": 8367
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.493896806653176e-07,
+      "loss": 0.0692,
+      "step": 8368
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.482551033755212e-07,
+      "loss": 0.0723,
+      "step": 8369
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.471216658172274e-07,
+      "loss": 0.0643,
+      "step": 8370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.4598936812711e-07,
+      "loss": 0.0624,
+      "step": 8371
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.448582104417033e-07,
+      "loss": 0.0643,
+      "step": 8372
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.437281928974114e-07,
+      "loss": 0.0673,
+      "step": 8373
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.425993156304943e-07,
+      "loss": 0.0587,
+      "step": 8374
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.414715787770753e-07,
+      "loss": 0.065,
+      "step": 8375
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.403449824731455e-07,
+      "loss": 0.0587,
+      "step": 8376
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.392195268545519e-07,
+      "loss": 0.0633,
+      "step": 8377
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.38095212057006e-07,
+      "loss": 0.069,
+      "step": 8378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.369720382160848e-07,
+      "loss": 0.062,
+      "step": 8379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.358500054672233e-07,
+      "loss": 0.0652,
+      "step": 8380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.347291139457244e-07,
+      "loss": 0.0693,
+      "step": 8381
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.336093637867468e-07,
+      "loss": 0.0675,
+      "step": 8382
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.324907551253144e-07,
+      "loss": 0.0634,
+      "step": 8383
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.313732880963163e-07,
+      "loss": 0.0642,
+      "step": 8384
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.302569628345e-07,
+      "loss": 0.0627,
+      "step": 8385
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.291417794744747e-07,
+      "loss": 0.0636,
+      "step": 8386
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.280277381507182e-07,
+      "loss": 0.0705,
+      "step": 8387
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.269148389975643e-07,
+      "loss": 0.0639,
+      "step": 8388
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.258030821492088e-07,
+      "loss": 0.06,
+      "step": 8389
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.24692467739717e-07,
+      "loss": 0.0653,
+      "step": 8390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.23582995903007e-07,
+      "loss": 0.0724,
+      "step": 8391
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.224746667728653e-07,
+      "loss": 0.0641,
+      "step": 8392
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.213674804829394e-07,
+      "loss": 0.069,
+      "step": 8393
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.20261437166737e-07,
+      "loss": 0.0713,
+      "step": 8394
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.191565369576324e-07,
+      "loss": 0.0643,
+      "step": 8395
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.180527799888568e-07,
+      "loss": 0.0684,
+      "step": 8396
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.16950166393505e-07,
+      "loss": 0.0705,
+      "step": 8397
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.158486963045383e-07,
+      "loss": 0.0618,
+      "step": 8398
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.147483698547739e-07,
+      "loss": 0.0644,
+      "step": 8399
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.136491871768934e-07,
+      "loss": 0.0646,
+      "step": 8400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.12551148403444e-07,
+      "loss": 0.0686,
+      "step": 8401
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.114542536668288e-07,
+      "loss": 0.0674,
+      "step": 8402
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.103585030993164e-07,
+      "loss": 0.0676,
+      "step": 8403
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.092638968330387e-07,
+      "loss": 0.067,
+      "step": 8404
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.081704349999861e-07,
+      "loss": 0.065,
+      "step": 8405
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.070781177320138e-07,
+      "loss": 0.0611,
+      "step": 8406
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.05986945160839e-07,
+      "loss": 0.0607,
+      "step": 8407
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.048969174180396e-07,
+      "loss": 0.0704,
+      "step": 8408
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.038080346350538e-07,
+      "loss": 0.0676,
+      "step": 8409
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.027202969431865e-07,
+      "loss": 0.0642,
+      "step": 8410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.016337044735986e-07,
+      "loss": 0.0672,
+      "step": 8411
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.005482573573184e-07,
+      "loss": 0.0697,
+      "step": 8412
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.994639557252335e-07,
+      "loss": 0.0598,
+      "step": 8413
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.983807997080925e-07,
+      "loss": 0.0674,
+      "step": 8414
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.972987894365078e-07,
+      "loss": 0.0662,
+      "step": 8415
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.962179250409537e-07,
+      "loss": 0.0698,
+      "step": 8416
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.951382066517618e-07,
+      "loss": 0.0631,
+      "step": 8417
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.940596343991344e-07,
+      "loss": 0.0693,
+      "step": 8418
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.929822084131264e-07,
+      "loss": 0.0708,
+      "step": 8419
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.919059288236582e-07,
+      "loss": 0.0643,
+      "step": 8420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.908307957605151e-07,
+      "loss": 0.074,
+      "step": 8421
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.897568093533378e-07,
+      "loss": 0.0654,
+      "step": 8422
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.886839697316359e-07,
+      "loss": 0.0671,
+      "step": 8423
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.876122770247748e-07,
+      "loss": 0.0755,
+      "step": 8424
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.865417313619825e-07,
+      "loss": 0.0677,
+      "step": 8425
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.854723328723521e-07,
+      "loss": 0.0645,
+      "step": 8426
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.844040816848361e-07,
+      "loss": 0.0685,
+      "step": 8427
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.833369779282471e-07,
+      "loss": 0.063,
+      "step": 8428
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.822710217312632e-07,
+      "loss": 0.0715,
+      "step": 8429
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.812062132224204e-07,
+      "loss": 0.0655,
+      "step": 8430
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.801425525301173e-07,
+      "loss": 0.0616,
+      "step": 8431
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.790800397826167e-07,
+      "loss": 0.0695,
+      "step": 8432
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.780186751080396e-07,
+      "loss": 0.0682,
+      "step": 8433
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.769584586343679e-07,
+      "loss": 0.0683,
+      "step": 8434
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7589939048945156e-07,
+      "loss": 0.0681,
+      "step": 8435
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7484147080099297e-07,
+      "loss": 0.0705,
+      "step": 8436
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.737846996965645e-07,
+      "loss": 0.0637,
+      "step": 8437
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.727290773035931e-07,
+      "loss": 0.0664,
+      "step": 8438
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7167460374937133e-07,
+      "loss": 0.068,
+      "step": 8439
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7062127916105203e-07,
+      "loss": 0.0682,
+      "step": 8440
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6956910366565023e-07,
+      "loss": 0.0742,
+      "step": 8441
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6851807739003996e-07,
+      "loss": 0.0672,
+      "step": 8442
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6746820046096097e-07,
+      "loss": 0.0707,
+      "step": 8443
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.664194730050098e-07,
+      "loss": 0.0658,
+      "step": 8444
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.653718951486463e-07,
+      "loss": 0.0662,
+      "step": 8445
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6432546701819405e-07,
+      "loss": 0.0648,
+      "step": 8446
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.632801887398319e-07,
+      "loss": 0.0618,
+      "step": 8447
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6223606043960924e-07,
+      "loss": 0.0659,
+      "step": 8448
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.611930822434274e-07,
+      "loss": 0.0701,
+      "step": 8449
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.601512542770537e-07,
+      "loss": 0.066,
+      "step": 8450
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.591105766661186e-07,
+      "loss": 0.0623,
+      "step": 8451
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.580710495361085e-07,
+      "loss": 0.0655,
+      "step": 8452
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5703267301237533e-07,
+      "loss": 0.0681,
+      "step": 8453
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.559954472201311e-07,
+      "loss": 0.0614,
+      "step": 8454
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549593722844492e-07,
+      "loss": 0.0647,
+      "step": 8455
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.539244483302607e-07,
+      "loss": 0.0655,
+      "step": 8456
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.528906754823659e-07,
+      "loss": 0.0678,
+      "step": 8457
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.518580538654183e-07,
+      "loss": 0.0673,
+      "step": 8458
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5082658360393826e-07,
+      "loss": 0.0589,
+      "step": 8459
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.497962648223031e-07,
+      "loss": 0.0624,
+      "step": 8460
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4876709764475223e-07,
+      "loss": 0.0675,
+      "step": 8461
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4773908219538976e-07,
+      "loss": 0.0665,
+      "step": 8462
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4671221859817536e-07,
+      "loss": 0.0678,
+      "step": 8463
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.456865069769334e-07,
+      "loss": 0.0698,
+      "step": 8464
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.446619474553504e-07,
+      "loss": 0.0625,
+      "step": 8465
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4363854015696983e-07,
+      "loss": 0.0641,
+      "step": 8466
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.426162852051985e-07,
+      "loss": 0.0697,
+      "step": 8467
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.415951827233056e-07,
+      "loss": 0.0662,
+      "step": 8468
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4057523283441815e-07,
+      "loss": 0.0708,
+      "step": 8469
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3955643566152896e-07,
+      "loss": 0.0643,
+      "step": 8470
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3853879132748746e-07,
+      "loss": 0.0706,
+      "step": 8471
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.375222999550044e-07,
+      "loss": 0.0659,
+      "step": 8472
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3650696166665397e-07,
+      "loss": 0.0628,
+      "step": 8473
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.354927765848693e-07,
+      "loss": 0.0631,
+      "step": 8474
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.344797448319449e-07,
+      "loss": 0.0631,
+      "step": 8475
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3346786653003845e-07,
+      "loss": 0.0665,
+      "step": 8476
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.324571418011636e-07,
+      "loss": 0.06,
+      "step": 8477
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3144757076719833e-07,
+      "loss": 0.0697,
+      "step": 8478
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.304391535498842e-07,
+      "loss": 0.0649,
+      "step": 8479
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.294318902708172e-07,
+      "loss": 0.0661,
+      "step": 8480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2842578105145696e-07,
+      "loss": 0.0582,
+      "step": 8481
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.274208260131274e-07,
+      "loss": 0.0639,
+      "step": 8482
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2641702527700723e-07,
+      "loss": 0.0641,
+      "step": 8483
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2541437896414186e-07,
+      "loss": 0.0614,
+      "step": 8484
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.244128871954323e-07,
+      "loss": 0.0672,
+      "step": 8485
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.234125500916431e-07,
+      "loss": 0.0695,
+      "step": 8486
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2241336777340103e-07,
+      "loss": 0.0728,
+      "step": 8487
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.214153403611898e-07,
+      "loss": 0.062,
+      "step": 8488
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2041846797535426e-07,
+      "loss": 0.0622,
+      "step": 8489
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1942275073610594e-07,
+      "loss": 0.0603,
+      "step": 8490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1842818876351e-07,
+      "loss": 0.0681,
+      "step": 8491
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.174347821774938e-07,
+      "loss": 0.0578,
+      "step": 8492
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1644253109784926e-07,
+      "loss": 0.0681,
+      "step": 8493
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.15451435644223e-07,
+      "loss": 0.0669,
+      "step": 8494
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1446149593612926e-07,
+      "loss": 0.0705,
+      "step": 8495
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1347271209293714e-07,
+      "loss": 0.0691,
+      "step": 8496
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.124850842338779e-07,
+      "loss": 0.0739,
+      "step": 8497
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.114986124780451e-07,
+      "loss": 0.0682,
+      "step": 8498
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.105132969443915e-07,
+      "loss": 0.0661,
+      "step": 8499
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0952913775172854e-07,
+      "loss": 0.0639,
+      "step": 8500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.085461350187336e-07,
+      "loss": 0.0677,
+      "step": 8501
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0756428886393975e-07,
+      "loss": 0.065,
+      "step": 8502
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0658359940573986e-07,
+      "loss": 0.063,
+      "step": 8503
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0560406676239504e-07,
+      "loss": 0.0646,
+      "step": 8504
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.046256910520152e-07,
+      "loss": 0.0696,
+      "step": 8505
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.036484723925826e-07,
+      "loss": 0.0698,
+      "step": 8506
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.02672410901932e-07,
+      "loss": 0.0597,
+      "step": 8507
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.016975066977602e-07,
+      "loss": 0.0635,
+      "step": 8508
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.007237598976266e-07,
+      "loss": 0.0689,
+      "step": 8509
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.997511706189494e-07,
+      "loss": 0.071,
+      "step": 8510
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.987797389790071e-07,
+      "loss": 0.0696,
+      "step": 8511
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.978094650949393e-07,
+      "loss": 0.0677,
+      "step": 8512
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9684034908374694e-07,
+      "loss": 0.0636,
+      "step": 8513
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9587239106228636e-07,
+      "loss": 0.0709,
+      "step": 8514
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9490559114728323e-07,
+      "loss": 0.0687,
+      "step": 8515
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.939399494553142e-07,
+      "loss": 0.0634,
+      "step": 8516
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.929754661028229e-07,
+      "loss": 0.061,
+      "step": 8517
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9201214120611066e-07,
+      "loss": 0.0747,
+      "step": 8518
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.910499748813368e-07,
+      "loss": 0.0594,
+      "step": 8519
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9008896724452627e-07,
+      "loss": 0.0717,
+      "step": 8520
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.891291184115609e-07,
+      "loss": 0.0676,
+      "step": 8521
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.881704284981824e-07,
+      "loss": 0.0636,
+      "step": 8522
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8721289761999404e-07,
+      "loss": 0.0671,
+      "step": 8523
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8625652589245997e-07,
+      "loss": 0.0678,
+      "step": 8524
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8530131343090026e-07,
+      "loss": 0.0632,
+      "step": 8525
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8434726035050275e-07,
+      "loss": 0.0643,
+      "step": 8526
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.833943667663087e-07,
+      "loss": 0.074,
+      "step": 8527
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.824426327932229e-07,
+      "loss": 0.0613,
+      "step": 8528
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8149205854601026e-07,
+      "loss": 0.0688,
+      "step": 8529
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.805426441392923e-07,
+      "loss": 0.0719,
+      "step": 8530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7959438968755646e-07,
+      "loss": 0.0704,
+      "step": 8531
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7864729530514563e-07,
+      "loss": 0.0648,
+      "step": 8532
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.777013611062641e-07,
+      "loss": 0.069,
+      "step": 8533
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.767565872049783e-07,
+      "loss": 0.0698,
+      "step": 8534
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7581297371521166e-07,
+      "loss": 0.0658,
+      "step": 8535
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.748705207507486e-07,
+      "loss": 0.0693,
+      "step": 8536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7392922842523713e-07,
+      "loss": 0.0657,
+      "step": 8537
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.729890968521799e-07,
+      "loss": 0.0711,
+      "step": 8538
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.720501261449405e-07,
+      "loss": 0.0712,
+      "step": 8539
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.711123164167474e-07,
+      "loss": 0.0662,
+      "step": 8540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7017566778068226e-07,
+      "loss": 0.066,
+      "step": 8541
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6924018034969364e-07,
+      "loss": 0.067,
+      "step": 8542
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.683058542365847e-07,
+      "loss": 0.0655,
+      "step": 8543
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6737268955401974e-07,
+      "loss": 0.0669,
+      "step": 8544
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6644068641452537e-07,
+      "loss": 0.0665,
+      "step": 8545
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6550984493048614e-07,
+      "loss": 0.0667,
+      "step": 8546
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.645801652141456e-07,
+      "loss": 0.0673,
+      "step": 8547
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.636516473776097e-07,
+      "loss": 0.0674,
+      "step": 8548
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.627242915328433e-07,
+      "loss": 0.072,
+      "step": 8549
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.617980977916691e-07,
+      "loss": 0.067,
+      "step": 8550
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6087306626577335e-07,
+      "loss": 0.0661,
+      "step": 8551
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5994919706669796e-07,
+      "loss": 0.0721,
+      "step": 8552
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5902649030585045e-07,
+      "loss": 0.0671,
+      "step": 8553
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.58104946094493e-07,
+      "loss": 0.0734,
+      "step": 8554
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5718456454374773e-07,
+      "loss": 0.0693,
+      "step": 8555
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.562653457646015e-07,
+      "loss": 0.0619,
+      "step": 8556
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5534728986789556e-07,
+      "loss": 0.0647,
+      "step": 8557
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5443039696433143e-07,
+      "loss": 0.0628,
+      "step": 8558
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5351466716447515e-07,
+      "loss": 0.0641,
+      "step": 8559
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.526001005787472e-07,
+      "loss": 0.0683,
+      "step": 8560
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5168669731743044e-07,
+      "loss": 0.0724,
+      "step": 8561
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5077445749066795e-07,
+      "loss": 0.0655,
+      "step": 8562
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4986338120845843e-07,
+      "loss": 0.0672,
+      "step": 8563
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4895346858066723e-07,
+      "loss": 0.063,
+      "step": 8564
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.480447197170134e-07,
+      "loss": 0.0662,
+      "step": 8565
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.47137134727078e-07,
+      "loss": 0.0699,
+      "step": 8566
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4623071372030135e-07,
+      "loss": 0.0616,
+      "step": 8567
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4532545680598384e-07,
+      "loss": 0.0678,
+      "step": 8568
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.444213640932847e-07,
+      "loss": 0.0699,
+      "step": 8569
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4351843569122355e-07,
+      "loss": 0.0663,
+      "step": 8570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4261667170867983e-07,
+      "loss": 0.061,
+      "step": 8571
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4171607225439106e-07,
+      "loss": 0.07,
+      "step": 8572
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.40816637436957e-07,
+      "loss": 0.0641,
+      "step": 8573
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3991836736483207e-07,
+      "loss": 0.0686,
+      "step": 8574
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.390212621463362e-07,
+      "loss": 0.0689,
+      "step": 8575
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3812532188964743e-07,
+      "loss": 0.0609,
+      "step": 8576
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.372305467027992e-07,
+      "loss": 0.0625,
+      "step": 8577
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.363369366936875e-07,
+      "loss": 0.0621,
+      "step": 8578
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3544449197006944e-07,
+      "loss": 0.0702,
+      "step": 8579
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.345532126395579e-07,
+      "loss": 0.07,
+      "step": 8580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3366309880962787e-07,
+      "loss": 0.063,
+      "step": 8581
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3277415058761476e-07,
+      "loss": 0.0714,
+      "step": 8582
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.318863680807072e-07,
+      "loss": 0.0685,
+      "step": 8583
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3099975139596285e-07,
+      "loss": 0.0665,
+      "step": 8584
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.301143006402918e-07,
+      "loss": 0.0624,
+      "step": 8585
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2923001592046424e-07,
+      "loss": 0.0658,
+      "step": 8586
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.283468973431125e-07,
+      "loss": 0.0746,
+      "step": 8587
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.27464945014726e-07,
+      "loss": 0.0651,
+      "step": 8588
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2658415904165407e-07,
+      "loss": 0.0714,
+      "step": 8589
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2570453953010616e-07,
+      "loss": 0.0643,
+      "step": 8590
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.248260865861519e-07,
+      "loss": 0.0664,
+      "step": 8591
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.239488003157143e-07,
+      "loss": 0.0651,
+      "step": 8592
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2307268082458544e-07,
+      "loss": 0.066,
+      "step": 8593
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2219772821840853e-07,
+      "loss": 0.0659,
+      "step": 8594
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.213239426026915e-07,
+      "loss": 0.0652,
+      "step": 8595
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2045132408279665e-07,
+      "loss": 0.0684,
+      "step": 8596
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1957987276394766e-07,
+      "loss": 0.068,
+      "step": 8597
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1870958875123147e-07,
+      "loss": 0.065,
+      "step": 8598
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1784047214958756e-07,
+      "loss": 0.0636,
+      "step": 8599
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1697252306381653e-07,
+      "loss": 0.0745,
+      "step": 8600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.161057415985835e-07,
+      "loss": 0.0673,
+      "step": 8601
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1524012785840605e-07,
+      "loss": 0.0699,
+      "step": 8602
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.143756819476629e-07,
+      "loss": 0.0657,
+      "step": 8603
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.135124039705939e-07,
+      "loss": 0.0681,
+      "step": 8604
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.126502940312959e-07,
+      "loss": 0.068,
+      "step": 8605
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.117893522337279e-07,
+      "loss": 0.0689,
+      "step": 8606
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.109295786817046e-07,
+      "loss": 0.0637,
+      "step": 8607
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1007097347889867e-07,
+      "loss": 0.0644,
+      "step": 8608
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0921353672884847e-07,
+      "loss": 0.0647,
+      "step": 8609
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0835726853494565e-07,
+      "loss": 0.0656,
+      "step": 8610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0750216900044093e-07,
+      "loss": 0.0656,
+      "step": 8611
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0664823822844857e-07,
+      "loss": 0.0636,
+      "step": 8612
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.057954763219384e-07,
+      "loss": 0.0654,
+      "step": 8613
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0494388338373815e-07,
+      "loss": 0.0658,
+      "step": 8614
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0409345951653903e-07,
+      "loss": 0.0674,
+      "step": 8615
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0324420482288675e-07,
+      "loss": 0.07,
+      "step": 8616
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0239611940519056e-07,
+      "loss": 0.0671,
+      "step": 8617
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.015492033657141e-07,
+      "loss": 0.0681,
+      "step": 8618
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0070345680658255e-07,
+      "loss": 0.0719,
+      "step": 8619
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.998588798297797e-07,
+      "loss": 0.0619,
+      "step": 8620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.990154725371497e-07,
+      "loss": 0.0656,
+      "step": 8621
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9817323503039007e-07,
+      "loss": 0.0631,
+      "step": 8622
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.973321674110663e-07,
+      "loss": 0.0654,
+      "step": 8623
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.964922697805961e-07,
+      "loss": 0.0693,
+      "step": 8624
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.956535422402551e-07,
+      "loss": 0.0643,
+      "step": 8625
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.948159848911847e-07,
+      "loss": 0.0699,
+      "step": 8626
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9397959783437956e-07,
+      "loss": 0.0631,
+      "step": 8627
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9314438117069355e-07,
+      "loss": 0.0713,
+      "step": 8628
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.923103350008427e-07,
+      "loss": 0.0682,
+      "step": 8629
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9147745942539774e-07,
+      "loss": 0.0678,
+      "step": 8630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.906457545447927e-07,
+      "loss": 0.0668,
+      "step": 8631
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8981522045931633e-07,
+      "loss": 0.0656,
+      "step": 8632
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.889858572691173e-07,
+      "loss": 0.0702,
+      "step": 8633
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.881576650742068e-07,
+      "loss": 0.0661,
+      "step": 8634
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8733064397444944e-07,
+      "loss": 0.0665,
+      "step": 8635
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.865047940695709e-07,
+      "loss": 0.0659,
+      "step": 8636
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8568011545915596e-07,
+      "loss": 0.0628,
+      "step": 8637
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8485660824264847e-07,
+      "loss": 0.0683,
+      "step": 8638
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.840342725193501e-07,
+      "loss": 0.0642,
+      "step": 8639
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.832131083884215e-07,
+      "loss": 0.067,
+      "step": 8640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.823931159488813e-07,
+      "loss": 0.0637,
+      "step": 8641
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8157429529960923e-07,
+      "loss": 0.0674,
+      "step": 8642
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8075664653934076e-07,
+      "loss": 0.0685,
+      "step": 8643
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.799401697666726e-07,
+      "loss": 0.0672,
+      "step": 8644
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.791248650800582e-07,
+      "loss": 0.061,
+      "step": 8645
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7831073257781117e-07,
+      "loss": 0.0717,
+      "step": 8646
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.774977723581018e-07,
+      "loss": 0.0709,
+      "step": 8647
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.766859845189618e-07,
+      "loss": 0.0658,
+      "step": 8648
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7587536915827937e-07,
+      "loss": 0.0713,
+      "step": 8649
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.750659263738009e-07,
+      "loss": 0.0629,
+      "step": 8650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7425765626313496e-07,
+      "loss": 0.0631,
+      "step": 8651
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.734505589237424e-07,
+      "loss": 0.0694,
+      "step": 8652
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.72644634452951e-07,
+      "loss": 0.062,
+      "step": 8653
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7183988294793964e-07,
+      "loss": 0.074,
+      "step": 8654
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7103630450574734e-07,
+      "loss": 0.0604,
+      "step": 8655
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.702338992232767e-07,
+      "loss": 0.0649,
+      "step": 8656
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6943266719728354e-07,
+      "loss": 0.0707,
+      "step": 8657
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.686326085243829e-07,
+      "loss": 0.0687,
+      "step": 8658
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6783372330105085e-07,
+      "loss": 0.0645,
+      "step": 8659
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6703601162361924e-07,
+      "loss": 0.0672,
+      "step": 8660
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6623947358827786e-07,
+      "loss": 0.0648,
+      "step": 8661
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6544410929108086e-07,
+      "loss": 0.0644,
+      "step": 8662
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.646499188279328e-07,
+      "loss": 0.0603,
+      "step": 8663
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.638569022946036e-07,
+      "loss": 0.0662,
+      "step": 8664
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.630650597867168e-07,
+      "loss": 0.0658,
+      "step": 8665
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6227439139975495e-07,
+      "loss": 0.0651,
+      "step": 8666
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.614848972290629e-07,
+      "loss": 0.065,
+      "step": 8667
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6069657736984e-07,
+      "loss": 0.0621,
+      "step": 8668
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5990943191714467e-07,
+      "loss": 0.0667,
+      "step": 8669
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5912346096589545e-07,
+      "loss": 0.0582,
+      "step": 8670
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5833866461086655e-07,
+      "loss": 0.0662,
+      "step": 8671
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.575550429466922e-07,
+      "loss": 0.0696,
+      "step": 8672
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5677259606786686e-07,
+      "loss": 0.0641,
+      "step": 8673
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.559913240687384e-07,
+      "loss": 0.0667,
+      "step": 8674
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.552112270435181e-07,
+      "loss": 0.0645,
+      "step": 8675
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5443230508627293e-07,
+      "loss": 0.0722,
+      "step": 8676
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.536545582909278e-07,
+      "loss": 0.0617,
+      "step": 8677
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5287798675126764e-07,
+      "loss": 0.0665,
+      "step": 8678
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5210259056093535e-07,
+      "loss": 0.066,
+      "step": 8679
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.513283698134283e-07,
+      "loss": 0.0695,
+      "step": 8680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5055532460210974e-07,
+      "loss": 0.0691,
+      "step": 8681
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.497834550201927e-07,
+      "loss": 0.0687,
+      "step": 8682
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4901276116075514e-07,
+      "loss": 0.0645,
+      "step": 8683
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.482432431167292e-07,
+      "loss": 0.0626,
+      "step": 8684
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4747490098090853e-07,
+      "loss": 0.0673,
+      "step": 8685
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.467077348459401e-07,
+      "loss": 0.0653,
+      "step": 8686
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.459417448043344e-07,
+      "loss": 0.0664,
+      "step": 8687
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.451769309484564e-07,
+      "loss": 0.0702,
+      "step": 8688
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.444132933705323e-07,
+      "loss": 0.0699,
+      "step": 8689
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4365083216264297e-07,
+      "loss": 0.0654,
+      "step": 8690
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4288954741673033e-07,
+      "loss": 0.0653,
+      "step": 8691
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.421294392245932e-07,
+      "loss": 0.0659,
+      "step": 8692
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.413705076778883e-07,
+      "loss": 0.0641,
+      "step": 8693
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.406127528681301e-07,
+      "loss": 0.0629,
+      "step": 8694
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3985617488669344e-07,
+      "loss": 0.0665,
+      "step": 8695
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3910077382480856e-07,
+      "loss": 0.0704,
+      "step": 8696
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3834654977356487e-07,
+      "loss": 0.0642,
+      "step": 8697
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.375935028239118e-07,
+      "loss": 0.065,
+      "step": 8698
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.368416330666523e-07,
+      "loss": 0.0684,
+      "step": 8699
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3609094059245162e-07,
+      "loss": 0.0696,
+      "step": 8700
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3534142549183182e-07,
+      "loss": 0.0683,
+      "step": 8701
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3459308785517054e-07,
+      "loss": 0.0644,
+      "step": 8702
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3384592777270786e-07,
+      "loss": 0.0638,
+      "step": 8703
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.330999453345384e-07,
+      "loss": 0.0629,
+      "step": 8704
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.323551406306146e-07,
+      "loss": 0.0669,
+      "step": 8705
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.316115137507513e-07,
+      "loss": 0.061,
+      "step": 8706
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.308690647846157e-07,
+      "loss": 0.0751,
+      "step": 8707
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.30127793821735e-07,
+      "loss": 0.0678,
+      "step": 8708
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2938770095149777e-07,
+      "loss": 0.0615,
+      "step": 8709
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.286487862631448e-07,
+      "loss": 0.0649,
+      "step": 8710
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2791104984577817e-07,
+      "loss": 0.062,
+      "step": 8711
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2717449178835892e-07,
+      "loss": 0.0677,
+      "step": 8712
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.264391121797016e-07,
+      "loss": 0.0695,
+      "step": 8713
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2570491110848413e-07,
+      "loss": 0.066,
+      "step": 8714
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2497188866323683e-07,
+      "loss": 0.0736,
+      "step": 8715
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.242400449323523e-07,
+      "loss": 0.0675,
+      "step": 8716
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2350938000407997e-07,
+      "loss": 0.0601,
+      "step": 8717
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.22779893966526e-07,
+      "loss": 0.0727,
+      "step": 8718
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.220515869076534e-07,
+      "loss": 0.0624,
+      "step": 8719
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2132445891528742e-07,
+      "loss": 0.074,
+      "step": 8720
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2059851007710464e-07,
+      "loss": 0.0644,
+      "step": 8721
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1987374048064613e-07,
+      "loss": 0.0688,
+      "step": 8722
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1915015021330644e-07,
+      "loss": 0.0659,
+      "step": 8723
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1842773936233797e-07,
+      "loss": 0.0701,
+      "step": 8724
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.177065080148555e-07,
+      "loss": 0.0699,
+      "step": 8725
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.16986456257825e-07,
+      "loss": 0.0655,
+      "step": 8726
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.162675841780737e-07,
+      "loss": 0.0695,
+      "step": 8727
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.155498918622878e-07,
+      "loss": 0.0695,
+      "step": 8728
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1483337939700921e-07,
+      "loss": 0.0656,
+      "step": 8729
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1411804686863657e-07,
+      "loss": 0.0606,
+      "step": 8730
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.134038943634298e-07,
+      "loss": 0.0692,
+      "step": 8731
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.126909219675033e-07,
+      "loss": 0.0631,
+      "step": 8732
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1197912976682945e-07,
+      "loss": 0.0665,
+      "step": 8733
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1126851784724066e-07,
+      "loss": 0.067,
+      "step": 8734
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1055908629442401e-07,
+      "loss": 0.0627,
+      "step": 8735
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.098508351939288e-07,
+      "loss": 0.0672,
+      "step": 8736
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.091437646311567e-07,
+      "loss": 0.0644,
+      "step": 8737
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0843787469136735e-07,
+      "loss": 0.0621,
+      "step": 8738
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.077331654596848e-07,
+      "loss": 0.0597,
+      "step": 8739
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0702963702108225e-07,
+      "loss": 0.0639,
+      "step": 8740
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0632728946039293e-07,
+      "loss": 0.0678,
+      "step": 8741
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0562612286231354e-07,
+      "loss": 0.0713,
+      "step": 8742
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0492613731139087e-07,
+      "loss": 0.0704,
+      "step": 8743
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.042273328920319e-07,
+      "loss": 0.0638,
+      "step": 8744
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.035297096885025e-07,
+      "loss": 0.0684,
+      "step": 8745
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.028332677849254e-07,
+      "loss": 0.0674,
+      "step": 8746
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0213800726527789e-07,
+      "loss": 0.0612,
+      "step": 8747
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0144392821340065e-07,
+      "loss": 0.0609,
+      "step": 8748
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0075103071298784e-07,
+      "loss": 0.069,
+      "step": 8749
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0005931484758935e-07,
+      "loss": 0.0651,
+      "step": 8750
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.993687807006195e-07,
+      "loss": 0.0618,
+      "step": 8751
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.986794283553417e-07,
+      "loss": 0.0622,
+      "step": 8752
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9799125789488392e-07,
+      "loss": 0.066,
+      "step": 8753
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9730426940222757e-07,
+      "loss": 0.0657,
+      "step": 8754
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9661846296021193e-07,
+      "loss": 0.0683,
+      "step": 8755
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9593383865153637e-07,
+      "loss": 0.0647,
+      "step": 8756
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9525039655875377e-07,
+      "loss": 0.0636,
+      "step": 8757
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.945681367642771e-07,
+      "loss": 0.0667,
+      "step": 8758
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9388705935037833e-07,
+      "loss": 0.0631,
+      "step": 8759
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9320716439918175e-07,
+      "loss": 0.0627,
+      "step": 8760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9252845199267178e-07,
+      "loss": 0.0634,
+      "step": 8761
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9185092221269296e-07,
+      "loss": 0.0676,
+      "step": 8762
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9117457514094216e-07,
+      "loss": 0.0667,
+      "step": 8763
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.904994108589786e-07,
+      "loss": 0.0624,
+      "step": 8764
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.898254294482149e-07,
+      "loss": 0.063,
+      "step": 8765
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8915263098992164e-07,
+      "loss": 0.0641,
+      "step": 8766
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8848101556523057e-07,
+      "loss": 0.0643,
+      "step": 8767
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8781058325512692e-07,
+      "loss": 0.0681,
+      "step": 8768
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8714133414045155e-07,
+      "loss": 0.0651,
+      "step": 8769
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8647326830190882e-07,
+      "loss": 0.071,
+      "step": 8770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.858063858200565e-07,
+      "loss": 0.0731,
+      "step": 8771
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8514068677530805e-07,
+      "loss": 0.0627,
+      "step": 8772
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8447617124793814e-07,
+      "loss": 0.0667,
+      "step": 8773
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8381283931807604e-07,
+      "loss": 0.0661,
+      "step": 8774
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8315069106570772e-07,
+      "loss": 0.0633,
+      "step": 8775
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8248972657068154e-07,
+      "loss": 0.066,
+      "step": 8776
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.81829945912696e-07,
+      "loss": 0.0658,
+      "step": 8777
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8117134917131185e-07,
+      "loss": 0.0648,
+      "step": 8778
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.805139364259456e-07,
+      "loss": 0.0607,
+      "step": 8779
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7985770775587051e-07,
+      "loss": 0.0663,
+      "step": 8780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7920266324021663e-07,
+      "loss": 0.0702,
+      "step": 8781
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7854880295797406e-07,
+      "loss": 0.0621,
+      "step": 8782
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.778961269879842e-07,
+      "loss": 0.0629,
+      "step": 8783
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.772446354089541e-07,
+      "loss": 0.0626,
+      "step": 8784
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.765943282994409e-07,
+      "loss": 0.0682,
+      "step": 8785
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7594520573786078e-07,
+      "loss": 0.0658,
+      "step": 8786
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7529726780249e-07,
+      "loss": 0.0667,
+      "step": 8787
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7465051457145721e-07,
+      "loss": 0.0657,
+      "step": 8788
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7400494612275222e-07,
+      "loss": 0.0714,
+      "step": 8789
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7336056253422052e-07,
+      "loss": 0.0638,
+      "step": 8790
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.727173638835633e-07,
+      "loss": 0.0663,
+      "step": 8791
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7207535024834186e-07,
+      "loss": 0.0614,
+      "step": 8792
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7143452170597097e-07,
+      "loss": 0.0664,
+      "step": 8793
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7079487833372544e-07,
+      "loss": 0.067,
+      "step": 8794
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7015642020873803e-07,
+      "loss": 0.0606,
+      "step": 8795
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.695191474079938e-07,
+      "loss": 0.0705,
+      "step": 8796
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6888306000833798e-07,
+      "loss": 0.062,
+      "step": 8797
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6824815808647476e-07,
+      "loss": 0.0636,
+      "step": 8798
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.676144417189618e-07,
+      "loss": 0.0671,
+      "step": 8799
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6698191098221572e-07,
+      "loss": 0.0658,
+      "step": 8800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6635056595250998e-07,
+      "loss": 0.0611,
+      "step": 8801
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6572040670597366e-07,
+      "loss": 0.0653,
+      "step": 8802
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.650914333185971e-07,
+      "loss": 0.0683,
+      "step": 8803
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6446364586622078e-07,
+      "loss": 0.0677,
+      "step": 8804
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.638370444245474e-07,
+      "loss": 0.0581,
+      "step": 8805
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6321162906913657e-07,
+      "loss": 0.0668,
+      "step": 8806
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.625873998754024e-07,
+      "loss": 0.0648,
+      "step": 8807
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.619643569186169e-07,
+      "loss": 0.0708,
+      "step": 8808
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6134250027390997e-07,
+      "loss": 0.0719,
+      "step": 8809
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6072183001626607e-07,
+      "loss": 0.0681,
+      "step": 8810
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.601023462205298e-07,
+      "loss": 0.0671,
+      "step": 8811
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5948404896140134e-07,
+      "loss": 0.0637,
+      "step": 8812
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5886693831343557e-07,
+      "loss": 0.0609,
+      "step": 8813
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.582510143510496e-07,
+      "loss": 0.0669,
+      "step": 8814
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5763627714851182e-07,
+      "loss": 0.0666,
+      "step": 8815
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5702272677994955e-07,
+      "loss": 0.066,
+      "step": 8816
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.564103633193492e-07,
+      "loss": 0.0603,
+      "step": 8817
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5579918684055062e-07,
+      "loss": 0.065,
+      "step": 8818
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5518919741725147e-07,
+      "loss": 0.0667,
+      "step": 8819
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5458039512300848e-07,
+      "loss": 0.0672,
+      "step": 8820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5397278003123407e-07,
+      "loss": 0.0646,
+      "step": 8821
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5336635221519402e-07,
+      "loss": 0.0661,
+      "step": 8822
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5276111174801766e-07,
+      "loss": 0.0681,
+      "step": 8823
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5215705870268437e-07,
+      "loss": 0.064,
+      "step": 8824
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.515541931520359e-07,
+      "loss": 0.0636,
+      "step": 8825
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.509525151687663e-07,
+      "loss": 0.0671,
+      "step": 8826
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5035202482542977e-07,
+      "loss": 0.0613,
+      "step": 8827
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4975272219443615e-07,
+      "loss": 0.0648,
+      "step": 8828
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.49154607348051e-07,
+      "loss": 0.0699,
+      "step": 8829
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.485576803583977e-07,
+      "loss": 0.0672,
+      "step": 8830
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4796194129745643e-07,
+      "loss": 0.0675,
+      "step": 8831
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4736739023706426e-07,
+      "loss": 0.0633,
+      "step": 8832
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4677402724891266e-07,
+      "loss": 0.0694,
+      "step": 8833
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.461818524045555e-07,
+      "loss": 0.065,
+      "step": 8834
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4559086577539684e-07,
+      "loss": 0.0645,
+      "step": 8835
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4500106743270183e-07,
+      "loss": 0.0685,
+      "step": 8836
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4441245744759026e-07,
+      "loss": 0.0645,
+      "step": 8837
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438250358910387e-07,
+      "loss": 0.058,
+      "step": 8838
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.432388028338816e-07,
+      "loss": 0.0673,
+      "step": 8839
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4265375834681017e-07,
+      "loss": 0.0662,
+      "step": 8840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4206990250037022e-07,
+      "loss": 0.066,
+      "step": 8841
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4148723536496543e-07,
+      "loss": 0.0676,
+      "step": 8842
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.409057570108585e-07,
+      "loss": 0.0658,
+      "step": 8843
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4032546750816334e-07,
+      "loss": 0.0586,
+      "step": 8844
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3974636692685616e-07,
+      "loss": 0.0651,
+      "step": 8845
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3916845533676558e-07,
+      "loss": 0.0646,
+      "step": 8846
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.385917328075803e-07,
+      "loss": 0.0649,
+      "step": 8847
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3801619940884358e-07,
+      "loss": 0.0646,
+      "step": 8848
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3744185520995546e-07,
+      "loss": 0.0651,
+      "step": 8849
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3686870028017164e-07,
+      "loss": 0.0716,
+      "step": 8850
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3629673468860794e-07,
+      "loss": 0.0701,
+      "step": 8851
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3572595850423253e-07,
+      "loss": 0.071,
+      "step": 8852
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.351563717958726e-07,
+      "loss": 0.068,
+      "step": 8853
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3458797463221208e-07,
+      "loss": 0.0761,
+      "step": 8854
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3402076708178834e-07,
+      "loss": 0.0692,
+      "step": 8855
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3345474921300006e-07,
+      "loss": 0.0668,
+      "step": 8856
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3288992109409927e-07,
+      "loss": 0.0719,
+      "step": 8857
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3232628279319594e-07,
+      "loss": 0.0649,
+      "step": 8858
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3176383437825457e-07,
+      "loss": 0.0699,
+      "step": 8859
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3120257591709873e-07,
+      "loss": 0.068,
+      "step": 8860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.306425074774076e-07,
+      "loss": 0.0756,
+      "step": 8861
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3008362912671602e-07,
+      "loss": 0.063,
+      "step": 8862
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2952594093241454e-07,
+      "loss": 0.0695,
+      "step": 8863
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2896944296175386e-07,
+      "loss": 0.0664,
+      "step": 8864
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.284141352818391e-07,
+      "loss": 0.0653,
+      "step": 8865
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.27860017959629e-07,
+      "loss": 0.0688,
+      "step": 8866
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.273070910619434e-07,
+      "loss": 0.0589,
+      "step": 8867
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2675535465545452e-07,
+      "loss": 0.0624,
+      "step": 8868
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2620480880669695e-07,
+      "loss": 0.063,
+      "step": 8869
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2565545358205423e-07,
+      "loss": 0.0662,
+      "step": 8870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2510728904777115e-07,
+      "loss": 0.0666,
+      "step": 8871
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2456031526994816e-07,
+      "loss": 0.0666,
+      "step": 8872
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.240145323145414e-07,
+      "loss": 0.0649,
+      "step": 8873
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2346994024736268e-07,
+      "loss": 0.0652,
+      "step": 8874
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2292653913408393e-07,
+      "loss": 0.0645,
+      "step": 8875
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2238432904022824e-07,
+      "loss": 0.0682,
+      "step": 8876
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2184331003117666e-07,
+      "loss": 0.0591,
+      "step": 8877
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2130348217217036e-07,
+      "loss": 0.0643,
+      "step": 8878
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2076484552830393e-07,
+      "loss": 0.0619,
+      "step": 8879
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2022740016452538e-07,
+      "loss": 0.0659,
+      "step": 8880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1969114614564627e-07,
+      "loss": 0.0601,
+      "step": 8881
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1915608353632702e-07,
+      "loss": 0.064,
+      "step": 8882
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.186222124010894e-07,
+      "loss": 0.0663,
+      "step": 8883
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1808953280430968e-07,
+      "loss": 0.0713,
+      "step": 8884
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1755804481021982e-07,
+      "loss": 0.0697,
+      "step": 8885
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1702774848291076e-07,
+      "loss": 0.0644,
+      "step": 8886
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1649864388632692e-07,
+      "loss": 0.0679,
+      "step": 8887
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1597073108426837e-07,
+      "loss": 0.074,
+      "step": 8888
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.154440101403953e-07,
+      "loss": 0.068,
+      "step": 8889
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1491848111822024e-07,
+      "loss": 0.0706,
+      "step": 8890
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1439414408111471e-07,
+      "loss": 0.0657,
+      "step": 8891
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1387099909230593e-07,
+      "loss": 0.0631,
+      "step": 8892
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1334904621487564e-07,
+      "loss": 0.0672,
+      "step": 8893
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1282828551176461e-07,
+      "loss": 0.0597,
+      "step": 8894
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1230871704576707e-07,
+      "loss": 0.0633,
+      "step": 8895
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1179034087953622e-07,
+      "loss": 0.0568,
+      "step": 8896
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1127315707557873e-07,
+      "loss": 0.069,
+      "step": 8897
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1075716569625916e-07,
+      "loss": 0.0687,
+      "step": 8898
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1024236680379774e-07,
+      "loss": 0.0603,
+      "step": 8899
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0972876046027148e-07,
+      "loss": 0.0651,
+      "step": 8900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0921634672761306e-07,
+      "loss": 0.0684,
+      "step": 8901
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0870512566761194e-07,
+      "loss": 0.0664,
+      "step": 8902
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0819509734191324e-07,
+      "loss": 0.0678,
+      "step": 8903
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.076862618120167e-07,
+      "loss": 0.0554,
+      "step": 8904
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0717861913928207e-07,
+      "loss": 0.0681,
+      "step": 8905
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0667216938492264e-07,
+      "loss": 0.066,
+      "step": 8906
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0616691261000733e-07,
+      "loss": 0.0664,
+      "step": 8907
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0566284887546296e-07,
+      "loss": 0.0648,
+      "step": 8908
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0515997824207091e-07,
+      "loss": 0.0746,
+      "step": 8909
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0465830077047046e-07,
+      "loss": 0.0678,
+      "step": 8910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0415781652115431e-07,
+      "loss": 0.06,
+      "step": 8911
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0365852555447642e-07,
+      "loss": 0.0589,
+      "step": 8912
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0316042793063863e-07,
+      "loss": 0.0628,
+      "step": 8913
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0266352370970734e-07,
+      "loss": 0.0592,
+      "step": 8914
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0216781295160128e-07,
+      "loss": 0.0616,
+      "step": 8915
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0167329571609264e-07,
+      "loss": 0.0681,
+      "step": 8916
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0117997206281483e-07,
+      "loss": 0.0703,
+      "step": 8917
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.006878420512536e-07,
+      "loss": 0.0654,
+      "step": 8918
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0019690574075258e-07,
+      "loss": 0.0637,
+      "step": 8919
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.97071631905111e-08,
+      "loss": 0.0667,
+      "step": 8920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.921861445958525e-08,
+      "loss": 0.0736,
+      "step": 8921
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.873125960688346e-08,
+      "loss": 0.0641,
+      "step": 8922
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.824509869117649e-08,
+      "loss": 0.0628,
+      "step": 8923
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.776013177108523e-08,
+      "loss": 0.0644,
+      "step": 8924
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.727635890508957e-08,
+      "loss": 0.069,
+      "step": 8925
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.679378015152618e-08,
+      "loss": 0.0636,
+      "step": 8926
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.631239556858519e-08,
+      "loss": 0.0608,
+      "step": 8927
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.583220521431458e-08,
+      "loss": 0.06,
+      "step": 8928
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.535320914661694e-08,
+      "loss": 0.07,
+      "step": 8929
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.487540742325163e-08,
+      "loss": 0.0582,
+      "step": 8930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.439880010183589e-08,
+      "loss": 0.0718,
+      "step": 8931
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.392338723983929e-08,
+      "loss": 0.066,
+      "step": 8932
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.344916889458933e-08,
+      "loss": 0.0693,
+      "step": 8933
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.297614512327025e-08,
+      "loss": 0.0695,
+      "step": 8934
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.250431598291864e-08,
+      "loss": 0.0714,
+      "step": 8935
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.203368153043346e-08,
+      "loss": 0.069,
+      "step": 8936
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.156424182256374e-08,
+      "loss": 0.0726,
+      "step": 8937
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.10959969159153e-08,
+      "loss": 0.0685,
+      "step": 8938
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.062894686695412e-08,
+      "loss": 0.0691,
+      "step": 8939
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.016309173199622e-08,
+      "loss": 0.067,
+      "step": 8940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.96984315672178e-08,
+      "loss": 0.0697,
+      "step": 8941
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.92349664286496e-08,
+      "loss": 0.0674,
+      "step": 8942
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.8772696372178e-08,
+      "loss": 0.0666,
+      "step": 8943
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.831162145354511e-08,
+      "loss": 0.0619,
+      "step": 8944
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.785174172834976e-08,
+      "loss": 0.0663,
+      "step": 8945
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.73930572520465e-08,
+      "loss": 0.0613,
+      "step": 8946
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.693556807994441e-08,
+      "loss": 0.0679,
+      "step": 8947
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.647927426721048e-08,
+      "loss": 0.0628,
+      "step": 8948
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.602417586886513e-08,
+      "loss": 0.0642,
+      "step": 8949
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.557027293978892e-08,
+      "loss": 0.0631,
+      "step": 8950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.511756553471252e-08,
+      "loss": 0.0606,
+      "step": 8951
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.466605370822556e-08,
+      "loss": 0.0655,
+      "step": 8952
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.421573751477563e-08,
+      "loss": 0.0647,
+      "step": 8953
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.37666170086615e-08,
+      "loss": 0.0639,
+      "step": 8954
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.331869224403988e-08,
+      "loss": 0.0617,
+      "step": 8955
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.287196327492531e-08,
+      "loss": 0.0684,
+      "step": 8956
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.24264301551847e-08,
+      "loss": 0.0681,
+      "step": 8957
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.198209293854287e-08,
+      "loss": 0.0619,
+      "step": 8958
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.153895167858028e-08,
+      "loss": 0.0691,
+      "step": 8959
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.109700642873086e-08,
+      "loss": 0.0663,
+      "step": 8960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.065625724228976e-08,
+      "loss": 0.0709,
+      "step": 8961
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.021670417240112e-08,
+      "loss": 0.068,
+      "step": 8962
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.977834727206924e-08,
+      "loss": 0.0628,
+      "step": 8963
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.934118659415402e-08,
+      "loss": 0.0699,
+      "step": 8964
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.890522219136998e-08,
+      "loss": 0.0664,
+      "step": 8965
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.847045411628507e-08,
+      "loss": 0.0645,
+      "step": 8966
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.803688242132845e-08,
+      "loss": 0.0626,
+      "step": 8967
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.760450715878055e-08,
+      "loss": 0.0617,
+      "step": 8968
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.717332838077851e-08,
+      "loss": 0.0606,
+      "step": 8969
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.674334613931744e-08,
+      "loss": 0.0694,
+      "step": 8970
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.631456048624475e-08,
+      "loss": 0.0658,
+      "step": 8971
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.588697147326685e-08,
+      "loss": 0.0659,
+      "step": 8972
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.54605791519436e-08,
+      "loss": 0.066,
+      "step": 8973
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.503538357368944e-08,
+      "loss": 0.0664,
+      "step": 8974
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.461138478977892e-08,
+      "loss": 0.0683,
+      "step": 8975
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.41885828513389e-08,
+      "loss": 0.0638,
+      "step": 8976
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.376697780935083e-08,
+      "loss": 0.065,
+      "step": 8977
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.334656971465626e-08,
+      "loss": 0.0666,
+      "step": 8978
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.292735861794798e-08,
+      "loss": 0.0625,
+      "step": 8979
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.250934456977665e-08,
+      "loss": 0.0708,
+      "step": 8980
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.209252762054863e-08,
+      "loss": 0.0615,
+      "step": 8981
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.167690782052483e-08,
+      "loss": 0.0709,
+      "step": 8982
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.126248521982294e-08,
+      "loss": 0.0619,
+      "step": 8983
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.084925986841517e-08,
+      "loss": 0.0696,
+      "step": 8984
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.04372318161306e-08,
+      "loss": 0.061,
+      "step": 8985
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.00264011126539e-08,
+      "loss": 0.0621,
+      "step": 8986
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.961676780752435e-08,
+      "loss": 0.0719,
+      "step": 8987
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.920833195013576e-08,
+      "loss": 0.068,
+      "step": 8988
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.880109358974096e-08,
+      "loss": 0.0621,
+      "step": 8989
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.839505277544622e-08,
+      "loss": 0.0627,
+      "step": 8990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.799020955621238e-08,
+      "loss": 0.0672,
+      "step": 8991
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.758656398085817e-08,
+      "loss": 0.0634,
+      "step": 8992
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.718411609805575e-08,
+      "loss": 0.0578,
+      "step": 8993
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.678286595633631e-08,
+      "loss": 0.0612,
+      "step": 8994
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.638281360408339e-08,
+      "loss": 0.066,
+      "step": 8995
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.598395908953504e-08,
+      "loss": 0.0685,
+      "step": 8996
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.558630246078946e-08,
+      "loss": 0.0608,
+      "step": 8997
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.518984376579718e-08,
+      "loss": 0.0754,
+      "step": 8998
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.47945830523633e-08,
+      "loss": 0.0597,
+      "step": 8999
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.440052036815081e-08,
+      "loss": 0.0653,
+      "step": 9000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.400765576067835e-08,
+      "loss": 0.0646,
+      "step": 9001
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.361598927731916e-08,
+      "loss": 0.0636,
+      "step": 9002
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.322552096530099e-08,
+      "loss": 0.0641,
+      "step": 9003
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.283625087170731e-08,
+      "loss": 0.0653,
+      "step": 9004
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.244817904348166e-08,
+      "loss": 0.0689,
+      "step": 9005
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.206130552741552e-08,
+      "loss": 0.0616,
+      "step": 9006
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.167563037016156e-08,
+      "loss": 0.0714,
+      "step": 9007
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.129115361822591e-08,
+      "loss": 0.0594,
+      "step": 9008
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.090787531797037e-08,
+      "loss": 0.063,
+      "step": 9009
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.052579551561133e-08,
+      "loss": 0.0614,
+      "step": 9010
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.014491425722413e-08,
+      "loss": 0.0644,
+      "step": 9011
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.976523158873426e-08,
+      "loss": 0.0657,
+      "step": 9012
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9386747555926215e-08,
+      "loss": 0.0616,
+      "step": 9013
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.900946220444126e-08,
+      "loss": 0.0621,
+      "step": 9014
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.863337557977078e-08,
+      "loss": 0.0634,
+      "step": 9015
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.825848772726739e-08,
+      "loss": 0.0652,
+      "step": 9016
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.788479869213493e-08,
+      "loss": 0.0685,
+      "step": 9017
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7512308519436236e-08,
+      "loss": 0.0635,
+      "step": 9018
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7141017254086496e-08,
+      "loss": 0.066,
+      "step": 9019
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6770924940857677e-08,
+      "loss": 0.0673,
+      "step": 9020
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6402031624377404e-08,
+      "loss": 0.0638,
+      "step": 9021
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6034337349127885e-08,
+      "loss": 0.0638,
+      "step": 9022
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.566784215944809e-08,
+      "loss": 0.059,
+      "step": 9023
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.530254609953045e-08,
+      "loss": 0.0661,
+      "step": 9024
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.493844921342528e-08,
+      "loss": 0.0664,
+      "step": 9025
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.457555154503635e-08,
+      "loss": 0.0718,
+      "step": 9026
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.421385313812311e-08,
+      "loss": 0.0697,
+      "step": 9027
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3853354036300655e-08,
+      "loss": 0.0692,
+      "step": 9028
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3494054283038666e-08,
+      "loss": 0.0673,
+      "step": 9029
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3135953921665816e-08,
+      "loss": 0.0665,
+      "step": 9030
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.27790529953609e-08,
+      "loss": 0.071,
+      "step": 9031
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.242335154716061e-08,
+      "loss": 0.0656,
+      "step": 9032
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.206884961995839e-08,
+      "loss": 0.062,
+      "step": 9033
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1715547256501186e-08,
+      "loss": 0.0641,
+      "step": 9034
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.136344449939046e-08,
+      "loss": 0.0719,
+      "step": 9035
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1012541391084466e-08,
+      "loss": 0.066,
+      "step": 9036
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.066283797389826e-08,
+      "loss": 0.0595,
+      "step": 9037
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.031433428999921e-08,
+      "loss": 0.0657,
+      "step": 9038
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.996703038141149e-08,
+      "loss": 0.069,
+      "step": 9039
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.962092629001492e-08,
+      "loss": 0.0659,
+      "step": 9040
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9276022057543896e-08,
+      "loss": 0.067,
+      "step": 9041
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.893231772558848e-08,
+      "loss": 0.067,
+      "step": 9042
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8589813335594424e-08,
+      "loss": 0.06,
+      "step": 9043
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.82485089288609e-08,
+      "loss": 0.0636,
+      "step": 9044
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.790840454654611e-08,
+      "loss": 0.0615,
+      "step": 9045
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.756950022965834e-08,
+      "loss": 0.0683,
+      "step": 9046
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.723179601906713e-08,
+      "loss": 0.0709,
+      "step": 9047
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.689529195549325e-08,
+      "loss": 0.0684,
+      "step": 9048
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6559988079512006e-08,
+      "loss": 0.0655,
+      "step": 9049
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.622588443155884e-08,
+      "loss": 0.0685,
+      "step": 9050
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.589298105191819e-08,
+      "loss": 0.0737,
+      "step": 9051
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.556127798073573e-08,
+      "loss": 0.0684,
+      "step": 9052
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.523077525800834e-08,
+      "loss": 0.066,
+      "step": 9053
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.490147292358971e-08,
+      "loss": 0.0631,
+      "step": 9054
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.457337101718806e-08,
+      "loss": 0.0614,
+      "step": 9055
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.424646957836842e-08,
+      "loss": 0.064,
+      "step": 9056
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.392076864654815e-08,
+      "loss": 0.0664,
+      "step": 9057
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.35962682610025e-08,
+      "loss": 0.0669,
+      "step": 9058
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.327296846086238e-08,
+      "loss": 0.0698,
+      "step": 9059
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.295086928510994e-08,
+      "loss": 0.0608,
+      "step": 9060
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.262997077258857e-08,
+      "loss": 0.0688,
+      "step": 9061
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.231027296199064e-08,
+      "loss": 0.0704,
+      "step": 9062
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1991775891868645e-08,
+      "loss": 0.0644,
+      "step": 9063
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.167447960062743e-08,
+      "loss": 0.0749,
+      "step": 9064
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.135838412652749e-08,
+      "loss": 0.0668,
+      "step": 9065
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.104348950768611e-08,
+      "loss": 0.0646,
+      "step": 9066
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0729795782072924e-08,
+      "loss": 0.0641,
+      "step": 9067
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.041730298751656e-08,
+      "loss": 0.0624,
+      "step": 9068
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.010601116169799e-08,
+      "loss": 0.0638,
+      "step": 9069
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9795920342153846e-08,
+      "loss": 0.0654,
+      "step": 9070
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9487030566275345e-08,
+      "loss": 0.0692,
+      "step": 9071
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.917934187131045e-08,
+      "loss": 0.0679,
+      "step": 9072
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.887285429436283e-08,
+      "loss": 0.0687,
+      "step": 9073
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8567567872388464e-08,
+      "loss": 0.0643,
+      "step": 9074
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.826348264220014e-08,
+      "loss": 0.0666,
+      "step": 9075
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7960598640466304e-08,
+      "loss": 0.0693,
+      "step": 9076
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7658915903709956e-08,
+      "loss": 0.0657,
+      "step": 9077
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.735843446830867e-08,
+      "loss": 0.0698,
+      "step": 9078
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.705915437049679e-08,
+      "loss": 0.07,
+      "step": 9079
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.676107564636211e-08,
+      "loss": 0.0663,
+      "step": 9080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.646419833184811e-08,
+      "loss": 0.0653,
+      "step": 9081
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.616852246275393e-08,
+      "loss": 0.0681,
+      "step": 9082
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.587404807473327e-08,
+      "loss": 0.0678,
+      "step": 9083
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.55807752032955e-08,
+      "loss": 0.0652,
+      "step": 9084
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.528870388380457e-08,
+      "loss": 0.0687,
+      "step": 9085
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.499783415147895e-08,
+      "loss": 0.067,
+      "step": 9086
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.470816604139282e-08,
+      "loss": 0.0645,
+      "step": 9087
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4419699588477126e-08,
+      "loss": 0.0715,
+      "step": 9088
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.413243482751516e-08,
+      "loss": 0.0689,
+      "step": 9089
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.384637179314587e-08,
+      "loss": 0.0684,
+      "step": 9090
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3561510519866116e-08,
+      "loss": 0.0708,
+      "step": 9091
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.327785104202397e-08,
+      "loss": 0.0618,
+      "step": 9092
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.299539339382429e-08,
+      "loss": 0.0679,
+      "step": 9093
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.271413760932762e-08,
+      "loss": 0.065,
+      "step": 9094
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.243408372244905e-08,
+      "loss": 0.0668,
+      "step": 9095
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.215523176695823e-08,
+      "loss": 0.069,
+      "step": 9096
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.187758177648159e-08,
+      "loss": 0.0666,
+      "step": 9097
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.160113378449792e-08,
+      "loss": 0.0658,
+      "step": 9098
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.132588782434165e-08,
+      "loss": 0.0631,
+      "step": 9099
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.105184392920513e-08,
+      "loss": 0.069,
+      "step": 9100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.077900213213303e-08,
+      "loss": 0.0674,
+      "step": 9101
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.05073624660257e-08,
+      "loss": 0.0661,
+      "step": 9102
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.023692496363917e-08,
+      "loss": 0.0712,
+      "step": 9103
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.996768965758179e-08,
+      "loss": 0.0638,
+      "step": 9104
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9699656580322033e-08,
+      "loss": 0.062,
+      "step": 9105
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9432825764178497e-08,
+      "loss": 0.0675,
+      "step": 9106
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.916719724132655e-08,
+      "loss": 0.0662,
+      "step": 9107
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.890277104379946e-08,
+      "loss": 0.065,
+      "step": 9108
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.86395472034795e-08,
+      "loss": 0.0673,
+      "step": 9109
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8377525752109057e-08,
+      "loss": 0.0666,
+      "step": 9110
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.811670672128397e-08,
+      "loss": 0.0624,
+      "step": 9111
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7857090142454634e-08,
+      "loss": 0.068,
+      "step": 9112
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.759867604692601e-08,
+      "loss": 0.0708,
+      "step": 9113
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7341464465859835e-08,
+      "loss": 0.0685,
+      "step": 9114
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.70854554302713e-08,
+      "loss": 0.0702,
+      "step": 9115
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.683064897103127e-08,
+      "loss": 0.0655,
+      "step": 9116
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6577045118864052e-08,
+      "loss": 0.0722,
+      "step": 9117
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6324643904351853e-08,
+      "loss": 0.0643,
+      "step": 9118
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6073445357930326e-08,
+      "loss": 0.0696,
+      "step": 9119
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5823449509889687e-08,
+      "loss": 0.0641,
+      "step": 9120
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5574656390374707e-08,
+      "loss": 0.0612,
+      "step": 9121
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5327066029386948e-08,
+      "loss": 0.0665,
+      "step": 9122
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5080678456781414e-08,
+      "loss": 0.0703,
+      "step": 9123
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4835493702267677e-08,
+      "loss": 0.0717,
+      "step": 9124
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4591511795413194e-08,
+      "loss": 0.068,
+      "step": 9125
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4348732765637762e-08,
+      "loss": 0.0709,
+      "step": 9126
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4107156642214636e-08,
+      "loss": 0.0642,
+      "step": 9127
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3866783454277175e-08,
+      "loss": 0.0695,
+      "step": 9128
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3627613230807757e-08,
+      "loss": 0.0691,
+      "step": 9129
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3389646000648856e-08,
+      "loss": 0.067,
+      "step": 9130
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.315288179249531e-08,
+      "loss": 0.0601,
+      "step": 9131
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.291732063489538e-08,
+      "loss": 0.0752,
+      "step": 9132
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2682962556255238e-08,
+      "loss": 0.065,
+      "step": 9133
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2449807584834504e-08,
+      "loss": 0.0659,
+      "step": 9134
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2217855748748463e-08,
+      "loss": 0.0639,
+      "step": 9135
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1987107075965853e-08,
+      "loss": 0.0704,
+      "step": 9136
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1757561594313303e-08,
+      "loss": 0.0708,
+      "step": 9137
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1529219331467564e-08,
+      "loss": 0.0694,
+      "step": 9138
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1302080314965502e-08,
+      "loss": 0.0593,
+      "step": 9139
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.10761445721952e-08,
+      "loss": 0.064,
+      "step": 9140
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0851412130401538e-08,
+      "loss": 0.068,
+      "step": 9141
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.062788301668395e-08,
+      "loss": 0.0576,
+      "step": 9142
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0405557257995312e-08,
+      "loss": 0.0626,
+      "step": 9143
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.018443488114641e-08,
+      "loss": 0.072,
+      "step": 9144
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9964515912799244e-08,
+      "loss": 0.0607,
+      "step": 9145
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.974580037947371e-08,
+      "loss": 0.0686,
+      "step": 9146
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9528288307543165e-08,
+      "loss": 0.0725,
+      "step": 9147
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9311979723236618e-08,
+      "loss": 0.0677,
+      "step": 9148
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9096874652636545e-08,
+      "loss": 0.0623,
+      "step": 9149
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.888297312168108e-08,
+      "loss": 0.0636,
+      "step": 9150
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.867027515616515e-08,
+      "loss": 0.0669,
+      "step": 9151
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.845878078173491e-08,
+      "loss": 0.065,
+      "step": 9152
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8248490023894395e-08,
+      "loss": 0.068,
+      "step": 9153
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.80394029080011e-08,
+      "loss": 0.0683,
+      "step": 9154
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7831519459267087e-08,
+      "loss": 0.0689,
+      "step": 9155
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7624839702760078e-08,
+      "loss": 0.0678,
+      "step": 9156
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7419363663402357e-08,
+      "loss": 0.0719,
+      "step": 9157
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7215091365971882e-08,
+      "loss": 0.0608,
+      "step": 9158
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.701202283510006e-08,
+      "loss": 0.0645,
+      "step": 9159
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6810158095273975e-08,
+      "loss": 0.0674,
+      "step": 9160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6609497170834154e-08,
+      "loss": 0.0661,
+      "step": 9161
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6410040085979018e-08,
+      "loss": 0.0673,
+      "step": 9162
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6211786864758218e-08,
+      "loss": 0.0697,
+      "step": 9163
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6014737531079295e-08,
+      "loss": 0.07,
+      "step": 9164
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5818892108702133e-08,
+      "loss": 0.065,
+      "step": 9165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5624250621244508e-08,
+      "loss": 0.0639,
+      "step": 9166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.543081309217431e-08,
+      "loss": 0.0627,
+      "step": 9167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5238579544819556e-08,
+      "loss": 0.066,
+      "step": 9168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.504755000235947e-08,
+      "loss": 0.0702,
+      "step": 9169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4857724487830072e-08,
+      "loss": 0.0647,
+      "step": 9170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4669103024119723e-08,
+      "loss": 0.0623,
+      "step": 9171
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4481685633973563e-08,
+      "loss": 0.071,
+      "step": 9172
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4295472339992401e-08,
+      "loss": 0.0612,
+      "step": 9173
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4110463164630494e-08,
+      "loss": 0.0673,
+      "step": 9174
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3926658130195558e-08,
+      "loss": 0.0668,
+      "step": 9175
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3744057258853195e-08,
+      "loss": 0.0614,
+      "step": 9176
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3562660572621344e-08,
+      "loss": 0.0679,
+      "step": 9177
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3382468093373624e-08,
+      "loss": 0.066,
+      "step": 9178
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3203479842838207e-08,
+      "loss": 0.0684,
+      "step": 9179
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.302569584259894e-08,
+      "loss": 0.0648,
+      "step": 9180
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2849116114093118e-08,
+      "loss": 0.0645,
+      "step": 9181
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.267374067861371e-08,
+      "loss": 0.0681,
+      "step": 9182
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2499569557309354e-08,
+      "loss": 0.063,
+      "step": 9183
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2326602771179918e-08,
+      "loss": 0.0627,
+      "step": 9184
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2154840341084273e-08,
+      "loss": 0.067,
+      "step": 9185
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1984282287733628e-08,
+      "loss": 0.0611,
+      "step": 9186
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1814928631694867e-08,
+      "loss": 0.0719,
+      "step": 9187
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.164677939339054e-08,
+      "loss": 0.063,
+      "step": 9188
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.147983459309443e-08,
+      "loss": 0.0629,
+      "step": 9189
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1314094250938212e-08,
+      "loss": 0.0707,
+      "step": 9190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1149558386908122e-08,
+      "loss": 0.0686,
+      "step": 9191
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0986227020844953e-08,
+      "loss": 0.0585,
+      "step": 9192
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0824100172442953e-08,
+      "loss": 0.0598,
+      "step": 9193
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0663177861252039e-08,
+      "loss": 0.0642,
+      "step": 9194
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0503460106677799e-08,
+      "loss": 0.0639,
+      "step": 9195
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.034494692797927e-08,
+      "loss": 0.0603,
+      "step": 9196
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0187638344270057e-08,
+      "loss": 0.0671,
+      "step": 9197
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0031534374519425e-08,
+      "loss": 0.0665,
+      "step": 9198
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.876635037552318e-09,
+      "loss": 0.0614,
+      "step": 9199
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.722940352044907e-09,
+      "loss": 0.0625,
+      "step": 9200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.570450336532366e-09,
+      "loss": 0.0682,
+      "step": 9201
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.419165009401099e-09,
+      "loss": 0.0707,
+      "step": 9202
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.26908438889429e-09,
+      "loss": 0.0641,
+      "step": 9203
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.120208493109683e-09,
+      "loss": 0.064,
+      "step": 9204
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.972537339999588e-09,
+      "loss": 0.0698,
+      "step": 9205
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.826070947369759e-09,
+      "loss": 0.0643,
+      "step": 9206
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.680809332881623e-09,
+      "loss": 0.065,
+      "step": 9207
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.536752514053393e-09,
+      "loss": 0.0674,
+      "step": 9208
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.393900508253393e-09,
+      "loss": 0.0621,
+      "step": 9209
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.25225333270896e-09,
+      "loss": 0.0624,
+      "step": 9210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.11181100450087e-09,
+      "loss": 0.0604,
+      "step": 9211
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.972573540563355e-09,
+      "loss": 0.0655,
+      "step": 9212
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.834540957686321e-09,
+      "loss": 0.0666,
+      "step": 9213
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.69771327251423e-09,
+      "loss": 0.0634,
+      "step": 9214
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.562090501547214e-09,
+      "loss": 0.0637,
+      "step": 9215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.427672661138862e-09,
+      "loss": 0.07,
+      "step": 9216
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.294459767497319e-09,
+      "loss": 0.0648,
+      "step": 9217
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.162451836685291e-09,
+      "loss": 0.0657,
+      "step": 9218
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.031648884623377e-09,
+      "loss": 0.0622,
+      "step": 9219
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.902050927082294e-09,
+      "loss": 0.0707,
+      "step": 9220
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.7736579796895406e-09,
+      "loss": 0.0706,
+      "step": 9221
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.646470057928289e-09,
+      "loss": 0.0662,
+      "step": 9222
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.520487177134049e-09,
+      "loss": 0.0648,
+      "step": 9223
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.3957093525002235e-09,
+      "loss": 0.0586,
+      "step": 9224
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.272136599071443e-09,
+      "loss": 0.072,
+      "step": 9225
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.149768931749123e-09,
+      "loss": 0.0684,
+      "step": 9226
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.028606365289236e-09,
+      "loss": 0.0661,
+      "step": 9227
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.9086489143023175e-09,
+      "loss": 0.0679,
+      "step": 9228
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.789896593251243e-09,
+      "loss": 0.0643,
+      "step": 9229
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.672349416459e-09,
+      "loss": 0.063,
+      "step": 9230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.556007398097585e-09,
+      "loss": 0.0624,
+      "step": 9231
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.4408705521968865e-09,
+      "loss": 0.0648,
+      "step": 9232
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.326938892640243e-09,
+      "loss": 0.063,
+      "step": 9233
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.2142124331655555e-09,
+      "loss": 0.065,
+      "step": 9234
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.102691187366393e-09,
+      "loss": 0.0652,
+      "step": 9235
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.9923751686908885e-09,
+      "loss": 0.0647,
+      "step": 9236
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.883264390440623e-09,
+      "loss": 0.0659,
+      "step": 9237
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.77535886577396e-09,
+      "loss": 0.0646,
+      "step": 9238
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.668658607701604e-09,
+      "loss": 0.0653,
+      "step": 9239
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.563163629088818e-09,
+      "loss": 0.0634,
+      "step": 9240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.458873942658759e-09,
+      "loss": 0.0643,
+      "step": 9241
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.355789560985812e-09,
+      "loss": 0.0682,
+      "step": 9242
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.253910496501146e-09,
+      "loss": 0.063,
+      "step": 9243
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.153236761488266e-09,
+      "loss": 0.0733,
+      "step": 9244
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.053768368088573e-09,
+      "loss": 0.0633,
+      "step": 9245
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.955505328295806e-09,
+      "loss": 0.0678,
+      "step": 9246
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.858447653958264e-09,
+      "loss": 0.0615,
+      "step": 9247
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.762595356781029e-09,
+      "loss": 0.0601,
+      "step": 9248
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.667948448320413e-09,
+      "loss": 0.0688,
+      "step": 9249
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5745069399917285e-09,
+      "loss": 0.0636,
+      "step": 9250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4822708430604093e-09,
+      "loss": 0.0631,
+      "step": 9251
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3912401686497787e-09,
+      "loss": 0.069,
+      "step": 9252
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3014149277355024e-09,
+      "loss": 0.0669,
+      "step": 9253
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2127951311511363e-09,
+      "loss": 0.0654,
+      "step": 9254
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.125380789581467e-09,
+      "loss": 0.0614,
+      "step": 9255
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0391719135669517e-09,
+      "loss": 0.0713,
+      "step": 9256
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.954168513503719e-09,
+      "loss": 0.0632,
+      "step": 9257
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.870370599642458e-09,
+      "loss": 0.0677,
+      "step": 9258
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.787778182086198e-09,
+      "loss": 0.0666,
+      "step": 9259
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7063912707958605e-09,
+      "loss": 0.0642,
+      "step": 9260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6262098755847064e-09,
+      "loss": 0.0601,
+      "step": 9261
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.547234006121668e-09,
+      "loss": 0.0637,
+      "step": 9262
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4694636719291286e-09,
+      "loss": 0.0719,
+      "step": 9263
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.392898882386252e-09,
+      "loss": 0.0723,
+      "step": 9264
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3175396467245428e-09,
+      "loss": 0.0661,
+      "step": 9265
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.243385974031176e-09,
+      "loss": 0.0675,
+      "step": 9266
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1704378732489984e-09,
+      "loss": 0.0642,
+      "step": 9267
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0986953531720867e-09,
+      "loss": 0.0724,
+      "step": 9268
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0281584224546293e-09,
+      "loss": 0.0681,
+      "step": 9269
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.958827089599824e-09,
+      "loss": 0.0715,
+      "step": 9270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.890701362968761e-09,
+      "loss": 0.0656,
+      "step": 9271
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8237812507770901e-09,
+      "loss": 0.063,
+      "step": 9272
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7580667610928025e-09,
+      "loss": 0.06,
+      "step": 9273
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.69355790184067e-09,
+      "loss": 0.0649,
+      "step": 9274
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6302546808000253e-09,
+      "loss": 0.0683,
+      "step": 9275
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5681571056036515e-09,
+      "loss": 0.065,
+      "step": 9276
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.507265183738893e-09,
+      "loss": 0.0616,
+      "step": 9277
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4475789225498748e-09,
+      "loss": 0.0632,
+      "step": 9278
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3890983292330628e-09,
+      "loss": 0.0696,
+      "step": 9279
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3318234108394834e-09,
+      "loss": 0.0696,
+      "step": 9280
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2757541742769442e-09,
+      "loss": 0.0655,
+      "step": 9281
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2208906263067033e-09,
+      "loss": 0.0663,
+      "step": 9282
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1672327735423595e-09,
+      "loss": 0.0693,
+      "step": 9283
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1147806224565128e-09,
+      "loss": 0.0654,
+      "step": 9284
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0635341793729935e-09,
+      "loss": 0.0665,
+      "step": 9285
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0134934504713034e-09,
+      "loss": 0.0672,
+      "step": 9286
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.646584417855042e-10,
+      "loss": 0.066,
+      "step": 9287
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.170291592042191e-10,
+      "loss": 0.0656,
+      "step": 9288
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.706056084717418e-10,
+      "loss": 0.0716,
+      "step": 9289
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.253877951847067e-10,
+      "loss": 0.0676,
+      "step": 9290
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.813757247965292e-10,
+      "loss": 0.0685,
+      "step": 9291
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.385694026140755e-10,
+      "loss": 0.0622,
+      "step": 9292
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.969688337998825e-10,
+      "loss": 0.0592,
+      "step": 9293
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.565740233688278e-10,
+      "loss": 0.0671,
+      "step": 9294
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.173849761925699e-10,
+      "loss": 0.0719,
+      "step": 9295
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.794016969973282e-10,
+      "loss": 0.0667,
+      "step": 9296
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.426241903627727e-10,
+      "loss": 0.0668,
+      "step": 9297
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.070524607231342e-10,
+      "loss": 0.0608,
+      "step": 9298
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.726865123694247e-10,
+      "loss": 0.0681,
+      "step": 9299
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3952634944388617e-10,
+      "loss": 0.0635,
+      "step": 9300
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.075719759455421e-10,
+      "loss": 0.0656,
+      "step": 9301
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7682339572797656e-10,
+      "loss": 0.0666,
+      "step": 9302
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.4728061249933443e-10,
+      "loss": 0.0645,
+      "step": 9303
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.189436298212112e-10,
+      "loss": 0.0622,
+      "step": 9304
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.918124511108733e-10,
+      "loss": 0.0666,
+      "step": 9305
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.65887079640148e-10,
+      "loss": 0.0618,
+      "step": 9306
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.411675185354234e-10,
+      "loss": 0.0575,
+      "step": 9307
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1765377077653804e-10,
+      "loss": 0.0654,
+      "step": 9308
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9534583919900152e-10,
+      "loss": 0.0764,
+      "step": 9309
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7424372649399446e-10,
+      "loss": 0.0632,
+      "step": 9310
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5434743520503782e-10,
+      "loss": 0.0649,
+      "step": 9311
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3565696773132352e-10,
+      "loss": 0.0658,
+      "step": 9312
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.181723263277146e-10,
+      "loss": 0.066,
+      "step": 9313
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0189351310030404e-10,
+      "loss": 0.0702,
+      "step": 9314
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.682053001529688e-11,
+      "loss": 0.0618,
+      "step": 9315
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.295337888790777e-11,
+      "loss": 0.0656,
+      "step": 9316
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.029206139013255e-11,
+      "loss": 0.0619,
+      "step": 9317
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8836579050748344e-11,
+      "loss": 0.0678,
+      "step": 9318
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.8586933248652146e-11,
+      "loss": 0.0619,
+      "step": 9319
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.954312522174263e-11,
+      "loss": 0.0687,
+      "step": 9320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.170515605914858e-11,
+      "loss": 0.069,
+      "step": 9321
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.507302670789024e-11,
+      "loss": 0.0646,
+      "step": 9322
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.646737965107733e-12,
+      "loss": 0.0703,
+      "step": 9323
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.426290486942876e-12,
+      "loss": 0.0705,
+      "step": 9324
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.41168478187781e-12,
+      "loss": 0.0637,
+      "step": 9325
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.029212140656881e-13,
+      "loss": 0.0599,
+      "step": 9326
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.0688,
+      "step": 9327
+    },
+    {
+      "epoch": 3.0,
+      "step": 9327,
+      "total_flos": 2.3915576951790633e+18,
+      "train_loss": 0.11992915521680485,
+      "train_runtime": 26228.1209,
+      "train_samples_per_second": 45.517,
+      "train_steps_per_second": 0.356
+    }
+  ],
+  "max_steps": 9327,
+  "num_train_epochs": 3,
+  "total_flos": 2.3915576951790633e+18,
+  "trial_name": null,
+  "trial_params": null
+}