diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36061 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 25.0,
+  "eval_steps": 3000,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9995000000000004e-05,
+      "loss": 0.1285,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999e-05,
+      "loss": 0.3432,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9985e-05,
+      "loss": 0.3338,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9980000000000005e-05,
+      "loss": 0.2971,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9975e-05,
+      "loss": 0.4566,
+      "step": 5
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9970000000000004e-05,
+      "loss": 0.5346,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9965e-05,
+      "loss": 0.3419,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.996e-05,
+      "loss": 0.463,
+      "step": 8
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9955e-05,
+      "loss": 0.3006,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995e-05,
+      "loss": 0.182,
+      "step": 10
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9945e-05,
+      "loss": 0.5121,
+      "step": 11
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994e-05,
+      "loss": 0.2336,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9935e-05,
+      "loss": 0.5231,
+      "step": 13
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9930000000000004e-05,
+      "loss": 0.5099,
+      "step": 14
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9925000000000006e-05,
+      "loss": 0.3736,
+      "step": 15
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.992e-05,
+      "loss": 0.4328,
+      "step": 16
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9915000000000005e-05,
+      "loss": 0.4428,
+      "step": 17
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.991000000000001e-05,
+      "loss": 0.2561,
+      "step": 18
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9904999999999997e-05,
+      "loss": 0.4564,
+      "step": 19
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99e-05,
+      "loss": 0.2965,
+      "step": 20
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9895e-05,
+      "loss": 0.2254,
+      "step": 21
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.989e-05,
+      "loss": 0.3985,
+      "step": 22
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9885e-05,
+      "loss": 0.4021,
+      "step": 23
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.988e-05,
+      "loss": 0.3768,
+      "step": 24
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9875000000000005e-05,
+      "loss": 0.1655,
+      "step": 25
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.987e-05,
+      "loss": 0.2646,
+      "step": 26
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9865000000000004e-05,
+      "loss": 0.4063,
+      "step": 27
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9860000000000006e-05,
+      "loss": 0.4548,
+      "step": 28
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9855e-05,
+      "loss": 0.1685,
+      "step": 29
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9850000000000005e-05,
+      "loss": 0.7555,
+      "step": 30
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9845e-05,
+      "loss": 0.3011,
+      "step": 31
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9839999999999996e-05,
+      "loss": 0.3521,
+      "step": 32
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9835e-05,
+      "loss": 0.2528,
+      "step": 33
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983e-05,
+      "loss": 0.3435,
+      "step": 34
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9825000000000004e-05,
+      "loss": 0.4145,
+      "step": 35
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982e-05,
+      "loss": 0.4301,
+      "step": 36
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9815e-05,
+      "loss": 0.2195,
+      "step": 37
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9810000000000005e-05,
+      "loss": 0.235,
+      "step": 38
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9805e-05,
+      "loss": 0.5224,
+      "step": 39
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.197,
+      "step": 40
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9795000000000006e-05,
+      "loss": 0.4239,
+      "step": 41
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979e-05,
+      "loss": 0.4302,
+      "step": 42
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9785e-05,
+      "loss": 0.2282,
+      "step": 43
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978e-05,
+      "loss": 0.1763,
+      "step": 44
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9775e-05,
+      "loss": 0.4236,
+      "step": 45
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.977e-05,
+      "loss": 0.4492,
+      "step": 46
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9765e-05,
+      "loss": 0.4814,
+      "step": 47
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9760000000000004e-05,
+      "loss": 0.374,
+      "step": 48
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9755e-05,
+      "loss": 0.2636,
+      "step": 49
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.975e-05,
+      "loss": 0.2385,
+      "step": 50
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9745000000000005e-05,
+      "loss": 0.2099,
+      "step": 51
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.974e-05,
+      "loss": 0.3527,
+      "step": 52
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9735e-05,
+      "loss": 0.4626,
+      "step": 53
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9730000000000006e-05,
+      "loss": 0.157,
+      "step": 54
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9725e-05,
+      "loss": 0.2053,
+      "step": 55
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972e-05,
+      "loss": 0.399,
+      "step": 56
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9715e-05,
+      "loss": 0.3766,
+      "step": 57
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971e-05,
+      "loss": 0.361,
+      "step": 58
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9705e-05,
+      "loss": 0.2062,
+      "step": 59
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.97e-05,
+      "loss": 0.2278,
+      "step": 60
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9695000000000004e-05,
+      "loss": 0.3808,
+      "step": 61
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.969e-05,
+      "loss": 0.6601,
+      "step": 62
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9685e-05,
+      "loss": 0.3966,
+      "step": 63
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9680000000000005e-05,
+      "loss": 0.4347,
+      "step": 64
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967500000000001e-05,
+      "loss": 0.6115,
+      "step": 65
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.967e-05,
+      "loss": 0.4005,
+      "step": 66
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9665e-05,
+      "loss": 0.4407,
+      "step": 67
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966e-05,
+      "loss": 0.2405,
+      "step": 68
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9655e-05,
+      "loss": 0.3027,
+      "step": 69
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.965e-05,
+      "loss": 0.2923,
+      "step": 70
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9645e-05,
+      "loss": 0.4011,
+      "step": 71
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.964e-05,
+      "loss": 0.2542,
+      "step": 72
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9635e-05,
+      "loss": 0.2347,
+      "step": 73
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.963e-05,
+      "loss": 0.4032,
+      "step": 74
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9625000000000006e-05,
+      "loss": 0.3212,
+      "step": 75
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.962e-05,
+      "loss": 0.2636,
+      "step": 76
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9615000000000004e-05,
+      "loss": 0.2318,
+      "step": 77
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.961000000000001e-05,
+      "loss": 1.0812,
+      "step": 78
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9604999999999996e-05,
+      "loss": 0.4528,
+      "step": 79
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.96e-05,
+      "loss": 0.1548,
+      "step": 80
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9595e-05,
+      "loss": 0.4125,
+      "step": 81
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959e-05,
+      "loss": 0.2965,
+      "step": 82
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9585e-05,
+      "loss": 0.3892,
+      "step": 83
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.958e-05,
+      "loss": 0.2076,
+      "step": 84
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9575000000000005e-05,
+      "loss": 0.186,
+      "step": 85
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.957e-05,
+      "loss": 0.3577,
+      "step": 86
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.9565e-05,
+      "loss": 0.2461,
+      "step": 87
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9560000000000006e-05,
+      "loss": 0.2581,
+      "step": 88
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9555e-05,
+      "loss": 0.1551,
+      "step": 89
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9550000000000004e-05,
+      "loss": 0.4558,
+      "step": 90
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9545e-05,
+      "loss": 0.1652,
+      "step": 91
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9539999999999996e-05,
+      "loss": 0.3322,
+      "step": 92
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9535e-05,
+      "loss": 0.3678,
+      "step": 93
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.953e-05,
+      "loss": 0.3536,
+      "step": 94
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9525000000000004e-05,
+      "loss": 0.2243,
+      "step": 95
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.952e-05,
+      "loss": 0.1453,
+      "step": 96
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9515e-05,
+      "loss": 0.4241,
+      "step": 97
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9510000000000005e-05,
+      "loss": 0.5273,
+      "step": 98
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9505e-05,
+      "loss": 0.3262,
+      "step": 99
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.95e-05,
+      "loss": 0.25,
+      "step": 100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9495000000000006e-05,
+      "loss": 0.2154,
+      "step": 101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949e-05,
+      "loss": 0.3913,
+      "step": 102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9485e-05,
+      "loss": 0.5388,
+      "step": 103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.948e-05,
+      "loss": 0.3297,
+      "step": 104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9475e-05,
+      "loss": 0.4287,
+      "step": 105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947e-05,
+      "loss": 0.4871,
+      "step": 106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9465e-05,
+      "loss": 0.3373,
+      "step": 107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946e-05,
+      "loss": 0.171,
+      "step": 108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9455e-05,
+      "loss": 0.3428,
+      "step": 109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945e-05,
+      "loss": 0.2957,
+      "step": 110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9445000000000004e-05,
+      "loss": 0.4428,
+      "step": 111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.944e-05,
+      "loss": 0.2148,
+      "step": 112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9435e-05,
+      "loss": 0.1659,
+      "step": 113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9430000000000005e-05,
+      "loss": 0.3082,
+      "step": 114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9425e-05,
+      "loss": 0.2167,
+      "step": 115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942e-05,
+      "loss": 0.7734,
+      "step": 116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.9415e-05,
+      "loss": 0.2923,
+      "step": 117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941e-05,
+      "loss": 0.2043,
+      "step": 118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9405e-05,
+      "loss": 0.3325,
+      "step": 119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.94e-05,
+      "loss": 0.4744,
+      "step": 120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9395e-05,
+      "loss": 0.4442,
+      "step": 121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.939e-05,
+      "loss": 0.3927,
+      "step": 122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9385e-05,
+      "loss": 0.3186,
+      "step": 123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.9380000000000004e-05,
+      "loss": 0.2289,
+      "step": 124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937500000000001e-05,
+      "loss": 0.5687,
+      "step": 125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.937e-05,
+      "loss": 0.2492,
+      "step": 126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9365000000000005e-05,
+      "loss": 0.5131,
+      "step": 127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936e-05,
+      "loss": 0.2755,
+      "step": 128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9355e-05,
+      "loss": 0.3599,
+      "step": 129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.935e-05,
+      "loss": 0.189,
+      "step": 130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9345e-05,
+      "loss": 0.2644,
+      "step": 131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.934e-05,
+      "loss": 0.6113,
+      "step": 132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9335e-05,
+      "loss": 0.3174,
+      "step": 133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.933e-05,
+      "loss": 0.4224,
+      "step": 134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9325000000000005e-05,
+      "loss": 0.1645,
+      "step": 135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.932e-05,
+      "loss": 0.1888,
+      "step": 136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9315000000000004e-05,
+      "loss": 0.4115,
+      "step": 137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9310000000000006e-05,
+      "loss": 0.2566,
+      "step": 138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9305e-05,
+      "loss": 0.4836,
+      "step": 139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.93e-05,
+      "loss": 0.1507,
+      "step": 140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9295e-05,
+      "loss": 0.351,
+      "step": 141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9289999999999996e-05,
+      "loss": 0.3851,
+      "step": 142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9285e-05,
+      "loss": 0.2143,
+      "step": 143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928e-05,
+      "loss": 0.1551,
+      "step": 144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9275000000000004e-05,
+      "loss": 0.4425,
+      "step": 145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.927e-05,
+      "loss": 0.5692,
+      "step": 146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9265e-05,
+      "loss": 0.3714,
+      "step": 147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9260000000000005e-05,
+      "loss": 0.155,
+      "step": 148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9255e-05,
+      "loss": 0.3835,
+      "step": 149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9250000000000004e-05,
+      "loss": 0.4837,
+      "step": 150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9245000000000006e-05,
+      "loss": 0.2139,
+      "step": 151
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9239999999999995e-05,
+      "loss": 0.2575,
+      "step": 152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9235e-05,
+      "loss": 0.3627,
+      "step": 153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923e-05,
+      "loss": 0.2587,
+      "step": 154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9225e-05,
+      "loss": 0.3901,
+      "step": 155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.922e-05,
+      "loss": 0.3092,
+      "step": 156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9215e-05,
+      "loss": 0.4186,
+      "step": 157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9210000000000004e-05,
+      "loss": 0.3926,
+      "step": 158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9205e-05,
+      "loss": 0.2344,
+      "step": 159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.92e-05,
+      "loss": 0.3967,
+      "step": 160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9195000000000005e-05,
+      "loss": 0.1756,
+      "step": 161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.919e-05,
+      "loss": 0.2687,
+      "step": 162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9185000000000003e-05,
+      "loss": 0.395,
+      "step": 163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918e-05,
+      "loss": 0.2391,
+      "step": 164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9175e-05,
+      "loss": 0.3528,
+      "step": 165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.917e-05,
+      "loss": 0.507,
+      "step": 166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9165e-05,
+      "loss": 0.23,
+      "step": 167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.916e-05,
+      "loss": 0.4235,
+      "step": 168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9155e-05,
+      "loss": 0.2516,
+      "step": 169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.915e-05,
+      "loss": 0.1857,
+      "step": 170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9145000000000004e-05,
+      "loss": 0.2839,
+      "step": 171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.914e-05,
+      "loss": 0.3275,
+      "step": 172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9135e-05,
+      "loss": 0.2265,
+      "step": 173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9130000000000005e-05,
+      "loss": 0.4699,
+      "step": 174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912500000000001e-05,
+      "loss": 0.52,
+      "step": 175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.9119999999999996e-05,
+      "loss": 0.1683,
+      "step": 176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.9115e-05,
+      "loss": 0.4142,
+      "step": 177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911e-05,
+      "loss": 0.4943,
+      "step": 178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9105e-05,
+      "loss": 0.1797,
+      "step": 179
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.91e-05,
+      "loss": 0.3931,
+      "step": 180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9095e-05,
+      "loss": 0.144,
+      "step": 181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9090000000000005e-05,
+      "loss": 0.461,
+      "step": 182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9085e-05,
+      "loss": 0.2163,
+      "step": 183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9080000000000004e-05,
+      "loss": 0.2321,
+      "step": 184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9075000000000006e-05,
+      "loss": 0.3433,
+      "step": 185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.907e-05,
+      "loss": 0.4397,
+      "step": 186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9065000000000005e-05,
+      "loss": 0.1621,
+      "step": 187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906e-05,
+      "loss": 0.1784,
+      "step": 188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.9054999999999996e-05,
+      "loss": 0.3732,
+      "step": 189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.905e-05,
+      "loss": 0.5768,
+      "step": 190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9045e-05,
+      "loss": 0.3771,
+      "step": 191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9040000000000004e-05,
+      "loss": 0.3831,
+      "step": 192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9035e-05,
+      "loss": 0.2963,
+      "step": 193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.903e-05,
+      "loss": 0.2845,
+      "step": 194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9025000000000005e-05,
+      "loss": 0.404,
+      "step": 195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.902e-05,
+      "loss": 0.3699,
+      "step": 196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9015e-05,
+      "loss": 0.3798,
+      "step": 197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9010000000000006e-05,
+      "loss": 0.4422,
+      "step": 198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9005e-05,
+      "loss": 0.5241,
+      "step": 199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9e-05,
+      "loss": 0.1749,
+      "step": 200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.8995e-05,
+      "loss": 0.5123,
+      "step": 201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899e-05,
+      "loss": 0.4451,
+      "step": 202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8985e-05,
+      "loss": 0.4027,
+      "step": 203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898e-05,
+      "loss": 0.3135,
+      "step": 204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8975000000000004e-05,
+      "loss": 0.2027,
+      "step": 205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.897e-05,
+      "loss": 0.1476,
+      "step": 206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.8965e-05,
+      "loss": 0.2469,
+      "step": 207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.8960000000000005e-05,
+      "loss": 0.6093,
+      "step": 208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.8955e-05,
+      "loss": 0.3019,
+      "step": 209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.895e-05,
+      "loss": 0.2645,
+      "step": 210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8945000000000006e-05,
+      "loss": 0.3196,
+      "step": 211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894e-05,
+      "loss": 0.4245,
+      "step": 212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8935e-05,
+      "loss": 0.3728,
+      "step": 213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893e-05,
+      "loss": 0.3417,
+      "step": 214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8925e-05,
+      "loss": 0.3078,
+      "step": 215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.892e-05,
+      "loss": 0.4321,
+      "step": 216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8915e-05,
+      "loss": 0.2928,
+      "step": 217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.8910000000000003e-05,
+      "loss": 0.3856,
+      "step": 218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.8905e-05,
+      "loss": 0.1561,
+      "step": 219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.89e-05,
+      "loss": 0.9382,
+      "step": 220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8895000000000004e-05,
+      "loss": 0.715,
+      "step": 221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.889000000000001e-05,
+      "loss": 0.2042,
+      "step": 222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8885e-05,
+      "loss": 0.3821,
+      "step": 223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888e-05,
+      "loss": 0.2381,
+      "step": 224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8875e-05,
+      "loss": 0.2923,
+      "step": 225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887e-05,
+      "loss": 0.2623,
+      "step": 226
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8865e-05,
+      "loss": 0.3072,
+      "step": 227
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.886e-05,
+      "loss": 0.4156,
+      "step": 228
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8855e-05,
+      "loss": 0.5985,
+      "step": 229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.885e-05,
+      "loss": 0.1719,
+      "step": 230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8845e-05,
+      "loss": 0.408,
+      "step": 231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8840000000000006e-05,
+      "loss": 0.2775,
+      "step": 232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8835e-05,
+      "loss": 0.4121,
+      "step": 233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8830000000000004e-05,
+      "loss": 0.3919,
+      "step": 234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882500000000001e-05,
+      "loss": 0.3609,
+      "step": 235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8819999999999996e-05,
+      "loss": 0.548,
+      "step": 236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8815e-05,
+      "loss": 0.2861,
+      "step": 237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.881e-05,
+      "loss": 0.41,
+      "step": 238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8805e-05,
+      "loss": 0.2673,
+      "step": 239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.88e-05,
+      "loss": 0.6881,
+      "step": 240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8795e-05,
+      "loss": 0.4097,
+      "step": 241
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8790000000000005e-05,
+      "loss": 0.252,
+      "step": 242
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8785e-05,
+      "loss": 0.2188,
+      "step": 243
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.878e-05,
+      "loss": 0.3312,
+      "step": 244
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8775000000000006e-05,
+      "loss": 0.3615,
+      "step": 245
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.877e-05,
+      "loss": 0.1939,
+      "step": 246
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8765000000000004e-05,
+      "loss": 0.2062,
+      "step": 247
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.876e-05,
+      "loss": 0.5097,
+      "step": 248
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8754999999999996e-05,
+      "loss": 0.2443,
+      "step": 249
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.875e-05,
+      "loss": 0.4485,
+      "step": 250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8745e-05,
+      "loss": 0.2659,
+      "step": 251
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8740000000000003e-05,
+      "loss": 0.1235,
+      "step": 252
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8735e-05,
+      "loss": 0.1849,
+      "step": 253
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.873e-05,
+      "loss": 0.288,
+      "step": 254
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.8725000000000004e-05,
+      "loss": 0.3657,
+      "step": 255
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.872e-05,
+      "loss": 0.5219,
+      "step": 256
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.8715e-05,
+      "loss": 0.344,
+      "step": 257
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.8710000000000005e-05,
+      "loss": 0.3364,
+      "step": 258
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.8705e-05,
+      "loss": 0.193,
+      "step": 259
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.8700000000000004e-05,
+      "loss": 0.3849,
+      "step": 260
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8695e-05,
+      "loss": 0.1877,
+      "step": 261
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.869e-05,
+      "loss": 0.2312,
+      "step": 262
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.8685e-05,
+      "loss": 0.4275,
+      "step": 263
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868e-05,
+      "loss": 0.3551,
+      "step": 264
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.8675e-05,
+      "loss": 0.2802,
+      "step": 265
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.867e-05,
+      "loss": 0.2879,
+      "step": 266
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.8665e-05,
+      "loss": 0.2513,
+      "step": 267
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.8660000000000004e-05,
+      "loss": 0.2619,
+      "step": 268
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.8655e-05,
+      "loss": 0.4957,
+      "step": 269
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865e-05,
+      "loss": 0.6091,
+      "step": 270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.8645000000000005e-05,
+      "loss": 0.4097,
+      "step": 271
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864000000000001e-05,
+      "loss": 0.3927,
+      "step": 272
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.8635e-05,
+      "loss": 0.3724,
+      "step": 273
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.863e-05,
+      "loss": 0.3313,
+      "step": 274
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8625e-05,
+      "loss": 0.2487,
+      "step": 275
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.862e-05,
+      "loss": 0.1564,
+      "step": 276
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8615e-05,
+      "loss": 0.2536,
+      "step": 277
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.861e-05,
+      "loss": 0.5395,
+      "step": 278
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.8605e-05,
+      "loss": 0.3284,
+      "step": 279
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.86e-05,
+      "loss": 0.2666,
+      "step": 280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.8595000000000004e-05,
+      "loss": 0.3246,
+      "step": 281
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8590000000000007e-05,
+      "loss": 0.3642,
+      "step": 282
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8585e-05,
+      "loss": 0.4872,
+      "step": 283
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8580000000000005e-05,
+      "loss": 0.2135,
+      "step": 284
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.8575e-05,
+      "loss": 0.2281,
+      "step": 285
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.8569999999999997e-05,
+      "loss": 0.417,
+      "step": 286
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.8565e-05,
+      "loss": 0.3346,
+      "step": 287
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856e-05,
+      "loss": 0.3389,
+      "step": 288
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.8555e-05,
+      "loss": 0.1737,
+      "step": 289
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.855e-05,
+      "loss": 0.2286,
+      "step": 290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.8545e-05,
+      "loss": 0.6892,
+      "step": 291
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.8540000000000005e-05,
+      "loss": 0.1649,
+      "step": 292
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.8535e-05,
+      "loss": 0.1359,
+      "step": 293
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.8530000000000004e-05,
+      "loss": 0.1605,
+      "step": 294
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.8525000000000006e-05,
+      "loss": 0.266,
+      "step": 295
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852e-05,
+      "loss": 0.1546,
+      "step": 296
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8515e-05,
+      "loss": 0.2445,
+      "step": 297
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.851e-05,
+      "loss": 0.1851,
+      "step": 298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8504999999999996e-05,
+      "loss": 0.1986,
+      "step": 299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.85e-05,
+      "loss": 0.1286,
+      "step": 300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8495e-05,
+      "loss": 0.2977,
+      "step": 301
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8490000000000004e-05,
+      "loss": 0.4141,
+      "step": 302
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8485e-05,
+      "loss": 0.4175,
+      "step": 303
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.848e-05,
+      "loss": 0.5086,
+      "step": 304
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8475000000000005e-05,
+      "loss": 0.4574,
+      "step": 305
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.847e-05,
+      "loss": 0.2957,
+      "step": 306
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.8465000000000004e-05,
+      "loss": 0.6006,
+      "step": 307
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.8460000000000006e-05,
+      "loss": 0.3806,
+      "step": 308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8454999999999995e-05,
+      "loss": 0.2417,
+      "step": 309
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845e-05,
+      "loss": 0.2721,
+      "step": 310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8445e-05,
+      "loss": 0.1563,
+      "step": 311
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.844e-05,
+      "loss": 0.3864,
+      "step": 312
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8435e-05,
+      "loss": 0.3947,
+      "step": 313
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.843e-05,
+      "loss": 0.5931,
+      "step": 314
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8425000000000004e-05,
+      "loss": 0.2198,
+      "step": 315
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.842e-05,
+      "loss": 0.3263,
+      "step": 316
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.8415e-05,
+      "loss": 0.2346,
+      "step": 317
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.8410000000000005e-05,
+      "loss": 0.3188,
+      "step": 318
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.8405e-05,
+      "loss": 0.3629,
+      "step": 319
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.84e-05,
+      "loss": 0.3546,
+      "step": 320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.8395e-05,
+      "loss": 0.291,
+      "step": 321
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839e-05,
+      "loss": 0.1589,
+      "step": 322
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8385e-05,
+      "loss": 0.4872,
+      "step": 323
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.838e-05,
+      "loss": 0.411,
+      "step": 324
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8375e-05,
+      "loss": 0.3248,
+      "step": 325
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.837e-05,
+      "loss": 0.1926,
+      "step": 326
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.8365e-05,
+      "loss": 0.3031,
+      "step": 327
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.8360000000000004e-05,
+      "loss": 0.2073,
+      "step": 328
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.8355e-05,
+      "loss": 0.1667,
+      "step": 329
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.835e-05,
+      "loss": 0.4271,
+      "step": 330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.8345000000000005e-05,
+      "loss": 0.2795,
+      "step": 331
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834000000000001e-05,
+      "loss": 0.1861,
+      "step": 332
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.8334999999999996e-05,
+      "loss": 0.4875,
+      "step": 333
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.833e-05,
+      "loss": 0.2104,
+      "step": 334
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8325e-05,
+      "loss": 0.5816,
+      "step": 335
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.832e-05,
+      "loss": 0.1687,
+      "step": 336
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8315e-05,
+      "loss": 0.3454,
+      "step": 337
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.831e-05,
+      "loss": 0.2306,
+      "step": 338
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8305e-05,
+      "loss": 0.3652,
+      "step": 339
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.83e-05,
+      "loss": 0.4242,
+      "step": 340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8295000000000003e-05,
+      "loss": 0.4581,
+      "step": 341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8290000000000006e-05,
+      "loss": 0.2052,
+      "step": 342
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8285e-05,
+      "loss": 0.2195,
+      "step": 343
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8280000000000004e-05,
+      "loss": 0.3154,
+      "step": 344
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8275e-05,
+      "loss": 0.2745,
+      "step": 345
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8269999999999996e-05,
+      "loss": 0.5247,
+      "step": 346
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8265e-05,
+      "loss": 0.2813,
+      "step": 347
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826e-05,
+      "loss": 0.265,
+      "step": 348
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8255000000000004e-05,
+      "loss": 0.4018,
+      "step": 349
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.825e-05,
+      "loss": 0.2793,
+      "step": 350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.8245e-05,
+      "loss": 0.2379,
+      "step": 351
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.8240000000000005e-05,
+      "loss": 0.2108,
+      "step": 352
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.8235e-05,
+      "loss": 0.2337,
+      "step": 353
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.823e-05,
+      "loss": 0.3922,
+      "step": 354
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.8225000000000006e-05,
+      "loss": 0.2223,
+      "step": 355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822e-05,
+      "loss": 0.2691,
+      "step": 356
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.8215e-05,
+      "loss": 0.4151,
+      "step": 357
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.821e-05,
+      "loss": 0.4069,
+      "step": 358
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8205e-05,
+      "loss": 0.1918,
+      "step": 359
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.82e-05,
+      "loss": 0.1593,
+      "step": 360
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8195e-05,
+      "loss": 0.8291,
+      "step": 361
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8190000000000004e-05,
+      "loss": 0.2617,
+      "step": 362
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8185e-05,
+      "loss": 0.3895,
+      "step": 363
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.818e-05,
+      "loss": 0.2347,
+      "step": 364
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8175000000000005e-05,
+      "loss": 0.423,
+      "step": 365
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817e-05,
+      "loss": 0.4341,
+      "step": 366
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.8165e-05,
+      "loss": 0.2364,
+      "step": 367
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.8160000000000006e-05,
+      "loss": 0.2921,
+      "step": 368
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8155e-05,
+      "loss": 0.3628,
+      "step": 369
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.815e-05,
+      "loss": 0.3398,
+      "step": 370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8145e-05,
+      "loss": 0.4922,
+      "step": 371
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.814e-05,
+      "loss": 0.1635,
+      "step": 372
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8135e-05,
+      "loss": 0.4581,
+      "step": 373
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.813e-05,
+      "loss": 0.2601,
+      "step": 374
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8125e-05,
+      "loss": 0.3451,
+      "step": 375
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.812e-05,
+      "loss": 0.398,
+      "step": 376
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8115e-05,
+      "loss": 0.2816,
+      "step": 377
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8110000000000004e-05,
+      "loss": 0.4832,
+      "step": 378
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810500000000001e-05,
+      "loss": 0.195,
+      "step": 379
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.81e-05,
+      "loss": 0.4073,
+      "step": 380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8095000000000005e-05,
+      "loss": 0.3971,
+      "step": 381
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.809e-05,
+      "loss": 0.2488,
+      "step": 382
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8085e-05,
+      "loss": 0.4384,
+      "step": 383
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.808e-05,
+      "loss": 0.3641,
+      "step": 384
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8075e-05,
+      "loss": 0.4822,
+      "step": 385
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.807e-05,
+      "loss": 0.3763,
+      "step": 386
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8065e-05,
+      "loss": 0.2844,
+      "step": 387
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.806e-05,
+      "loss": 0.4352,
+      "step": 388
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8055000000000006e-05,
+      "loss": 0.1786,
+      "step": 389
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.805e-05,
+      "loss": 0.1934,
+      "step": 390
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8045000000000004e-05,
+      "loss": 0.2536,
+      "step": 391
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.804000000000001e-05,
+      "loss": 0.309,
+      "step": 392
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8035e-05,
+      "loss": 0.3739,
+      "step": 393
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.803e-05,
+      "loss": 0.3191,
+      "step": 394
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8025e-05,
+      "loss": 0.324,
+      "step": 395
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.802e-05,
+      "loss": 0.2079,
+      "step": 396
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8015e-05,
+      "loss": 0.343,
+      "step": 397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.801e-05,
+      "loss": 0.4273,
+      "step": 398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8005000000000005e-05,
+      "loss": 0.2937,
+      "step": 399
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.8e-05,
+      "loss": 0.2382,
+      "step": 400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7995e-05,
+      "loss": 0.1909,
+      "step": 401
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7990000000000006e-05,
+      "loss": 0.1734,
+      "step": 402
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7985e-05,
+      "loss": 0.1627,
+      "step": 403
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7980000000000004e-05,
+      "loss": 0.3099,
+      "step": 404
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7975000000000007e-05,
+      "loss": 0.1919,
+      "step": 405
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7969999999999996e-05,
+      "loss": 0.1412,
+      "step": 406
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7965e-05,
+      "loss": 0.208,
+      "step": 407
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796e-05,
+      "loss": 0.3554,
+      "step": 408
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7955e-05,
+      "loss": 0.3278,
+      "step": 409
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.795e-05,
+      "loss": 0.3675,
+      "step": 410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.7945e-05,
+      "loss": 0.2423,
+      "step": 411
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.7940000000000004e-05,
+      "loss": 0.4138,
+      "step": 412
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.7935e-05,
+      "loss": 0.1602,
+      "step": 413
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.793e-05,
+      "loss": 0.1154,
+      "step": 414
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.7925000000000005e-05,
+      "loss": 1.1579,
+      "step": 415
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792e-05,
+      "loss": 0.2921,
+      "step": 416
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.7915000000000004e-05,
+      "loss": 0.2756,
+      "step": 417
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.791e-05,
+      "loss": 0.1548,
+      "step": 418
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.7905e-05,
+      "loss": 0.2239,
+      "step": 419
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.79e-05,
+      "loss": 0.3136,
+      "step": 420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.7895e-05,
+      "loss": 0.1586,
+      "step": 421
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.789e-05,
+      "loss": 0.6804,
+      "step": 422
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.7885e-05,
+      "loss": 0.5067,
+      "step": 423
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.788e-05,
+      "loss": 0.4381,
+      "step": 424
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.7875000000000004e-05,
+      "loss": 0.4275,
+      "step": 425
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787e-05,
+      "loss": 0.2294,
+      "step": 426
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7865e-05,
+      "loss": 0.3967,
+      "step": 427
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7860000000000005e-05,
+      "loss": 0.2469,
+      "step": 428
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785500000000001e-05,
+      "loss": 0.2241,
+      "step": 429
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785e-05,
+      "loss": 0.1819,
+      "step": 430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.7845e-05,
+      "loss": 0.1665,
+      "step": 431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784e-05,
+      "loss": 0.3498,
+      "step": 432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.7835e-05,
+      "loss": 0.3999,
+      "step": 433
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.783e-05,
+      "loss": 0.3258,
+      "step": 434
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.7825e-05,
+      "loss": 0.2113,
+      "step": 435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.782e-05,
+      "loss": 0.3336,
+      "step": 436
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7815e-05,
+      "loss": 0.2155,
+      "step": 437
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7810000000000004e-05,
+      "loss": 0.1649,
+      "step": 438
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.7805000000000006e-05,
+      "loss": 0.5063,
+      "step": 439
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.78e-05,
+      "loss": 0.2629,
+      "step": 440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.7795000000000005e-05,
+      "loss": 0.1958,
+      "step": 441
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779e-05,
+      "loss": 0.317,
+      "step": 442
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.7784999999999996e-05,
+      "loss": 0.5024,
+      "step": 443
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.778e-05,
+      "loss": 0.3084,
+      "step": 444
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.7775e-05,
+      "loss": 0.2583,
+      "step": 445
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.777e-05,
+      "loss": 0.4089,
+      "step": 446
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.7765e-05,
+      "loss": 0.3285,
+      "step": 447
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.776e-05,
+      "loss": 0.2792,
+      "step": 448
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.7755000000000005e-05,
+      "loss": 0.3462,
+      "step": 449
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.775e-05,
+      "loss": 0.6242,
+      "step": 450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7745000000000004e-05,
+      "loss": 0.3534,
+      "step": 451
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7740000000000006e-05,
+      "loss": 0.1181,
+      "step": 452
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7735e-05,
+      "loss": 0.6767,
+      "step": 453
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.773e-05,
+      "loss": 0.7082,
+      "step": 454
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7725e-05,
+      "loss": 0.346,
+      "step": 455
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7719999999999996e-05,
+      "loss": 0.3114,
+      "step": 456
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7715e-05,
+      "loss": 0.386,
+      "step": 457
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.771e-05,
+      "loss": 0.3685,
+      "step": 458
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.7705000000000004e-05,
+      "loss": 0.8115,
+      "step": 459
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.77e-05,
+      "loss": 0.3613,
+      "step": 460
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7695e-05,
+      "loss": 0.2762,
+      "step": 461
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7690000000000005e-05,
+      "loss": 0.2646,
+      "step": 462
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7685e-05,
+      "loss": 0.318,
+      "step": 463
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7680000000000003e-05,
+      "loss": 0.1977,
+      "step": 464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.7675000000000006e-05,
+      "loss": 0.2257,
+      "step": 465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.7669999999999995e-05,
+      "loss": 0.4197,
+      "step": 466
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7665e-05,
+      "loss": 0.1389,
+      "step": 467
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766e-05,
+      "loss": 0.2679,
+      "step": 468
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7655e-05,
+      "loss": 0.2622,
+      "step": 469
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.765e-05,
+      "loss": 0.3097,
+      "step": 470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7645e-05,
+      "loss": 0.3208,
+      "step": 471
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7640000000000004e-05,
+      "loss": 0.3781,
+      "step": 472
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7635e-05,
+      "loss": 0.4351,
+      "step": 473
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.763e-05,
+      "loss": 0.3293,
+      "step": 474
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7625000000000005e-05,
+      "loss": 0.1293,
+      "step": 475
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762e-05,
+      "loss": 0.356,
+      "step": 476
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.7615e-05,
+      "loss": 0.1804,
+      "step": 477
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.761e-05,
+      "loss": 0.275,
+      "step": 478
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7605e-05,
+      "loss": 0.2013,
+      "step": 479
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.76e-05,
+      "loss": 0.185,
+      "step": 480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7595e-05,
+      "loss": 0.2278,
+      "step": 481
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.759e-05,
+      "loss": 0.3011,
+      "step": 482
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.7585e-05,
+      "loss": 0.3125,
+      "step": 483
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.758e-05,
+      "loss": 0.3609,
+      "step": 484
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7575000000000004e-05,
+      "loss": 0.3069,
+      "step": 485
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757e-05,
+      "loss": 0.1969,
+      "step": 486
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7565e-05,
+      "loss": 0.2419,
+      "step": 487
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7560000000000005e-05,
+      "loss": 0.225,
+      "step": 488
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755500000000001e-05,
+      "loss": 0.3952,
+      "step": 489
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.7549999999999996e-05,
+      "loss": 0.3058,
+      "step": 490
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.7545e-05,
+      "loss": 0.1397,
+      "step": 491
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754e-05,
+      "loss": 0.4229,
+      "step": 492
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.7535e-05,
+      "loss": 0.1837,
+      "step": 493
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.753e-05,
+      "loss": 0.2899,
+      "step": 494
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.7525e-05,
+      "loss": 0.1582,
+      "step": 495
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.752e-05,
+      "loss": 0.2489,
+      "step": 496
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.7515e-05,
+      "loss": 0.2457,
+      "step": 497
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.751e-05,
+      "loss": 0.3108,
+      "step": 498
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.7505000000000006e-05,
+      "loss": 0.2415,
+      "step": 499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.75e-05,
+      "loss": 0.4864,
+      "step": 500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.7495000000000004e-05,
+      "loss": 0.2747,
+      "step": 501
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.749000000000001e-05,
+      "loss": 0.2753,
+      "step": 502
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.7484999999999996e-05,
+      "loss": 0.2676,
+      "step": 503
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.748e-05,
+      "loss": 0.519,
+      "step": 504
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.7475e-05,
+      "loss": 0.1641,
+      "step": 505
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.747e-05,
+      "loss": 0.2388,
+      "step": 506
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.7465e-05,
+      "loss": 0.2406,
+      "step": 507
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.746e-05,
+      "loss": 0.321,
+      "step": 508
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.7455000000000005e-05,
+      "loss": 0.2911,
+      "step": 509
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745e-05,
+      "loss": 0.2589,
+      "step": 510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7445e-05,
+      "loss": 0.4161,
+      "step": 511
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7440000000000006e-05,
+      "loss": 0.197,
+      "step": 512
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7435e-05,
+      "loss": 0.2775,
+      "step": 513
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7430000000000004e-05,
+      "loss": 0.2292,
+      "step": 514
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7425e-05,
+      "loss": 0.4308,
+      "step": 515
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.742e-05,
+      "loss": 0.1431,
+      "step": 516
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7415e-05,
+      "loss": 0.1623,
+      "step": 517
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.741e-05,
+      "loss": 0.2128,
+      "step": 518
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7405000000000003e-05,
+      "loss": 0.1995,
+      "step": 519
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.74e-05,
+      "loss": 0.1989,
+      "step": 520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7395e-05,
+      "loss": 0.1677,
+      "step": 521
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7390000000000004e-05,
+      "loss": 0.3216,
+      "step": 522
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7385e-05,
+      "loss": 0.2131,
+      "step": 523
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.738e-05,
+      "loss": 0.1926,
+      "step": 524
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.7375000000000005e-05,
+      "loss": 0.116,
+      "step": 525
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737e-05,
+      "loss": 0.129,
+      "step": 526
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.7365e-05,
+      "loss": 0.5607,
+      "step": 527
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.736e-05,
+      "loss": 0.3864,
+      "step": 528
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.7355e-05,
+      "loss": 0.137,
+      "step": 529
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.735e-05,
+      "loss": 0.5119,
+      "step": 530
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.7345e-05,
+      "loss": 0.5951,
+      "step": 531
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.734e-05,
+      "loss": 0.3638,
+      "step": 532
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.7335e-05,
+      "loss": 0.4133,
+      "step": 533
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.733e-05,
+      "loss": 0.3498,
+      "step": 534
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.7325000000000004e-05,
+      "loss": 0.1605,
+      "step": 535
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.732000000000001e-05,
+      "loss": 0.4576,
+      "step": 536
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7315e-05,
+      "loss": 0.2813,
+      "step": 537
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7310000000000005e-05,
+      "loss": 0.2811,
+      "step": 538
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.7305e-05,
+      "loss": 0.6759,
+      "step": 539
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.73e-05,
+      "loss": 0.4413,
+      "step": 540
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.7295e-05,
+      "loss": 0.3489,
+      "step": 541
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.729e-05,
+      "loss": 0.1644,
+      "step": 542
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.7285e-05,
+      "loss": 0.3661,
+      "step": 543
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.728e-05,
+      "loss": 0.2105,
+      "step": 544
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.7275e-05,
+      "loss": 0.1431,
+      "step": 545
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.7270000000000006e-05,
+      "loss": 0.3887,
+      "step": 546
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7265e-05,
+      "loss": 0.22,
+      "step": 547
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7260000000000004e-05,
+      "loss": 0.3937,
+      "step": 548
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725500000000001e-05,
+      "loss": 0.2516,
+      "step": 549
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725e-05,
+      "loss": 0.2362,
+      "step": 550
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7245e-05,
+      "loss": 0.3951,
+      "step": 551
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724e-05,
+      "loss": 0.4749,
+      "step": 552
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7235e-05,
+      "loss": 0.1472,
+      "step": 553
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.723e-05,
+      "loss": 0.2896,
+      "step": 554
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7225e-05,
+      "loss": 0.4259,
+      "step": 555
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7220000000000004e-05,
+      "loss": 0.2504,
+      "step": 556
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7215e-05,
+      "loss": 0.3231,
+      "step": 557
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.721e-05,
+      "loss": 0.4463,
+      "step": 558
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7205000000000005e-05,
+      "loss": 0.3125,
+      "step": 559
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.72e-05,
+      "loss": 0.2814,
+      "step": 560
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7195000000000004e-05,
+      "loss": 0.1927,
+      "step": 561
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7190000000000006e-05,
+      "loss": 0.1865,
+      "step": 562
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7184999999999995e-05,
+      "loss": 0.1367,
+      "step": 563
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.718e-05,
+      "loss": 0.3304,
+      "step": 564
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7175e-05,
+      "loss": 0.3056,
+      "step": 565
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.717e-05,
+      "loss": 0.4243,
+      "step": 566
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.7165e-05,
+      "loss": 0.1464,
+      "step": 567
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.716e-05,
+      "loss": 0.3077,
+      "step": 568
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.7155000000000004e-05,
+      "loss": 0.4069,
+      "step": 569
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.715e-05,
+      "loss": 0.1367,
+      "step": 570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.7145e-05,
+      "loss": 0.3026,
+      "step": 571
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.7140000000000005e-05,
+      "loss": 0.3837,
+      "step": 572
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.7135e-05,
+      "loss": 0.2914,
+      "step": 573
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.7130000000000004e-05,
+      "loss": 0.4488,
+      "step": 574
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.7125e-05,
+      "loss": 0.4857,
+      "step": 575
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712e-05,
+      "loss": 0.5345,
+      "step": 576
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.7115e-05,
+      "loss": 0.3238,
+      "step": 577
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.711e-05,
+      "loss": 0.2408,
+      "step": 578
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.7105e-05,
+      "loss": 0.2476,
+      "step": 579
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.71e-05,
+      "loss": 0.3123,
+      "step": 580
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.7095e-05,
+      "loss": 0.3867,
+      "step": 581
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.7090000000000004e-05,
+      "loss": 0.2234,
+      "step": 582
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.7085e-05,
+      "loss": 0.4886,
+      "step": 583
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708e-05,
+      "loss": 0.1696,
+      "step": 584
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7075000000000005e-05,
+      "loss": 0.3511,
+      "step": 585
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707000000000001e-05,
+      "loss": 0.3591,
+      "step": 586
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7064999999999997e-05,
+      "loss": 0.3509,
+      "step": 587
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.706e-05,
+      "loss": 0.1643,
+      "step": 588
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7055e-05,
+      "loss": 0.3154,
+      "step": 589
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.705e-05,
+      "loss": 0.1859,
+      "step": 590
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.7045e-05,
+      "loss": 0.423,
+      "step": 591
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.704e-05,
+      "loss": 0.1574,
+      "step": 592
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.7035e-05,
+      "loss": 0.3342,
+      "step": 593
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.703e-05,
+      "loss": 0.5308,
+      "step": 594
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7025000000000004e-05,
+      "loss": 0.4901,
+      "step": 595
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7020000000000006e-05,
+      "loss": 0.188,
+      "step": 596
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7015e-05,
+      "loss": 0.4129,
+      "step": 597
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7010000000000005e-05,
+      "loss": 0.1978,
+      "step": 598
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.7005e-05,
+      "loss": 0.1521,
+      "step": 599
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 0.2486,
+      "step": 600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6995e-05,
+      "loss": 0.1646,
+      "step": 601
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.699e-05,
+      "loss": 0.1878,
+      "step": 602
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.6985e-05,
+      "loss": 0.1148,
+      "step": 603
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.698e-05,
+      "loss": 0.2011,
+      "step": 604
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6975e-05,
+      "loss": 0.2471,
+      "step": 605
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6970000000000005e-05,
+      "loss": 0.3073,
+      "step": 606
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.6965e-05,
+      "loss": 0.3269,
+      "step": 607
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.6960000000000004e-05,
+      "loss": 0.3666,
+      "step": 608
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6955000000000006e-05,
+      "loss": 0.4325,
+      "step": 609
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695e-05,
+      "loss": 0.1958,
+      "step": 610
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6945e-05,
+      "loss": 0.1829,
+      "step": 611
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.694e-05,
+      "loss": 0.2069,
+      "step": 612
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6934999999999996e-05,
+      "loss": 0.1436,
+      "step": 613
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.693e-05,
+      "loss": 0.5529,
+      "step": 614
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6925e-05,
+      "loss": 0.1851,
+      "step": 615
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6920000000000004e-05,
+      "loss": 0.3368,
+      "step": 616
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6915e-05,
+      "loss": 0.3021,
+      "step": 617
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.691e-05,
+      "loss": 0.3459,
+      "step": 618
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.6905000000000005e-05,
+      "loss": 0.2836,
+      "step": 619
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.69e-05,
+      "loss": 0.4397,
+      "step": 620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.6895e-05,
+      "loss": 0.3548,
+      "step": 621
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.6890000000000006e-05,
+      "loss": 0.1577,
+      "step": 622
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6884999999999995e-05,
+      "loss": 0.2571,
+      "step": 623
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.688e-05,
+      "loss": 0.4337,
+      "step": 624
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6875e-05,
+      "loss": 0.3559,
+      "step": 625
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.687e-05,
+      "loss": 0.3605,
+      "step": 626
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6865e-05,
+      "loss": 0.3232,
+      "step": 627
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.686e-05,
+      "loss": 0.1923,
+      "step": 628
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6855000000000004e-05,
+      "loss": 0.3535,
+      "step": 629
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.685e-05,
+      "loss": 0.391,
+      "step": 630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6845e-05,
+      "loss": 0.3177,
+      "step": 631
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6840000000000005e-05,
+      "loss": 0.4109,
+      "step": 632
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.6835e-05,
+      "loss": 0.355,
+      "step": 633
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683e-05,
+      "loss": 0.3096,
+      "step": 634
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.6825000000000006e-05,
+      "loss": 0.2778,
+      "step": 635
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.682e-05,
+      "loss": 0.3986,
+      "step": 636
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.6815e-05,
+      "loss": 0.1316,
+      "step": 637
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.681e-05,
+      "loss": 0.2947,
+      "step": 638
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6805e-05,
+      "loss": 0.1853,
+      "step": 639
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.68e-05,
+      "loss": 0.1767,
+      "step": 640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.6795e-05,
+      "loss": 0.2242,
+      "step": 641
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.6790000000000003e-05,
+      "loss": 0.3228,
+      "step": 642
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.6785e-05,
+      "loss": 0.262,
+      "step": 643
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678e-05,
+      "loss": 0.1706,
+      "step": 644
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6775000000000004e-05,
+      "loss": 0.4403,
+      "step": 645
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.677000000000001e-05,
+      "loss": 0.2187,
+      "step": 646
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.6765e-05,
+      "loss": 0.1649,
+      "step": 647
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.676e-05,
+      "loss": 0.3233,
+      "step": 648
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.6755e-05,
+      "loss": 0.2533,
+      "step": 649
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.675e-05,
+      "loss": 0.157,
+      "step": 650
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.6745e-05,
+      "loss": 0.2198,
+      "step": 651
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.674e-05,
+      "loss": 0.1924,
+      "step": 652
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.6735e-05,
+      "loss": 0.213,
+      "step": 653
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.673e-05,
+      "loss": 0.468,
+      "step": 654
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.6725e-05,
+      "loss": 0.1999,
+      "step": 655
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.6720000000000006e-05,
+      "loss": 0.2046,
+      "step": 656
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.6715e-05,
+      "loss": 0.4882,
+      "step": 657
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.6710000000000004e-05,
+      "loss": 0.474,
+      "step": 658
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.670500000000001e-05,
+      "loss": 0.2926,
+      "step": 659
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.6699999999999996e-05,
+      "loss": 0.3351,
+      "step": 660
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.6695e-05,
+      "loss": 0.2524,
+      "step": 661
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.669e-05,
+      "loss": 0.4165,
+      "step": 662
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.6685e-05,
+      "loss": 0.2623,
+      "step": 663
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.668e-05,
+      "loss": 0.3169,
+      "step": 664
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.6675e-05,
+      "loss": 0.2244,
+      "step": 665
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.6670000000000005e-05,
+      "loss": 0.4057,
+      "step": 666
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.6665e-05,
+      "loss": 0.2955,
+      "step": 667
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666e-05,
+      "loss": 0.3978,
+      "step": 668
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6655000000000006e-05,
+      "loss": 0.1889,
+      "step": 669
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.665e-05,
+      "loss": 0.2073,
+      "step": 670
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.6645000000000004e-05,
+      "loss": 0.1687,
+      "step": 671
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.664e-05,
+      "loss": 0.3286,
+      "step": 672
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.6634999999999996e-05,
+      "loss": 0.2286,
+      "step": 673
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.663e-05,
+      "loss": 0.2582,
+      "step": 674
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.6625e-05,
+      "loss": 0.1304,
+      "step": 675
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.6620000000000003e-05,
+      "loss": 0.3222,
+      "step": 676
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.6615e-05,
+      "loss": 0.1801,
+      "step": 677
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.661e-05,
+      "loss": 0.2844,
+      "step": 678
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6605000000000004e-05,
+      "loss": 0.2713,
+      "step": 679
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.66e-05,
+      "loss": 0.4297,
+      "step": 680
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.6595e-05,
+      "loss": 0.3766,
+      "step": 681
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.6590000000000005e-05,
+      "loss": 0.458,
+      "step": 682
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6585e-05,
+      "loss": 0.3027,
+      "step": 683
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.658e-05,
+      "loss": 0.3336,
+      "step": 684
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6575e-05,
+      "loss": 0.1273,
+      "step": 685
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.657e-05,
+      "loss": 0.3881,
+      "step": 686
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.6565e-05,
+      "loss": 0.1421,
+      "step": 687
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.656e-05,
+      "loss": 0.3209,
+      "step": 688
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.6555e-05,
+      "loss": 0.2121,
+      "step": 689
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.655e-05,
+      "loss": 0.2481,
+      "step": 690
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6545e-05,
+      "loss": 0.1546,
+      "step": 691
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6540000000000004e-05,
+      "loss": 0.4866,
+      "step": 692
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.6535e-05,
+      "loss": 0.22,
+      "step": 693
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653e-05,
+      "loss": 0.524,
+      "step": 694
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6525000000000005e-05,
+      "loss": 0.6278,
+      "step": 695
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652e-05,
+      "loss": 0.4508,
+      "step": 696
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6515e-05,
+      "loss": 0.1236,
+      "step": 697
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.651e-05,
+      "loss": 0.153,
+      "step": 698
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.6505e-05,
+      "loss": 0.2203,
+      "step": 699
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.65e-05,
+      "loss": 0.1822,
+      "step": 700
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.6495e-05,
+      "loss": 0.2418,
+      "step": 701
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649e-05,
+      "loss": 0.2971,
+      "step": 702
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.6485000000000006e-05,
+      "loss": 0.3674,
+      "step": 703
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648e-05,
+      "loss": 0.4231,
+      "step": 704
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.6475000000000004e-05,
+      "loss": 0.2436,
+      "step": 705
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.6470000000000007e-05,
+      "loss": 0.2025,
+      "step": 706
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.6465e-05,
+      "loss": 0.2831,
+      "step": 707
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.646e-05,
+      "loss": 0.3786,
+      "step": 708
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.6455e-05,
+      "loss": 0.272,
+      "step": 709
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6449999999999997e-05,
+      "loss": 0.3684,
+      "step": 710
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6445e-05,
+      "loss": 0.3038,
+      "step": 711
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.644e-05,
+      "loss": 0.2826,
+      "step": 712
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6435000000000004e-05,
+      "loss": 0.2359,
+      "step": 713
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.643e-05,
+      "loss": 0.2096,
+      "step": 714
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6425e-05,
+      "loss": 0.2536,
+      "step": 715
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6420000000000005e-05,
+      "loss": 0.4817,
+      "step": 716
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6415e-05,
+      "loss": 0.2442,
+      "step": 717
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6410000000000004e-05,
+      "loss": 0.2573,
+      "step": 718
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6405000000000006e-05,
+      "loss": 0.152,
+      "step": 719
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 0.1829,
+      "step": 720
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6395e-05,
+      "loss": 0.2628,
+      "step": 721
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.639e-05,
+      "loss": 0.2693,
+      "step": 722
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.6385e-05,
+      "loss": 0.429,
+      "step": 723
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.638e-05,
+      "loss": 0.1369,
+      "step": 724
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.6375e-05,
+      "loss": 0.3751,
+      "step": 725
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.6370000000000004e-05,
+      "loss": 0.2491,
+      "step": 726
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.6365e-05,
+      "loss": 0.3314,
+      "step": 727
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636e-05,
+      "loss": 0.2205,
+      "step": 728
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.6355000000000005e-05,
+      "loss": 0.2928,
+      "step": 729
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635e-05,
+      "loss": 0.2267,
+      "step": 730
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.6345000000000003e-05,
+      "loss": 0.3456,
+      "step": 731
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.634e-05,
+      "loss": 0.1961,
+      "step": 732
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.6335e-05,
+      "loss": 0.3346,
+      "step": 733
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.633e-05,
+      "loss": 0.4079,
+      "step": 734
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.6325e-05,
+      "loss": 0.2396,
+      "step": 735
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.632e-05,
+      "loss": 0.474,
+      "step": 736
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.6315e-05,
+      "loss": 0.2553,
+      "step": 737
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.631e-05,
+      "loss": 0.2367,
+      "step": 738
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.6305000000000004e-05,
+      "loss": 0.1473,
+      "step": 739
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.63e-05,
+      "loss": 0.441,
+      "step": 740
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.6295e-05,
+      "loss": 0.2828,
+      "step": 741
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.6290000000000005e-05,
+      "loss": 0.4787,
+      "step": 742
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628500000000001e-05,
+      "loss": 0.3052,
+      "step": 743
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.6279999999999996e-05,
+      "loss": 0.2214,
+      "step": 744
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.6275e-05,
+      "loss": 0.1372,
+      "step": 745
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.627e-05,
+      "loss": 0.3145,
+      "step": 746
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.6265e-05,
+      "loss": 0.2419,
+      "step": 747
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.626e-05,
+      "loss": 0.2134,
+      "step": 748
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.6255e-05,
+      "loss": 0.2462,
+      "step": 749
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625e-05,
+      "loss": 0.2898,
+      "step": 750
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.6245e-05,
+      "loss": 0.2592,
+      "step": 751
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.6240000000000004e-05,
+      "loss": 0.2164,
+      "step": 752
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6235000000000006e-05,
+      "loss": 0.2385,
+      "step": 753
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.623e-05,
+      "loss": 0.3989,
+      "step": 754
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.6225000000000005e-05,
+      "loss": 0.1592,
+      "step": 755
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.622000000000001e-05,
+      "loss": 0.2368,
+      "step": 756
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.6214999999999996e-05,
+      "loss": 0.1056,
+      "step": 757
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.621e-05,
+      "loss": 0.441,
+      "step": 758
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6205e-05,
+      "loss": 0.2938,
+      "step": 759
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.62e-05,
+      "loss": 0.1384,
+      "step": 760
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.6195e-05,
+      "loss": 0.3453,
+      "step": 761
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619e-05,
+      "loss": 0.31,
+      "step": 762
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.6185000000000005e-05,
+      "loss": 0.3082,
+      "step": 763
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.618e-05,
+      "loss": 0.3492,
+      "step": 764
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6175000000000003e-05,
+      "loss": 0.2969,
+      "step": 765
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6170000000000006e-05,
+      "loss": 0.3576,
+      "step": 766
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6165e-05,
+      "loss": 0.1775,
+      "step": 767
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6160000000000004e-05,
+      "loss": 0.3766,
+      "step": 768
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6155e-05,
+      "loss": 0.1222,
+      "step": 769
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6149999999999996e-05,
+      "loss": 0.1619,
+      "step": 770
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6145e-05,
+      "loss": 0.293,
+      "step": 771
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.614e-05,
+      "loss": 0.1746,
+      "step": 772
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6135000000000004e-05,
+      "loss": 0.1584,
+      "step": 773
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.613e-05,
+      "loss": 0.1577,
+      "step": 774
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6125e-05,
+      "loss": 0.2592,
+      "step": 775
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6120000000000005e-05,
+      "loss": 0.3324,
+      "step": 776
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.6115e-05,
+      "loss": 0.3724,
+      "step": 777
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611e-05,
+      "loss": 0.4033,
+      "step": 778
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.6105000000000006e-05,
+      "loss": 0.3041,
+      "step": 779
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.61e-05,
+      "loss": 0.2603,
+      "step": 780
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.6095e-05,
+      "loss": 0.2231,
+      "step": 781
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.609e-05,
+      "loss": 0.2785,
+      "step": 782
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6085e-05,
+      "loss": 0.2576,
+      "step": 783
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.608e-05,
+      "loss": 0.1476,
+      "step": 784
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6075e-05,
+      "loss": 0.3166,
+      "step": 785
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6070000000000004e-05,
+      "loss": 0.2515,
+      "step": 786
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6065e-05,
+      "loss": 0.2111,
+      "step": 787
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606e-05,
+      "loss": 0.2455,
+      "step": 788
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.6055000000000005e-05,
+      "loss": 0.3504,
+      "step": 789
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605e-05,
+      "loss": 0.3375,
+      "step": 790
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.6045e-05,
+      "loss": 0.099,
+      "step": 791
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.6040000000000006e-05,
+      "loss": 0.4889,
+      "step": 792
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.6035e-05,
+      "loss": 0.3773,
+      "step": 793
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.603e-05,
+      "loss": 0.395,
+      "step": 794
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.6025e-05,
+      "loss": 0.1553,
+      "step": 795
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.602e-05,
+      "loss": 0.2494,
+      "step": 796
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.6015e-05,
+      "loss": 0.1823,
+      "step": 797
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.601e-05,
+      "loss": 0.12,
+      "step": 798
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.6005e-05,
+      "loss": 0.2782,
+      "step": 799
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.6e-05,
+      "loss": 0.3908,
+      "step": 800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.5995e-05,
+      "loss": 0.352,
+      "step": 801
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.5990000000000004e-05,
+      "loss": 0.1401,
+      "step": 802
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.598500000000001e-05,
+      "loss": 0.4964,
+      "step": 803
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.598e-05,
+      "loss": 0.1369,
+      "step": 804
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.5975e-05,
+      "loss": 0.2905,
+      "step": 805
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.597e-05,
+      "loss": 0.2107,
+      "step": 806
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.5965e-05,
+      "loss": 0.2872,
+      "step": 807
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.596e-05,
+      "loss": 0.3889,
+      "step": 808
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.5955e-05,
+      "loss": 0.3379,
+      "step": 809
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.595e-05,
+      "loss": 0.221,
+      "step": 810
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.5945e-05,
+      "loss": 0.1541,
+      "step": 811
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594e-05,
+      "loss": 0.1205,
+      "step": 812
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.5935000000000006e-05,
+      "loss": 0.3322,
+      "step": 813
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.593e-05,
+      "loss": 0.2089,
+      "step": 814
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.5925000000000004e-05,
+      "loss": 0.3227,
+      "step": 815
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.592000000000001e-05,
+      "loss": 0.1515,
+      "step": 816
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.5914999999999996e-05,
+      "loss": 0.1499,
+      "step": 817
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.591e-05,
+      "loss": 0.368,
+      "step": 818
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.5905e-05,
+      "loss": 0.4388,
+      "step": 819
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.59e-05,
+      "loss": 0.1432,
+      "step": 820
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.5895e-05,
+      "loss": 0.1556,
+      "step": 821
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589e-05,
+      "loss": 0.1667,
+      "step": 822
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.5885000000000004e-05,
+      "loss": 0.12,
+      "step": 823
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588e-05,
+      "loss": 0.3466,
+      "step": 824
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.5875e-05,
+      "loss": 0.2401,
+      "step": 825
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.5870000000000005e-05,
+      "loss": 0.303,
+      "step": 826
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.5865e-05,
+      "loss": 0.3943,
+      "step": 827
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.5860000000000004e-05,
+      "loss": 0.1657,
+      "step": 828
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.5855e-05,
+      "loss": 0.2513,
+      "step": 829
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.5849999999999996e-05,
+      "loss": 0.2418,
+      "step": 830
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.5845e-05,
+      "loss": 0.1739,
+      "step": 831
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.584e-05,
+      "loss": 0.2532,
+      "step": 832
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.5835e-05,
+      "loss": 0.437,
+      "step": 833
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.583e-05,
+      "loss": 0.1776,
+      "step": 834
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.5825e-05,
+      "loss": 0.2816,
+      "step": 835
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.5820000000000004e-05,
+      "loss": 0.2871,
+      "step": 836
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.5815e-05,
+      "loss": 0.3183,
+      "step": 837
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.581e-05,
+      "loss": 0.4972,
+      "step": 838
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.5805000000000005e-05,
+      "loss": 0.3298,
+      "step": 839
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.58e-05,
+      "loss": 0.2668,
+      "step": 840
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.5795e-05,
+      "loss": 0.091,
+      "step": 841
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.579e-05,
+      "loss": 0.31,
+      "step": 842
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.5785e-05,
+      "loss": 0.3621,
+      "step": 843
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.578e-05,
+      "loss": 0.2681,
+      "step": 844
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.5775e-05,
+      "loss": 0.1871,
+      "step": 845
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.577e-05,
+      "loss": 0.4439,
+      "step": 846
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.5765e-05,
+      "loss": 0.2848,
+      "step": 847
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.576e-05,
+      "loss": 0.2463,
+      "step": 848
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.5755000000000004e-05,
+      "loss": 0.4768,
+      "step": 849
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.575e-05,
+      "loss": 0.2841,
+      "step": 850
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.5745e-05,
+      "loss": 0.3017,
+      "step": 851
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.5740000000000005e-05,
+      "loss": 0.3466,
+      "step": 852
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.5735e-05,
+      "loss": 0.1418,
+      "step": 853
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.573e-05,
+      "loss": 0.238,
+      "step": 854
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.5725e-05,
+      "loss": 0.1448,
+      "step": 855
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.572e-05,
+      "loss": 0.3548,
+      "step": 856
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.5715e-05,
+      "loss": 0.6272,
+      "step": 857
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.571e-05,
+      "loss": 0.4371,
+      "step": 858
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.5705e-05,
+      "loss": 0.3282,
+      "step": 859
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.5700000000000005e-05,
+      "loss": 0.2555,
+      "step": 860
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5695e-05,
+      "loss": 0.2816,
+      "step": 861
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5690000000000004e-05,
+      "loss": 0.3576,
+      "step": 862
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5685000000000006e-05,
+      "loss": 0.124,
+      "step": 863
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.568e-05,
+      "loss": 0.4011,
+      "step": 864
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5675e-05,
+      "loss": 0.1204,
+      "step": 865
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.567e-05,
+      "loss": 0.362,
+      "step": 866
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.5664999999999996e-05,
+      "loss": 0.3052,
+      "step": 867
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.566e-05,
+      "loss": 0.3695,
+      "step": 868
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5655e-05,
+      "loss": 0.367,
+      "step": 869
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5650000000000004e-05,
+      "loss": 0.1846,
+      "step": 870
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.5645e-05,
+      "loss": 0.1497,
+      "step": 871
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.564e-05,
+      "loss": 0.4353,
+      "step": 872
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.5635000000000005e-05,
+      "loss": 0.226,
+      "step": 873
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.563e-05,
+      "loss": 0.3458,
+      "step": 874
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.5625000000000004e-05,
+      "loss": 0.1691,
+      "step": 875
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.5620000000000006e-05,
+      "loss": 0.1304,
+      "step": 876
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.5615e-05,
+      "loss": 0.2217,
+      "step": 877
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.561e-05,
+      "loss": 0.1999,
+      "step": 878
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.5605e-05,
+      "loss": 0.4365,
+      "step": 879
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.56e-05,
+      "loss": 0.2221,
+      "step": 880
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.5595e-05,
+      "loss": 0.2715,
+      "step": 881
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.559e-05,
+      "loss": 0.1407,
+      "step": 882
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.5585000000000004e-05,
+      "loss": 0.3228,
+      "step": 883
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.558e-05,
+      "loss": 0.3544,
+      "step": 884
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.5575e-05,
+      "loss": 0.4459,
+      "step": 885
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.5570000000000005e-05,
+      "loss": 0.1591,
+      "step": 886
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.5565e-05,
+      "loss": 0.1275,
+      "step": 887
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.556e-05,
+      "loss": 0.2358,
+      "step": 888
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.5555000000000006e-05,
+      "loss": 0.2136,
+      "step": 889
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.555e-05,
+      "loss": 0.1612,
+      "step": 890
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.5545e-05,
+      "loss": 0.1786,
+      "step": 891
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.554e-05,
+      "loss": 0.1233,
+      "step": 892
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.5535e-05,
+      "loss": 0.2124,
+      "step": 893
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.553e-05,
+      "loss": 0.3278,
+      "step": 894
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.5525e-05,
+      "loss": 0.2868,
+      "step": 895
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.5520000000000004e-05,
+      "loss": 0.1188,
+      "step": 896
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.5515e-05,
+      "loss": 0.1414,
+      "step": 897
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.551e-05,
+      "loss": 0.1637,
+      "step": 898
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.5505000000000005e-05,
+      "loss": 0.3187,
+      "step": 899
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.550000000000001e-05,
+      "loss": 0.4106,
+      "step": 900
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.5495e-05,
+      "loss": 0.4103,
+      "step": 901
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.549e-05,
+      "loss": 0.3277,
+      "step": 902
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.5485e-05,
+      "loss": 0.2859,
+      "step": 903
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.548e-05,
+      "loss": 0.174,
+      "step": 904
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.5475e-05,
+      "loss": 0.3997,
+      "step": 905
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.547e-05,
+      "loss": 0.1805,
+      "step": 906
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.5465e-05,
+      "loss": 0.4128,
+      "step": 907
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.546e-05,
+      "loss": 0.3403,
+      "step": 908
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.5455000000000004e-05,
+      "loss": 0.1449,
+      "step": 909
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.5450000000000006e-05,
+      "loss": 0.1652,
+      "step": 910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.5445e-05,
+      "loss": 0.3162,
+      "step": 911
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.5440000000000005e-05,
+      "loss": 0.1455,
+      "step": 912
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.543500000000001e-05,
+      "loss": 0.4442,
+      "step": 913
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.5429999999999996e-05,
+      "loss": 0.1228,
+      "step": 914
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.5425e-05,
+      "loss": 0.1962,
+      "step": 915
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.542e-05,
+      "loss": 0.3838,
+      "step": 916
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.5415e-05,
+      "loss": 0.2746,
+      "step": 917
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.541e-05,
+      "loss": 0.2201,
+      "step": 918
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.5405e-05,
+      "loss": 0.4354,
+      "step": 919
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 0.3854,
+      "step": 920
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.5395e-05,
+      "loss": 0.2401,
+      "step": 921
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.539e-05,
+      "loss": 0.2538,
+      "step": 922
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.5385000000000006e-05,
+      "loss": 0.1333,
+      "step": 923
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.538e-05,
+      "loss": 0.3147,
+      "step": 924
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.5375000000000004e-05,
+      "loss": 0.5098,
+      "step": 925
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.537e-05,
+      "loss": 0.1819,
+      "step": 926
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.5364999999999996e-05,
+      "loss": 0.2986,
+      "step": 927
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.536e-05,
+      "loss": 0.2938,
+      "step": 928
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.5355e-05,
+      "loss": 0.1293,
+      "step": 929
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.5350000000000004e-05,
+      "loss": 0.3343,
+      "step": 930
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.5345e-05,
+      "loss": 0.161,
+      "step": 931
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.534e-05,
+      "loss": 0.2054,
+      "step": 932
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.5335000000000005e-05,
+      "loss": 0.1862,
+      "step": 933
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.533e-05,
+      "loss": 0.1929,
+      "step": 934
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.5325e-05,
+      "loss": 0.2433,
+      "step": 935
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.5320000000000006e-05,
+      "loss": 0.5474,
+      "step": 936
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.5315e-05,
+      "loss": 0.2999,
+      "step": 937
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.531e-05,
+      "loss": 0.1979,
+      "step": 938
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.5305e-05,
+      "loss": 0.1139,
+      "step": 939
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.53e-05,
+      "loss": 0.1959,
+      "step": 940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.5295e-05,
+      "loss": 0.4542,
+      "step": 941
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.529e-05,
+      "loss": 0.1425,
+      "step": 942
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.5285000000000003e-05,
+      "loss": 0.1477,
+      "step": 943
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.528e-05,
+      "loss": 0.2822,
+      "step": 944
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5275e-05,
+      "loss": 0.1532,
+      "step": 945
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5270000000000004e-05,
+      "loss": 0.1624,
+      "step": 946
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5265e-05,
+      "loss": 0.4238,
+      "step": 947
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.526e-05,
+      "loss": 0.4586,
+      "step": 948
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5255000000000005e-05,
+      "loss": 0.1837,
+      "step": 949
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.525e-05,
+      "loss": 0.1576,
+      "step": 950
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.5245e-05,
+      "loss": 0.3438,
+      "step": 951
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.524e-05,
+      "loss": 0.1807,
+      "step": 952
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.5235e-05,
+      "loss": 0.5946,
+      "step": 953
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.523e-05,
+      "loss": 0.17,
+      "step": 954
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.5225e-05,
+      "loss": 0.3349,
+      "step": 955
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.522e-05,
+      "loss": 0.3321,
+      "step": 956
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.5215e-05,
+      "loss": 0.4887,
+      "step": 957
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.521e-05,
+      "loss": 0.1378,
+      "step": 958
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.5205000000000004e-05,
+      "loss": 0.2305,
+      "step": 959
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 0.4908,
+      "step": 960
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.5195e-05,
+      "loss": 0.0878,
+      "step": 961
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.519e-05,
+      "loss": 0.3701,
+      "step": 962
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.5185e-05,
+      "loss": 0.1414,
+      "step": 963
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.518e-05,
+      "loss": 0.1763,
+      "step": 964
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.5175e-05,
+      "loss": 0.1744,
+      "step": 965
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.517e-05,
+      "loss": 0.2458,
+      "step": 966
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.5165e-05,
+      "loss": 0.1662,
+      "step": 967
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.516e-05,
+      "loss": 0.1686,
+      "step": 968
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.5155e-05,
+      "loss": 0.1176,
+      "step": 969
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.5150000000000006e-05,
+      "loss": 0.3373,
+      "step": 970
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.5145e-05,
+      "loss": 0.2557,
+      "step": 971
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.5140000000000004e-05,
+      "loss": 0.338,
+      "step": 972
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.5135000000000007e-05,
+      "loss": 0.3702,
+      "step": 973
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.5129999999999996e-05,
+      "loss": 0.2167,
+      "step": 974
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.5125e-05,
+      "loss": 0.1672,
+      "step": 975
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.512e-05,
+      "loss": 0.3301,
+      "step": 976
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.5115e-05,
+      "loss": 0.1157,
+      "step": 977
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.511e-05,
+      "loss": 0.1499,
+      "step": 978
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.5105e-05,
+      "loss": 0.0961,
+      "step": 979
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.5100000000000004e-05,
+      "loss": 0.1654,
+      "step": 980
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.5095e-05,
+      "loss": 0.3468,
+      "step": 981
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.509e-05,
+      "loss": 0.2164,
+      "step": 982
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.5085000000000005e-05,
+      "loss": 0.1833,
+      "step": 983
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.508e-05,
+      "loss": 0.2524,
+      "step": 984
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.5075000000000004e-05,
+      "loss": 0.2069,
+      "step": 985
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.507e-05,
+      "loss": 0.1312,
+      "step": 986
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.5064999999999995e-05,
+      "loss": 0.12,
+      "step": 987
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.506e-05,
+      "loss": 0.212,
+      "step": 988
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.5055e-05,
+      "loss": 0.3476,
+      "step": 989
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.505e-05,
+      "loss": 0.3502,
+      "step": 990
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.5045e-05,
+      "loss": 0.247,
+      "step": 991
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.504e-05,
+      "loss": 0.3731,
+      "step": 992
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.5035000000000004e-05,
+      "loss": 0.481,
+      "step": 993
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.503e-05,
+      "loss": 0.2644,
+      "step": 994
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.5025e-05,
+      "loss": 0.4434,
+      "step": 995
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.5020000000000005e-05,
+      "loss": 0.3231,
+      "step": 996
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.5015e-05,
+      "loss": 0.2255,
+      "step": 997
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.501e-05,
+      "loss": 0.3101,
+      "step": 998
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.5005e-05,
+      "loss": 0.278,
+      "step": 999
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.5e-05,
+      "loss": 0.3495,
+      "step": 1000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.4995e-05,
+      "loss": 0.2339,
+      "step": 1001
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.499e-05,
+      "loss": 0.1867,
+      "step": 1002
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.4985e-05,
+      "loss": 0.5057,
+      "step": 1003
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.498e-05,
+      "loss": 0.4808,
+      "step": 1004
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.4975e-05,
+      "loss": 0.1158,
+      "step": 1005
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.4970000000000004e-05,
+      "loss": 0.1392,
+      "step": 1006
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.4965e-05,
+      "loss": 0.2662,
+      "step": 1007
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.496e-05,
+      "loss": 0.1217,
+      "step": 1008
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.4955000000000005e-05,
+      "loss": 0.2234,
+      "step": 1009
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.495000000000001e-05,
+      "loss": 0.2342,
+      "step": 1010
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.4944999999999997e-05,
+      "loss": 0.279,
+      "step": 1011
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.494e-05,
+      "loss": 0.1406,
+      "step": 1012
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.4935e-05,
+      "loss": 0.0868,
+      "step": 1013
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.493e-05,
+      "loss": 0.3665,
+      "step": 1014
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.4925e-05,
+      "loss": 0.2083,
+      "step": 1015
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.492e-05,
+      "loss": 0.1793,
+      "step": 1016
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.4915e-05,
+      "loss": 0.2917,
+      "step": 1017
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.491e-05,
+      "loss": 0.1751,
+      "step": 1018
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.4905000000000004e-05,
+      "loss": 0.2423,
+      "step": 1019
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.4900000000000006e-05,
+      "loss": 0.2627,
+      "step": 1020
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.4895e-05,
+      "loss": 0.2755,
+      "step": 1021
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.4890000000000005e-05,
+      "loss": 0.2574,
+      "step": 1022
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.4885e-05,
+      "loss": 0.1429,
+      "step": 1023
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.4879999999999996e-05,
+      "loss": 0.1265,
+      "step": 1024
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.4875e-05,
+      "loss": 0.3543,
+      "step": 1025
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.487e-05,
+      "loss": 0.1477,
+      "step": 1026
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.4865000000000004e-05,
+      "loss": 0.6565,
+      "step": 1027
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.486e-05,
+      "loss": 0.4173,
+      "step": 1028
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.4855e-05,
+      "loss": 0.3228,
+      "step": 1029
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.4850000000000005e-05,
+      "loss": 0.5635,
+      "step": 1030
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.4845e-05,
+      "loss": 0.127,
+      "step": 1031
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.4840000000000003e-05,
+      "loss": 0.2578,
+      "step": 1032
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.4835000000000006e-05,
+      "loss": 0.2507,
+      "step": 1033
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.483e-05,
+      "loss": 0.2576,
+      "step": 1034
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.4825e-05,
+      "loss": 0.2552,
+      "step": 1035
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.482e-05,
+      "loss": 0.2325,
+      "step": 1036
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.4815e-05,
+      "loss": 0.3029,
+      "step": 1037
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.481e-05,
+      "loss": 0.2376,
+      "step": 1038
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.4805e-05,
+      "loss": 0.5769,
+      "step": 1039
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 0.2918,
+      "step": 1040
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.4795e-05,
+      "loss": 0.1197,
+      "step": 1041
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.479e-05,
+      "loss": 0.4995,
+      "step": 1042
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.4785000000000005e-05,
+      "loss": 0.1081,
+      "step": 1043
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.478e-05,
+      "loss": 0.2629,
+      "step": 1044
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.4775e-05,
+      "loss": 0.2458,
+      "step": 1045
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.4770000000000006e-05,
+      "loss": 0.1957,
+      "step": 1046
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.4765e-05,
+      "loss": 0.4092,
+      "step": 1047
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.476e-05,
+      "loss": 0.2675,
+      "step": 1048
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.4755e-05,
+      "loss": 0.1126,
+      "step": 1049
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.475e-05,
+      "loss": 0.3309,
+      "step": 1050
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.4745e-05,
+      "loss": 0.4501,
+      "step": 1051
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.474e-05,
+      "loss": 0.4349,
+      "step": 1052
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.4735000000000004e-05,
+      "loss": 0.2899,
+      "step": 1053
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.473e-05,
+      "loss": 0.3238,
+      "step": 1054
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.4725e-05,
+      "loss": 0.1634,
+      "step": 1055
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.4720000000000005e-05,
+      "loss": 0.0913,
+      "step": 1056
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.471500000000001e-05,
+      "loss": 0.2485,
+      "step": 1057
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.471e-05,
+      "loss": 0.1364,
+      "step": 1058
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.4705e-05,
+      "loss": 0.1018,
+      "step": 1059
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.47e-05,
+      "loss": 0.1012,
+      "step": 1060
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.4695e-05,
+      "loss": 0.3972,
+      "step": 1061
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.469e-05,
+      "loss": 0.2834,
+      "step": 1062
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.4685e-05,
+      "loss": 0.3788,
+      "step": 1063
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.468e-05,
+      "loss": 0.1415,
+      "step": 1064
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.4675e-05,
+      "loss": 0.2212,
+      "step": 1065
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.4670000000000003e-05,
+      "loss": 0.1281,
+      "step": 1066
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.4665000000000006e-05,
+      "loss": 0.3422,
+      "step": 1067
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.466e-05,
+      "loss": 0.144,
+      "step": 1068
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.4655000000000004e-05,
+      "loss": 0.2359,
+      "step": 1069
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.465000000000001e-05,
+      "loss": 0.3435,
+      "step": 1070
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.4644999999999996e-05,
+      "loss": 0.264,
+      "step": 1071
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.464e-05,
+      "loss": 0.2754,
+      "step": 1072
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.4635e-05,
+      "loss": 0.3791,
+      "step": 1073
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.463e-05,
+      "loss": 0.1526,
+      "step": 1074
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.4625e-05,
+      "loss": 0.1509,
+      "step": 1075
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.462e-05,
+      "loss": 0.1977,
+      "step": 1076
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4615000000000005e-05,
+      "loss": 0.3451,
+      "step": 1077
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.461e-05,
+      "loss": 0.3156,
+      "step": 1078
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.4605e-05,
+      "loss": 0.307,
+      "step": 1079
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 0.2321,
+      "step": 1080
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.4595e-05,
+      "loss": 0.1554,
+      "step": 1081
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.4590000000000004e-05,
+      "loss": 0.2943,
+      "step": 1082
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.4585e-05,
+      "loss": 0.1608,
+      "step": 1083
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.4579999999999996e-05,
+      "loss": 0.1641,
+      "step": 1084
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.4575e-05,
+      "loss": 0.3279,
+      "step": 1085
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.457e-05,
+      "loss": 0.1655,
+      "step": 1086
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.4565000000000004e-05,
+      "loss": 0.2367,
+      "step": 1087
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.456e-05,
+      "loss": 0.1541,
+      "step": 1088
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.4555e-05,
+      "loss": 0.2955,
+      "step": 1089
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.4550000000000005e-05,
+      "loss": 0.3235,
+      "step": 1090
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.4545e-05,
+      "loss": 0.2088,
+      "step": 1091
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.454e-05,
+      "loss": 0.4172,
+      "step": 1092
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.4535000000000006e-05,
+      "loss": 0.2841,
+      "step": 1093
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.453e-05,
+      "loss": 0.1348,
+      "step": 1094
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.4525e-05,
+      "loss": 0.2953,
+      "step": 1095
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.452e-05,
+      "loss": 0.1375,
+      "step": 1096
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.4515e-05,
+      "loss": 0.1152,
+      "step": 1097
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.451e-05,
+      "loss": 0.268,
+      "step": 1098
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.4505e-05,
+      "loss": 0.3306,
+      "step": 1099
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.45e-05,
+      "loss": 0.1663,
+      "step": 1100
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.4495e-05,
+      "loss": 0.158,
+      "step": 1101
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.449e-05,
+      "loss": 0.5986,
+      "step": 1102
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.4485000000000004e-05,
+      "loss": 0.2222,
+      "step": 1103
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.448e-05,
+      "loss": 0.265,
+      "step": 1104
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.4475e-05,
+      "loss": 0.3913,
+      "step": 1105
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.4470000000000005e-05,
+      "loss": 0.3639,
+      "step": 1106
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.4465e-05,
+      "loss": 0.1447,
+      "step": 1107
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.446e-05,
+      "loss": 0.3232,
+      "step": 1108
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.4455e-05,
+      "loss": 0.2868,
+      "step": 1109
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.445e-05,
+      "loss": 0.2371,
+      "step": 1110
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.4445e-05,
+      "loss": 0.419,
+      "step": 1111
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.444e-05,
+      "loss": 0.1105,
+      "step": 1112
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.4435e-05,
+      "loss": 0.2097,
+      "step": 1113
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.443e-05,
+      "loss": 0.1534,
+      "step": 1114
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.4425e-05,
+      "loss": 0.3226,
+      "step": 1115
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.4420000000000004e-05,
+      "loss": 0.3008,
+      "step": 1116
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.441500000000001e-05,
+      "loss": 0.3417,
+      "step": 1117
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.441e-05,
+      "loss": 0.2229,
+      "step": 1118
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.4405e-05,
+      "loss": 0.1226,
+      "step": 1119
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.44e-05,
+      "loss": 0.2524,
+      "step": 1120
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.4395e-05,
+      "loss": 0.1997,
+      "step": 1121
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.439e-05,
+      "loss": 0.1735,
+      "step": 1122
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.4385e-05,
+      "loss": 0.4729,
+      "step": 1123
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.438e-05,
+      "loss": 0.2847,
+      "step": 1124
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.4375e-05,
+      "loss": 0.3992,
+      "step": 1125
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.437e-05,
+      "loss": 0.2955,
+      "step": 1126
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.4365000000000005e-05,
+      "loss": 0.2445,
+      "step": 1127
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.436e-05,
+      "loss": 0.2523,
+      "step": 1128
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.4355000000000004e-05,
+      "loss": 0.1608,
+      "step": 1129
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.4350000000000006e-05,
+      "loss": 0.2523,
+      "step": 1130
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.4345e-05,
+      "loss": 0.4919,
+      "step": 1131
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.434e-05,
+      "loss": 0.2518,
+      "step": 1132
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.4335e-05,
+      "loss": 0.4321,
+      "step": 1133
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.4329999999999997e-05,
+      "loss": 0.181,
+      "step": 1134
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.4325e-05,
+      "loss": 0.2807,
+      "step": 1135
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.432e-05,
+      "loss": 0.3556,
+      "step": 1136
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.4315000000000004e-05,
+      "loss": 0.1357,
+      "step": 1137
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.431e-05,
+      "loss": 0.3101,
+      "step": 1138
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.4305e-05,
+      "loss": 0.1587,
+      "step": 1139
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.4300000000000005e-05,
+      "loss": 0.2647,
+      "step": 1140
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.4295e-05,
+      "loss": 0.2397,
+      "step": 1141
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.4290000000000004e-05,
+      "loss": 0.2589,
+      "step": 1142
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.4285000000000006e-05,
+      "loss": 0.1729,
+      "step": 1143
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.4279999999999995e-05,
+      "loss": 0.2662,
+      "step": 1144
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.4275e-05,
+      "loss": 0.2894,
+      "step": 1145
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.427e-05,
+      "loss": 0.2199,
+      "step": 1146
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.4265e-05,
+      "loss": 0.2203,
+      "step": 1147
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.426e-05,
+      "loss": 0.4355,
+      "step": 1148
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.4255e-05,
+      "loss": 0.2515,
+      "step": 1149
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.4250000000000004e-05,
+      "loss": 0.192,
+      "step": 1150
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.4245e-05,
+      "loss": 0.3158,
+      "step": 1151
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.424e-05,
+      "loss": 0.1809,
+      "step": 1152
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.4235000000000005e-05,
+      "loss": 0.1707,
+      "step": 1153
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.423e-05,
+      "loss": 0.134,
+      "step": 1154
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.4225000000000003e-05,
+      "loss": 0.3423,
+      "step": 1155
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.422e-05,
+      "loss": 0.3034,
+      "step": 1156
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.4215e-05,
+      "loss": 0.3217,
+      "step": 1157
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.421e-05,
+      "loss": 0.1715,
+      "step": 1158
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.4205e-05,
+      "loss": 0.2688,
+      "step": 1159
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.42e-05,
+      "loss": 0.3909,
+      "step": 1160
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4195e-05,
+      "loss": 0.1145,
+      "step": 1161
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.419e-05,
+      "loss": 0.0936,
+      "step": 1162
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.4185000000000004e-05,
+      "loss": 0.4105,
+      "step": 1163
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.418e-05,
+      "loss": 0.2621,
+      "step": 1164
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.4175e-05,
+      "loss": 0.2444,
+      "step": 1165
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.4170000000000005e-05,
+      "loss": 0.2231,
+      "step": 1166
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.416500000000001e-05,
+      "loss": 0.2628,
+      "step": 1167
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.4159999999999996e-05,
+      "loss": 0.1621,
+      "step": 1168
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.4155e-05,
+      "loss": 0.196,
+      "step": 1169
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.415e-05,
+      "loss": 0.219,
+      "step": 1170
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.4145e-05,
+      "loss": 0.2163,
+      "step": 1171
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.414e-05,
+      "loss": 0.1389,
+      "step": 1172
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.4135e-05,
+      "loss": 0.4736,
+      "step": 1173
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.413e-05,
+      "loss": 0.2487,
+      "step": 1174
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4125e-05,
+      "loss": 0.12,
+      "step": 1175
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4120000000000004e-05,
+      "loss": 0.1955,
+      "step": 1176
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4115000000000006e-05,
+      "loss": 0.1263,
+      "step": 1177
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.411e-05,
+      "loss": 0.1331,
+      "step": 1178
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.4105000000000005e-05,
+      "loss": 0.1307,
+      "step": 1179
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.41e-05,
+      "loss": 0.2697,
+      "step": 1180
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.4094999999999996e-05,
+      "loss": 0.2485,
+      "step": 1181
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.409e-05,
+      "loss": 0.2228,
+      "step": 1182
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.4085e-05,
+      "loss": 0.1576,
+      "step": 1183
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.408e-05,
+      "loss": 0.1786,
+      "step": 1184
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.4075e-05,
+      "loss": 0.3179,
+      "step": 1185
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.407e-05,
+      "loss": 0.3756,
+      "step": 1186
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.4065000000000005e-05,
+      "loss": 0.1811,
+      "step": 1187
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.406e-05,
+      "loss": 0.3712,
+      "step": 1188
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.4055e-05,
+      "loss": 0.2076,
+      "step": 1189
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.4050000000000006e-05,
+      "loss": 0.2516,
+      "step": 1190
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.4045e-05,
+      "loss": 0.3398,
+      "step": 1191
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.404e-05,
+      "loss": 0.3741,
+      "step": 1192
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.4035e-05,
+      "loss": 0.2945,
+      "step": 1193
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.403e-05,
+      "loss": 0.4141,
+      "step": 1194
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.4025e-05,
+      "loss": 0.3433,
+      "step": 1195
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.402e-05,
+      "loss": 0.214,
+      "step": 1196
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.4015000000000004e-05,
+      "loss": 0.17,
+      "step": 1197
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.401e-05,
+      "loss": 0.2674,
+      "step": 1198
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.4005e-05,
+      "loss": 0.2838,
+      "step": 1199
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 0.2188,
+      "step": 1200
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.3995e-05,
+      "loss": 0.1975,
+      "step": 1201
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.399e-05,
+      "loss": 0.198,
+      "step": 1202
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.3985000000000006e-05,
+      "loss": 0.1563,
+      "step": 1203
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.398e-05,
+      "loss": 0.2662,
+      "step": 1204
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.3975e-05,
+      "loss": 0.4196,
+      "step": 1205
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.397e-05,
+      "loss": 0.1821,
+      "step": 1206
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.3965e-05,
+      "loss": 0.4274,
+      "step": 1207
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.396e-05,
+      "loss": 0.2215,
+      "step": 1208
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5.3955e-05,
+      "loss": 0.168,
+      "step": 1209
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5.3950000000000004e-05,
+      "loss": 0.2995,
+      "step": 1210
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.3945e-05,
+      "loss": 0.1323,
+      "step": 1211
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.394e-05,
+      "loss": 0.2793,
+      "step": 1212
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.3935000000000005e-05,
+      "loss": 0.3837,
+      "step": 1213
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.393000000000001e-05,
+      "loss": 0.3899,
+      "step": 1214
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.3925e-05,
+      "loss": 0.2947,
+      "step": 1215
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 5.392e-05,
+      "loss": 0.2598,
+      "step": 1216
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 5.3915e-05,
+      "loss": 0.2038,
+      "step": 1217
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.391e-05,
+      "loss": 0.1745,
+      "step": 1218
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.3905e-05,
+      "loss": 0.1601,
+      "step": 1219
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.39e-05,
+      "loss": 0.1318,
+      "step": 1220
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.3895e-05,
+      "loss": 0.1711,
+      "step": 1221
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.389e-05,
+      "loss": 0.2816,
+      "step": 1222
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.3885e-05,
+      "loss": 0.2694,
+      "step": 1223
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.3880000000000006e-05,
+      "loss": 0.4006,
+      "step": 1224
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.3875e-05,
+      "loss": 0.2076,
+      "step": 1225
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.3870000000000004e-05,
+      "loss": 0.3128,
+      "step": 1226
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.386500000000001e-05,
+      "loss": 0.3515,
+      "step": 1227
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.3859999999999996e-05,
+      "loss": 0.1995,
+      "step": 1228
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.3855e-05,
+      "loss": 0.1389,
+      "step": 1229
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.385e-05,
+      "loss": 0.2078,
+      "step": 1230
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.3845e-05,
+      "loss": 0.1307,
+      "step": 1231
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.384e-05,
+      "loss": 0.2922,
+      "step": 1232
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.3835e-05,
+      "loss": 0.1065,
+      "step": 1233
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.3830000000000005e-05,
+      "loss": 0.4569,
+      "step": 1234
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.3825e-05,
+      "loss": 0.1795,
+      "step": 1235
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.382e-05,
+      "loss": 0.2269,
+      "step": 1236
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.3815000000000006e-05,
+      "loss": 0.4409,
+      "step": 1237
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.381e-05,
+      "loss": 0.2886,
+      "step": 1238
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.3805000000000004e-05,
+      "loss": 0.3213,
+      "step": 1239
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.38e-05,
+      "loss": 0.3623,
+      "step": 1240
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.3794999999999996e-05,
+      "loss": 0.1124,
+      "step": 1241
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.379e-05,
+      "loss": 0.2432,
+      "step": 1242
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 5.3785e-05,
+      "loss": 0.1034,
+      "step": 1243
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 5.3780000000000003e-05,
+      "loss": 0.2814,
+      "step": 1244
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.3775e-05,
+      "loss": 0.0994,
+      "step": 1245
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.377e-05,
+      "loss": 0.2199,
+      "step": 1246
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.3765000000000004e-05,
+      "loss": 0.1983,
+      "step": 1247
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.376e-05,
+      "loss": 0.3544,
+      "step": 1248
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.3755e-05,
+      "loss": 0.2299,
+      "step": 1249
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.3750000000000005e-05,
+      "loss": 0.1359,
+      "step": 1250
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.3745e-05,
+      "loss": 0.2697,
+      "step": 1251
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.3740000000000004e-05,
+      "loss": 0.3325,
+      "step": 1252
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.3735e-05,
+      "loss": 0.1637,
+      "step": 1253
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.373e-05,
+      "loss": 0.3007,
+      "step": 1254
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.3725e-05,
+      "loss": 0.4445,
+      "step": 1255
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.372e-05,
+      "loss": 0.1645,
+      "step": 1256
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.3715e-05,
+      "loss": 0.1873,
+      "step": 1257
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.371e-05,
+      "loss": 0.4373,
+      "step": 1258
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.3705e-05,
+      "loss": 0.2325,
+      "step": 1259
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.3700000000000004e-05,
+      "loss": 0.2602,
+      "step": 1260
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.3695e-05,
+      "loss": 0.4495,
+      "step": 1261
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.369e-05,
+      "loss": 0.1003,
+      "step": 1262
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.3685000000000005e-05,
+      "loss": 0.3184,
+      "step": 1263
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.368000000000001e-05,
+      "loss": 0.1378,
+      "step": 1264
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.3675e-05,
+      "loss": 0.285,
+      "step": 1265
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.367e-05,
+      "loss": 0.218,
+      "step": 1266
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.3665e-05,
+      "loss": 0.193,
+      "step": 1267
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.366e-05,
+      "loss": 0.4031,
+      "step": 1268
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.3655e-05,
+      "loss": 0.2524,
+      "step": 1269
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.365e-05,
+      "loss": 0.1082,
+      "step": 1270
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.3645e-05,
+      "loss": 0.2519,
+      "step": 1271
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.364e-05,
+      "loss": 0.2252,
+      "step": 1272
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.3635000000000004e-05,
+      "loss": 0.2296,
+      "step": 1273
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.3630000000000007e-05,
+      "loss": 0.2106,
+      "step": 1274
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.3625e-05,
+      "loss": 0.2475,
+      "step": 1275
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.3620000000000005e-05,
+      "loss": 0.1019,
+      "step": 1276
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.3615e-05,
+      "loss": 0.5145,
+      "step": 1277
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.361e-05,
+      "loss": 0.2218,
+      "step": 1278
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.3605e-05,
+      "loss": 0.2003,
+      "step": 1279
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.36e-05,
+      "loss": 0.3444,
+      "step": 1280
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.3595e-05,
+      "loss": 0.2108,
+      "step": 1281
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.359e-05,
+      "loss": 0.3062,
+      "step": 1282
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.3585e-05,
+      "loss": 0.2481,
+      "step": 1283
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.3580000000000005e-05,
+      "loss": 0.2436,
+      "step": 1284
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.3575e-05,
+      "loss": 0.1077,
+      "step": 1285
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.3570000000000004e-05,
+      "loss": 0.3784,
+      "step": 1286
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.3565000000000006e-05,
+      "loss": 0.2441,
+      "step": 1287
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.356e-05,
+      "loss": 0.3512,
+      "step": 1288
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.3555e-05,
+      "loss": 0.3156,
+      "step": 1289
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.355e-05,
+      "loss": 0.1103,
+      "step": 1290
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.3544999999999996e-05,
+      "loss": 0.4687,
+      "step": 1291
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.354e-05,
+      "loss": 0.1929,
+      "step": 1292
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.3535e-05,
+      "loss": 0.3433,
+      "step": 1293
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.3530000000000004e-05,
+      "loss": 0.2379,
+      "step": 1294
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.3525e-05,
+      "loss": 0.3766,
+      "step": 1295
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.352e-05,
+      "loss": 0.1096,
+      "step": 1296
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.3515000000000005e-05,
+      "loss": 0.2015,
+      "step": 1297
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.351e-05,
+      "loss": 0.2658,
+      "step": 1298
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.3505000000000004e-05,
+      "loss": 0.2716,
+      "step": 1299
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 5.3500000000000006e-05,
+      "loss": 0.348,
+      "step": 1300
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 5.3494999999999995e-05,
+      "loss": 0.2891,
+      "step": 1301
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 5.349e-05,
+      "loss": 0.2342,
+      "step": 1302
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 5.3485e-05,
+      "loss": 0.2474,
+      "step": 1303
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 5.348e-05,
+      "loss": 0.3315,
+      "step": 1304
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 5.3475e-05,
+      "loss": 0.5229,
+      "step": 1305
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 5.347e-05,
+      "loss": 0.2622,
+      "step": 1306
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 5.3465000000000004e-05,
+      "loss": 0.2506,
+      "step": 1307
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 5.346e-05,
+      "loss": 0.1485,
+      "step": 1308
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 5.3455e-05,
+      "loss": 0.4087,
+      "step": 1309
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 5.3450000000000005e-05,
+      "loss": 0.1444,
+      "step": 1310
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 5.3445e-05,
+      "loss": 0.2956,
+      "step": 1311
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 5.344e-05,
+      "loss": 0.3182,
+      "step": 1312
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 5.3435e-05,
+      "loss": 0.1665,
+      "step": 1313
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 5.343e-05,
+      "loss": 0.1466,
+      "step": 1314
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 5.3425e-05,
+      "loss": 0.105,
+      "step": 1315
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 5.342e-05,
+      "loss": 0.2926,
+      "step": 1316
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 5.3415e-05,
+      "loss": 0.2307,
+      "step": 1317
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 5.341e-05,
+      "loss": 0.361,
+      "step": 1318
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 5.3405e-05,
+      "loss": 0.3712,
+      "step": 1319
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 0.273,
+      "step": 1320
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 5.3395e-05,
+      "loss": 0.3898,
+      "step": 1321
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 5.339e-05,
+      "loss": 0.152,
+      "step": 1322
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 5.3385000000000005e-05,
+      "loss": 0.0943,
+      "step": 1323
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 5.338000000000001e-05,
+      "loss": 0.2666,
+      "step": 1324
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 5.3374999999999996e-05,
+      "loss": 0.2697,
+      "step": 1325
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 5.337e-05,
+      "loss": 0.5582,
+      "step": 1326
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 5.3365e-05,
+      "loss": 0.1377,
+      "step": 1327
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 5.336e-05,
+      "loss": 0.2905,
+      "step": 1328
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 5.3355e-05,
+      "loss": 0.2331,
+      "step": 1329
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 5.335e-05,
+      "loss": 0.1353,
+      "step": 1330
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 5.3345e-05,
+      "loss": 0.1403,
+      "step": 1331
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 5.334e-05,
+      "loss": 0.1135,
+      "step": 1332
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 5.3335000000000003e-05,
+      "loss": 0.2661,
+      "step": 1333
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 5.3330000000000006e-05,
+      "loss": 0.2331,
+      "step": 1334
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 5.3325e-05,
+      "loss": 0.1751,
+      "step": 1335
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 5.3320000000000004e-05,
+      "loss": 0.1302,
+      "step": 1336
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 5.3315e-05,
+      "loss": 0.1553,
+      "step": 1337
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 5.3309999999999996e-05,
+      "loss": 0.1284,
+      "step": 1338
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 5.3305e-05,
+      "loss": 0.2646,
+      "step": 1339
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 5.33e-05,
+      "loss": 0.154,
+      "step": 1340
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 5.3295e-05,
+      "loss": 0.1727,
+      "step": 1341
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 5.329e-05,
+      "loss": 0.2906,
+      "step": 1342
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 5.3285e-05,
+      "loss": 0.1171,
+      "step": 1343
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 5.3280000000000005e-05,
+      "loss": 0.1445,
+      "step": 1344
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 5.3275e-05,
+      "loss": 0.1558,
+      "step": 1345
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 5.327e-05,
+      "loss": 0.1638,
+      "step": 1346
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 5.3265000000000006e-05,
+      "loss": 0.1186,
+      "step": 1347
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.326e-05,
+      "loss": 0.0956,
+      "step": 1348
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.3255e-05,
+      "loss": 0.2991,
+      "step": 1349
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.325e-05,
+      "loss": 0.1683,
+      "step": 1350
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 5.3244999999999996e-05,
+      "loss": 0.2053,
+      "step": 1351
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 5.324e-05,
+      "loss": 0.1595,
+      "step": 1352
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.3235e-05,
+      "loss": 0.1621,
+      "step": 1353
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.3230000000000004e-05,
+      "loss": 0.1173,
+      "step": 1354
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 5.3225e-05,
+      "loss": 0.2436,
+      "step": 1355
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 5.322e-05,
+      "loss": 0.2477,
+      "step": 1356
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 5.3215000000000005e-05,
+      "loss": 0.1294,
+      "step": 1357
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 5.321e-05,
+      "loss": 0.1827,
+      "step": 1358
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 5.3205e-05,
+      "loss": 0.2548,
+      "step": 1359
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 0.1203,
+      "step": 1360
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 5.3195e-05,
+      "loss": 0.2817,
+      "step": 1361
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 5.319e-05,
+      "loss": 0.2685,
+      "step": 1362
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 5.3185e-05,
+      "loss": 0.1395,
+      "step": 1363
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 5.318e-05,
+      "loss": 0.1916,
+      "step": 1364
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.3175e-05,
+      "loss": 0.1222,
+      "step": 1365
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.317e-05,
+      "loss": 0.4515,
+      "step": 1366
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 5.3165000000000003e-05,
+      "loss": 0.3843,
+      "step": 1367
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 5.316e-05,
+      "loss": 0.1044,
+      "step": 1368
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 5.3155e-05,
+      "loss": 0.2592,
+      "step": 1369
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 5.3150000000000004e-05,
+      "loss": 0.1249,
+      "step": 1370
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 5.3145e-05,
+      "loss": 0.1615,
+      "step": 1371
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.314e-05,
+      "loss": 0.1172,
+      "step": 1372
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.3135e-05,
+      "loss": 0.4464,
+      "step": 1373
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.313e-05,
+      "loss": 0.1868,
+      "step": 1374
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 5.3125e-05,
+      "loss": 0.2168,
+      "step": 1375
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 5.312e-05,
+      "loss": 0.1934,
+      "step": 1376
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 5.3115e-05,
+      "loss": 0.2973,
+      "step": 1377
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 5.311e-05,
+      "loss": 0.2323,
+      "step": 1378
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 5.3105e-05,
+      "loss": 0.2446,
+      "step": 1379
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 5.31e-05,
+      "loss": 0.2054,
+      "step": 1380
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 5.3095000000000006e-05,
+      "loss": 0.3519,
+      "step": 1381
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 5.309e-05,
+      "loss": 0.1088,
+      "step": 1382
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 5.3085000000000004e-05,
+      "loss": 0.309,
+      "step": 1383
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 5.308000000000001e-05,
+      "loss": 0.29,
+      "step": 1384
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 5.3075e-05,
+      "loss": 0.3722,
+      "step": 1385
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 5.307e-05,
+      "loss": 0.1399,
+      "step": 1386
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 5.3065e-05,
+      "loss": 0.0815,
+      "step": 1387
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 5.306e-05,
+      "loss": 0.3226,
+      "step": 1388
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.3055e-05,
+      "loss": 0.3268,
+      "step": 1389
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.305e-05,
+      "loss": 0.3131,
+      "step": 1390
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.3045000000000005e-05,
+      "loss": 0.2186,
+      "step": 1391
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.304e-05,
+      "loss": 0.2289,
+      "step": 1392
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.3035e-05,
+      "loss": 0.2558,
+      "step": 1393
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.3030000000000006e-05,
+      "loss": 0.3149,
+      "step": 1394
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.3025e-05,
+      "loss": 0.163,
+      "step": 1395
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 5.3020000000000004e-05,
+      "loss": 0.2364,
+      "step": 1396
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 5.3015000000000007e-05,
+      "loss": 0.2301,
+      "step": 1397
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.3009999999999996e-05,
+      "loss": 0.1806,
+      "step": 1398
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.3005e-05,
+      "loss": 0.329,
+      "step": 1399
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.3e-05,
+      "loss": 0.1762,
+      "step": 1400
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 5.2995e-05,
+      "loss": 0.309,
+      "step": 1401
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 5.299e-05,
+      "loss": 0.1377,
+      "step": 1402
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.2985e-05,
+      "loss": 0.3962,
+      "step": 1403
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.2980000000000004e-05,
+      "loss": 0.1652,
+      "step": 1404
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.2975e-05,
+      "loss": 0.1525,
+      "step": 1405
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 5.297e-05,
+      "loss": 0.315,
+      "step": 1406
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 5.2965000000000005e-05,
+      "loss": 0.4086,
+      "step": 1407
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.296e-05,
+      "loss": 0.3483,
+      "step": 1408
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.2955000000000004e-05,
+      "loss": 0.2276,
+      "step": 1409
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.295e-05,
+      "loss": 0.2098,
+      "step": 1410
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.2945e-05,
+      "loss": 0.2457,
+      "step": 1411
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.294e-05,
+      "loss": 0.1249,
+      "step": 1412
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.2935e-05,
+      "loss": 0.1617,
+      "step": 1413
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.293e-05,
+      "loss": 0.1119,
+      "step": 1414
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.2925e-05,
+      "loss": 0.1395,
+      "step": 1415
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.292e-05,
+      "loss": 0.3787,
+      "step": 1416
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.2915000000000004e-05,
+      "loss": 0.2742,
+      "step": 1417
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 5.291e-05,
+      "loss": 0.2166,
+      "step": 1418
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 5.2905e-05,
+      "loss": 0.278,
+      "step": 1419
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.2900000000000005e-05,
+      "loss": 0.3054,
+      "step": 1420
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.289500000000001e-05,
+      "loss": 0.2308,
+      "step": 1421
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.289e-05,
+      "loss": 0.1881,
+      "step": 1422
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.2885e-05,
+      "loss": 0.2937,
+      "step": 1423
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.288e-05,
+      "loss": 0.2566,
+      "step": 1424
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.2875e-05,
+      "loss": 0.1063,
+      "step": 1425
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.287e-05,
+      "loss": 0.2995,
+      "step": 1426
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.2865e-05,
+      "loss": 0.3382,
+      "step": 1427
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.286e-05,
+      "loss": 0.3343,
+      "step": 1428
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.2855e-05,
+      "loss": 0.117,
+      "step": 1429
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 5.2850000000000004e-05,
+      "loss": 0.2209,
+      "step": 1430
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 5.2845000000000006e-05,
+      "loss": 0.1139,
+      "step": 1431
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.284e-05,
+      "loss": 0.1152,
+      "step": 1432
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.2835000000000005e-05,
+      "loss": 0.5124,
+      "step": 1433
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.283e-05,
+      "loss": 0.1878,
+      "step": 1434
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.2824999999999997e-05,
+      "loss": 0.1341,
+      "step": 1435
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.282e-05,
+      "loss": 0.2641,
+      "step": 1436
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.2815e-05,
+      "loss": 0.1208,
+      "step": 1437
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.281e-05,
+      "loss": 0.2441,
+      "step": 1438
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5.2805e-05,
+      "loss": 0.2089,
+      "step": 1439
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5.28e-05,
+      "loss": 0.2531,
+      "step": 1440
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5.2795000000000005e-05,
+      "loss": 0.1729,
+      "step": 1441
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.279e-05,
+      "loss": 0.1392,
+      "step": 1442
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.2785000000000004e-05,
+      "loss": 0.1136,
+      "step": 1443
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.2780000000000006e-05,
+      "loss": 0.3371,
+      "step": 1444
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.2775e-05,
+      "loss": 0.3155,
+      "step": 1445
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.277e-05,
+      "loss": 0.4882,
+      "step": 1446
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.2765e-05,
+      "loss": 0.3839,
+      "step": 1447
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.2759999999999996e-05,
+      "loss": 0.1167,
+      "step": 1448
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.2755e-05,
+      "loss": 0.1465,
+      "step": 1449
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.275e-05,
+      "loss": 0.2196,
+      "step": 1450
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.2745000000000004e-05,
+      "loss": 0.1927,
+      "step": 1451
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.274e-05,
+      "loss": 0.2144,
+      "step": 1452
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.2735e-05,
+      "loss": 0.1194,
+      "step": 1453
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.2730000000000005e-05,
+      "loss": 0.338,
+      "step": 1454
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.2725e-05,
+      "loss": 0.2639,
+      "step": 1455
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.2720000000000003e-05,
+      "loss": 0.102,
+      "step": 1456
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.2715000000000006e-05,
+      "loss": 0.1418,
+      "step": 1457
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.2709999999999995e-05,
+      "loss": 0.1153,
+      "step": 1458
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.2705e-05,
+      "loss": 0.3345,
+      "step": 1459
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.27e-05,
+      "loss": 0.1548,
+      "step": 1460
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.2695e-05,
+      "loss": 0.1664,
+      "step": 1461
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.269e-05,
+      "loss": 0.0891,
+      "step": 1462
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.2685e-05,
+      "loss": 0.3752,
+      "step": 1463
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.2680000000000004e-05,
+      "loss": 0.1061,
+      "step": 1464
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.2675e-05,
+      "loss": 0.2846,
+      "step": 1465
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.267e-05,
+      "loss": 0.2331,
+      "step": 1466
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.2665000000000005e-05,
+      "loss": 0.1679,
+      "step": 1467
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.266e-05,
+      "loss": 0.295,
+      "step": 1468
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.2655e-05,
+      "loss": 0.2211,
+      "step": 1469
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.265e-05,
+      "loss": 0.1678,
+      "step": 1470
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.2645e-05,
+      "loss": 0.2369,
+      "step": 1471
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.264e-05,
+      "loss": 0.1505,
+      "step": 1472
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.2635e-05,
+      "loss": 0.2364,
+      "step": 1473
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.263e-05,
+      "loss": 0.1231,
+      "step": 1474
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.2625e-05,
+      "loss": 0.2376,
+      "step": 1475
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.262e-05,
+      "loss": 0.1326,
+      "step": 1476
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.2615000000000004e-05,
+      "loss": 0.25,
+      "step": 1477
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.261e-05,
+      "loss": 0.1287,
+      "step": 1478
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.2605e-05,
+      "loss": 0.3274,
+      "step": 1479
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 0.2745,
+      "step": 1480
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.259500000000001e-05,
+      "loss": 0.2292,
+      "step": 1481
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.2589999999999996e-05,
+      "loss": 0.2455,
+      "step": 1482
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.2585e-05,
+      "loss": 0.1864,
+      "step": 1483
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.258e-05,
+      "loss": 0.3113,
+      "step": 1484
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.2575e-05,
+      "loss": 0.1843,
+      "step": 1485
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.257e-05,
+      "loss": 0.1288,
+      "step": 1486
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.2565e-05,
+      "loss": 0.282,
+      "step": 1487
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.256e-05,
+      "loss": 0.2616,
+      "step": 1488
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.2555e-05,
+      "loss": 0.2318,
+      "step": 1489
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.2550000000000003e-05,
+      "loss": 0.2341,
+      "step": 1490
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.2545000000000006e-05,
+      "loss": 0.2497,
+      "step": 1491
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.254e-05,
+      "loss": 0.2502,
+      "step": 1492
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.2535000000000004e-05,
+      "loss": 0.2131,
+      "step": 1493
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.253e-05,
+      "loss": 0.1242,
+      "step": 1494
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.2524999999999996e-05,
+      "loss": 0.2542,
+      "step": 1495
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.252e-05,
+      "loss": 0.1242,
+      "step": 1496
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.2515e-05,
+      "loss": 0.2824,
+      "step": 1497
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.251e-05,
+      "loss": 0.1529,
+      "step": 1498
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.2505e-05,
+      "loss": 0.1955,
+      "step": 1499
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.25e-05,
+      "loss": 0.1231,
+      "step": 1500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.2495000000000005e-05,
+      "loss": 0.2355,
+      "step": 1501
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.249e-05,
+      "loss": 0.3414,
+      "step": 1502
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.2485e-05,
+      "loss": 0.2797,
+      "step": 1503
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.2480000000000006e-05,
+      "loss": 0.1737,
+      "step": 1504
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.2475e-05,
+      "loss": 0.2185,
+      "step": 1505
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.2470000000000004e-05,
+      "loss": 0.3437,
+      "step": 1506
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.2465e-05,
+      "loss": 0.2992,
+      "step": 1507
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.2459999999999996e-05,
+      "loss": 0.2456,
+      "step": 1508
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.2455e-05,
+      "loss": 0.2282,
+      "step": 1509
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.245e-05,
+      "loss": 0.1407,
+      "step": 1510
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.2445000000000004e-05,
+      "loss": 0.3564,
+      "step": 1511
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.244e-05,
+      "loss": 0.4183,
+      "step": 1512
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.2435e-05,
+      "loss": 0.2721,
+      "step": 1513
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.2430000000000005e-05,
+      "loss": 0.1154,
+      "step": 1514
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.2425e-05,
+      "loss": 0.1315,
+      "step": 1515
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.242e-05,
+      "loss": 0.1904,
+      "step": 1516
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.2415000000000006e-05,
+      "loss": 0.1888,
+      "step": 1517
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.241e-05,
+      "loss": 0.2319,
+      "step": 1518
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.2405e-05,
+      "loss": 0.2028,
+      "step": 1519
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.24e-05,
+      "loss": 0.4611,
+      "step": 1520
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.2395e-05,
+      "loss": 0.1426,
+      "step": 1521
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.239e-05,
+      "loss": 0.1861,
+      "step": 1522
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.2385e-05,
+      "loss": 0.0746,
+      "step": 1523
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.238e-05,
+      "loss": 0.1562,
+      "step": 1524
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.2375e-05,
+      "loss": 0.1292,
+      "step": 1525
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.237e-05,
+      "loss": 0.324,
+      "step": 1526
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.2365000000000004e-05,
+      "loss": 0.1285,
+      "step": 1527
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.236e-05,
+      "loss": 0.1247,
+      "step": 1528
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.2355e-05,
+      "loss": 0.1602,
+      "step": 1529
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.2350000000000005e-05,
+      "loss": 0.1867,
+      "step": 1530
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.2345e-05,
+      "loss": 0.2505,
+      "step": 1531
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.234e-05,
+      "loss": 0.151,
+      "step": 1532
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.2335e-05,
+      "loss": 0.2495,
+      "step": 1533
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.233e-05,
+      "loss": 0.3428,
+      "step": 1534
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.2325e-05,
+      "loss": 0.1382,
+      "step": 1535
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.232e-05,
+      "loss": 0.3062,
+      "step": 1536
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.2315e-05,
+      "loss": 0.253,
+      "step": 1537
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.231e-05,
+      "loss": 0.1216,
+      "step": 1538
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.2305e-05,
+      "loss": 0.2966,
+      "step": 1539
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.2300000000000004e-05,
+      "loss": 0.2422,
+      "step": 1540
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.229500000000001e-05,
+      "loss": 0.2539,
+      "step": 1541
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.229e-05,
+      "loss": 0.2732,
+      "step": 1542
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.2285e-05,
+      "loss": 0.1106,
+      "step": 1543
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.228e-05,
+      "loss": 0.1397,
+      "step": 1544
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.2275e-05,
+      "loss": 0.2247,
+      "step": 1545
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.227e-05,
+      "loss": 0.5682,
+      "step": 1546
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.2265e-05,
+      "loss": 0.0819,
+      "step": 1547
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.2260000000000004e-05,
+      "loss": 0.1161,
+      "step": 1548
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.2255e-05,
+      "loss": 0.3117,
+      "step": 1549
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.225e-05,
+      "loss": 0.1322,
+      "step": 1550
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.2245000000000005e-05,
+      "loss": 0.2284,
+      "step": 1551
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.224e-05,
+      "loss": 0.4592,
+      "step": 1552
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.2235000000000004e-05,
+      "loss": 0.2934,
+      "step": 1553
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.2230000000000006e-05,
+      "loss": 0.215,
+      "step": 1554
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.2224999999999995e-05,
+      "loss": 0.3639,
+      "step": 1555
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.222e-05,
+      "loss": 0.1493,
+      "step": 1556
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.2215e-05,
+      "loss": 0.3135,
+      "step": 1557
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.221e-05,
+      "loss": 0.3961,
+      "step": 1558
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.2205e-05,
+      "loss": 0.1209,
+      "step": 1559
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.22e-05,
+      "loss": 0.2355,
+      "step": 1560
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.2195000000000004e-05,
+      "loss": 0.2264,
+      "step": 1561
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.219e-05,
+      "loss": 0.1108,
+      "step": 1562
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.2185e-05,
+      "loss": 0.235,
+      "step": 1563
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.2180000000000005e-05,
+      "loss": 0.359,
+      "step": 1564
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.2175e-05,
+      "loss": 0.1867,
+      "step": 1565
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.2170000000000004e-05,
+      "loss": 0.1217,
+      "step": 1566
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.2165e-05,
+      "loss": 0.2288,
+      "step": 1567
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.216e-05,
+      "loss": 0.1253,
+      "step": 1568
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.2155e-05,
+      "loss": 0.164,
+      "step": 1569
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.215e-05,
+      "loss": 0.2763,
+      "step": 1570
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.2145e-05,
+      "loss": 0.4075,
+      "step": 1571
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.214e-05,
+      "loss": 0.2501,
+      "step": 1572
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.2135e-05,
+      "loss": 0.2178,
+      "step": 1573
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.2130000000000004e-05,
+      "loss": 0.2844,
+      "step": 1574
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.2125e-05,
+      "loss": 0.151,
+      "step": 1575
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.212e-05,
+      "loss": 0.2554,
+      "step": 1576
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.2115000000000005e-05,
+      "loss": 0.3667,
+      "step": 1577
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.211000000000001e-05,
+      "loss": 0.1129,
+      "step": 1578
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.2105e-05,
+      "loss": 0.0901,
+      "step": 1579
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.21e-05,
+      "loss": 0.2171,
+      "step": 1580
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.2095e-05,
+      "loss": 0.1494,
+      "step": 1581
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.209e-05,
+      "loss": 0.4425,
+      "step": 1582
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.2085e-05,
+      "loss": 0.1032,
+      "step": 1583
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.208e-05,
+      "loss": 0.2064,
+      "step": 1584
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.2075e-05,
+      "loss": 0.2883,
+      "step": 1585
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.207e-05,
+      "loss": 0.2063,
+      "step": 1586
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.2065000000000004e-05,
+      "loss": 0.2213,
+      "step": 1587
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.2060000000000006e-05,
+      "loss": 0.2343,
+      "step": 1588
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.2055e-05,
+      "loss": 0.2888,
+      "step": 1589
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.2050000000000005e-05,
+      "loss": 0.4454,
+      "step": 1590
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.2045e-05,
+      "loss": 0.1341,
+      "step": 1591
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.2039999999999996e-05,
+      "loss": 0.2368,
+      "step": 1592
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.2035e-05,
+      "loss": 0.1176,
+      "step": 1593
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.203e-05,
+      "loss": 0.4003,
+      "step": 1594
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.2025e-05,
+      "loss": 0.1335,
+      "step": 1595
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.202e-05,
+      "loss": 0.122,
+      "step": 1596
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.2015e-05,
+      "loss": 0.5714,
+      "step": 1597
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.2010000000000005e-05,
+      "loss": 0.1301,
+      "step": 1598
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.2005e-05,
+      "loss": 0.2538,
+      "step": 1599
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.1919,
+      "step": 1600
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.1995000000000006e-05,
+      "loss": 0.1118,
+      "step": 1601
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.199e-05,
+      "loss": 0.3103,
+      "step": 1602
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.1985e-05,
+      "loss": 0.1212,
+      "step": 1603
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.198e-05,
+      "loss": 0.0925,
+      "step": 1604
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.1974999999999996e-05,
+      "loss": 0.2353,
+      "step": 1605
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.197e-05,
+      "loss": 0.3595,
+      "step": 1606
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.1965e-05,
+      "loss": 0.2897,
+      "step": 1607
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.1960000000000004e-05,
+      "loss": 0.0738,
+      "step": 1608
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.1955e-05,
+      "loss": 0.1757,
+      "step": 1609
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.195e-05,
+      "loss": 0.2866,
+      "step": 1610
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.1945000000000005e-05,
+      "loss": 0.2221,
+      "step": 1611
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.194e-05,
+      "loss": 0.2595,
+      "step": 1612
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.1935e-05,
+      "loss": 0.1832,
+      "step": 1613
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.1930000000000006e-05,
+      "loss": 0.2308,
+      "step": 1614
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.1924999999999995e-05,
+      "loss": 0.0832,
+      "step": 1615
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.192e-05,
+      "loss": 0.0858,
+      "step": 1616
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.1915e-05,
+      "loss": 0.237,
+      "step": 1617
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.191e-05,
+      "loss": 0.1897,
+      "step": 1618
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.1905e-05,
+      "loss": 0.3509,
+      "step": 1619
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.19e-05,
+      "loss": 0.1468,
+      "step": 1620
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.1895000000000004e-05,
+      "loss": 0.2015,
+      "step": 1621
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.189e-05,
+      "loss": 0.2603,
+      "step": 1622
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.1885e-05,
+      "loss": 0.2762,
+      "step": 1623
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.1880000000000005e-05,
+      "loss": 0.3435,
+      "step": 1624
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.1875e-05,
+      "loss": 0.4843,
+      "step": 1625
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.187e-05,
+      "loss": 0.1594,
+      "step": 1626
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.1865000000000006e-05,
+      "loss": 0.2337,
+      "step": 1627
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.186e-05,
+      "loss": 0.2499,
+      "step": 1628
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.1855e-05,
+      "loss": 0.2528,
+      "step": 1629
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.185e-05,
+      "loss": 0.1033,
+      "step": 1630
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.1845e-05,
+      "loss": 0.1396,
+      "step": 1631
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.184e-05,
+      "loss": 0.4738,
+      "step": 1632
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.1835e-05,
+      "loss": 0.1957,
+      "step": 1633
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.1830000000000004e-05,
+      "loss": 0.1559,
+      "step": 1634
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.1825e-05,
+      "loss": 0.2432,
+      "step": 1635
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.182e-05,
+      "loss": 0.111,
+      "step": 1636
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.1815000000000005e-05,
+      "loss": 0.2778,
+      "step": 1637
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.181000000000001e-05,
+      "loss": 0.1404,
+      "step": 1638
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.1805e-05,
+      "loss": 0.3359,
+      "step": 1639
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.18e-05,
+      "loss": 0.1223,
+      "step": 1640
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.1795e-05,
+      "loss": 0.1173,
+      "step": 1641
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.179e-05,
+      "loss": 0.2639,
+      "step": 1642
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.1785e-05,
+      "loss": 0.164,
+      "step": 1643
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.178e-05,
+      "loss": 0.2741,
+      "step": 1644
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.1775e-05,
+      "loss": 0.1519,
+      "step": 1645
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.177e-05,
+      "loss": 0.1034,
+      "step": 1646
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.1765e-05,
+      "loss": 0.1707,
+      "step": 1647
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.1760000000000006e-05,
+      "loss": 0.2516,
+      "step": 1648
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.1755e-05,
+      "loss": 0.1237,
+      "step": 1649
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.1750000000000004e-05,
+      "loss": 0.0925,
+      "step": 1650
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.174500000000001e-05,
+      "loss": 0.342,
+      "step": 1651
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.1739999999999996e-05,
+      "loss": 0.1951,
+      "step": 1652
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.1735e-05,
+      "loss": 0.172,
+      "step": 1653
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.173e-05,
+      "loss": 0.2988,
+      "step": 1654
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.1725e-05,
+      "loss": 0.2475,
+      "step": 1655
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.172e-05,
+      "loss": 0.2086,
+      "step": 1656
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.1715e-05,
+      "loss": 0.3958,
+      "step": 1657
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.1710000000000005e-05,
+      "loss": 0.3804,
+      "step": 1658
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.1705e-05,
+      "loss": 0.102,
+      "step": 1659
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.17e-05,
+      "loss": 0.2614,
+      "step": 1660
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.1695000000000006e-05,
+      "loss": 0.243,
+      "step": 1661
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.169e-05,
+      "loss": 0.4406,
+      "step": 1662
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.1685000000000004e-05,
+      "loss": 0.2867,
+      "step": 1663
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.168e-05,
+      "loss": 0.2102,
+      "step": 1664
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.1674999999999996e-05,
+      "loss": 0.3837,
+      "step": 1665
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.167e-05,
+      "loss": 0.1076,
+      "step": 1666
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.1665e-05,
+      "loss": 0.2477,
+      "step": 1667
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.1660000000000003e-05,
+      "loss": 0.2794,
+      "step": 1668
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.1655e-05,
+      "loss": 0.2645,
+      "step": 1669
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.165e-05,
+      "loss": 0.1692,
+      "step": 1670
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.1645000000000004e-05,
+      "loss": 0.2635,
+      "step": 1671
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.164e-05,
+      "loss": 0.3125,
+      "step": 1672
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.1635e-05,
+      "loss": 0.0898,
+      "step": 1673
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.1630000000000005e-05,
+      "loss": 0.2098,
+      "step": 1674
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.1625e-05,
+      "loss": 0.2505,
+      "step": 1675
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.162e-05,
+      "loss": 0.1826,
+      "step": 1676
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.1615e-05,
+      "loss": 0.1256,
+      "step": 1677
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.161e-05,
+      "loss": 0.1811,
+      "step": 1678
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.1605e-05,
+      "loss": 0.2339,
+      "step": 1679
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.16e-05,
+      "loss": 0.244,
+      "step": 1680
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.1595e-05,
+      "loss": 0.1418,
+      "step": 1681
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.159e-05,
+      "loss": 0.4044,
+      "step": 1682
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.1585e-05,
+      "loss": 0.2145,
+      "step": 1683
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.1580000000000004e-05,
+      "loss": 0.2163,
+      "step": 1684
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.1575e-05,
+      "loss": 0.2025,
+      "step": 1685
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.157e-05,
+      "loss": 0.2607,
+      "step": 1686
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.1565000000000005e-05,
+      "loss": 0.1369,
+      "step": 1687
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.156e-05,
+      "loss": 0.4026,
+      "step": 1688
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.1555e-05,
+      "loss": 0.1308,
+      "step": 1689
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.155e-05,
+      "loss": 0.3683,
+      "step": 1690
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.1545e-05,
+      "loss": 0.1706,
+      "step": 1691
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.154e-05,
+      "loss": 0.1962,
+      "step": 1692
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.1535e-05,
+      "loss": 0.3808,
+      "step": 1693
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.153e-05,
+      "loss": 0.155,
+      "step": 1694
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.1525e-05,
+      "loss": 0.1968,
+      "step": 1695
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.152e-05,
+      "loss": 0.1512,
+      "step": 1696
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.1515000000000004e-05,
+      "loss": 0.264,
+      "step": 1697
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.1510000000000007e-05,
+      "loss": 0.2791,
+      "step": 1698
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.1505e-05,
+      "loss": 0.2229,
+      "step": 1699
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.15e-05,
+      "loss": 0.1053,
+      "step": 1700
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.1495e-05,
+      "loss": 0.308,
+      "step": 1701
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.1489999999999997e-05,
+      "loss": 0.2728,
+      "step": 1702
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.1485e-05,
+      "loss": 0.2765,
+      "step": 1703
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.148e-05,
+      "loss": 0.1428,
+      "step": 1704
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.1475e-05,
+      "loss": 0.3661,
+      "step": 1705
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.147e-05,
+      "loss": 0.2468,
+      "step": 1706
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.1465e-05,
+      "loss": 0.253,
+      "step": 1707
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.1460000000000005e-05,
+      "loss": 0.2665,
+      "step": 1708
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.1455e-05,
+      "loss": 0.2479,
+      "step": 1709
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.1450000000000004e-05,
+      "loss": 0.1104,
+      "step": 1710
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.1445000000000006e-05,
+      "loss": 0.2592,
+      "step": 1711
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.1439999999999995e-05,
+      "loss": 0.1912,
+      "step": 1712
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.1435e-05,
+      "loss": 0.2654,
+      "step": 1713
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.143e-05,
+      "loss": 0.2217,
+      "step": 1714
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.1425e-05,
+      "loss": 0.3442,
+      "step": 1715
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.142e-05,
+      "loss": 0.0975,
+      "step": 1716
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.1415e-05,
+      "loss": 0.1179,
+      "step": 1717
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.1410000000000004e-05,
+      "loss": 0.3601,
+      "step": 1718
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.1405e-05,
+      "loss": 0.2999,
+      "step": 1719
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.14e-05,
+      "loss": 0.1714,
+      "step": 1720
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.1395000000000005e-05,
+      "loss": 0.1527,
+      "step": 1721
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.139e-05,
+      "loss": 0.1088,
+      "step": 1722
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.1385000000000004e-05,
+      "loss": 0.1448,
+      "step": 1723
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.138e-05,
+      "loss": 0.2955,
+      "step": 1724
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.1375e-05,
+      "loss": 0.0935,
+      "step": 1725
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.137e-05,
+      "loss": 0.2462,
+      "step": 1726
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.1365e-05,
+      "loss": 0.2771,
+      "step": 1727
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.136e-05,
+      "loss": 0.191,
+      "step": 1728
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.1355e-05,
+      "loss": 0.4079,
+      "step": 1729
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.135e-05,
+      "loss": 0.3203,
+      "step": 1730
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.1345000000000004e-05,
+      "loss": 0.2211,
+      "step": 1731
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.134e-05,
+      "loss": 0.1076,
+      "step": 1732
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.1335e-05,
+      "loss": 0.0863,
+      "step": 1733
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.1330000000000005e-05,
+      "loss": 0.2514,
+      "step": 1734
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.132500000000001e-05,
+      "loss": 0.2988,
+      "step": 1735
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.1319999999999997e-05,
+      "loss": 0.3083,
+      "step": 1736
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.1315e-05,
+      "loss": 0.3462,
+      "step": 1737
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.131e-05,
+      "loss": 0.2298,
+      "step": 1738
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.1305e-05,
+      "loss": 0.1448,
+      "step": 1739
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.13e-05,
+      "loss": 0.2777,
+      "step": 1740
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.1295e-05,
+      "loss": 0.1041,
+      "step": 1741
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.129e-05,
+      "loss": 0.1406,
+      "step": 1742
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.1285e-05,
+      "loss": 0.1855,
+      "step": 1743
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.1280000000000004e-05,
+      "loss": 0.0898,
+      "step": 1744
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.1275000000000006e-05,
+      "loss": 0.1242,
+      "step": 1745
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.127e-05,
+      "loss": 0.0792,
+      "step": 1746
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.1265000000000005e-05,
+      "loss": 0.202,
+      "step": 1747
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.126e-05,
+      "loss": 0.2845,
+      "step": 1748
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.1254999999999996e-05,
+      "loss": 0.1228,
+      "step": 1749
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.125e-05,
+      "loss": 0.0952,
+      "step": 1750
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.1245e-05,
+      "loss": 0.1724,
+      "step": 1751
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.124e-05,
+      "loss": 0.1092,
+      "step": 1752
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.1235e-05,
+      "loss": 0.1223,
+      "step": 1753
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.123e-05,
+      "loss": 0.2207,
+      "step": 1754
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.1225000000000005e-05,
+      "loss": 0.2837,
+      "step": 1755
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.122e-05,
+      "loss": 0.1202,
+      "step": 1756
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.1215000000000003e-05,
+      "loss": 0.3205,
+      "step": 1757
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.1210000000000006e-05,
+      "loss": 0.1667,
+      "step": 1758
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.1205e-05,
+      "loss": 0.2589,
+      "step": 1759
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 0.2507,
+      "step": 1760
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.1195e-05,
+      "loss": 0.2205,
+      "step": 1761
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.1189999999999996e-05,
+      "loss": 0.1308,
+      "step": 1762
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.1185e-05,
+      "loss": 0.3132,
+      "step": 1763
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.118e-05,
+      "loss": 0.1366,
+      "step": 1764
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.1175000000000004e-05,
+      "loss": 0.0934,
+      "step": 1765
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.117e-05,
+      "loss": 0.2231,
+      "step": 1766
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.1165e-05,
+      "loss": 0.2454,
+      "step": 1767
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.1160000000000005e-05,
+      "loss": 0.1987,
+      "step": 1768
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.1155e-05,
+      "loss": 0.0957,
+      "step": 1769
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.115e-05,
+      "loss": 0.2268,
+      "step": 1770
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.1145000000000006e-05,
+      "loss": 0.3055,
+      "step": 1771
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.114e-05,
+      "loss": 0.1556,
+      "step": 1772
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.1135e-05,
+      "loss": 0.1501,
+      "step": 1773
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.113e-05,
+      "loss": 0.3637,
+      "step": 1774
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.1125e-05,
+      "loss": 0.131,
+      "step": 1775
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.112e-05,
+      "loss": 0.0851,
+      "step": 1776
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.1115e-05,
+      "loss": 0.0696,
+      "step": 1777
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.1110000000000004e-05,
+      "loss": 0.2604,
+      "step": 1778
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.1105e-05,
+      "loss": 0.1614,
+      "step": 1779
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.11e-05,
+      "loss": 0.0902,
+      "step": 1780
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.1095000000000005e-05,
+      "loss": 0.2251,
+      "step": 1781
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.109e-05,
+      "loss": 0.3254,
+      "step": 1782
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.1085e-05,
+      "loss": 0.2657,
+      "step": 1783
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.1080000000000006e-05,
+      "loss": 0.1319,
+      "step": 1784
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.1075e-05,
+      "loss": 0.1335,
+      "step": 1785
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.107e-05,
+      "loss": 0.1393,
+      "step": 1786
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.1065e-05,
+      "loss": 0.376,
+      "step": 1787
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.106e-05,
+      "loss": 0.0806,
+      "step": 1788
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.1055e-05,
+      "loss": 0.241,
+      "step": 1789
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.105e-05,
+      "loss": 0.3027,
+      "step": 1790
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.1045000000000003e-05,
+      "loss": 0.1889,
+      "step": 1791
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.104e-05,
+      "loss": 0.0842,
+      "step": 1792
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.1035e-05,
+      "loss": 0.2907,
+      "step": 1793
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.1030000000000004e-05,
+      "loss": 0.1198,
+      "step": 1794
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.102500000000001e-05,
+      "loss": 0.0959,
+      "step": 1795
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.102e-05,
+      "loss": 0.3084,
+      "step": 1796
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1015e-05,
+      "loss": 0.1379,
+      "step": 1797
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.101e-05,
+      "loss": 0.2126,
+      "step": 1798
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.1005e-05,
+      "loss": 0.321,
+      "step": 1799
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.1e-05,
+      "loss": 0.2908,
+      "step": 1800
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0995e-05,
+      "loss": 0.1475,
+      "step": 1801
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.099e-05,
+      "loss": 0.1283,
+      "step": 1802
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.0985e-05,
+      "loss": 0.257,
+      "step": 1803
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.098e-05,
+      "loss": 0.1986,
+      "step": 1804
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.0975000000000006e-05,
+      "loss": 0.2854,
+      "step": 1805
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.097e-05,
+      "loss": 0.2141,
+      "step": 1806
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.0965000000000004e-05,
+      "loss": 0.2258,
+      "step": 1807
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.096000000000001e-05,
+      "loss": 0.2704,
+      "step": 1808
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.0954999999999996e-05,
+      "loss": 0.1208,
+      "step": 1809
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.095e-05,
+      "loss": 0.4354,
+      "step": 1810
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.0945e-05,
+      "loss": 0.1108,
+      "step": 1811
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.094e-05,
+      "loss": 0.1937,
+      "step": 1812
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.0935e-05,
+      "loss": 0.1122,
+      "step": 1813
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.093e-05,
+      "loss": 0.094,
+      "step": 1814
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.0925000000000005e-05,
+      "loss": 0.378,
+      "step": 1815
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.092e-05,
+      "loss": 0.1471,
+      "step": 1816
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.0915e-05,
+      "loss": 0.1271,
+      "step": 1817
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.0910000000000006e-05,
+      "loss": 0.1186,
+      "step": 1818
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.0905e-05,
+      "loss": 0.2524,
+      "step": 1819
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.0900000000000004e-05,
+      "loss": 0.3883,
+      "step": 1820
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.0895e-05,
+      "loss": 0.4478,
+      "step": 1821
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.0889999999999996e-05,
+      "loss": 0.2664,
+      "step": 1822
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.0885e-05,
+      "loss": 0.4737,
+      "step": 1823
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.088e-05,
+      "loss": 0.1818,
+      "step": 1824
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.0875e-05,
+      "loss": 0.2609,
+      "step": 1825
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.087e-05,
+      "loss": 0.231,
+      "step": 1826
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.0865e-05,
+      "loss": 0.3187,
+      "step": 1827
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.0860000000000004e-05,
+      "loss": 0.2957,
+      "step": 1828
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.0855e-05,
+      "loss": 0.4216,
+      "step": 1829
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.085e-05,
+      "loss": 0.1091,
+      "step": 1830
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.0845000000000005e-05,
+      "loss": 0.1305,
+      "step": 1831
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.084e-05,
+      "loss": 0.1661,
+      "step": 1832
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.0835e-05,
+      "loss": 0.0911,
+      "step": 1833
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.083e-05,
+      "loss": 0.3147,
+      "step": 1834
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.0825e-05,
+      "loss": 0.2116,
+      "step": 1835
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.082e-05,
+      "loss": 0.1896,
+      "step": 1836
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.0815e-05,
+      "loss": 0.2861,
+      "step": 1837
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.081e-05,
+      "loss": 0.2695,
+      "step": 1838
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.0805e-05,
+      "loss": 0.1,
+      "step": 1839
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.08e-05,
+      "loss": 0.1475,
+      "step": 1840
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.0795000000000004e-05,
+      "loss": 0.2468,
+      "step": 1841
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.079e-05,
+      "loss": 0.1392,
+      "step": 1842
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.0785e-05,
+      "loss": 0.1673,
+      "step": 1843
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.0780000000000005e-05,
+      "loss": 0.181,
+      "step": 1844
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.0775e-05,
+      "loss": 0.3281,
+      "step": 1845
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.077e-05,
+      "loss": 0.1737,
+      "step": 1846
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.0765e-05,
+      "loss": 0.3084,
+      "step": 1847
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.076e-05,
+      "loss": 0.1299,
+      "step": 1848
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.0755e-05,
+      "loss": 0.3281,
+      "step": 1849
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.075e-05,
+      "loss": 0.0972,
+      "step": 1850
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.0745e-05,
+      "loss": 0.1228,
+      "step": 1851
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.074e-05,
+      "loss": 0.1198,
+      "step": 1852
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.0735e-05,
+      "loss": 0.2504,
+      "step": 1853
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.0730000000000004e-05,
+      "loss": 0.2165,
+      "step": 1854
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.0725000000000006e-05,
+      "loss": 0.0658,
+      "step": 1855
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.072e-05,
+      "loss": 0.1176,
+      "step": 1856
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.0715e-05,
+      "loss": 0.2118,
+      "step": 1857
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.071e-05,
+      "loss": 0.1993,
+      "step": 1858
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.0704999999999996e-05,
+      "loss": 0.3097,
+      "step": 1859
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.07e-05,
+      "loss": 0.2681,
+      "step": 1860
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.0695e-05,
+      "loss": 0.2208,
+      "step": 1861
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.069e-05,
+      "loss": 0.2233,
+      "step": 1862
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.0685e-05,
+      "loss": 0.3275,
+      "step": 1863
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.068e-05,
+      "loss": 0.2083,
+      "step": 1864
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.0675000000000005e-05,
+      "loss": 0.0826,
+      "step": 1865
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.067e-05,
+      "loss": 0.1748,
+      "step": 1866
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.0665000000000004e-05,
+      "loss": 0.22,
+      "step": 1867
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.0660000000000006e-05,
+      "loss": 0.3071,
+      "step": 1868
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.0654999999999995e-05,
+      "loss": 0.4272,
+      "step": 1869
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.065e-05,
+      "loss": 0.2343,
+      "step": 1870
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.0645e-05,
+      "loss": 0.2296,
+      "step": 1871
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.0639999999999996e-05,
+      "loss": 0.2822,
+      "step": 1872
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.0635e-05,
+      "loss": 0.2345,
+      "step": 1873
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.063e-05,
+      "loss": 0.1204,
+      "step": 1874
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.0625000000000004e-05,
+      "loss": 0.1083,
+      "step": 1875
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.062e-05,
+      "loss": 0.2495,
+      "step": 1876
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.0615e-05,
+      "loss": 0.1811,
+      "step": 1877
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.0610000000000005e-05,
+      "loss": 0.2985,
+      "step": 1878
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.0605e-05,
+      "loss": 0.2149,
+      "step": 1879
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 0.1691,
+      "step": 1880
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.0595000000000006e-05,
+      "loss": 0.2681,
+      "step": 1881
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.059e-05,
+      "loss": 0.3692,
+      "step": 1882
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.0585e-05,
+      "loss": 0.2273,
+      "step": 1883
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.058e-05,
+      "loss": 0.4009,
+      "step": 1884
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.0575e-05,
+      "loss": 0.1472,
+      "step": 1885
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.057e-05,
+      "loss": 0.1049,
+      "step": 1886
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.0565e-05,
+      "loss": 0.1291,
+      "step": 1887
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.0560000000000004e-05,
+      "loss": 0.3079,
+      "step": 1888
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.0555e-05,
+      "loss": 0.2589,
+      "step": 1889
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.055e-05,
+      "loss": 0.1746,
+      "step": 1890
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.0545000000000005e-05,
+      "loss": 0.081,
+      "step": 1891
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.054e-05,
+      "loss": 0.2052,
+      "step": 1892
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.0535e-05,
+      "loss": 0.339,
+      "step": 1893
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.053e-05,
+      "loss": 0.2865,
+      "step": 1894
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.0525e-05,
+      "loss": 0.1837,
+      "step": 1895
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.052e-05,
+      "loss": 0.2512,
+      "step": 1896
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.0515e-05,
+      "loss": 0.2632,
+      "step": 1897
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.051e-05,
+      "loss": 0.384,
+      "step": 1898
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.0505e-05,
+      "loss": 0.2176,
+      "step": 1899
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.05e-05,
+      "loss": 0.0837,
+      "step": 1900
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.0495000000000004e-05,
+      "loss": 0.1296,
+      "step": 1901
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.0490000000000006e-05,
+      "loss": 0.2592,
+      "step": 1902
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.0485e-05,
+      "loss": 0.3119,
+      "step": 1903
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.0480000000000005e-05,
+      "loss": 0.5112,
+      "step": 1904
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.047500000000001e-05,
+      "loss": 0.2416,
+      "step": 1905
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.0469999999999996e-05,
+      "loss": 0.1588,
+      "step": 1906
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.0465e-05,
+      "loss": 0.1805,
+      "step": 1907
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.046e-05,
+      "loss": 0.1223,
+      "step": 1908
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.0455e-05,
+      "loss": 0.1404,
+      "step": 1909
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.045e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.0445e-05,
+      "loss": 0.1038,
+      "step": 1911
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.0440000000000005e-05,
+      "loss": 0.253,
+      "step": 1912
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.0435e-05,
+      "loss": 0.3182,
+      "step": 1913
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.043e-05,
+      "loss": 0.0891,
+      "step": 1914
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.0425000000000006e-05,
+      "loss": 0.3189,
+      "step": 1915
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.042e-05,
+      "loss": 0.1105,
+      "step": 1916
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.0415000000000004e-05,
+      "loss": 0.2619,
+      "step": 1917
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.041e-05,
+      "loss": 0.1167,
+      "step": 1918
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.0404999999999996e-05,
+      "loss": 0.2534,
+      "step": 1919
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.04e-05,
+      "loss": 0.0729,
+      "step": 1920
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.0395e-05,
+      "loss": 0.0728,
+      "step": 1921
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.0390000000000004e-05,
+      "loss": 0.3938,
+      "step": 1922
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.0385e-05,
+      "loss": 0.2514,
+      "step": 1923
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.038e-05,
+      "loss": 0.0926,
+      "step": 1924
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.0375000000000005e-05,
+      "loss": 0.082,
+      "step": 1925
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.037e-05,
+      "loss": 0.1315,
+      "step": 1926
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.0365e-05,
+      "loss": 0.3183,
+      "step": 1927
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.0360000000000006e-05,
+      "loss": 0.3461,
+      "step": 1928
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.0355e-05,
+      "loss": 0.2878,
+      "step": 1929
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.035e-05,
+      "loss": 0.1528,
+      "step": 1930
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.0345e-05,
+      "loss": 0.0918,
+      "step": 1931
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.034e-05,
+      "loss": 0.3476,
+      "step": 1932
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.0335e-05,
+      "loss": 0.2086,
+      "step": 1933
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.033e-05,
+      "loss": 0.1468,
+      "step": 1934
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.0325000000000004e-05,
+      "loss": 0.1184,
+      "step": 1935
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.032e-05,
+      "loss": 0.2704,
+      "step": 1936
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.0315e-05,
+      "loss": 0.1646,
+      "step": 1937
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.0310000000000005e-05,
+      "loss": 0.1045,
+      "step": 1938
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.0305e-05,
+      "loss": 0.1983,
+      "step": 1939
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.03e-05,
+      "loss": 0.0964,
+      "step": 1940
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.0295000000000006e-05,
+      "loss": 0.1109,
+      "step": 1941
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.029e-05,
+      "loss": 0.1518,
+      "step": 1942
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.0285e-05,
+      "loss": 0.2371,
+      "step": 1943
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.028e-05,
+      "loss": 0.1106,
+      "step": 1944
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.0275e-05,
+      "loss": 0.2142,
+      "step": 1945
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.027e-05,
+      "loss": 0.3483,
+      "step": 1946
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.0265e-05,
+      "loss": 0.2424,
+      "step": 1947
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.026e-05,
+      "loss": 0.1175,
+      "step": 1948
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.0255e-05,
+      "loss": 0.1031,
+      "step": 1949
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.025e-05,
+      "loss": 0.1532,
+      "step": 1950
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.0245000000000004e-05,
+      "loss": 0.1213,
+      "step": 1951
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.024000000000001e-05,
+      "loss": 0.2434,
+      "step": 1952
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.0235e-05,
+      "loss": 0.1243,
+      "step": 1953
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.023e-05,
+      "loss": 0.3553,
+      "step": 1954
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.0225e-05,
+      "loss": 0.2131,
+      "step": 1955
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.022e-05,
+      "loss": 0.0892,
+      "step": 1956
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.0215e-05,
+      "loss": 0.4248,
+      "step": 1957
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.021e-05,
+      "loss": 0.187,
+      "step": 1958
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.0205e-05,
+      "loss": 0.2725,
+      "step": 1959
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.02e-05,
+      "loss": 0.2988,
+      "step": 1960
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.0195e-05,
+      "loss": 0.2701,
+      "step": 1961
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.0190000000000006e-05,
+      "loss": 0.1439,
+      "step": 1962
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.0185e-05,
+      "loss": 0.2554,
+      "step": 1963
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.0180000000000004e-05,
+      "loss": 0.2206,
+      "step": 1964
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.017500000000001e-05,
+      "loss": 0.2033,
+      "step": 1965
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.0169999999999996e-05,
+      "loss": 0.381,
+      "step": 1966
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.0165e-05,
+      "loss": 0.0951,
+      "step": 1967
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.016e-05,
+      "loss": 0.1684,
+      "step": 1968
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.0155e-05,
+      "loss": 0.2083,
+      "step": 1969
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.015e-05,
+      "loss": 0.1742,
+      "step": 1970
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.0145e-05,
+      "loss": 0.2421,
+      "step": 1971
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.0140000000000004e-05,
+      "loss": 0.15,
+      "step": 1972
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.0135e-05,
+      "loss": 0.2905,
+      "step": 1973
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.013e-05,
+      "loss": 0.2355,
+      "step": 1974
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.0125000000000005e-05,
+      "loss": 0.1525,
+      "step": 1975
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.012e-05,
+      "loss": 0.2161,
+      "step": 1976
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.0115000000000004e-05,
+      "loss": 0.2155,
+      "step": 1977
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.011e-05,
+      "loss": 0.2247,
+      "step": 1978
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.0104999999999995e-05,
+      "loss": 0.3207,
+      "step": 1979
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.01e-05,
+      "loss": 0.1134,
+      "step": 1980
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.0095e-05,
+      "loss": 0.1414,
+      "step": 1981
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.009e-05,
+      "loss": 0.2616,
+      "step": 1982
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.0085e-05,
+      "loss": 0.1982,
+      "step": 1983
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.008e-05,
+      "loss": 0.323,
+      "step": 1984
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.0075000000000004e-05,
+      "loss": 0.1903,
+      "step": 1985
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.007e-05,
+      "loss": 0.2068,
+      "step": 1986
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.0065e-05,
+      "loss": 0.2302,
+      "step": 1987
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.0060000000000005e-05,
+      "loss": 0.3184,
+      "step": 1988
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.0055e-05,
+      "loss": 0.3111,
+      "step": 1989
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.005e-05,
+      "loss": 0.2666,
+      "step": 1990
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.0045e-05,
+      "loss": 0.2668,
+      "step": 1991
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.004e-05,
+      "loss": 0.2806,
+      "step": 1992
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.0035e-05,
+      "loss": 0.226,
+      "step": 1993
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.003e-05,
+      "loss": 0.3246,
+      "step": 1994
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.0025e-05,
+      "loss": 0.1333,
+      "step": 1995
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.002e-05,
+      "loss": 0.2357,
+      "step": 1996
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.0015e-05,
+      "loss": 0.1257,
+      "step": 1997
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.0010000000000004e-05,
+      "loss": 0.3047,
+      "step": 1998
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.0005e-05,
+      "loss": 0.2135,
+      "step": 1999
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5e-05,
+      "loss": 0.0885,
+      "step": 2000
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.9995000000000005e-05,
+      "loss": 0.1758,
+      "step": 2001
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.999000000000001e-05,
+      "loss": 0.1893,
+      "step": 2002
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.9984999999999997e-05,
+      "loss": 0.1177,
+      "step": 2003
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.998e-05,
+      "loss": 0.0648,
+      "step": 2004
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.9975e-05,
+      "loss": 0.1842,
+      "step": 2005
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.997e-05,
+      "loss": 0.1095,
+      "step": 2006
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.9965e-05,
+      "loss": 0.2067,
+      "step": 2007
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.996e-05,
+      "loss": 0.1617,
+      "step": 2008
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.9955e-05,
+      "loss": 0.1133,
+      "step": 2009
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.995e-05,
+      "loss": 0.0779,
+      "step": 2010
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.9945000000000004e-05,
+      "loss": 0.3058,
+      "step": 2011
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.9940000000000006e-05,
+      "loss": 0.2637,
+      "step": 2012
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.9935e-05,
+      "loss": 0.2785,
+      "step": 2013
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.9930000000000005e-05,
+      "loss": 0.0972,
+      "step": 2014
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.9925e-05,
+      "loss": 0.1989,
+      "step": 2015
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.9919999999999996e-05,
+      "loss": 0.308,
+      "step": 2016
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.9915e-05,
+      "loss": 0.1546,
+      "step": 2017
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.991e-05,
+      "loss": 0.3242,
+      "step": 2018
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.9905e-05,
+      "loss": 0.2682,
+      "step": 2019
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.99e-05,
+      "loss": 0.0977,
+      "step": 2020
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.9895e-05,
+      "loss": 0.2581,
+      "step": 2021
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.9890000000000005e-05,
+      "loss": 0.2392,
+      "step": 2022
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.9885e-05,
+      "loss": 0.2433,
+      "step": 2023
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.9880000000000004e-05,
+      "loss": 0.1447,
+      "step": 2024
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.9875000000000006e-05,
+      "loss": 0.1036,
+      "step": 2025
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.987e-05,
+      "loss": 0.2974,
+      "step": 2026
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.9865e-05,
+      "loss": 0.1406,
+      "step": 2027
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.986e-05,
+      "loss": 0.2271,
+      "step": 2028
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.9854999999999996e-05,
+      "loss": 0.124,
+      "step": 2029
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.985e-05,
+      "loss": 0.0811,
+      "step": 2030
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.9845e-05,
+      "loss": 0.2265,
+      "step": 2031
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9840000000000004e-05,
+      "loss": 0.3051,
+      "step": 2032
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9835e-05,
+      "loss": 0.2469,
+      "step": 2033
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.983e-05,
+      "loss": 0.2254,
+      "step": 2034
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9825000000000005e-05,
+      "loss": 0.1953,
+      "step": 2035
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.982e-05,
+      "loss": 0.1667,
+      "step": 2036
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9815e-05,
+      "loss": 0.134,
+      "step": 2037
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9810000000000006e-05,
+      "loss": 0.3116,
+      "step": 2038
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.9804999999999995e-05,
+      "loss": 0.2821,
+      "step": 2039
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.98e-05,
+      "loss": 0.1839,
+      "step": 2040
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.9795e-05,
+      "loss": 0.2227,
+      "step": 2041
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.979e-05,
+      "loss": 0.1451,
+      "step": 2042
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.9785e-05,
+      "loss": 0.3433,
+      "step": 2043
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.978e-05,
+      "loss": 0.262,
+      "step": 2044
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.9775000000000004e-05,
+      "loss": 0.1477,
+      "step": 2045
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.977e-05,
+      "loss": 0.3168,
+      "step": 2046
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.9765e-05,
+      "loss": 0.1475,
+      "step": 2047
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.9760000000000005e-05,
+      "loss": 0.1657,
+      "step": 2048
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.9755e-05,
+      "loss": 0.3378,
+      "step": 2049
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.975e-05,
+      "loss": 0.3085,
+      "step": 2050
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.9745e-05,
+      "loss": 0.2586,
+      "step": 2051
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.974e-05,
+      "loss": 0.1991,
+      "step": 2052
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.9735e-05,
+      "loss": 0.1132,
+      "step": 2053
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.973e-05,
+      "loss": 0.1745,
+      "step": 2054
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.9725e-05,
+      "loss": 0.1993,
+      "step": 2055
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.972e-05,
+      "loss": 0.2373,
+      "step": 2056
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.9715e-05,
+      "loss": 0.2303,
+      "step": 2057
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.9710000000000003e-05,
+      "loss": 0.1065,
+      "step": 2058
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.9705e-05,
+      "loss": 0.1007,
+      "step": 2059
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.97e-05,
+      "loss": 0.1743,
+      "step": 2060
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.9695000000000004e-05,
+      "loss": 0.2785,
+      "step": 2061
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.969000000000001e-05,
+      "loss": 0.1154,
+      "step": 2062
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.9684999999999996e-05,
+      "loss": 0.0892,
+      "step": 2063
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.968e-05,
+      "loss": 0.2206,
+      "step": 2064
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.9675e-05,
+      "loss": 0.0939,
+      "step": 2065
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.967e-05,
+      "loss": 0.2571,
+      "step": 2066
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.9665e-05,
+      "loss": 0.1569,
+      "step": 2067
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.966e-05,
+      "loss": 0.174,
+      "step": 2068
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.9655000000000005e-05,
+      "loss": 0.3398,
+      "step": 2069
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.965e-05,
+      "loss": 0.0997,
+      "step": 2070
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.9645e-05,
+      "loss": 0.1828,
+      "step": 2071
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.9640000000000006e-05,
+      "loss": 0.1557,
+      "step": 2072
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.9635e-05,
+      "loss": 0.2756,
+      "step": 2073
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.9630000000000004e-05,
+      "loss": 0.2062,
+      "step": 2074
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.9625e-05,
+      "loss": 0.1127,
+      "step": 2075
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.9619999999999996e-05,
+      "loss": 0.2912,
+      "step": 2076
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.9615e-05,
+      "loss": 0.3752,
+      "step": 2077
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.961e-05,
+      "loss": 0.109,
+      "step": 2078
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.9605000000000004e-05,
+      "loss": 0.1905,
+      "step": 2079
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.96e-05,
+      "loss": 0.1305,
+      "step": 2080
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.9595e-05,
+      "loss": 0.3838,
+      "step": 2081
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.9590000000000005e-05,
+      "loss": 0.2786,
+      "step": 2082
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.9585e-05,
+      "loss": 0.2377,
+      "step": 2083
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.958e-05,
+      "loss": 0.1606,
+      "step": 2084
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.9575000000000006e-05,
+      "loss": 0.2531,
+      "step": 2085
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.957e-05,
+      "loss": 0.259,
+      "step": 2086
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.9565e-05,
+      "loss": 0.2745,
+      "step": 2087
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.956e-05,
+      "loss": 0.2331,
+      "step": 2088
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.9555e-05,
+      "loss": 0.4208,
+      "step": 2089
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.955e-05,
+      "loss": 0.1991,
+      "step": 2090
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.9545e-05,
+      "loss": 0.0785,
+      "step": 2091
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.9540000000000003e-05,
+      "loss": 0.177,
+      "step": 2092
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.9535e-05,
+      "loss": 0.3153,
+      "step": 2093
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.953e-05,
+      "loss": 0.1314,
+      "step": 2094
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.9525000000000004e-05,
+      "loss": 0.1212,
+      "step": 2095
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.952e-05,
+      "loss": 0.2348,
+      "step": 2096
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.9515e-05,
+      "loss": 0.3233,
+      "step": 2097
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.9510000000000005e-05,
+      "loss": 0.2547,
+      "step": 2098
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.9505e-05,
+      "loss": 0.122,
+      "step": 2099
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.95e-05,
+      "loss": 0.0977,
+      "step": 2100
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.9495e-05,
+      "loss": 0.1137,
+      "step": 2101
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.949e-05,
+      "loss": 0.2063,
+      "step": 2102
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.9485e-05,
+      "loss": 0.1116,
+      "step": 2103
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.948e-05,
+      "loss": 0.1677,
+      "step": 2104
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.9475e-05,
+      "loss": 0.3245,
+      "step": 2105
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.947e-05,
+      "loss": 0.0596,
+      "step": 2106
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.9465e-05,
+      "loss": 0.1067,
+      "step": 2107
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.9460000000000004e-05,
+      "loss": 0.1968,
+      "step": 2108
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.945500000000001e-05,
+      "loss": 0.0909,
+      "step": 2109
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.945e-05,
+      "loss": 0.2678,
+      "step": 2110
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.9445e-05,
+      "loss": 0.2005,
+      "step": 2111
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.944e-05,
+      "loss": 0.0902,
+      "step": 2112
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.9435e-05,
+      "loss": 0.1007,
+      "step": 2113
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.943e-05,
+      "loss": 0.2111,
+      "step": 2114
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.9425e-05,
+      "loss": 0.262,
+      "step": 2115
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.942e-05,
+      "loss": 0.229,
+      "step": 2116
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.9415e-05,
+      "loss": 0.2366,
+      "step": 2117
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.941e-05,
+      "loss": 0.0926,
+      "step": 2118
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.9405000000000006e-05,
+      "loss": 0.2794,
+      "step": 2119
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.94e-05,
+      "loss": 0.3409,
+      "step": 2120
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.9395000000000004e-05,
+      "loss": 0.2564,
+      "step": 2121
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.9390000000000007e-05,
+      "loss": 0.1508,
+      "step": 2122
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.9384999999999996e-05,
+      "loss": 0.3154,
+      "step": 2123
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.938e-05,
+      "loss": 0.1747,
+      "step": 2124
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.9375e-05,
+      "loss": 0.0886,
+      "step": 2125
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.9369999999999997e-05,
+      "loss": 0.5079,
+      "step": 2126
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.9365e-05,
+      "loss": 0.3149,
+      "step": 2127
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.936e-05,
+      "loss": 0.2993,
+      "step": 2128
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.9355000000000004e-05,
+      "loss": 0.1157,
+      "step": 2129
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.935e-05,
+      "loss": 0.2785,
+      "step": 2130
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.9345e-05,
+      "loss": 0.2383,
+      "step": 2131
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.9340000000000005e-05,
+      "loss": 0.2251,
+      "step": 2132
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.9335e-05,
+      "loss": 0.2983,
+      "step": 2133
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.9330000000000004e-05,
+      "loss": 0.1191,
+      "step": 2134
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.9325000000000006e-05,
+      "loss": 0.1298,
+      "step": 2135
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.9319999999999995e-05,
+      "loss": 0.3277,
+      "step": 2136
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.9315e-05,
+      "loss": 0.1082,
+      "step": 2137
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.931e-05,
+      "loss": 0.2206,
+      "step": 2138
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.9305e-05,
+      "loss": 0.0915,
+      "step": 2139
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.93e-05,
+      "loss": 0.2511,
+      "step": 2140
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.9295e-05,
+      "loss": 0.2903,
+      "step": 2141
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.9290000000000004e-05,
+      "loss": 0.1086,
+      "step": 2142
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.9285e-05,
+      "loss": 0.2415,
+      "step": 2143
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.928e-05,
+      "loss": 0.2615,
+      "step": 2144
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.9275000000000005e-05,
+      "loss": 0.2352,
+      "step": 2145
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.927e-05,
+      "loss": 0.3479,
+      "step": 2146
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.9265000000000004e-05,
+      "loss": 0.1282,
+      "step": 2147
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.926e-05,
+      "loss": 0.1846,
+      "step": 2148
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.9255e-05,
+      "loss": 0.2833,
+      "step": 2149
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.925e-05,
+      "loss": 0.3083,
+      "step": 2150
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.9245e-05,
+      "loss": 0.1048,
+      "step": 2151
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.924e-05,
+      "loss": 0.2472,
+      "step": 2152
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.9235e-05,
+      "loss": 0.0816,
+      "step": 2153
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.923e-05,
+      "loss": 0.2388,
+      "step": 2154
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.9225000000000004e-05,
+      "loss": 0.1413,
+      "step": 2155
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.922e-05,
+      "loss": 0.1727,
+      "step": 2156
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.9215e-05,
+      "loss": 0.0936,
+      "step": 2157
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.9210000000000005e-05,
+      "loss": 0.0902,
+      "step": 2158
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.920500000000001e-05,
+      "loss": 0.1352,
+      "step": 2159
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 0.2132,
+      "step": 2160
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.9195e-05,
+      "loss": 0.2384,
+      "step": 2161
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.919e-05,
+      "loss": 0.2618,
+      "step": 2162
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.9185e-05,
+      "loss": 0.2399,
+      "step": 2163
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.918e-05,
+      "loss": 0.2397,
+      "step": 2164
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.9175e-05,
+      "loss": 0.1211,
+      "step": 2165
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.917e-05,
+      "loss": 0.2541,
+      "step": 2166
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.9165e-05,
+      "loss": 0.2329,
+      "step": 2167
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.9160000000000004e-05,
+      "loss": 0.0938,
+      "step": 2168
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.9155000000000006e-05,
+      "loss": 0.1715,
+      "step": 2169
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.915e-05,
+      "loss": 0.2246,
+      "step": 2170
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.9145000000000005e-05,
+      "loss": 0.2672,
+      "step": 2171
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.914e-05,
+      "loss": 0.106,
+      "step": 2172
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.9134999999999996e-05,
+      "loss": 0.3806,
+      "step": 2173
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.913e-05,
+      "loss": 0.1168,
+      "step": 2174
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.9125e-05,
+      "loss": 0.3481,
+      "step": 2175
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.912e-05,
+      "loss": 0.1235,
+      "step": 2176
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.9115e-05,
+      "loss": 0.3546,
+      "step": 2177
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.911e-05,
+      "loss": 0.2454,
+      "step": 2178
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.9105000000000005e-05,
+      "loss": 0.0907,
+      "step": 2179
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.91e-05,
+      "loss": 0.0704,
+      "step": 2180
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.9095000000000003e-05,
+      "loss": 0.3596,
+      "step": 2181
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.9090000000000006e-05,
+      "loss": 0.0727,
+      "step": 2182
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.9085e-05,
+      "loss": 0.1178,
+      "step": 2183
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.908e-05,
+      "loss": 0.3168,
+      "step": 2184
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.9075e-05,
+      "loss": 0.108,
+      "step": 2185
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.9069999999999996e-05,
+      "loss": 0.1193,
+      "step": 2186
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.9065e-05,
+      "loss": 0.1523,
+      "step": 2187
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.906e-05,
+      "loss": 0.2613,
+      "step": 2188
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.9055000000000004e-05,
+      "loss": 0.2278,
+      "step": 2189
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.905e-05,
+      "loss": 0.1014,
+      "step": 2190
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.9045e-05,
+      "loss": 0.0948,
+      "step": 2191
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.9040000000000005e-05,
+      "loss": 0.1634,
+      "step": 2192
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.9035e-05,
+      "loss": 0.1082,
+      "step": 2193
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.903e-05,
+      "loss": 0.1063,
+      "step": 2194
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.9025000000000006e-05,
+      "loss": 0.092,
+      "step": 2195
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.9019999999999995e-05,
+      "loss": 0.2886,
+      "step": 2196
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.9015e-05,
+      "loss": 0.3264,
+      "step": 2197
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.901e-05,
+      "loss": 0.2601,
+      "step": 2198
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.9005e-05,
+      "loss": 0.2194,
+      "step": 2199
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.9e-05,
+      "loss": 0.1325,
+      "step": 2200
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.8995e-05,
+      "loss": 0.0984,
+      "step": 2201
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.8990000000000004e-05,
+      "loss": 0.0785,
+      "step": 2202
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.8985e-05,
+      "loss": 0.2829,
+      "step": 2203
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.898e-05,
+      "loss": 0.2218,
+      "step": 2204
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.8975000000000005e-05,
+      "loss": 0.322,
+      "step": 2205
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.897e-05,
+      "loss": 0.4104,
+      "step": 2206
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.8965e-05,
+      "loss": 0.2654,
+      "step": 2207
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.896e-05,
+      "loss": 0.184,
+      "step": 2208
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.8955e-05,
+      "loss": 0.0752,
+      "step": 2209
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.895e-05,
+      "loss": 0.1307,
+      "step": 2210
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.8945e-05,
+      "loss": 0.3503,
+      "step": 2211
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.894e-05,
+      "loss": 0.2002,
+      "step": 2212
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.8935e-05,
+      "loss": 0.0973,
+      "step": 2213
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.893e-05,
+      "loss": 0.1082,
+      "step": 2214
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.8925e-05,
+      "loss": 0.2097,
+      "step": 2215
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.892e-05,
+      "loss": 0.1469,
+      "step": 2216
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.8915e-05,
+      "loss": 0.0888,
+      "step": 2217
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.8910000000000004e-05,
+      "loss": 0.0799,
+      "step": 2218
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.890500000000001e-05,
+      "loss": 0.0874,
+      "step": 2219
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.8899999999999996e-05,
+      "loss": 0.1314,
+      "step": 2220
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.8895e-05,
+      "loss": 0.1352,
+      "step": 2221
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.889e-05,
+      "loss": 0.0974,
+      "step": 2222
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.8885e-05,
+      "loss": 0.1091,
+      "step": 2223
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.888e-05,
+      "loss": 0.0657,
+      "step": 2224
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.8875e-05,
+      "loss": 0.0969,
+      "step": 2225
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.887e-05,
+      "loss": 0.1111,
+      "step": 2226
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.8865e-05,
+      "loss": 0.2353,
+      "step": 2227
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.886e-05,
+      "loss": 0.2166,
+      "step": 2228
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.8855000000000006e-05,
+      "loss": 0.0916,
+      "step": 2229
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.885e-05,
+      "loss": 0.4167,
+      "step": 2230
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.8845000000000004e-05,
+      "loss": 0.2069,
+      "step": 2231
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.884e-05,
+      "loss": 0.2424,
+      "step": 2232
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.8834999999999996e-05,
+      "loss": 0.1964,
+      "step": 2233
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.883e-05,
+      "loss": 0.2438,
+      "step": 2234
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.8825e-05,
+      "loss": 0.0917,
+      "step": 2235
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.8820000000000004e-05,
+      "loss": 0.1499,
+      "step": 2236
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.8815e-05,
+      "loss": 0.1749,
+      "step": 2237
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.881e-05,
+      "loss": 0.3485,
+      "step": 2238
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.8805000000000005e-05,
+      "loss": 0.1517,
+      "step": 2239
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.88e-05,
+      "loss": 0.158,
+      "step": 2240
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.8795e-05,
+      "loss": 0.2356,
+      "step": 2241
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.8790000000000006e-05,
+      "loss": 0.1897,
+      "step": 2242
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.8785e-05,
+      "loss": 0.2049,
+      "step": 2243
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.878e-05,
+      "loss": 0.0734,
+      "step": 2244
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.8775e-05,
+      "loss": 0.1388,
+      "step": 2245
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.877e-05,
+      "loss": 0.2667,
+      "step": 2246
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.8765e-05,
+      "loss": 0.2084,
+      "step": 2247
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.876e-05,
+      "loss": 0.5259,
+      "step": 2248
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.8755e-05,
+      "loss": 0.1043,
+      "step": 2249
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.875e-05,
+      "loss": 0.1314,
+      "step": 2250
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.8745e-05,
+      "loss": 0.1755,
+      "step": 2251
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.8740000000000004e-05,
+      "loss": 0.1054,
+      "step": 2252
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.8735e-05,
+      "loss": 0.1056,
+      "step": 2253
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.873e-05,
+      "loss": 0.0754,
+      "step": 2254
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.8725000000000005e-05,
+      "loss": 0.1144,
+      "step": 2255
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.872000000000001e-05,
+      "loss": 0.1545,
+      "step": 2256
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.8715e-05,
+      "loss": 0.1368,
+      "step": 2257
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.871e-05,
+      "loss": 0.2113,
+      "step": 2258
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.8705e-05,
+      "loss": 0.1365,
+      "step": 2259
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.87e-05,
+      "loss": 0.2607,
+      "step": 2260
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.8695e-05,
+      "loss": 0.1042,
+      "step": 2261
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.869e-05,
+      "loss": 0.2598,
+      "step": 2262
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.8685e-05,
+      "loss": 0.1223,
+      "step": 2263
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.868e-05,
+      "loss": 0.0967,
+      "step": 2264
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.8675000000000004e-05,
+      "loss": 0.2294,
+      "step": 2265
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.867000000000001e-05,
+      "loss": 0.1961,
+      "step": 2266
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.8665e-05,
+      "loss": 0.2165,
+      "step": 2267
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.8660000000000005e-05,
+      "loss": 0.193,
+      "step": 2268
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.8655e-05,
+      "loss": 0.1324,
+      "step": 2269
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.865e-05,
+      "loss": 0.144,
+      "step": 2270
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.8645e-05,
+      "loss": 0.1051,
+      "step": 2271
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.864e-05,
+      "loss": 0.1226,
+      "step": 2272
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.8635e-05,
+      "loss": 0.3324,
+      "step": 2273
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.863e-05,
+      "loss": 0.0865,
+      "step": 2274
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.8625e-05,
+      "loss": 0.1038,
+      "step": 2275
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.8620000000000005e-05,
+      "loss": 0.4532,
+      "step": 2276
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.8615e-05,
+      "loss": 0.1249,
+      "step": 2277
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.8610000000000004e-05,
+      "loss": 0.134,
+      "step": 2278
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.8605000000000006e-05,
+      "loss": 0.3848,
+      "step": 2279
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.86e-05,
+      "loss": 0.1841,
+      "step": 2280
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.8595e-05,
+      "loss": 0.1508,
+      "step": 2281
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.859e-05,
+      "loss": 0.2626,
+      "step": 2282
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.8584999999999996e-05,
+      "loss": 0.0777,
+      "step": 2283
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.858e-05,
+      "loss": 0.1176,
+      "step": 2284
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.8575e-05,
+      "loss": 0.3387,
+      "step": 2285
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.8570000000000004e-05,
+      "loss": 0.2564,
+      "step": 2286
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.8565e-05,
+      "loss": 0.4088,
+      "step": 2287
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.856e-05,
+      "loss": 0.2215,
+      "step": 2288
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.8555000000000005e-05,
+      "loss": 0.1933,
+      "step": 2289
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.855e-05,
+      "loss": 0.1594,
+      "step": 2290
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.8545000000000004e-05,
+      "loss": 0.2295,
+      "step": 2291
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.8540000000000006e-05,
+      "loss": 0.0757,
+      "step": 2292
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.8534999999999995e-05,
+      "loss": 0.2016,
+      "step": 2293
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.853e-05,
+      "loss": 0.257,
+      "step": 2294
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.8525e-05,
+      "loss": 0.2161,
+      "step": 2295
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.852e-05,
+      "loss": 0.1045,
+      "step": 2296
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.8515e-05,
+      "loss": 0.358,
+      "step": 2297
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.851e-05,
+      "loss": 0.2829,
+      "step": 2298
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.8505000000000004e-05,
+      "loss": 0.2266,
+      "step": 2299
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.85e-05,
+      "loss": 0.0818,
+      "step": 2300
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.8495e-05,
+      "loss": 0.242,
+      "step": 2301
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.8490000000000005e-05,
+      "loss": 0.2826,
+      "step": 2302
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.8485e-05,
+      "loss": 0.3794,
+      "step": 2303
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.8480000000000003e-05,
+      "loss": 0.1026,
+      "step": 2304
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.8475e-05,
+      "loss": 0.1076,
+      "step": 2305
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.847e-05,
+      "loss": 0.1421,
+      "step": 2306
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.8465e-05,
+      "loss": 0.1527,
+      "step": 2307
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.846e-05,
+      "loss": 0.1892,
+      "step": 2308
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.8455e-05,
+      "loss": 0.0991,
+      "step": 2309
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.845e-05,
+      "loss": 0.0837,
+      "step": 2310
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.8445e-05,
+      "loss": 0.2805,
+      "step": 2311
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.8440000000000004e-05,
+      "loss": 0.2403,
+      "step": 2312
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.8435e-05,
+      "loss": 0.2239,
+      "step": 2313
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.843e-05,
+      "loss": 0.124,
+      "step": 2314
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.8425000000000005e-05,
+      "loss": 0.1714,
+      "step": 2315
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.842000000000001e-05,
+      "loss": 0.098,
+      "step": 2316
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.8414999999999996e-05,
+      "loss": 0.2899,
+      "step": 2317
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.841e-05,
+      "loss": 0.1489,
+      "step": 2318
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.8405e-05,
+      "loss": 0.176,
+      "step": 2319
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.84e-05,
+      "loss": 0.3006,
+      "step": 2320
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.8395e-05,
+      "loss": 0.274,
+      "step": 2321
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.839e-05,
+      "loss": 0.1329,
+      "step": 2322
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.8385e-05,
+      "loss": 0.2553,
+      "step": 2323
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.838e-05,
+      "loss": 0.0994,
+      "step": 2324
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.8375000000000004e-05,
+      "loss": 0.2736,
+      "step": 2325
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.8370000000000006e-05,
+      "loss": 0.3995,
+      "step": 2326
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.8365e-05,
+      "loss": 0.2598,
+      "step": 2327
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.8360000000000005e-05,
+      "loss": 0.1308,
+      "step": 2328
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.8355e-05,
+      "loss": 0.3206,
+      "step": 2329
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.8349999999999996e-05,
+      "loss": 0.1979,
+      "step": 2330
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.8345e-05,
+      "loss": 0.4997,
+      "step": 2331
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.834e-05,
+      "loss": 0.2726,
+      "step": 2332
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.8335e-05,
+      "loss": 0.1194,
+      "step": 2333
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.833e-05,
+      "loss": 0.1664,
+      "step": 2334
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.8325e-05,
+      "loss": 0.1671,
+      "step": 2335
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.8320000000000005e-05,
+      "loss": 0.3778,
+      "step": 2336
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.8315e-05,
+      "loss": 0.3036,
+      "step": 2337
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.831e-05,
+      "loss": 0.1332,
+      "step": 2338
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.8305000000000006e-05,
+      "loss": 0.2688,
+      "step": 2339
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.83e-05,
+      "loss": 0.1044,
+      "step": 2340
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.8295e-05,
+      "loss": 0.4063,
+      "step": 2341
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.829e-05,
+      "loss": 0.2258,
+      "step": 2342
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.8284999999999996e-05,
+      "loss": 0.2206,
+      "step": 2343
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.828e-05,
+      "loss": 0.2269,
+      "step": 2344
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.8275e-05,
+      "loss": 0.2207,
+      "step": 2345
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.8270000000000004e-05,
+      "loss": 0.2195,
+      "step": 2346
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.8265e-05,
+      "loss": 0.3057,
+      "step": 2347
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.826e-05,
+      "loss": 0.3242,
+      "step": 2348
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.8255000000000005e-05,
+      "loss": 0.1868,
+      "step": 2349
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.825e-05,
+      "loss": 0.0895,
+      "step": 2350
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.8245e-05,
+      "loss": 0.1553,
+      "step": 2351
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.8240000000000006e-05,
+      "loss": 0.2777,
+      "step": 2352
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.8234999999999995e-05,
+      "loss": 0.2143,
+      "step": 2353
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.823e-05,
+      "loss": 0.1408,
+      "step": 2354
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.8225e-05,
+      "loss": 0.1347,
+      "step": 2355
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.822e-05,
+      "loss": 0.2221,
+      "step": 2356
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.8215e-05,
+      "loss": 0.2009,
+      "step": 2357
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.821e-05,
+      "loss": 0.1387,
+      "step": 2358
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.8205000000000003e-05,
+      "loss": 0.309,
+      "step": 2359
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.82e-05,
+      "loss": 0.1286,
+      "step": 2360
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.8195e-05,
+      "loss": 0.1628,
+      "step": 2361
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.8190000000000004e-05,
+      "loss": 0.3083,
+      "step": 2362
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.8185e-05,
+      "loss": 0.1237,
+      "step": 2363
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.818e-05,
+      "loss": 0.0591,
+      "step": 2364
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.8175e-05,
+      "loss": 0.2773,
+      "step": 2365
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.817e-05,
+      "loss": 0.206,
+      "step": 2366
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.8165e-05,
+      "loss": 0.4586,
+      "step": 2367
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.816e-05,
+      "loss": 0.2822,
+      "step": 2368
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.8155e-05,
+      "loss": 0.234,
+      "step": 2369
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.815e-05,
+      "loss": 0.1674,
+      "step": 2370
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.8145e-05,
+      "loss": 0.1722,
+      "step": 2371
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.814e-05,
+      "loss": 0.2516,
+      "step": 2372
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.8135e-05,
+      "loss": 0.2103,
+      "step": 2373
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.813e-05,
+      "loss": 0.2639,
+      "step": 2374
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.8125000000000004e-05,
+      "loss": 0.1515,
+      "step": 2375
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.812000000000001e-05,
+      "loss": 0.1961,
+      "step": 2376
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.8115e-05,
+      "loss": 0.1809,
+      "step": 2377
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.811e-05,
+      "loss": 0.2191,
+      "step": 2378
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.8105e-05,
+      "loss": 0.0708,
+      "step": 2379
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.81e-05,
+      "loss": 0.2038,
+      "step": 2380
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.8095e-05,
+      "loss": 0.1412,
+      "step": 2381
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.809e-05,
+      "loss": 0.2116,
+      "step": 2382
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.8085e-05,
+      "loss": 0.2639,
+      "step": 2383
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.808e-05,
+      "loss": 0.1958,
+      "step": 2384
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.8075e-05,
+      "loss": 0.2142,
+      "step": 2385
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.8070000000000006e-05,
+      "loss": 0.1571,
+      "step": 2386
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.8065e-05,
+      "loss": 0.1903,
+      "step": 2387
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.8060000000000004e-05,
+      "loss": 0.3243,
+      "step": 2388
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.805500000000001e-05,
+      "loss": 0.1824,
+      "step": 2389
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.8049999999999996e-05,
+      "loss": 0.2239,
+      "step": 2390
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.8045e-05,
+      "loss": 0.3465,
+      "step": 2391
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.804e-05,
+      "loss": 0.2129,
+      "step": 2392
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.8035e-05,
+      "loss": 0.2464,
+      "step": 2393
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.803e-05,
+      "loss": 0.1673,
+      "step": 2394
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.8025e-05,
+      "loss": 0.1454,
+      "step": 2395
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.8020000000000004e-05,
+      "loss": 0.2265,
+      "step": 2396
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.8015e-05,
+      "loss": 0.2871,
+      "step": 2397
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.801e-05,
+      "loss": 0.1147,
+      "step": 2398
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.8005000000000005e-05,
+      "loss": 0.189,
+      "step": 2399
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.8e-05,
+      "loss": 0.2042,
+      "step": 2400
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.7995000000000004e-05,
+      "loss": 0.1174,
+      "step": 2401
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.799e-05,
+      "loss": 0.1739,
+      "step": 2402
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.7985e-05,
+      "loss": 0.4238,
+      "step": 2403
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.798e-05,
+      "loss": 0.088,
+      "step": 2404
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.7975e-05,
+      "loss": 0.3368,
+      "step": 2405
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.797e-05,
+      "loss": 0.2178,
+      "step": 2406
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.7965e-05,
+      "loss": 0.2585,
+      "step": 2407
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.796e-05,
+      "loss": 0.3115,
+      "step": 2408
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.7955000000000004e-05,
+      "loss": 0.1144,
+      "step": 2409
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.795e-05,
+      "loss": 0.1923,
+      "step": 2410
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.7945e-05,
+      "loss": 0.2071,
+      "step": 2411
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.7940000000000005e-05,
+      "loss": 0.2453,
+      "step": 2412
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.7935e-05,
+      "loss": 0.1211,
+      "step": 2413
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.793e-05,
+      "loss": 0.208,
+      "step": 2414
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.7925e-05,
+      "loss": 0.1972,
+      "step": 2415
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.792e-05,
+      "loss": 0.162,
+      "step": 2416
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.7915e-05,
+      "loss": 0.0986,
+      "step": 2417
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.791e-05,
+      "loss": 0.2711,
+      "step": 2418
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.7905e-05,
+      "loss": 0.1102,
+      "step": 2419
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.79e-05,
+      "loss": 0.0664,
+      "step": 2420
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.7895e-05,
+      "loss": 0.0805,
+      "step": 2421
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.7890000000000004e-05,
+      "loss": 0.1277,
+      "step": 2422
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.7885000000000007e-05,
+      "loss": 0.0857,
+      "step": 2423
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.788e-05,
+      "loss": 0.0446,
+      "step": 2424
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.7875000000000005e-05,
+      "loss": 0.1583,
+      "step": 2425
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.787e-05,
+      "loss": 0.2379,
+      "step": 2426
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.7864999999999997e-05,
+      "loss": 0.1955,
+      "step": 2427
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.786e-05,
+      "loss": 0.0715,
+      "step": 2428
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.7855e-05,
+      "loss": 0.0862,
+      "step": 2429
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.785e-05,
+      "loss": 0.2018,
+      "step": 2430
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.7845e-05,
+      "loss": 0.2593,
+      "step": 2431
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.784e-05,
+      "loss": 0.2446,
+      "step": 2432
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.7835000000000005e-05,
+      "loss": 0.1028,
+      "step": 2433
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.783e-05,
+      "loss": 0.2114,
+      "step": 2434
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.7825000000000004e-05,
+      "loss": 0.1799,
+      "step": 2435
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.7820000000000006e-05,
+      "loss": 0.1137,
+      "step": 2436
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.7815e-05,
+      "loss": 0.1017,
+      "step": 2437
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.781e-05,
+      "loss": 0.307,
+      "step": 2438
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.7805e-05,
+      "loss": 0.0974,
+      "step": 2439
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.7799999999999996e-05,
+      "loss": 0.089,
+      "step": 2440
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.7795e-05,
+      "loss": 0.1032,
+      "step": 2441
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.779e-05,
+      "loss": 0.0854,
+      "step": 2442
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.7785000000000004e-05,
+      "loss": 0.1497,
+      "step": 2443
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.778e-05,
+      "loss": 0.097,
+      "step": 2444
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.7775e-05,
+      "loss": 0.0916,
+      "step": 2445
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.7770000000000005e-05,
+      "loss": 0.104,
+      "step": 2446
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.7765e-05,
+      "loss": 0.4775,
+      "step": 2447
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.7760000000000004e-05,
+      "loss": 0.2951,
+      "step": 2448
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.7755000000000006e-05,
+      "loss": 0.2354,
+      "step": 2449
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.7749999999999995e-05,
+      "loss": 0.1362,
+      "step": 2450
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.7745e-05,
+      "loss": 0.2417,
+      "step": 2451
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.774e-05,
+      "loss": 0.3215,
+      "step": 2452
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.7735e-05,
+      "loss": 0.2119,
+      "step": 2453
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.773e-05,
+      "loss": 0.2377,
+      "step": 2454
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.7725e-05,
+      "loss": 0.0903,
+      "step": 2455
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.7720000000000004e-05,
+      "loss": 0.0808,
+      "step": 2456
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.7715e-05,
+      "loss": 0.2606,
+      "step": 2457
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.771e-05,
+      "loss": 0.1074,
+      "step": 2458
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.7705000000000005e-05,
+      "loss": 0.0795,
+      "step": 2459
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.77e-05,
+      "loss": 0.2096,
+      "step": 2460
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.7695e-05,
+      "loss": 0.2003,
+      "step": 2461
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.769e-05,
+      "loss": 0.3292,
+      "step": 2462
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.7685e-05,
+      "loss": 0.2467,
+      "step": 2463
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.768e-05,
+      "loss": 0.1443,
+      "step": 2464
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.7675e-05,
+      "loss": 0.3092,
+      "step": 2465
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.767e-05,
+      "loss": 0.2159,
+      "step": 2466
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.7665e-05,
+      "loss": 0.2134,
+      "step": 2467
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.766e-05,
+      "loss": 0.1239,
+      "step": 2468
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.7655000000000004e-05,
+      "loss": 0.1015,
+      "step": 2469
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.765e-05,
+      "loss": 0.1006,
+      "step": 2470
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.7645e-05,
+      "loss": 0.3096,
+      "step": 2471
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.7640000000000005e-05,
+      "loss": 0.2051,
+      "step": 2472
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.763500000000001e-05,
+      "loss": 0.0823,
+      "step": 2473
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.7629999999999996e-05,
+      "loss": 0.1933,
+      "step": 2474
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.7625e-05,
+      "loss": 0.2873,
+      "step": 2475
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.762e-05,
+      "loss": 0.1672,
+      "step": 2476
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.7615e-05,
+      "loss": 0.1134,
+      "step": 2477
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.761e-05,
+      "loss": 0.3106,
+      "step": 2478
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.7605e-05,
+      "loss": 0.2569,
+      "step": 2479
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.76e-05,
+      "loss": 0.0908,
+      "step": 2480
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.7595e-05,
+      "loss": 0.1271,
+      "step": 2481
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.7590000000000003e-05,
+      "loss": 0.2189,
+      "step": 2482
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.7585000000000006e-05,
+      "loss": 0.1467,
+      "step": 2483
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.758e-05,
+      "loss": 0.1688,
+      "step": 2484
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.7575000000000004e-05,
+      "loss": 0.2929,
+      "step": 2485
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.757e-05,
+      "loss": 0.1847,
+      "step": 2486
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.7564999999999996e-05,
+      "loss": 0.2276,
+      "step": 2487
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.756e-05,
+      "loss": 0.1827,
+      "step": 2488
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.7555e-05,
+      "loss": 0.2187,
+      "step": 2489
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.755e-05,
+      "loss": 0.0817,
+      "step": 2490
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.7545e-05,
+      "loss": 0.2635,
+      "step": 2491
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.754e-05,
+      "loss": 0.1063,
+      "step": 2492
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.7535000000000005e-05,
+      "loss": 0.245,
+      "step": 2493
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.753e-05,
+      "loss": 0.2457,
+      "step": 2494
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.7525e-05,
+      "loss": 0.1071,
+      "step": 2495
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.7520000000000006e-05,
+      "loss": 0.3732,
+      "step": 2496
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.7515e-05,
+      "loss": 0.2191,
+      "step": 2497
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.751e-05,
+      "loss": 0.2734,
+      "step": 2498
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.7505e-05,
+      "loss": 0.2813,
+      "step": 2499
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.7499999999999996e-05,
+      "loss": 0.3023,
+      "step": 2500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.7495e-05,
+      "loss": 0.313,
+      "step": 2501
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.749e-05,
+      "loss": 0.0962,
+      "step": 2502
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.7485000000000004e-05,
+      "loss": 0.0956,
+      "step": 2503
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.748e-05,
+      "loss": 0.265,
+      "step": 2504
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.7475e-05,
+      "loss": 0.1193,
+      "step": 2505
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.7470000000000005e-05,
+      "loss": 0.1498,
+      "step": 2506
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.7465e-05,
+      "loss": 0.1843,
+      "step": 2507
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.746e-05,
+      "loss": 0.0796,
+      "step": 2508
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.7455000000000006e-05,
+      "loss": 0.234,
+      "step": 2509
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.745e-05,
+      "loss": 0.222,
+      "step": 2510
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.7445e-05,
+      "loss": 0.246,
+      "step": 2511
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.744e-05,
+      "loss": 0.0756,
+      "step": 2512
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.7435e-05,
+      "loss": 0.1927,
+      "step": 2513
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.743e-05,
+      "loss": 0.0758,
+      "step": 2514
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.7425e-05,
+      "loss": 0.1933,
+      "step": 2515
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.742e-05,
+      "loss": 0.2014,
+      "step": 2516
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.7415e-05,
+      "loss": 0.2701,
+      "step": 2517
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.741e-05,
+      "loss": 0.3149,
+      "step": 2518
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.7405000000000004e-05,
+      "loss": 0.0892,
+      "step": 2519
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.74e-05,
+      "loss": 0.2176,
+      "step": 2520
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.7395e-05,
+      "loss": 0.1555,
+      "step": 2521
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.7390000000000005e-05,
+      "loss": 0.0749,
+      "step": 2522
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.7385e-05,
+      "loss": 0.228,
+      "step": 2523
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.738e-05,
+      "loss": 0.0974,
+      "step": 2524
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.7375e-05,
+      "loss": 0.3475,
+      "step": 2525
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.737e-05,
+      "loss": 0.3242,
+      "step": 2526
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.7365e-05,
+      "loss": 0.1582,
+      "step": 2527
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.736e-05,
+      "loss": 0.1838,
+      "step": 2528
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.7355e-05,
+      "loss": 0.1172,
+      "step": 2529
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.735e-05,
+      "loss": 0.1113,
+      "step": 2530
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.7345e-05,
+      "loss": 0.2221,
+      "step": 2531
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.7340000000000004e-05,
+      "loss": 0.1958,
+      "step": 2532
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.733500000000001e-05,
+      "loss": 0.2813,
+      "step": 2533
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.733e-05,
+      "loss": 0.3343,
+      "step": 2534
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.7325e-05,
+      "loss": 0.3295,
+      "step": 2535
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.732e-05,
+      "loss": 0.1964,
+      "step": 2536
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.7315e-05,
+      "loss": 0.1251,
+      "step": 2537
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.731e-05,
+      "loss": 0.1009,
+      "step": 2538
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.7305e-05,
+      "loss": 0.2609,
+      "step": 2539
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.73e-05,
+      "loss": 0.0751,
+      "step": 2540
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.7295e-05,
+      "loss": 0.2279,
+      "step": 2541
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.729e-05,
+      "loss": 0.2907,
+      "step": 2542
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.7285000000000006e-05,
+      "loss": 0.1296,
+      "step": 2543
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.728e-05,
+      "loss": 0.2142,
+      "step": 2544
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.7275000000000004e-05,
+      "loss": 0.2429,
+      "step": 2545
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.7270000000000007e-05,
+      "loss": 0.3477,
+      "step": 2546
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.7264999999999996e-05,
+      "loss": 0.2544,
+      "step": 2547
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.726e-05,
+      "loss": 0.2253,
+      "step": 2548
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.7255e-05,
+      "loss": 0.1053,
+      "step": 2549
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 0.0956,
+      "step": 2550
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.7245e-05,
+      "loss": 0.1557,
+      "step": 2551
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.724e-05,
+      "loss": 0.2718,
+      "step": 2552
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.7235000000000004e-05,
+      "loss": 0.2237,
+      "step": 2553
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.723e-05,
+      "loss": 0.3543,
+      "step": 2554
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.7225e-05,
+      "loss": 0.2058,
+      "step": 2555
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.7220000000000005e-05,
+      "loss": 0.1655,
+      "step": 2556
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.7215e-05,
+      "loss": 0.0898,
+      "step": 2557
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.7210000000000004e-05,
+      "loss": 0.0929,
+      "step": 2558
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.7205e-05,
+      "loss": 0.1774,
+      "step": 2559
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.7199999999999995e-05,
+      "loss": 0.0827,
+      "step": 2560
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.7195e-05,
+      "loss": 0.0736,
+      "step": 2561
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.719e-05,
+      "loss": 0.1237,
+      "step": 2562
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.7185e-05,
+      "loss": 0.1046,
+      "step": 2563
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.718e-05,
+      "loss": 0.1027,
+      "step": 2564
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.7175e-05,
+      "loss": 0.2106,
+      "step": 2565
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.7170000000000004e-05,
+      "loss": 0.165,
+      "step": 2566
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.7165e-05,
+      "loss": 0.2654,
+      "step": 2567
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.716e-05,
+      "loss": 0.0934,
+      "step": 2568
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.7155000000000005e-05,
+      "loss": 0.2393,
+      "step": 2569
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.715e-05,
+      "loss": 0.0757,
+      "step": 2570
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.7145e-05,
+      "loss": 0.1825,
+      "step": 2571
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.714e-05,
+      "loss": 0.1179,
+      "step": 2572
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.7135e-05,
+      "loss": 0.6095,
+      "step": 2573
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.713e-05,
+      "loss": 0.2511,
+      "step": 2574
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.7125e-05,
+      "loss": 0.0724,
+      "step": 2575
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.712e-05,
+      "loss": 0.2541,
+      "step": 2576
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.7115e-05,
+      "loss": 0.1579,
+      "step": 2577
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.711e-05,
+      "loss": 0.237,
+      "step": 2578
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.7105000000000004e-05,
+      "loss": 0.0728,
+      "step": 2579
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.7100000000000006e-05,
+      "loss": 0.1136,
+      "step": 2580
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.7095e-05,
+      "loss": 0.1478,
+      "step": 2581
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.7090000000000005e-05,
+      "loss": 0.1142,
+      "step": 2582
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.7085e-05,
+      "loss": 0.1535,
+      "step": 2583
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.7079999999999996e-05,
+      "loss": 0.1295,
+      "step": 2584
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.7075e-05,
+      "loss": 0.2351,
+      "step": 2585
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.707e-05,
+      "loss": 0.265,
+      "step": 2586
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.7065e-05,
+      "loss": 0.2354,
+      "step": 2587
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.706e-05,
+      "loss": 0.2309,
+      "step": 2588
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.7055e-05,
+      "loss": 0.3732,
+      "step": 2589
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.7050000000000005e-05,
+      "loss": 0.4318,
+      "step": 2590
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.7045e-05,
+      "loss": 0.0599,
+      "step": 2591
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.7040000000000004e-05,
+      "loss": 0.232,
+      "step": 2592
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.7035000000000006e-05,
+      "loss": 0.1281,
+      "step": 2593
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.703e-05,
+      "loss": 0.0818,
+      "step": 2594
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.7025e-05,
+      "loss": 0.087,
+      "step": 2595
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.702e-05,
+      "loss": 0.1346,
+      "step": 2596
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7014999999999996e-05,
+      "loss": 0.2484,
+      "step": 2597
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.701e-05,
+      "loss": 0.0809,
+      "step": 2598
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7005e-05,
+      "loss": 0.2284,
+      "step": 2599
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7000000000000004e-05,
+      "loss": 0.2876,
+      "step": 2600
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.6995e-05,
+      "loss": 0.2246,
+      "step": 2601
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.699e-05,
+      "loss": 0.0831,
+      "step": 2602
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.6985000000000005e-05,
+      "loss": 0.0909,
+      "step": 2603
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.698e-05,
+      "loss": 0.288,
+      "step": 2604
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.6975000000000003e-05,
+      "loss": 0.2104,
+      "step": 2605
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.6970000000000006e-05,
+      "loss": 0.2602,
+      "step": 2606
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.6964999999999995e-05,
+      "loss": 0.1635,
+      "step": 2607
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.696e-05,
+      "loss": 0.1372,
+      "step": 2608
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.6955e-05,
+      "loss": 0.2303,
+      "step": 2609
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.695e-05,
+      "loss": 0.0888,
+      "step": 2610
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.6945e-05,
+      "loss": 0.0856,
+      "step": 2611
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.694e-05,
+      "loss": 0.5085,
+      "step": 2612
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.6935000000000004e-05,
+      "loss": 0.3162,
+      "step": 2613
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.693e-05,
+      "loss": 0.0879,
+      "step": 2614
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.6925e-05,
+      "loss": 0.2206,
+      "step": 2615
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.6920000000000005e-05,
+      "loss": 0.1691,
+      "step": 2616
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.6915e-05,
+      "loss": 0.4644,
+      "step": 2617
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.691e-05,
+      "loss": 0.2649,
+      "step": 2618
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.6905e-05,
+      "loss": 0.1149,
+      "step": 2619
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.69e-05,
+      "loss": 0.3104,
+      "step": 2620
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.6895e-05,
+      "loss": 0.1802,
+      "step": 2621
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.689e-05,
+      "loss": 0.2382,
+      "step": 2622
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.6885e-05,
+      "loss": 0.2901,
+      "step": 2623
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.688e-05,
+      "loss": 0.0663,
+      "step": 2624
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.2313,
+      "step": 2625
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.6870000000000004e-05,
+      "loss": 0.1666,
+      "step": 2626
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.6865e-05,
+      "loss": 0.1059,
+      "step": 2627
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.686e-05,
+      "loss": 0.1721,
+      "step": 2628
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.6855000000000005e-05,
+      "loss": 0.2275,
+      "step": 2629
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.685000000000001e-05,
+      "loss": 0.0815,
+      "step": 2630
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.6845e-05,
+      "loss": 0.0809,
+      "step": 2631
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.684e-05,
+      "loss": 0.6719,
+      "step": 2632
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.6835e-05,
+      "loss": 0.2299,
+      "step": 2633
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.683e-05,
+      "loss": 0.1212,
+      "step": 2634
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.6825e-05,
+      "loss": 0.1947,
+      "step": 2635
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.682e-05,
+      "loss": 0.1959,
+      "step": 2636
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.6815e-05,
+      "loss": 0.1058,
+      "step": 2637
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.681e-05,
+      "loss": 0.2952,
+      "step": 2638
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.6805e-05,
+      "loss": 0.202,
+      "step": 2639
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 0.1901,
+      "step": 2640
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.6795e-05,
+      "loss": 0.0793,
+      "step": 2641
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.6790000000000004e-05,
+      "loss": 0.0871,
+      "step": 2642
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.678500000000001e-05,
+      "loss": 0.3878,
+      "step": 2643
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.6779999999999996e-05,
+      "loss": 0.077,
+      "step": 2644
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.6775e-05,
+      "loss": 0.1898,
+      "step": 2645
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.677e-05,
+      "loss": 0.0957,
+      "step": 2646
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.6765e-05,
+      "loss": 0.0854,
+      "step": 2647
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.676e-05,
+      "loss": 0.3359,
+      "step": 2648
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.6755e-05,
+      "loss": 0.3119,
+      "step": 2649
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.6750000000000005e-05,
+      "loss": 0.2626,
+      "step": 2650
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.6745e-05,
+      "loss": 0.2644,
+      "step": 2651
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.674e-05,
+      "loss": 0.2516,
+      "step": 2652
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.6735000000000006e-05,
+      "loss": 0.0818,
+      "step": 2653
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.673e-05,
+      "loss": 0.1776,
+      "step": 2654
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.6725000000000004e-05,
+      "loss": 0.3138,
+      "step": 2655
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.672e-05,
+      "loss": 0.2663,
+      "step": 2656
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.6714999999999996e-05,
+      "loss": 0.1519,
+      "step": 2657
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.671e-05,
+      "loss": 0.2123,
+      "step": 2658
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.6705e-05,
+      "loss": 0.0815,
+      "step": 2659
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 0.0834,
+      "step": 2660
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.6695e-05,
+      "loss": 0.2414,
+      "step": 2661
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.669e-05,
+      "loss": 0.0642,
+      "step": 2662
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.6685000000000004e-05,
+      "loss": 0.1385,
+      "step": 2663
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.668e-05,
+      "loss": 0.2159,
+      "step": 2664
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.6675e-05,
+      "loss": 0.101,
+      "step": 2665
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.6670000000000005e-05,
+      "loss": 0.2628,
+      "step": 2666
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.6665e-05,
+      "loss": 0.0884,
+      "step": 2667
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.666e-05,
+      "loss": 0.1368,
+      "step": 2668
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.6655e-05,
+      "loss": 0.2739,
+      "step": 2669
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.665e-05,
+      "loss": 0.2163,
+      "step": 2670
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.6645e-05,
+      "loss": 0.0756,
+      "step": 2671
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.664e-05,
+      "loss": 0.1463,
+      "step": 2672
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.6635e-05,
+      "loss": 0.203,
+      "step": 2673
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.663e-05,
+      "loss": 0.0684,
+      "step": 2674
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.6625e-05,
+      "loss": 0.0777,
+      "step": 2675
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.6620000000000004e-05,
+      "loss": 0.1828,
+      "step": 2676
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.6615e-05,
+      "loss": 0.0703,
+      "step": 2677
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.661e-05,
+      "loss": 0.1097,
+      "step": 2678
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.6605000000000005e-05,
+      "loss": 0.3046,
+      "step": 2679
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.66e-05,
+      "loss": 0.3368,
+      "step": 2680
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.6595e-05,
+      "loss": 0.2173,
+      "step": 2681
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.659e-05,
+      "loss": 0.2513,
+      "step": 2682
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.6585e-05,
+      "loss": 0.0826,
+      "step": 2683
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.658e-05,
+      "loss": 0.4251,
+      "step": 2684
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.6575e-05,
+      "loss": 0.1946,
+      "step": 2685
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.657e-05,
+      "loss": 0.211,
+      "step": 2686
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.6565e-05,
+      "loss": 0.1845,
+      "step": 2687
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.656e-05,
+      "loss": 0.2791,
+      "step": 2688
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.6555000000000004e-05,
+      "loss": 0.235,
+      "step": 2689
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.655000000000001e-05,
+      "loss": 0.1392,
+      "step": 2690
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.6545e-05,
+      "loss": 0.117,
+      "step": 2691
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.654e-05,
+      "loss": 0.1938,
+      "step": 2692
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6535e-05,
+      "loss": 0.2495,
+      "step": 2693
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.653e-05,
+      "loss": 0.1619,
+      "step": 2694
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6525e-05,
+      "loss": 0.3016,
+      "step": 2695
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.652e-05,
+      "loss": 0.2541,
+      "step": 2696
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6515e-05,
+      "loss": 0.0863,
+      "step": 2697
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.651e-05,
+      "loss": 0.3574,
+      "step": 2698
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6505e-05,
+      "loss": 0.1858,
+      "step": 2699
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 0.3768,
+      "step": 2700
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6495e-05,
+      "loss": 0.3981,
+      "step": 2701
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6490000000000004e-05,
+      "loss": 0.1842,
+      "step": 2702
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6485000000000006e-05,
+      "loss": 0.1004,
+      "step": 2703
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6479999999999995e-05,
+      "loss": 0.2925,
+      "step": 2704
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6475e-05,
+      "loss": 0.214,
+      "step": 2705
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.647e-05,
+      "loss": 0.2638,
+      "step": 2706
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6464999999999996e-05,
+      "loss": 0.1225,
+      "step": 2707
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.646e-05,
+      "loss": 0.0772,
+      "step": 2708
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6455e-05,
+      "loss": 0.286,
+      "step": 2709
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6450000000000004e-05,
+      "loss": 0.2537,
+      "step": 2710
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6445e-05,
+      "loss": 0.0971,
+      "step": 2711
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.644e-05,
+      "loss": 0.0833,
+      "step": 2712
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6435000000000005e-05,
+      "loss": 0.3033,
+      "step": 2713
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.643e-05,
+      "loss": 0.0949,
+      "step": 2714
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6425000000000004e-05,
+      "loss": 0.2235,
+      "step": 2715
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.642e-05,
+      "loss": 0.2495,
+      "step": 2716
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.6414999999999995e-05,
+      "loss": 0.116,
+      "step": 2717
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.641e-05,
+      "loss": 0.152,
+      "step": 2718
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.6405e-05,
+      "loss": 0.2992,
+      "step": 2719
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.64e-05,
+      "loss": 0.096,
+      "step": 2720
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6395e-05,
+      "loss": 0.1952,
+      "step": 2721
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.639e-05,
+      "loss": 0.2195,
+      "step": 2722
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6385000000000004e-05,
+      "loss": 0.1062,
+      "step": 2723
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.638e-05,
+      "loss": 0.1925,
+      "step": 2724
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6375e-05,
+      "loss": 0.1008,
+      "step": 2725
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6370000000000005e-05,
+      "loss": 0.1872,
+      "step": 2726
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6365e-05,
+      "loss": 0.2216,
+      "step": 2727
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.6359999999999997e-05,
+      "loss": 0.1815,
+      "step": 2728
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.6355e-05,
+      "loss": 0.1941,
+      "step": 2729
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.635e-05,
+      "loss": 0.2205,
+      "step": 2730
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.6345e-05,
+      "loss": 0.076,
+      "step": 2731
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.634e-05,
+      "loss": 0.0885,
+      "step": 2732
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6335e-05,
+      "loss": 0.1548,
+      "step": 2733
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.633e-05,
+      "loss": 0.2016,
+      "step": 2734
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6325e-05,
+      "loss": 0.1737,
+      "step": 2735
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6320000000000004e-05,
+      "loss": 0.2013,
+      "step": 2736
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6315e-05,
+      "loss": 0.114,
+      "step": 2737
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.631e-05,
+      "loss": 0.1775,
+      "step": 2738
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.6305000000000005e-05,
+      "loss": 0.2192,
+      "step": 2739
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.63e-05,
+      "loss": 0.0723,
+      "step": 2740
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.6294999999999996e-05,
+      "loss": 0.1813,
+      "step": 2741
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.629e-05,
+      "loss": 0.0891,
+      "step": 2742
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6285e-05,
+      "loss": 0.0918,
+      "step": 2743
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.628e-05,
+      "loss": 0.1023,
+      "step": 2744
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6275e-05,
+      "loss": 0.1107,
+      "step": 2745
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.627e-05,
+      "loss": 0.2022,
+      "step": 2746
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6265000000000005e-05,
+      "loss": 0.178,
+      "step": 2747
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.626e-05,
+      "loss": 0.1859,
+      "step": 2748
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6255000000000004e-05,
+      "loss": 0.1,
+      "step": 2749
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 0.2279,
+      "step": 2750
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.6245e-05,
+      "loss": 0.3125,
+      "step": 2751
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6240000000000005e-05,
+      "loss": 0.3593,
+      "step": 2752
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6235e-05,
+      "loss": 0.1084,
+      "step": 2753
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6229999999999996e-05,
+      "loss": 0.2859,
+      "step": 2754
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6225e-05,
+      "loss": 0.1603,
+      "step": 2755
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.622e-05,
+      "loss": 0.1645,
+      "step": 2756
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.6215000000000004e-05,
+      "loss": 0.1824,
+      "step": 2757
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.621e-05,
+      "loss": 0.1666,
+      "step": 2758
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6205e-05,
+      "loss": 0.2177,
+      "step": 2759
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 0.3059,
+      "step": 2760
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6195e-05,
+      "loss": 0.1898,
+      "step": 2761
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.619e-05,
+      "loss": 0.2443,
+      "step": 2762
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.6185000000000006e-05,
+      "loss": 0.1795,
+      "step": 2763
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.618e-05,
+      "loss": 0.363,
+      "step": 2764
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6175e-05,
+      "loss": 0.1302,
+      "step": 2765
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.617e-05,
+      "loss": 0.2153,
+      "step": 2766
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.6165e-05,
+      "loss": 0.1558,
+      "step": 2767
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.616e-05,
+      "loss": 0.1465,
+      "step": 2768
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6155e-05,
+      "loss": 0.1267,
+      "step": 2769
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6150000000000004e-05,
+      "loss": 0.3137,
+      "step": 2770
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6145e-05,
+      "loss": 0.1377,
+      "step": 2771
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.614e-05,
+      "loss": 0.196,
+      "step": 2772
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6135000000000005e-05,
+      "loss": 0.0963,
+      "step": 2773
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.613e-05,
+      "loss": 0.1827,
+      "step": 2774
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.6125e-05,
+      "loss": 0.0962,
+      "step": 2775
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6120000000000006e-05,
+      "loss": 0.0577,
+      "step": 2776
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6115e-05,
+      "loss": 0.2785,
+      "step": 2777
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.611e-05,
+      "loss": 0.1805,
+      "step": 2778
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.6105e-05,
+      "loss": 0.2798,
+      "step": 2779
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.61e-05,
+      "loss": 0.4232,
+      "step": 2780
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.6095e-05,
+      "loss": 0.1846,
+      "step": 2781
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.609e-05,
+      "loss": 0.2116,
+      "step": 2782
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.6085000000000003e-05,
+      "loss": 0.0734,
+      "step": 2783
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.608e-05,
+      "loss": 0.2065,
+      "step": 2784
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.6075e-05,
+      "loss": 0.3313,
+      "step": 2785
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6070000000000004e-05,
+      "loss": 0.1371,
+      "step": 2786
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.606500000000001e-05,
+      "loss": 0.0838,
+      "step": 2787
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.606e-05,
+      "loss": 0.2831,
+      "step": 2788
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.6055e-05,
+      "loss": 0.1974,
+      "step": 2789
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.605e-05,
+      "loss": 0.2935,
+      "step": 2790
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.6045e-05,
+      "loss": 0.0804,
+      "step": 2791
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.604e-05,
+      "loss": 0.0928,
+      "step": 2792
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.6035e-05,
+      "loss": 0.1761,
+      "step": 2793
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.603e-05,
+      "loss": 0.1697,
+      "step": 2794
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.6025e-05,
+      "loss": 0.2192,
+      "step": 2795
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.602e-05,
+      "loss": 0.1406,
+      "step": 2796
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.6015000000000006e-05,
+      "loss": 0.2698,
+      "step": 2797
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.601e-05,
+      "loss": 0.2073,
+      "step": 2798
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.6005000000000004e-05,
+      "loss": 0.1969,
+      "step": 2799
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.0793,
+      "step": 2800
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.5994999999999996e-05,
+      "loss": 0.156,
+      "step": 2801
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.599e-05,
+      "loss": 0.1218,
+      "step": 2802
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.5985e-05,
+      "loss": 0.3897,
+      "step": 2803
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.598e-05,
+      "loss": 0.0797,
+      "step": 2804
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5975e-05,
+      "loss": 0.1339,
+      "step": 2805
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.597e-05,
+      "loss": 0.1969,
+      "step": 2806
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5965000000000005e-05,
+      "loss": 0.0938,
+      "step": 2807
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.596e-05,
+      "loss": 0.2428,
+      "step": 2808
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5955e-05,
+      "loss": 0.2608,
+      "step": 2809
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.5950000000000006e-05,
+      "loss": 0.0845,
+      "step": 2810
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.5945e-05,
+      "loss": 0.2726,
+      "step": 2811
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5940000000000004e-05,
+      "loss": 0.1159,
+      "step": 2812
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5935e-05,
+      "loss": 0.2664,
+      "step": 2813
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5929999999999996e-05,
+      "loss": 0.1129,
+      "step": 2814
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5925e-05,
+      "loss": 0.2055,
+      "step": 2815
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.592e-05,
+      "loss": 0.2817,
+      "step": 2816
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.5915000000000003e-05,
+      "loss": 0.0933,
+      "step": 2817
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.591e-05,
+      "loss": 0.1134,
+      "step": 2818
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5905e-05,
+      "loss": 0.2687,
+      "step": 2819
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 0.1817,
+      "step": 2820
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5895e-05,
+      "loss": 0.1191,
+      "step": 2821
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.589e-05,
+      "loss": 0.1378,
+      "step": 2822
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.5885000000000005e-05,
+      "loss": 0.3201,
+      "step": 2823
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.588e-05,
+      "loss": 0.2421,
+      "step": 2824
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5875e-05,
+      "loss": 0.1288,
+      "step": 2825
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.587e-05,
+      "loss": 0.1057,
+      "step": 2826
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.5865e-05,
+      "loss": 0.175,
+      "step": 2827
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.586e-05,
+      "loss": 0.2066,
+      "step": 2828
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5855e-05,
+      "loss": 0.1005,
+      "step": 2829
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.585e-05,
+      "loss": 0.2149,
+      "step": 2830
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5845e-05,
+      "loss": 0.0841,
+      "step": 2831
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.584e-05,
+      "loss": 0.181,
+      "step": 2832
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5835000000000004e-05,
+      "loss": 0.1375,
+      "step": 2833
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.583e-05,
+      "loss": 0.1773,
+      "step": 2834
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.5825e-05,
+      "loss": 0.1812,
+      "step": 2835
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5820000000000005e-05,
+      "loss": 0.1666,
+      "step": 2836
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5815e-05,
+      "loss": 0.1768,
+      "step": 2837
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.581e-05,
+      "loss": 0.1432,
+      "step": 2838
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5805e-05,
+      "loss": 0.086,
+      "step": 2839
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.58e-05,
+      "loss": 0.0681,
+      "step": 2840
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5795e-05,
+      "loss": 0.1958,
+      "step": 2841
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.579e-05,
+      "loss": 0.1954,
+      "step": 2842
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.5785e-05,
+      "loss": 0.2403,
+      "step": 2843
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.578e-05,
+      "loss": 0.1493,
+      "step": 2844
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.5775e-05,
+      "loss": 0.0966,
+      "step": 2845
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5770000000000004e-05,
+      "loss": 0.1102,
+      "step": 2846
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5765000000000007e-05,
+      "loss": 0.0706,
+      "step": 2847
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.576e-05,
+      "loss": 0.2066,
+      "step": 2848
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.5755e-05,
+      "loss": 0.2203,
+      "step": 2849
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.575e-05,
+      "loss": 0.3305,
+      "step": 2850
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5744999999999997e-05,
+      "loss": 0.0896,
+      "step": 2851
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.574e-05,
+      "loss": 0.4206,
+      "step": 2852
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.5735e-05,
+      "loss": 0.3631,
+      "step": 2853
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.573e-05,
+      "loss": 0.0949,
+      "step": 2854
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5725e-05,
+      "loss": 0.0946,
+      "step": 2855
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.572e-05,
+      "loss": 0.1439,
+      "step": 2856
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5715000000000005e-05,
+      "loss": 0.2409,
+      "step": 2857
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.571e-05,
+      "loss": 0.0851,
+      "step": 2858
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5705000000000004e-05,
+      "loss": 0.0897,
+      "step": 2859
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 0.0819,
+      "step": 2860
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5694999999999995e-05,
+      "loss": 0.1608,
+      "step": 2861
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.569e-05,
+      "loss": 0.0864,
+      "step": 2862
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5685e-05,
+      "loss": 0.0751,
+      "step": 2863
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5679999999999996e-05,
+      "loss": 0.1705,
+      "step": 2864
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.5675e-05,
+      "loss": 0.2406,
+      "step": 2865
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.567e-05,
+      "loss": 0.1749,
+      "step": 2866
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5665000000000004e-05,
+      "loss": 0.2963,
+      "step": 2867
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.566e-05,
+      "loss": 0.1356,
+      "step": 2868
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5655e-05,
+      "loss": 0.0942,
+      "step": 2869
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5650000000000005e-05,
+      "loss": 0.1293,
+      "step": 2870
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5645e-05,
+      "loss": 0.2732,
+      "step": 2871
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5640000000000003e-05,
+      "loss": 0.3476,
+      "step": 2872
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5635e-05,
+      "loss": 0.1823,
+      "step": 2873
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5629999999999995e-05,
+      "loss": 0.0698,
+      "step": 2874
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.5625e-05,
+      "loss": 0.2334,
+      "step": 2875
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.562e-05,
+      "loss": 0.2575,
+      "step": 2876
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.5615e-05,
+      "loss": 0.1451,
+      "step": 2877
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.561e-05,
+      "loss": 0.115,
+      "step": 2878
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5605e-05,
+      "loss": 0.1898,
+      "step": 2879
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 0.267,
+      "step": 2880
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5595e-05,
+      "loss": 0.1688,
+      "step": 2881
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.559e-05,
+      "loss": 0.3981,
+      "step": 2882
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.5585000000000005e-05,
+      "loss": 0.2752,
+      "step": 2883
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.558e-05,
+      "loss": 0.1066,
+      "step": 2884
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.5575e-05,
+      "loss": 0.101,
+      "step": 2885
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.557e-05,
+      "loss": 0.0688,
+      "step": 2886
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.5565e-05,
+      "loss": 0.228,
+      "step": 2887
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.556e-05,
+      "loss": 0.1235,
+      "step": 2888
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.5555e-05,
+      "loss": 0.0692,
+      "step": 2889
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.555e-05,
+      "loss": 0.2519,
+      "step": 2890
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.5545e-05,
+      "loss": 0.3805,
+      "step": 2891
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.554e-05,
+      "loss": 0.2521,
+      "step": 2892
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.5535000000000004e-05,
+      "loss": 0.0923,
+      "step": 2893
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.553e-05,
+      "loss": 0.197,
+      "step": 2894
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.5525e-05,
+      "loss": 0.1837,
+      "step": 2895
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.5520000000000005e-05,
+      "loss": 0.1763,
+      "step": 2896
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.551500000000001e-05,
+      "loss": 0.2493,
+      "step": 2897
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.5509999999999996e-05,
+      "loss": 0.0756,
+      "step": 2898
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.5505e-05,
+      "loss": 0.224,
+      "step": 2899
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.55e-05,
+      "loss": 0.091,
+      "step": 2900
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.5495e-05,
+      "loss": 0.261,
+      "step": 2901
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.549e-05,
+      "loss": 0.1572,
+      "step": 2902
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.5485e-05,
+      "loss": 0.1513,
+      "step": 2903
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.548e-05,
+      "loss": 0.0754,
+      "step": 2904
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.5475e-05,
+      "loss": 0.2299,
+      "step": 2905
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.5470000000000003e-05,
+      "loss": 0.197,
+      "step": 2906
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.5465000000000006e-05,
+      "loss": 0.0982,
+      "step": 2907
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.546e-05,
+      "loss": 0.2825,
+      "step": 2908
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.5455000000000004e-05,
+      "loss": 0.3268,
+      "step": 2909
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.545e-05,
+      "loss": 0.246,
+      "step": 2910
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.5444999999999996e-05,
+      "loss": 0.0994,
+      "step": 2911
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.544e-05,
+      "loss": 0.1143,
+      "step": 2912
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.5435e-05,
+      "loss": 0.153,
+      "step": 2913
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.5430000000000004e-05,
+      "loss": 0.084,
+      "step": 2914
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.5425e-05,
+      "loss": 0.1665,
+      "step": 2915
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.542e-05,
+      "loss": 0.0606,
+      "step": 2916
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.5415000000000005e-05,
+      "loss": 0.1444,
+      "step": 2917
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.541e-05,
+      "loss": 0.0821,
+      "step": 2918
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.5405e-05,
+      "loss": 0.0831,
+      "step": 2919
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 0.2631,
+      "step": 2920
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.5395e-05,
+      "loss": 0.1749,
+      "step": 2921
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.539e-05,
+      "loss": 0.2426,
+      "step": 2922
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.5385e-05,
+      "loss": 0.1909,
+      "step": 2923
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.538e-05,
+      "loss": 0.1392,
+      "step": 2924
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.5375e-05,
+      "loss": 0.0852,
+      "step": 2925
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.537e-05,
+      "loss": 0.2156,
+      "step": 2926
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.5365000000000004e-05,
+      "loss": 0.2022,
+      "step": 2927
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.536e-05,
+      "loss": 0.1917,
+      "step": 2928
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.5355e-05,
+      "loss": 0.238,
+      "step": 2929
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.5350000000000005e-05,
+      "loss": 0.2139,
+      "step": 2930
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.5345e-05,
+      "loss": 0.1378,
+      "step": 2931
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.534e-05,
+      "loss": 0.095,
+      "step": 2932
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.5335000000000006e-05,
+      "loss": 0.1305,
+      "step": 2933
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.533e-05,
+      "loss": 0.266,
+      "step": 2934
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.5325e-05,
+      "loss": 0.0963,
+      "step": 2935
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.532e-05,
+      "loss": 0.0891,
+      "step": 2936
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.5315e-05,
+      "loss": 0.1607,
+      "step": 2937
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.531e-05,
+      "loss": 0.1627,
+      "step": 2938
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.5305e-05,
+      "loss": 0.2211,
+      "step": 2939
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.53e-05,
+      "loss": 0.0908,
+      "step": 2940
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.5295e-05,
+      "loss": 0.206,
+      "step": 2941
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.529e-05,
+      "loss": 0.1441,
+      "step": 2942
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.5285000000000004e-05,
+      "loss": 0.2317,
+      "step": 2943
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.528000000000001e-05,
+      "loss": 0.0937,
+      "step": 2944
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.5275e-05,
+      "loss": 0.1889,
+      "step": 2945
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.527e-05,
+      "loss": 0.084,
+      "step": 2946
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.5265e-05,
+      "loss": 0.0873,
+      "step": 2947
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.526e-05,
+      "loss": 0.1823,
+      "step": 2948
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.5255e-05,
+      "loss": 0.232,
+      "step": 2949
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.525e-05,
+      "loss": 0.1669,
+      "step": 2950
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.5245e-05,
+      "loss": 0.1809,
+      "step": 2951
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.524e-05,
+      "loss": 0.1973,
+      "step": 2952
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.5235e-05,
+      "loss": 0.0767,
+      "step": 2953
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.5230000000000006e-05,
+      "loss": 0.1136,
+      "step": 2954
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.5225e-05,
+      "loss": 0.0712,
+      "step": 2955
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.5220000000000004e-05,
+      "loss": 0.1868,
+      "step": 2956
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.521500000000001e-05,
+      "loss": 0.1708,
+      "step": 2957
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.5209999999999996e-05,
+      "loss": 0.2191,
+      "step": 2958
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.5205e-05,
+      "loss": 0.2576,
+      "step": 2959
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.52e-05,
+      "loss": 0.0972,
+      "step": 2960
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.5195e-05,
+      "loss": 0.2344,
+      "step": 2961
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.519e-05,
+      "loss": 0.257,
+      "step": 2962
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.5185e-05,
+      "loss": 0.1058,
+      "step": 2963
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.5180000000000004e-05,
+      "loss": 0.1723,
+      "step": 2964
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.5175e-05,
+      "loss": 0.1061,
+      "step": 2965
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.517e-05,
+      "loss": 0.1297,
+      "step": 2966
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.5165000000000005e-05,
+      "loss": 0.2161,
+      "step": 2967
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.516e-05,
+      "loss": 0.0995,
+      "step": 2968
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.5155000000000004e-05,
+      "loss": 0.1721,
+      "step": 2969
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.515e-05,
+      "loss": 0.0912,
+      "step": 2970
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.5144999999999996e-05,
+      "loss": 0.2089,
+      "step": 2971
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.514e-05,
+      "loss": 0.1404,
+      "step": 2972
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.5135e-05,
+      "loss": 0.1598,
+      "step": 2973
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.513e-05,
+      "loss": 0.0959,
+      "step": 2974
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.5125e-05,
+      "loss": 0.134,
+      "step": 2975
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.512e-05,
+      "loss": 0.3873,
+      "step": 2976
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.5115000000000004e-05,
+      "loss": 0.3487,
+      "step": 2977
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.511e-05,
+      "loss": 0.0645,
+      "step": 2978
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.5105e-05,
+      "loss": 0.0637,
+      "step": 2979
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 0.1012,
+      "step": 2980
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.5095e-05,
+      "loss": 0.0792,
+      "step": 2981
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.509e-05,
+      "loss": 0.1717,
+      "step": 2982
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.5085e-05,
+      "loss": 0.096,
+      "step": 2983
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.508e-05,
+      "loss": 0.0748,
+      "step": 2984
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.5075e-05,
+      "loss": 0.0895,
+      "step": 2985
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.507e-05,
+      "loss": 0.1898,
+      "step": 2986
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.5065e-05,
+      "loss": 0.1049,
+      "step": 2987
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.506e-05,
+      "loss": 0.1927,
+      "step": 2988
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.5055e-05,
+      "loss": 0.2673,
+      "step": 2989
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.5050000000000004e-05,
+      "loss": 0.2182,
+      "step": 2990
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.5045e-05,
+      "loss": 0.1858,
+      "step": 2991
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.504e-05,
+      "loss": 0.2325,
+      "step": 2992
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.5035000000000005e-05,
+      "loss": 0.2154,
+      "step": 2993
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.503e-05,
+      "loss": 0.2012,
+      "step": 2994
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.5025e-05,
+      "loss": 0.3672,
+      "step": 2995
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.502e-05,
+      "loss": 0.2308,
+      "step": 2996
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.5015e-05,
+      "loss": 0.2014,
+      "step": 2997
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.501e-05,
+      "loss": 0.2456,
+      "step": 2998
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.5005e-05,
+      "loss": 0.1607,
+      "step": 2999
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.5e-05,
+      "loss": 0.1911,
+      "step": 3000
+    },
+    {
+      "epoch": 12.5,
+      "eval_loss": 0.3716381788253784,
+      "eval_mean_accuracy": 0.7670995222097438,
+      "eval_mean_iou": 0.6657154696874261,
+      "eval_overall_accuracy": 0.8690043131510417,
+      "eval_per_category_accuracy": [
+        0.8967080262800052,
+        0.914781060700395,
+        0.4898094796488311
+      ],
+      "eval_per_category_iou": [
+        0.8087784253965665,
+        0.8084543420797378,
+        0.37991364158597407
+      ],
+      "eval_runtime": 167.2921,
+      "eval_samples_per_second": 0.717,
+      "eval_steps_per_second": 0.359,
+      "step": 3000
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.4995e-05,
+      "loss": 0.1322,
+      "step": 3001
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.499e-05,
+      "loss": 0.1534,
+      "step": 3002
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.4985000000000004e-05,
+      "loss": 0.0974,
+      "step": 3003
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.4980000000000006e-05,
+      "loss": 0.2808,
+      "step": 3004
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.4975e-05,
+      "loss": 0.1047,
+      "step": 3005
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.4970000000000005e-05,
+      "loss": 0.0859,
+      "step": 3006
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.4965e-05,
+      "loss": 0.3088,
+      "step": 3007
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.4959999999999996e-05,
+      "loss": 0.2116,
+      "step": 3008
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.4955e-05,
+      "loss": 0.2552,
+      "step": 3009
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.495e-05,
+      "loss": 0.174,
+      "step": 3010
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.4945e-05,
+      "loss": 0.1811,
+      "step": 3011
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.494e-05,
+      "loss": 0.3378,
+      "step": 3012
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.4935e-05,
+      "loss": 0.1028,
+      "step": 3013
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.4930000000000005e-05,
+      "loss": 0.2252,
+      "step": 3014
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.4925e-05,
+      "loss": 0.2392,
+      "step": 3015
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.4920000000000004e-05,
+      "loss": 0.1004,
+      "step": 3016
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.4915000000000006e-05,
+      "loss": 0.2174,
+      "step": 3017
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.491e-05,
+      "loss": 0.183,
+      "step": 3018
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.4905e-05,
+      "loss": 0.1694,
+      "step": 3019
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.49e-05,
+      "loss": 0.1204,
+      "step": 3020
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.4894999999999996e-05,
+      "loss": 0.2309,
+      "step": 3021
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.489e-05,
+      "loss": 0.2364,
+      "step": 3022
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.4885e-05,
+      "loss": 0.0656,
+      "step": 3023
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.4880000000000004e-05,
+      "loss": 0.1873,
+      "step": 3024
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.4875e-05,
+      "loss": 0.3083,
+      "step": 3025
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.487e-05,
+      "loss": 0.1656,
+      "step": 3026
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.4865000000000005e-05,
+      "loss": 0.2381,
+      "step": 3027
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.486e-05,
+      "loss": 0.1262,
+      "step": 3028
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.4855e-05,
+      "loss": 0.0881,
+      "step": 3029
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.4850000000000006e-05,
+      "loss": 0.2441,
+      "step": 3030
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.4844999999999995e-05,
+      "loss": 0.0798,
+      "step": 3031
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.484e-05,
+      "loss": 0.065,
+      "step": 3032
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.4835e-05,
+      "loss": 0.0818,
+      "step": 3033
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.483e-05,
+      "loss": 0.1155,
+      "step": 3034
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.4825e-05,
+      "loss": 0.0591,
+      "step": 3035
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.482e-05,
+      "loss": 0.1072,
+      "step": 3036
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.4815000000000004e-05,
+      "loss": 0.2545,
+      "step": 3037
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.481e-05,
+      "loss": 0.3032,
+      "step": 3038
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.4805e-05,
+      "loss": 0.2487,
+      "step": 3039
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 0.2031,
+      "step": 3040
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.4795e-05,
+      "loss": 0.1721,
+      "step": 3041
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.479e-05,
+      "loss": 0.3514,
+      "step": 3042
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.4785e-05,
+      "loss": 0.0703,
+      "step": 3043
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.478e-05,
+      "loss": 0.1131,
+      "step": 3044
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.4775e-05,
+      "loss": 0.0808,
+      "step": 3045
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.477e-05,
+      "loss": 0.3227,
+      "step": 3046
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.4765e-05,
+      "loss": 0.0955,
+      "step": 3047
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.476e-05,
+      "loss": 0.2011,
+      "step": 3048
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.4755e-05,
+      "loss": 0.0852,
+      "step": 3049
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.4750000000000004e-05,
+      "loss": 0.2308,
+      "step": 3050
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.4745e-05,
+      "loss": 0.2235,
+      "step": 3051
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.474e-05,
+      "loss": 0.098,
+      "step": 3052
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.4735000000000005e-05,
+      "loss": 0.2721,
+      "step": 3053
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.473000000000001e-05,
+      "loss": 0.1709,
+      "step": 3054
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.4724999999999996e-05,
+      "loss": 0.564,
+      "step": 3055
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.472e-05,
+      "loss": 0.1652,
+      "step": 3056
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.4715e-05,
+      "loss": 0.1061,
+      "step": 3057
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.471e-05,
+      "loss": 0.1682,
+      "step": 3058
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.4705e-05,
+      "loss": 0.1437,
+      "step": 3059
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.47e-05,
+      "loss": 0.1938,
+      "step": 3060
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.4695e-05,
+      "loss": 0.2014,
+      "step": 3061
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.469e-05,
+      "loss": 0.2996,
+      "step": 3062
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.4685e-05,
+      "loss": 0.0692,
+      "step": 3063
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.4680000000000006e-05,
+      "loss": 0.1392,
+      "step": 3064
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.4675e-05,
+      "loss": 0.1178,
+      "step": 3065
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.4670000000000004e-05,
+      "loss": 0.2545,
+      "step": 3066
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.4665e-05,
+      "loss": 0.1103,
+      "step": 3067
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.4659999999999996e-05,
+      "loss": 0.1119,
+      "step": 3068
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.4655e-05,
+      "loss": 0.0779,
+      "step": 3069
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.465e-05,
+      "loss": 0.084,
+      "step": 3070
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.4645e-05,
+      "loss": 0.6345,
+      "step": 3071
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.464e-05,
+      "loss": 0.1367,
+      "step": 3072
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.4635e-05,
+      "loss": 0.2088,
+      "step": 3073
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.4630000000000005e-05,
+      "loss": 0.2635,
+      "step": 3074
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.4625e-05,
+      "loss": 0.2077,
+      "step": 3075
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.462e-05,
+      "loss": 0.1934,
+      "step": 3076
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.4615000000000006e-05,
+      "loss": 0.0654,
+      "step": 3077
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.461e-05,
+      "loss": 0.0952,
+      "step": 3078
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.4605e-05,
+      "loss": 0.0833,
+      "step": 3079
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.46e-05,
+      "loss": 0.2317,
+      "step": 3080
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.4595e-05,
+      "loss": 0.1562,
+      "step": 3081
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.459e-05,
+      "loss": 0.1808,
+      "step": 3082
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.4585e-05,
+      "loss": 0.2313,
+      "step": 3083
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.4580000000000003e-05,
+      "loss": 0.2159,
+      "step": 3084
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.4575e-05,
+      "loss": 0.0844,
+      "step": 3085
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.457e-05,
+      "loss": 0.0731,
+      "step": 3086
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.4565000000000004e-05,
+      "loss": 0.2196,
+      "step": 3087
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.456e-05,
+      "loss": 0.1818,
+      "step": 3088
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.4555e-05,
+      "loss": 0.2084,
+      "step": 3089
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.4550000000000005e-05,
+      "loss": 0.3904,
+      "step": 3090
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.4545e-05,
+      "loss": 0.2533,
+      "step": 3091
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.454e-05,
+      "loss": 0.1768,
+      "step": 3092
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.4535e-05,
+      "loss": 0.2705,
+      "step": 3093
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.453e-05,
+      "loss": 0.2167,
+      "step": 3094
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.4525e-05,
+      "loss": 0.3055,
+      "step": 3095
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.452e-05,
+      "loss": 0.1923,
+      "step": 3096
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.4515e-05,
+      "loss": 0.1884,
+      "step": 3097
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.451e-05,
+      "loss": 0.1259,
+      "step": 3098
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.4505e-05,
+      "loss": 0.0789,
+      "step": 3099
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 0.2793,
+      "step": 3100
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.449500000000001e-05,
+      "loss": 0.3088,
+      "step": 3101
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.449e-05,
+      "loss": 0.2187,
+      "step": 3102
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.4485e-05,
+      "loss": 0.0945,
+      "step": 3103
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.448e-05,
+      "loss": 0.0864,
+      "step": 3104
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.4475e-05,
+      "loss": 0.1056,
+      "step": 3105
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.447e-05,
+      "loss": 0.1473,
+      "step": 3106
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.4465e-05,
+      "loss": 0.0715,
+      "step": 3107
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.446e-05,
+      "loss": 0.1865,
+      "step": 3108
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.4455e-05,
+      "loss": 0.3115,
+      "step": 3109
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.445e-05,
+      "loss": 0.1677,
+      "step": 3110
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.4445000000000006e-05,
+      "loss": 0.0944,
+      "step": 3111
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.444e-05,
+      "loss": 0.293,
+      "step": 3112
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.4435000000000004e-05,
+      "loss": 0.2591,
+      "step": 3113
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.4430000000000007e-05,
+      "loss": 0.1659,
+      "step": 3114
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.4424999999999996e-05,
+      "loss": 0.0671,
+      "step": 3115
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.442e-05,
+      "loss": 0.1784,
+      "step": 3116
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.4415e-05,
+      "loss": 0.3057,
+      "step": 3117
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.441e-05,
+      "loss": 0.2303,
+      "step": 3118
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.4405e-05,
+      "loss": 0.1654,
+      "step": 3119
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.44e-05,
+      "loss": 0.1872,
+      "step": 3120
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.4395000000000004e-05,
+      "loss": 0.2167,
+      "step": 3121
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.439e-05,
+      "loss": 0.2371,
+      "step": 3122
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.4385e-05,
+      "loss": 0.2213,
+      "step": 3123
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.4380000000000005e-05,
+      "loss": 0.1748,
+      "step": 3124
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.4375e-05,
+      "loss": 0.167,
+      "step": 3125
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.4370000000000004e-05,
+      "loss": 0.0778,
+      "step": 3126
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.4365000000000006e-05,
+      "loss": 0.2804,
+      "step": 3127
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.4359999999999995e-05,
+      "loss": 0.1589,
+      "step": 3128
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.4355e-05,
+      "loss": 0.112,
+      "step": 3129
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.435e-05,
+      "loss": 0.0792,
+      "step": 3130
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.4345e-05,
+      "loss": 0.3276,
+      "step": 3131
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.434e-05,
+      "loss": 0.0839,
+      "step": 3132
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.4335e-05,
+      "loss": 0.2772,
+      "step": 3133
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.4330000000000004e-05,
+      "loss": 0.0772,
+      "step": 3134
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.4325e-05,
+      "loss": 0.2014,
+      "step": 3135
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.432e-05,
+      "loss": 0.2082,
+      "step": 3136
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.4315000000000005e-05,
+      "loss": 0.0814,
+      "step": 3137
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.431e-05,
+      "loss": 0.1114,
+      "step": 3138
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.4305000000000004e-05,
+      "loss": 0.2046,
+      "step": 3139
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.43e-05,
+      "loss": 0.1974,
+      "step": 3140
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.4295e-05,
+      "loss": 0.1798,
+      "step": 3141
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.429e-05,
+      "loss": 0.3526,
+      "step": 3142
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.4285e-05,
+      "loss": 0.1753,
+      "step": 3143
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.428e-05,
+      "loss": 0.0976,
+      "step": 3144
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.4275e-05,
+      "loss": 0.1063,
+      "step": 3145
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.427e-05,
+      "loss": 0.2765,
+      "step": 3146
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.4265000000000004e-05,
+      "loss": 0.2365,
+      "step": 3147
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.426e-05,
+      "loss": 0.1656,
+      "step": 3148
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.4255e-05,
+      "loss": 0.2086,
+      "step": 3149
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.4250000000000005e-05,
+      "loss": 0.1881,
+      "step": 3150
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.424500000000001e-05,
+      "loss": 0.2091,
+      "step": 3151
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.4239999999999997e-05,
+      "loss": 0.0857,
+      "step": 3152
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.4235e-05,
+      "loss": 0.1238,
+      "step": 3153
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.423e-05,
+      "loss": 0.0883,
+      "step": 3154
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.4225e-05,
+      "loss": 0.1382,
+      "step": 3155
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.422e-05,
+      "loss": 0.248,
+      "step": 3156
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.4215e-05,
+      "loss": 0.258,
+      "step": 3157
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.421e-05,
+      "loss": 0.1478,
+      "step": 3158
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.4205e-05,
+      "loss": 0.0589,
+      "step": 3159
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 0.1108,
+      "step": 3160
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.4195000000000006e-05,
+      "loss": 0.1127,
+      "step": 3161
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.419e-05,
+      "loss": 0.1644,
+      "step": 3162
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.4185000000000005e-05,
+      "loss": 0.0885,
+      "step": 3163
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.418e-05,
+      "loss": 0.238,
+      "step": 3164
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.4174999999999996e-05,
+      "loss": 0.1434,
+      "step": 3165
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.417e-05,
+      "loss": 0.0705,
+      "step": 3166
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.4165e-05,
+      "loss": 0.281,
+      "step": 3167
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.416e-05,
+      "loss": 0.0891,
+      "step": 3168
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.4155e-05,
+      "loss": 0.0677,
+      "step": 3169
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.415e-05,
+      "loss": 0.2833,
+      "step": 3170
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.4145000000000005e-05,
+      "loss": 0.2338,
+      "step": 3171
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.414e-05,
+      "loss": 0.275,
+      "step": 3172
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.4135000000000003e-05,
+      "loss": 0.096,
+      "step": 3173
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.4130000000000006e-05,
+      "loss": 0.0737,
+      "step": 3174
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.4125e-05,
+      "loss": 0.2004,
+      "step": 3175
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.412e-05,
+      "loss": 0.1508,
+      "step": 3176
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.4115e-05,
+      "loss": 0.321,
+      "step": 3177
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.4109999999999996e-05,
+      "loss": 0.2691,
+      "step": 3178
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.4105e-05,
+      "loss": 0.1245,
+      "step": 3179
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.41e-05,
+      "loss": 0.1162,
+      "step": 3180
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.4095000000000004e-05,
+      "loss": 0.1609,
+      "step": 3181
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.409e-05,
+      "loss": 0.1319,
+      "step": 3182
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.4085e-05,
+      "loss": 0.1013,
+      "step": 3183
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.4080000000000005e-05,
+      "loss": 0.112,
+      "step": 3184
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.4075e-05,
+      "loss": 0.282,
+      "step": 3185
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.407e-05,
+      "loss": 0.0901,
+      "step": 3186
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.4065000000000006e-05,
+      "loss": 0.0931,
+      "step": 3187
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.4059999999999995e-05,
+      "loss": 0.2288,
+      "step": 3188
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.4055e-05,
+      "loss": 0.0791,
+      "step": 3189
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.405e-05,
+      "loss": 0.1431,
+      "step": 3190
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.4045e-05,
+      "loss": 0.1052,
+      "step": 3191
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.404e-05,
+      "loss": 0.0707,
+      "step": 3192
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.4035e-05,
+      "loss": 0.0884,
+      "step": 3193
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.4030000000000004e-05,
+      "loss": 0.1768,
+      "step": 3194
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.4025e-05,
+      "loss": 0.0394,
+      "step": 3195
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.402e-05,
+      "loss": 0.5304,
+      "step": 3196
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.4015000000000005e-05,
+      "loss": 0.2021,
+      "step": 3197
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.401e-05,
+      "loss": 0.3312,
+      "step": 3198
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.4005e-05,
+      "loss": 0.1705,
+      "step": 3199
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.4e-05,
+      "loss": 0.3063,
+      "step": 3200
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.3995e-05,
+      "loss": 0.0988,
+      "step": 3201
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.399e-05,
+      "loss": 0.2462,
+      "step": 3202
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.3985e-05,
+      "loss": 0.2035,
+      "step": 3203
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.398e-05,
+      "loss": 0.1364,
+      "step": 3204
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.3975e-05,
+      "loss": 0.236,
+      "step": 3205
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.397e-05,
+      "loss": 0.1913,
+      "step": 3206
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.3965000000000003e-05,
+      "loss": 0.1182,
+      "step": 3207
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.396e-05,
+      "loss": 0.1941,
+      "step": 3208
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.3955e-05,
+      "loss": 0.2635,
+      "step": 3209
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.3950000000000004e-05,
+      "loss": 0.1945,
+      "step": 3210
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.394500000000001e-05,
+      "loss": 0.2089,
+      "step": 3211
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.3939999999999996e-05,
+      "loss": 0.1328,
+      "step": 3212
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.3935e-05,
+      "loss": 0.0898,
+      "step": 3213
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.393e-05,
+      "loss": 0.2803,
+      "step": 3214
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.3925e-05,
+      "loss": 0.2521,
+      "step": 3215
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.392e-05,
+      "loss": 0.2478,
+      "step": 3216
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.3915e-05,
+      "loss": 0.2391,
+      "step": 3217
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.391e-05,
+      "loss": 0.1436,
+      "step": 3218
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.3905e-05,
+      "loss": 0.1545,
+      "step": 3219
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.39e-05,
+      "loss": 0.1111,
+      "step": 3220
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.3895000000000006e-05,
+      "loss": 0.2761,
+      "step": 3221
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.389e-05,
+      "loss": 0.2448,
+      "step": 3222
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.3885000000000004e-05,
+      "loss": 0.2779,
+      "step": 3223
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.388e-05,
+      "loss": 0.0991,
+      "step": 3224
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.3874999999999996e-05,
+      "loss": 0.1703,
+      "step": 3225
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.387e-05,
+      "loss": 0.1789,
+      "step": 3226
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.3865e-05,
+      "loss": 0.175,
+      "step": 3227
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.386e-05,
+      "loss": 0.2114,
+      "step": 3228
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.3855e-05,
+      "loss": 0.2076,
+      "step": 3229
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.385e-05,
+      "loss": 0.0859,
+      "step": 3230
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.3845000000000005e-05,
+      "loss": 0.1743,
+      "step": 3231
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.384e-05,
+      "loss": 0.3211,
+      "step": 3232
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.3835e-05,
+      "loss": 0.0784,
+      "step": 3233
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.3830000000000006e-05,
+      "loss": 0.1881,
+      "step": 3234
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.3825e-05,
+      "loss": 0.1042,
+      "step": 3235
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.382e-05,
+      "loss": 0.0958,
+      "step": 3236
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.3815e-05,
+      "loss": 0.1657,
+      "step": 3237
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.3809999999999996e-05,
+      "loss": 0.0673,
+      "step": 3238
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.3805e-05,
+      "loss": 0.1706,
+      "step": 3239
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.38e-05,
+      "loss": 0.0688,
+      "step": 3240
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.3795e-05,
+      "loss": 0.094,
+      "step": 3241
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.379e-05,
+      "loss": 0.1706,
+      "step": 3242
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.3785e-05,
+      "loss": 0.0703,
+      "step": 3243
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.3780000000000004e-05,
+      "loss": 0.1462,
+      "step": 3244
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.3775e-05,
+      "loss": 0.0962,
+      "step": 3245
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.377e-05,
+      "loss": 0.2117,
+      "step": 3246
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.3765000000000005e-05,
+      "loss": 0.1668,
+      "step": 3247
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.376e-05,
+      "loss": 0.0976,
+      "step": 3248
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.3755e-05,
+      "loss": 0.3436,
+      "step": 3249
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.375e-05,
+      "loss": 0.0563,
+      "step": 3250
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.3745e-05,
+      "loss": 0.1335,
+      "step": 3251
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.374e-05,
+      "loss": 0.0965,
+      "step": 3252
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.3735e-05,
+      "loss": 0.1446,
+      "step": 3253
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.373e-05,
+      "loss": 0.1083,
+      "step": 3254
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.3725e-05,
+      "loss": 0.103,
+      "step": 3255
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.372e-05,
+      "loss": 0.2521,
+      "step": 3256
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.3715000000000004e-05,
+      "loss": 0.0741,
+      "step": 3257
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.371e-05,
+      "loss": 0.0948,
+      "step": 3258
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.3705e-05,
+      "loss": 0.062,
+      "step": 3259
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.3700000000000005e-05,
+      "loss": 0.0924,
+      "step": 3260
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.3695e-05,
+      "loss": 0.247,
+      "step": 3261
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.369e-05,
+      "loss": 0.391,
+      "step": 3262
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.3685e-05,
+      "loss": 0.3951,
+      "step": 3263
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.368e-05,
+      "loss": 0.1748,
+      "step": 3264
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.3675e-05,
+      "loss": 0.2068,
+      "step": 3265
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.367e-05,
+      "loss": 0.056,
+      "step": 3266
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.3665e-05,
+      "loss": 0.2996,
+      "step": 3267
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.3660000000000005e-05,
+      "loss": 0.1509,
+      "step": 3268
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.3655e-05,
+      "loss": 0.0879,
+      "step": 3269
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.3650000000000004e-05,
+      "loss": 0.0569,
+      "step": 3270
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.3645000000000006e-05,
+      "loss": 0.2172,
+      "step": 3271
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.364e-05,
+      "loss": 0.3694,
+      "step": 3272
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.3635e-05,
+      "loss": 0.1774,
+      "step": 3273
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.363e-05,
+      "loss": 0.0921,
+      "step": 3274
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.3624999999999997e-05,
+      "loss": 0.0629,
+      "step": 3275
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.362e-05,
+      "loss": 0.0891,
+      "step": 3276
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.3615e-05,
+      "loss": 0.3066,
+      "step": 3277
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.3610000000000004e-05,
+      "loss": 0.0809,
+      "step": 3278
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.3605e-05,
+      "loss": 0.0692,
+      "step": 3279
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.36e-05,
+      "loss": 0.0753,
+      "step": 3280
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.3595000000000005e-05,
+      "loss": 0.236,
+      "step": 3281
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.359e-05,
+      "loss": 0.2101,
+      "step": 3282
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.3585000000000004e-05,
+      "loss": 0.0683,
+      "step": 3283
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.3580000000000006e-05,
+      "loss": 0.203,
+      "step": 3284
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.3574999999999995e-05,
+      "loss": 0.2585,
+      "step": 3285
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.357e-05,
+      "loss": 0.0622,
+      "step": 3286
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.3565e-05,
+      "loss": 0.304,
+      "step": 3287
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.356e-05,
+      "loss": 0.13,
+      "step": 3288
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.3555e-05,
+      "loss": 0.1699,
+      "step": 3289
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.355e-05,
+      "loss": 0.1568,
+      "step": 3290
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.3545000000000004e-05,
+      "loss": 0.104,
+      "step": 3291
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.354e-05,
+      "loss": 0.0786,
+      "step": 3292
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.3535e-05,
+      "loss": 0.0939,
+      "step": 3293
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.3530000000000005e-05,
+      "loss": 0.3466,
+      "step": 3294
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.3525e-05,
+      "loss": 0.4253,
+      "step": 3295
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.3520000000000003e-05,
+      "loss": 0.0758,
+      "step": 3296
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.3515e-05,
+      "loss": 0.2228,
+      "step": 3297
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.351e-05,
+      "loss": 0.0764,
+      "step": 3298
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.3505e-05,
+      "loss": 0.2133,
+      "step": 3299
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.35e-05,
+      "loss": 0.2197,
+      "step": 3300
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.3495e-05,
+      "loss": 0.1787,
+      "step": 3301
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.349e-05,
+      "loss": 0.1133,
+      "step": 3302
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.3485e-05,
+      "loss": 0.2968,
+      "step": 3303
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.3480000000000004e-05,
+      "loss": 0.1016,
+      "step": 3304
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.3475e-05,
+      "loss": 0.1655,
+      "step": 3305
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.347e-05,
+      "loss": 0.1939,
+      "step": 3306
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.3465000000000005e-05,
+      "loss": 0.2213,
+      "step": 3307
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.346000000000001e-05,
+      "loss": 0.1734,
+      "step": 3308
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.3454999999999996e-05,
+      "loss": 0.1695,
+      "step": 3309
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.345e-05,
+      "loss": 0.3398,
+      "step": 3310
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.3445e-05,
+      "loss": 0.0798,
+      "step": 3311
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.344e-05,
+      "loss": 0.195,
+      "step": 3312
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.3435e-05,
+      "loss": 0.2884,
+      "step": 3313
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.343e-05,
+      "loss": 0.0663,
+      "step": 3314
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.3425e-05,
+      "loss": 0.0674,
+      "step": 3315
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.342e-05,
+      "loss": 0.1144,
+      "step": 3316
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.3415000000000004e-05,
+      "loss": 0.1759,
+      "step": 3317
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.3410000000000006e-05,
+      "loss": 0.1596,
+      "step": 3318
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.3405e-05,
+      "loss": 0.2828,
+      "step": 3319
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 0.1225,
+      "step": 3320
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.3395e-05,
+      "loss": 0.1203,
+      "step": 3321
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.3389999999999996e-05,
+      "loss": 0.2733,
+      "step": 3322
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.3385e-05,
+      "loss": 0.2028,
+      "step": 3323
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.338e-05,
+      "loss": 0.1009,
+      "step": 3324
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.3375e-05,
+      "loss": 0.1879,
+      "step": 3325
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.337e-05,
+      "loss": 0.2061,
+      "step": 3326
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.3365e-05,
+      "loss": 0.0781,
+      "step": 3327
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.3360000000000005e-05,
+      "loss": 0.0974,
+      "step": 3328
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.3355e-05,
+      "loss": 0.1087,
+      "step": 3329
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.335e-05,
+      "loss": 0.137,
+      "step": 3330
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.3345000000000006e-05,
+      "loss": 0.1376,
+      "step": 3331
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.334e-05,
+      "loss": 0.3171,
+      "step": 3332
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.3335e-05,
+      "loss": 0.1425,
+      "step": 3333
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.333e-05,
+      "loss": 0.0989,
+      "step": 3334
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.3324999999999996e-05,
+      "loss": 0.3193,
+      "step": 3335
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.332e-05,
+      "loss": 0.2404,
+      "step": 3336
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.3315e-05,
+      "loss": 0.0744,
+      "step": 3337
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.3310000000000004e-05,
+      "loss": 0.0731,
+      "step": 3338
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.3305e-05,
+      "loss": 0.1732,
+      "step": 3339
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.33e-05,
+      "loss": 0.2364,
+      "step": 3340
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.3295000000000005e-05,
+      "loss": 0.1789,
+      "step": 3341
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.329e-05,
+      "loss": 0.2053,
+      "step": 3342
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.3285e-05,
+      "loss": 0.0968,
+      "step": 3343
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.3280000000000006e-05,
+      "loss": 0.1418,
+      "step": 3344
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.3274999999999995e-05,
+      "loss": 0.2245,
+      "step": 3345
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.327e-05,
+      "loss": 0.2444,
+      "step": 3346
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.3265e-05,
+      "loss": 0.1568,
+      "step": 3347
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.326e-05,
+      "loss": 0.0959,
+      "step": 3348
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.3255e-05,
+      "loss": 0.2655,
+      "step": 3349
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.325e-05,
+      "loss": 0.2186,
+      "step": 3350
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.3245000000000004e-05,
+      "loss": 0.0812,
+      "step": 3351
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.324e-05,
+      "loss": 0.1395,
+      "step": 3352
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.3235e-05,
+      "loss": 0.2288,
+      "step": 3353
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.3230000000000005e-05,
+      "loss": 0.1415,
+      "step": 3354
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.3225e-05,
+      "loss": 0.2491,
+      "step": 3355
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.322e-05,
+      "loss": 0.2136,
+      "step": 3356
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.3215e-05,
+      "loss": 0.0551,
+      "step": 3357
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.321e-05,
+      "loss": 0.1712,
+      "step": 3358
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.3205e-05,
+      "loss": 0.1714,
+      "step": 3359
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.32e-05,
+      "loss": 0.2322,
+      "step": 3360
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.3195e-05,
+      "loss": 0.1101,
+      "step": 3361
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.319e-05,
+      "loss": 0.0792,
+      "step": 3362
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.3185e-05,
+      "loss": 0.1566,
+      "step": 3363
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.318e-05,
+      "loss": 0.068,
+      "step": 3364
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.3175e-05,
+      "loss": 0.1806,
+      "step": 3365
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.317e-05,
+      "loss": 0.0856,
+      "step": 3366
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.3165000000000004e-05,
+      "loss": 0.1052,
+      "step": 3367
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.316000000000001e-05,
+      "loss": 0.1786,
+      "step": 3368
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.3154999999999996e-05,
+      "loss": 0.0911,
+      "step": 3369
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.315e-05,
+      "loss": 0.1481,
+      "step": 3370
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.3145e-05,
+      "loss": 0.3259,
+      "step": 3371
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.314e-05,
+      "loss": 0.1189,
+      "step": 3372
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.3135e-05,
+      "loss": 0.0694,
+      "step": 3373
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.313e-05,
+      "loss": 0.3149,
+      "step": 3374
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.3125e-05,
+      "loss": 0.2875,
+      "step": 3375
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.312e-05,
+      "loss": 0.1912,
+      "step": 3376
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.3115e-05,
+      "loss": 0.172,
+      "step": 3377
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.3110000000000006e-05,
+      "loss": 0.0806,
+      "step": 3378
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.3105e-05,
+      "loss": 0.1001,
+      "step": 3379
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 0.1292,
+      "step": 3380
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.309500000000001e-05,
+      "loss": 0.2329,
+      "step": 3381
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.3089999999999996e-05,
+      "loss": 0.1064,
+      "step": 3382
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.3085e-05,
+      "loss": 0.0885,
+      "step": 3383
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.308e-05,
+      "loss": 0.1304,
+      "step": 3384
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.3075e-05,
+      "loss": 0.0839,
+      "step": 3385
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.307e-05,
+      "loss": 0.2922,
+      "step": 3386
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.3065e-05,
+      "loss": 0.0856,
+      "step": 3387
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.3060000000000004e-05,
+      "loss": 0.1986,
+      "step": 3388
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.3055e-05,
+      "loss": 0.4678,
+      "step": 3389
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.305e-05,
+      "loss": 0.2014,
+      "step": 3390
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.3045000000000005e-05,
+      "loss": 0.311,
+      "step": 3391
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.304e-05,
+      "loss": 0.1013,
+      "step": 3392
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.3035000000000004e-05,
+      "loss": 0.0971,
+      "step": 3393
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.303e-05,
+      "loss": 0.1535,
+      "step": 3394
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.3024999999999995e-05,
+      "loss": 0.1044,
+      "step": 3395
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.302e-05,
+      "loss": 0.0956,
+      "step": 3396
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.3015e-05,
+      "loss": 0.0678,
+      "step": 3397
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.301e-05,
+      "loss": 0.0929,
+      "step": 3398
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.3005e-05,
+      "loss": 0.1815,
+      "step": 3399
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.3e-05,
+      "loss": 0.2322,
+      "step": 3400
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.2995000000000004e-05,
+      "loss": 0.2865,
+      "step": 3401
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.299e-05,
+      "loss": 0.1207,
+      "step": 3402
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.2985e-05,
+      "loss": 0.1053,
+      "step": 3403
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.2980000000000005e-05,
+      "loss": 0.2795,
+      "step": 3404
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.2975e-05,
+      "loss": 0.2475,
+      "step": 3405
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.297e-05,
+      "loss": 0.0968,
+      "step": 3406
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.2965e-05,
+      "loss": 0.0816,
+      "step": 3407
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.296e-05,
+      "loss": 0.1056,
+      "step": 3408
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.2955e-05,
+      "loss": 0.2115,
+      "step": 3409
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.295e-05,
+      "loss": 0.0582,
+      "step": 3410
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.2945e-05,
+      "loss": 0.3129,
+      "step": 3411
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.294e-05,
+      "loss": 0.132,
+      "step": 3412
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.2935e-05,
+      "loss": 0.0865,
+      "step": 3413
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.2930000000000004e-05,
+      "loss": 0.2407,
+      "step": 3414
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.2925e-05,
+      "loss": 0.1608,
+      "step": 3415
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.292e-05,
+      "loss": 0.1145,
+      "step": 3416
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.2915000000000005e-05,
+      "loss": 0.1023,
+      "step": 3417
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.291e-05,
+      "loss": 0.0921,
+      "step": 3418
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.2905e-05,
+      "loss": 0.1141,
+      "step": 3419
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.29e-05,
+      "loss": 0.127,
+      "step": 3420
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.2895e-05,
+      "loss": 0.1223,
+      "step": 3421
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.289e-05,
+      "loss": 0.0634,
+      "step": 3422
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.2885e-05,
+      "loss": 0.0917,
+      "step": 3423
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.288e-05,
+      "loss": 0.0614,
+      "step": 3424
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.2875e-05,
+      "loss": 0.1555,
+      "step": 3425
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.287e-05,
+      "loss": 0.1358,
+      "step": 3426
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.2865000000000004e-05,
+      "loss": 0.1748,
+      "step": 3427
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.2860000000000006e-05,
+      "loss": 0.3322,
+      "step": 3428
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.2855e-05,
+      "loss": 0.1942,
+      "step": 3429
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.285e-05,
+      "loss": 0.1926,
+      "step": 3430
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.2845e-05,
+      "loss": 0.1645,
+      "step": 3431
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.2839999999999996e-05,
+      "loss": 0.1618,
+      "step": 3432
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.2835e-05,
+      "loss": 0.0753,
+      "step": 3433
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.283e-05,
+      "loss": 0.0811,
+      "step": 3434
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.2825000000000004e-05,
+      "loss": 0.1894,
+      "step": 3435
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.282e-05,
+      "loss": 0.2434,
+      "step": 3436
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.2815e-05,
+      "loss": 0.164,
+      "step": 3437
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.2810000000000005e-05,
+      "loss": 0.0601,
+      "step": 3438
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.2805e-05,
+      "loss": 0.2899,
+      "step": 3439
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 0.1587,
+      "step": 3440
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.2795000000000006e-05,
+      "loss": 0.2163,
+      "step": 3441
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.2789999999999995e-05,
+      "loss": 0.1038,
+      "step": 3442
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.2785e-05,
+      "loss": 0.2575,
+      "step": 3443
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.278e-05,
+      "loss": 0.272,
+      "step": 3444
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.2775e-05,
+      "loss": 0.1537,
+      "step": 3445
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.277e-05,
+      "loss": 0.1959,
+      "step": 3446
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.2765e-05,
+      "loss": 0.1723,
+      "step": 3447
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.2760000000000004e-05,
+      "loss": 0.189,
+      "step": 3448
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.2755e-05,
+      "loss": 0.2465,
+      "step": 3449
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.275e-05,
+      "loss": 0.1495,
+      "step": 3450
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.2745000000000005e-05,
+      "loss": 0.1339,
+      "step": 3451
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.274e-05,
+      "loss": 0.0715,
+      "step": 3452
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.2735e-05,
+      "loss": 0.1104,
+      "step": 3453
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.273e-05,
+      "loss": 0.1207,
+      "step": 3454
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.2725e-05,
+      "loss": 0.1526,
+      "step": 3455
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.272e-05,
+      "loss": 0.2669,
+      "step": 3456
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.2715e-05,
+      "loss": 0.0993,
+      "step": 3457
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.271e-05,
+      "loss": 0.0636,
+      "step": 3458
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.2705e-05,
+      "loss": 0.3053,
+      "step": 3459
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.27e-05,
+      "loss": 0.0537,
+      "step": 3460
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.2695000000000004e-05,
+      "loss": 0.2974,
+      "step": 3461
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.269e-05,
+      "loss": 0.2313,
+      "step": 3462
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.2685e-05,
+      "loss": 0.153,
+      "step": 3463
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.2680000000000005e-05,
+      "loss": 0.1365,
+      "step": 3464
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.267500000000001e-05,
+      "loss": 0.2447,
+      "step": 3465
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.2669999999999996e-05,
+      "loss": 0.5336,
+      "step": 3466
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.2665e-05,
+      "loss": 0.256,
+      "step": 3467
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.266e-05,
+      "loss": 0.208,
+      "step": 3468
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.2655e-05,
+      "loss": 0.3141,
+      "step": 3469
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.265e-05,
+      "loss": 0.1073,
+      "step": 3470
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.2645e-05,
+      "loss": 0.193,
+      "step": 3471
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.264e-05,
+      "loss": 0.0935,
+      "step": 3472
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.2635e-05,
+      "loss": 0.3612,
+      "step": 3473
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.2630000000000004e-05,
+      "loss": 0.0801,
+      "step": 3474
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.2625000000000006e-05,
+      "loss": 0.1813,
+      "step": 3475
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.262e-05,
+      "loss": 0.2605,
+      "step": 3476
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.2615000000000004e-05,
+      "loss": 0.0904,
+      "step": 3477
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.261e-05,
+      "loss": 0.2145,
+      "step": 3478
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.2604999999999996e-05,
+      "loss": 0.2559,
+      "step": 3479
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.26e-05,
+      "loss": 0.1472,
+      "step": 3480
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.2595e-05,
+      "loss": 0.2025,
+      "step": 3481
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.259e-05,
+      "loss": 0.3363,
+      "step": 3482
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.2585e-05,
+      "loss": 0.3265,
+      "step": 3483
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.258e-05,
+      "loss": 0.1479,
+      "step": 3484
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.2575000000000005e-05,
+      "loss": 0.0756,
+      "step": 3485
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.257e-05,
+      "loss": 0.1551,
+      "step": 3486
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.2565e-05,
+      "loss": 0.2534,
+      "step": 3487
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.2560000000000006e-05,
+      "loss": 0.262,
+      "step": 3488
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.2555e-05,
+      "loss": 0.1837,
+      "step": 3489
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.255e-05,
+      "loss": 0.181,
+      "step": 3490
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.2545e-05,
+      "loss": 0.0764,
+      "step": 3491
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.2539999999999996e-05,
+      "loss": 0.1285,
+      "step": 3492
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.2535e-05,
+      "loss": 0.3146,
+      "step": 3493
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.253e-05,
+      "loss": 0.1284,
+      "step": 3494
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.2525000000000004e-05,
+      "loss": 0.0805,
+      "step": 3495
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.252e-05,
+      "loss": 0.145,
+      "step": 3496
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.2515e-05,
+      "loss": 0.2925,
+      "step": 3497
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.2510000000000005e-05,
+      "loss": 0.095,
+      "step": 3498
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.2505e-05,
+      "loss": 0.1969,
+      "step": 3499
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.25e-05,
+      "loss": 0.2568,
+      "step": 3500
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.2495000000000006e-05,
+      "loss": 0.0733,
+      "step": 3501
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.249e-05,
+      "loss": 0.0645,
+      "step": 3502
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.2485e-05,
+      "loss": 0.0858,
+      "step": 3503
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.248e-05,
+      "loss": 0.1801,
+      "step": 3504
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.2475e-05,
+      "loss": 0.0731,
+      "step": 3505
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.247e-05,
+      "loss": 0.2584,
+      "step": 3506
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.2465e-05,
+      "loss": 0.0941,
+      "step": 3507
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.2460000000000003e-05,
+      "loss": 0.1927,
+      "step": 3508
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.2455e-05,
+      "loss": 0.2941,
+      "step": 3509
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.245e-05,
+      "loss": 0.2037,
+      "step": 3510
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.2445000000000004e-05,
+      "loss": 0.0515,
+      "step": 3511
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.244e-05,
+      "loss": 0.2517,
+      "step": 3512
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.2435e-05,
+      "loss": 0.099,
+      "step": 3513
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.2430000000000005e-05,
+      "loss": 0.0683,
+      "step": 3514
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.2425e-05,
+      "loss": 0.055,
+      "step": 3515
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.242e-05,
+      "loss": 0.2323,
+      "step": 3516
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.2415e-05,
+      "loss": 0.1243,
+      "step": 3517
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.241e-05,
+      "loss": 0.2301,
+      "step": 3518
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.2405e-05,
+      "loss": 0.0672,
+      "step": 3519
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.24e-05,
+      "loss": 0.2309,
+      "step": 3520
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.2395e-05,
+      "loss": 0.0754,
+      "step": 3521
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.239e-05,
+      "loss": 0.1645,
+      "step": 3522
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.2385e-05,
+      "loss": 0.0795,
+      "step": 3523
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.2380000000000004e-05,
+      "loss": 0.1068,
+      "step": 3524
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.237500000000001e-05,
+      "loss": 0.0799,
+      "step": 3525
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.237e-05,
+      "loss": 0.3191,
+      "step": 3526
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.2365e-05,
+      "loss": 0.1227,
+      "step": 3527
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.236e-05,
+      "loss": 0.2246,
+      "step": 3528
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.2355e-05,
+      "loss": 0.1465,
+      "step": 3529
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.235e-05,
+      "loss": 0.1542,
+      "step": 3530
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.2345e-05,
+      "loss": 0.0937,
+      "step": 3531
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.234e-05,
+      "loss": 0.3002,
+      "step": 3532
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.2335e-05,
+      "loss": 0.1703,
+      "step": 3533
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.233e-05,
+      "loss": 0.2352,
+      "step": 3534
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.2325000000000006e-05,
+      "loss": 0.3072,
+      "step": 3535
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.232e-05,
+      "loss": 0.2151,
+      "step": 3536
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.2315000000000004e-05,
+      "loss": 0.209,
+      "step": 3537
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.2310000000000007e-05,
+      "loss": 0.0863,
+      "step": 3538
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.2304999999999996e-05,
+      "loss": 0.2354,
+      "step": 3539
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.23e-05,
+      "loss": 0.1332,
+      "step": 3540
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.2295e-05,
+      "loss": 0.165,
+      "step": 3541
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.2289999999999997e-05,
+      "loss": 0.1013,
+      "step": 3542
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.2285e-05,
+      "loss": 0.2507,
+      "step": 3543
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.228e-05,
+      "loss": 0.0884,
+      "step": 3544
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.2275000000000004e-05,
+      "loss": 0.1738,
+      "step": 3545
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.227e-05,
+      "loss": 0.1138,
+      "step": 3546
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.2265e-05,
+      "loss": 0.076,
+      "step": 3547
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.2260000000000005e-05,
+      "loss": 0.2029,
+      "step": 3548
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.2255e-05,
+      "loss": 0.1731,
+      "step": 3549
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.2250000000000004e-05,
+      "loss": 0.1844,
+      "step": 3550
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.2245e-05,
+      "loss": 0.1841,
+      "step": 3551
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.2239999999999995e-05,
+      "loss": 0.2278,
+      "step": 3552
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.2235e-05,
+      "loss": 0.1111,
+      "step": 3553
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.223e-05,
+      "loss": 0.1294,
+      "step": 3554
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.2225e-05,
+      "loss": 0.2002,
+      "step": 3555
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.222e-05,
+      "loss": 0.1802,
+      "step": 3556
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.2215e-05,
+      "loss": 0.1692,
+      "step": 3557
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.2210000000000004e-05,
+      "loss": 0.077,
+      "step": 3558
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.2205e-05,
+      "loss": 0.2192,
+      "step": 3559
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.22e-05,
+      "loss": 0.242,
+      "step": 3560
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.2195000000000005e-05,
+      "loss": 0.1814,
+      "step": 3561
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.219e-05,
+      "loss": 0.0928,
+      "step": 3562
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.2185e-05,
+      "loss": 0.0723,
+      "step": 3563
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.218e-05,
+      "loss": 0.0895,
+      "step": 3564
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.2175e-05,
+      "loss": 0.2013,
+      "step": 3565
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.217e-05,
+      "loss": 0.1244,
+      "step": 3566
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.2165e-05,
+      "loss": 0.0671,
+      "step": 3567
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.216e-05,
+      "loss": 0.1056,
+      "step": 3568
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.2155e-05,
+      "loss": 0.081,
+      "step": 3569
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.215e-05,
+      "loss": 0.2078,
+      "step": 3570
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.2145000000000004e-05,
+      "loss": 0.1363,
+      "step": 3571
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.214e-05,
+      "loss": 0.075,
+      "step": 3572
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.2135e-05,
+      "loss": 0.0719,
+      "step": 3573
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.2130000000000005e-05,
+      "loss": 0.0732,
+      "step": 3574
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.2125e-05,
+      "loss": 0.1291,
+      "step": 3575
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.2119999999999997e-05,
+      "loss": 0.185,
+      "step": 3576
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.2115e-05,
+      "loss": 0.1725,
+      "step": 3577
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.211e-05,
+      "loss": 0.091,
+      "step": 3578
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.2105e-05,
+      "loss": 0.2338,
+      "step": 3579
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.21e-05,
+      "loss": 0.2536,
+      "step": 3580
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.2095e-05,
+      "loss": 0.073,
+      "step": 3581
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.209e-05,
+      "loss": 0.177,
+      "step": 3582
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.2085e-05,
+      "loss": 0.188,
+      "step": 3583
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.2080000000000004e-05,
+      "loss": 0.3196,
+      "step": 3584
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.2075000000000006e-05,
+      "loss": 0.2418,
+      "step": 3585
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.207e-05,
+      "loss": 0.1008,
+      "step": 3586
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.2065e-05,
+      "loss": 0.1129,
+      "step": 3587
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.206e-05,
+      "loss": 0.174,
+      "step": 3588
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.2054999999999996e-05,
+      "loss": 0.1457,
+      "step": 3589
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.205e-05,
+      "loss": 0.0634,
+      "step": 3590
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.2045e-05,
+      "loss": 0.1176,
+      "step": 3591
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.204e-05,
+      "loss": 0.163,
+      "step": 3592
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.2035e-05,
+      "loss": 0.2262,
+      "step": 3593
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.203e-05,
+      "loss": 0.1526,
+      "step": 3594
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.2025000000000005e-05,
+      "loss": 0.0848,
+      "step": 3595
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.202e-05,
+      "loss": 0.1417,
+      "step": 3596
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.2015000000000003e-05,
+      "loss": 0.0812,
+      "step": 3597
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.2010000000000006e-05,
+      "loss": 0.201,
+      "step": 3598
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.2004999999999995e-05,
+      "loss": 0.3443,
+      "step": 3599
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.2e-05,
+      "loss": 0.2102,
+      "step": 3600
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.1995e-05,
+      "loss": 0.0848,
+      "step": 3601
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.199e-05,
+      "loss": 0.0682,
+      "step": 3602
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.1985e-05,
+      "loss": 0.1892,
+      "step": 3603
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 4.198e-05,
+      "loss": 0.0574,
+      "step": 3604
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 4.1975000000000004e-05,
+      "loss": 0.0782,
+      "step": 3605
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 4.197e-05,
+      "loss": 0.1757,
+      "step": 3606
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 4.1965e-05,
+      "loss": 0.1615,
+      "step": 3607
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 4.1960000000000005e-05,
+      "loss": 0.0904,
+      "step": 3608
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 4.1955e-05,
+      "loss": 0.2364,
+      "step": 3609
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 4.195e-05,
+      "loss": 0.1056,
+      "step": 3610
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 4.1945e-05,
+      "loss": 0.2236,
+      "step": 3611
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 4.194e-05,
+      "loss": 0.1011,
+      "step": 3612
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 4.1935e-05,
+      "loss": 0.4103,
+      "step": 3613
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 4.193e-05,
+      "loss": 0.2544,
+      "step": 3614
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 4.1925e-05,
+      "loss": 0.3143,
+      "step": 3615
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 4.192e-05,
+      "loss": 0.1883,
+      "step": 3616
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 4.1915e-05,
+      "loss": 0.2263,
+      "step": 3617
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 4.1910000000000004e-05,
+      "loss": 0.2403,
+      "step": 3618
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 4.1905e-05,
+      "loss": 0.1672,
+      "step": 3619
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 4.19e-05,
+      "loss": 0.22,
+      "step": 3620
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 4.1895000000000005e-05,
+      "loss": 0.1276,
+      "step": 3621
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 4.189000000000001e-05,
+      "loss": 0.0455,
+      "step": 3622
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 4.1884999999999996e-05,
+      "loss": 0.0836,
+      "step": 3623
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 4.188e-05,
+      "loss": 0.1793,
+      "step": 3624
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 4.1875e-05,
+      "loss": 0.2186,
+      "step": 3625
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 4.187e-05,
+      "loss": 0.1127,
+      "step": 3626
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 4.1865e-05,
+      "loss": 0.0964,
+      "step": 3627
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 4.186e-05,
+      "loss": 0.196,
+      "step": 3628
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 4.1855e-05,
+      "loss": 0.1862,
+      "step": 3629
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 4.185e-05,
+      "loss": 0.1585,
+      "step": 3630
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 4.1845000000000003e-05,
+      "loss": 0.112,
+      "step": 3631
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 4.1840000000000006e-05,
+      "loss": 0.0709,
+      "step": 3632
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 4.1835e-05,
+      "loss": 0.0656,
+      "step": 3633
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 4.1830000000000004e-05,
+      "loss": 0.2173,
+      "step": 3634
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 4.182500000000001e-05,
+      "loss": 0.0673,
+      "step": 3635
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 4.1819999999999996e-05,
+      "loss": 0.2561,
+      "step": 3636
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 4.1815e-05,
+      "loss": 0.1744,
+      "step": 3637
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 4.181e-05,
+      "loss": 0.2648,
+      "step": 3638
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 4.1805e-05,
+      "loss": 0.0789,
+      "step": 3639
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 4.18e-05,
+      "loss": 0.0972,
+      "step": 3640
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 4.1795e-05,
+      "loss": 0.1003,
+      "step": 3641
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 4.1790000000000005e-05,
+      "loss": 0.2356,
+      "step": 3642
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 4.1785e-05,
+      "loss": 0.1611,
+      "step": 3643
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 4.178e-05,
+      "loss": 0.1698,
+      "step": 3644
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 4.1775000000000006e-05,
+      "loss": 0.146,
+      "step": 3645
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 4.177e-05,
+      "loss": 0.0906,
+      "step": 3646
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 4.1765000000000004e-05,
+      "loss": 0.204,
+      "step": 3647
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 4.176e-05,
+      "loss": 0.056,
+      "step": 3648
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 4.1754999999999996e-05,
+      "loss": 0.2209,
+      "step": 3649
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 4.175e-05,
+      "loss": 0.0576,
+      "step": 3650
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 4.1745e-05,
+      "loss": 0.2466,
+      "step": 3651
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 4.1740000000000004e-05,
+      "loss": 0.1045,
+      "step": 3652
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 4.1735e-05,
+      "loss": 0.0654,
+      "step": 3653
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 4.173e-05,
+      "loss": 0.2119,
+      "step": 3654
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 4.1725000000000005e-05,
+      "loss": 0.1124,
+      "step": 3655
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 4.172e-05,
+      "loss": 0.0496,
+      "step": 3656
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 4.1715e-05,
+      "loss": 0.093,
+      "step": 3657
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 4.1710000000000006e-05,
+      "loss": 0.2806,
+      "step": 3658
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 4.1705e-05,
+      "loss": 0.0826,
+      "step": 3659
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 4.17e-05,
+      "loss": 0.0939,
+      "step": 3660
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 4.1695e-05,
+      "loss": 0.1229,
+      "step": 3661
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 4.169e-05,
+      "loss": 0.3593,
+      "step": 3662
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 4.1685e-05,
+      "loss": 0.0849,
+      "step": 3663
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 4.168e-05,
+      "loss": 0.197,
+      "step": 3664
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 4.1675e-05,
+      "loss": 0.0727,
+      "step": 3665
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 4.167e-05,
+      "loss": 0.2063,
+      "step": 3666
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 4.1665e-05,
+      "loss": 0.0485,
+      "step": 3667
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 4.1660000000000004e-05,
+      "loss": 0.2845,
+      "step": 3668
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 4.1655e-05,
+      "loss": 0.0832,
+      "step": 3669
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 4.165e-05,
+      "loss": 0.2285,
+      "step": 3670
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 4.1645000000000005e-05,
+      "loss": 0.1592,
+      "step": 3671
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 4.164e-05,
+      "loss": 0.1037,
+      "step": 3672
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 4.1635e-05,
+      "loss": 0.0977,
+      "step": 3673
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 4.163e-05,
+      "loss": 0.1824,
+      "step": 3674
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 4.1625e-05,
+      "loss": 0.1738,
+      "step": 3675
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 4.162e-05,
+      "loss": 0.0977,
+      "step": 3676
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 4.1615e-05,
+      "loss": 0.0737,
+      "step": 3677
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 4.161e-05,
+      "loss": 0.1147,
+      "step": 3678
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 4.1605e-05,
+      "loss": 0.2263,
+      "step": 3679
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 4.16e-05,
+      "loss": 0.2011,
+      "step": 3680
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 4.1595000000000004e-05,
+      "loss": 0.2901,
+      "step": 3681
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 4.159000000000001e-05,
+      "loss": 0.0827,
+      "step": 3682
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 4.1585e-05,
+      "loss": 0.2586,
+      "step": 3683
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 4.158e-05,
+      "loss": 0.053,
+      "step": 3684
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 4.1575e-05,
+      "loss": 0.2188,
+      "step": 3685
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 4.157e-05,
+      "loss": 0.0924,
+      "step": 3686
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 4.1565e-05,
+      "loss": 0.0611,
+      "step": 3687
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 4.156e-05,
+      "loss": 0.2222,
+      "step": 3688
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 4.1555e-05,
+      "loss": 0.2676,
+      "step": 3689
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 4.155e-05,
+      "loss": 0.0831,
+      "step": 3690
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 4.1545e-05,
+      "loss": 0.1276,
+      "step": 3691
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 4.1540000000000005e-05,
+      "loss": 0.0817,
+      "step": 3692
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 4.1535e-05,
+      "loss": 0.1641,
+      "step": 3693
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 4.1530000000000004e-05,
+      "loss": 0.224,
+      "step": 3694
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 4.1525000000000006e-05,
+      "loss": 0.0867,
+      "step": 3695
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 4.1519999999999995e-05,
+      "loss": 0.191,
+      "step": 3696
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 4.1515e-05,
+      "loss": 0.1922,
+      "step": 3697
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 4.151e-05,
+      "loss": 0.0573,
+      "step": 3698
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 4.1504999999999996e-05,
+      "loss": 0.1578,
+      "step": 3699
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 4.15e-05,
+      "loss": 0.0855,
+      "step": 3700
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 4.1495e-05,
+      "loss": 0.1458,
+      "step": 3701
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 4.1490000000000004e-05,
+      "loss": 0.0789,
+      "step": 3702
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 4.1485e-05,
+      "loss": 0.0818,
+      "step": 3703
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 4.148e-05,
+      "loss": 0.0978,
+      "step": 3704
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 4.1475000000000005e-05,
+      "loss": 0.0695,
+      "step": 3705
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 4.147e-05,
+      "loss": 0.1339,
+      "step": 3706
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 4.1465000000000004e-05,
+      "loss": 0.1031,
+      "step": 3707
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 4.146e-05,
+      "loss": 0.076,
+      "step": 3708
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 4.1454999999999995e-05,
+      "loss": 0.0677,
+      "step": 3709
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 4.145e-05,
+      "loss": 0.1747,
+      "step": 3710
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 4.1445e-05,
+      "loss": 0.1659,
+      "step": 3711
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 4.144e-05,
+      "loss": 0.2881,
+      "step": 3712
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 4.1435e-05,
+      "loss": 0.3401,
+      "step": 3713
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 4.143e-05,
+      "loss": 0.2614,
+      "step": 3714
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 4.1425000000000004e-05,
+      "loss": 0.2047,
+      "step": 3715
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 4.142e-05,
+      "loss": 0.0818,
+      "step": 3716
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 4.1415e-05,
+      "loss": 0.2112,
+      "step": 3717
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 4.1410000000000005e-05,
+      "loss": 0.1409,
+      "step": 3718
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.1405e-05,
+      "loss": 0.1819,
+      "step": 3719
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.14e-05,
+      "loss": 0.3257,
+      "step": 3720
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.1395e-05,
+      "loss": 0.0887,
+      "step": 3721
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.139e-05,
+      "loss": 0.1105,
+      "step": 3722
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.1385e-05,
+      "loss": 0.1306,
+      "step": 3723
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.138e-05,
+      "loss": 0.19,
+      "step": 3724
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.1375e-05,
+      "loss": 0.1148,
+      "step": 3725
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.137e-05,
+      "loss": 0.3053,
+      "step": 3726
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.1365e-05,
+      "loss": 0.2331,
+      "step": 3727
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.1360000000000004e-05,
+      "loss": 0.1185,
+      "step": 3728
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.1355e-05,
+      "loss": 0.1666,
+      "step": 3729
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.135e-05,
+      "loss": 0.2015,
+      "step": 3730
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.1345000000000005e-05,
+      "loss": 0.0879,
+      "step": 3731
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.134e-05,
+      "loss": 0.1521,
+      "step": 3732
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.1334999999999996e-05,
+      "loss": 0.2489,
+      "step": 3733
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.133e-05,
+      "loss": 0.3245,
+      "step": 3734
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.1325e-05,
+      "loss": 0.1143,
+      "step": 3735
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.132e-05,
+      "loss": 0.2432,
+      "step": 3736
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.1315e-05,
+      "loss": 0.141,
+      "step": 3737
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.131e-05,
+      "loss": 0.1436,
+      "step": 3738
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.1305e-05,
+      "loss": 0.2865,
+      "step": 3739
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.13e-05,
+      "loss": 0.1802,
+      "step": 3740
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.1295000000000004e-05,
+      "loss": 0.1719,
+      "step": 3741
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.1290000000000006e-05,
+      "loss": 0.0791,
+      "step": 3742
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.1285e-05,
+      "loss": 0.1179,
+      "step": 3743
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.128e-05,
+      "loss": 0.0753,
+      "step": 3744
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.1275e-05,
+      "loss": 0.1588,
+      "step": 3745
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.1269999999999996e-05,
+      "loss": 0.214,
+      "step": 3746
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.1265e-05,
+      "loss": 0.2147,
+      "step": 3747
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.126e-05,
+      "loss": 0.1042,
+      "step": 3748
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.1255e-05,
+      "loss": 0.2103,
+      "step": 3749
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.125e-05,
+      "loss": 0.078,
+      "step": 3750
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.1245e-05,
+      "loss": 0.4006,
+      "step": 3751
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.1240000000000005e-05,
+      "loss": 0.0598,
+      "step": 3752
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.1235e-05,
+      "loss": 0.109,
+      "step": 3753
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.123e-05,
+      "loss": 0.336,
+      "step": 3754
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.1225000000000006e-05,
+      "loss": 0.2462,
+      "step": 3755
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.122e-05,
+      "loss": 0.0716,
+      "step": 3756
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.1215e-05,
+      "loss": 0.1405,
+      "step": 3757
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.121e-05,
+      "loss": 0.1921,
+      "step": 3758
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.1204999999999996e-05,
+      "loss": 0.1232,
+      "step": 3759
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.12e-05,
+      "loss": 0.1659,
+      "step": 3760
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.1195e-05,
+      "loss": 0.0743,
+      "step": 3761
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.1190000000000004e-05,
+      "loss": 0.086,
+      "step": 3762
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.1185e-05,
+      "loss": 0.0747,
+      "step": 3763
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.118e-05,
+      "loss": 0.1861,
+      "step": 3764
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.1175000000000005e-05,
+      "loss": 0.1419,
+      "step": 3765
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.117e-05,
+      "loss": 0.1565,
+      "step": 3766
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.1165e-05,
+      "loss": 0.1537,
+      "step": 3767
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.1160000000000006e-05,
+      "loss": 0.1366,
+      "step": 3768
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.1155e-05,
+      "loss": 0.08,
+      "step": 3769
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.115e-05,
+      "loss": 0.2087,
+      "step": 3770
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.1145e-05,
+      "loss": 0.1746,
+      "step": 3771
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.114e-05,
+      "loss": 0.0741,
+      "step": 3772
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.1135e-05,
+      "loss": 0.1792,
+      "step": 3773
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.113e-05,
+      "loss": 0.3043,
+      "step": 3774
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.1125000000000004e-05,
+      "loss": 0.2053,
+      "step": 3775
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.112e-05,
+      "loss": 0.0749,
+      "step": 3776
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.1115e-05,
+      "loss": 0.0823,
+      "step": 3777
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.1110000000000005e-05,
+      "loss": 0.2086,
+      "step": 3778
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.1105e-05,
+      "loss": 0.0812,
+      "step": 3779
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.11e-05,
+      "loss": 0.2371,
+      "step": 3780
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.1095e-05,
+      "loss": 0.1524,
+      "step": 3781
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.109e-05,
+      "loss": 0.2873,
+      "step": 3782
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.1085e-05,
+      "loss": 0.1863,
+      "step": 3783
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.108e-05,
+      "loss": 0.1436,
+      "step": 3784
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.1075e-05,
+      "loss": 0.2043,
+      "step": 3785
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.107e-05,
+      "loss": 0.1481,
+      "step": 3786
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.1065e-05,
+      "loss": 0.2362,
+      "step": 3787
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.106e-05,
+      "loss": 0.2039,
+      "step": 3788
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.1055000000000006e-05,
+      "loss": 0.0978,
+      "step": 3789
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.105e-05,
+      "loss": 0.1259,
+      "step": 3790
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.1045000000000004e-05,
+      "loss": 0.1848,
+      "step": 3791
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.104000000000001e-05,
+      "loss": 0.1471,
+      "step": 3792
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.1034999999999996e-05,
+      "loss": 0.3671,
+      "step": 3793
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.103e-05,
+      "loss": 0.0919,
+      "step": 3794
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.1025e-05,
+      "loss": 0.0976,
+      "step": 3795
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.102e-05,
+      "loss": 0.2326,
+      "step": 3796
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.1015e-05,
+      "loss": 0.234,
+      "step": 3797
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.101e-05,
+      "loss": 0.1703,
+      "step": 3798
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.1005000000000005e-05,
+      "loss": 0.126,
+      "step": 3799
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.1e-05,
+      "loss": 0.3235,
+      "step": 3800
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.0995e-05,
+      "loss": 0.235,
+      "step": 3801
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.0990000000000006e-05,
+      "loss": 0.1782,
+      "step": 3802
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.0985e-05,
+      "loss": 0.0933,
+      "step": 3803
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.0980000000000004e-05,
+      "loss": 0.064,
+      "step": 3804
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.0975e-05,
+      "loss": 0.2044,
+      "step": 3805
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.0969999999999996e-05,
+      "loss": 0.2958,
+      "step": 3806
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.0965e-05,
+      "loss": 0.2612,
+      "step": 3807
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.096e-05,
+      "loss": 0.0916,
+      "step": 3808
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.0955000000000003e-05,
+      "loss": 0.2393,
+      "step": 3809
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.095e-05,
+      "loss": 0.1484,
+      "step": 3810
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.0945e-05,
+      "loss": 0.118,
+      "step": 3811
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.0940000000000004e-05,
+      "loss": 0.0629,
+      "step": 3812
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.0935e-05,
+      "loss": 0.1165,
+      "step": 3813
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.093e-05,
+      "loss": 0.0819,
+      "step": 3814
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.0925000000000005e-05,
+      "loss": 0.1412,
+      "step": 3815
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.092e-05,
+      "loss": 0.1862,
+      "step": 3816
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.0915e-05,
+      "loss": 0.0901,
+      "step": 3817
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.091e-05,
+      "loss": 0.0874,
+      "step": 3818
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.0905e-05,
+      "loss": 0.1766,
+      "step": 3819
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.09e-05,
+      "loss": 0.1448,
+      "step": 3820
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.0895e-05,
+      "loss": 0.0947,
+      "step": 3821
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.089e-05,
+      "loss": 0.3029,
+      "step": 3822
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.0885e-05,
+      "loss": 0.2,
+      "step": 3823
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.088e-05,
+      "loss": 0.1405,
+      "step": 3824
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.0875000000000004e-05,
+      "loss": 0.1185,
+      "step": 3825
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.087e-05,
+      "loss": 0.1317,
+      "step": 3826
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.0865e-05,
+      "loss": 0.2229,
+      "step": 3827
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.0860000000000005e-05,
+      "loss": 0.2794,
+      "step": 3828
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.0855e-05,
+      "loss": 0.324,
+      "step": 3829
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.085e-05,
+      "loss": 0.0948,
+      "step": 3830
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.0845e-05,
+      "loss": 0.1837,
+      "step": 3831
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.084e-05,
+      "loss": 0.16,
+      "step": 3832
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.0835e-05,
+      "loss": 0.4436,
+      "step": 3833
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.083e-05,
+      "loss": 0.0802,
+      "step": 3834
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.0825e-05,
+      "loss": 0.1243,
+      "step": 3835
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.082e-05,
+      "loss": 0.1725,
+      "step": 3836
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.0815e-05,
+      "loss": 0.24,
+      "step": 3837
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.0810000000000004e-05,
+      "loss": 0.1625,
+      "step": 3838
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.0805000000000007e-05,
+      "loss": 0.125,
+      "step": 3839
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.08e-05,
+      "loss": 0.1604,
+      "step": 3840
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.0795e-05,
+      "loss": 0.1824,
+      "step": 3841
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.079e-05,
+      "loss": 0.0653,
+      "step": 3842
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.0784999999999997e-05,
+      "loss": 0.0835,
+      "step": 3843
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.078e-05,
+      "loss": 0.0939,
+      "step": 3844
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.0775e-05,
+      "loss": 0.2036,
+      "step": 3845
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.077e-05,
+      "loss": 0.1403,
+      "step": 3846
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.0765e-05,
+      "loss": 0.2944,
+      "step": 3847
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.076e-05,
+      "loss": 0.1852,
+      "step": 3848
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.0755000000000005e-05,
+      "loss": 0.1938,
+      "step": 3849
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.075e-05,
+      "loss": 0.1529,
+      "step": 3850
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.0745000000000004e-05,
+      "loss": 0.2243,
+      "step": 3851
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.0740000000000006e-05,
+      "loss": 0.3293,
+      "step": 3852
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.0734999999999995e-05,
+      "loss": 0.2933,
+      "step": 3853
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.073e-05,
+      "loss": 0.1612,
+      "step": 3854
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.0725e-05,
+      "loss": 0.1269,
+      "step": 3855
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.0719999999999996e-05,
+      "loss": 0.0876,
+      "step": 3856
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.0715e-05,
+      "loss": 0.3066,
+      "step": 3857
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.071e-05,
+      "loss": 0.1938,
+      "step": 3858
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.0705000000000004e-05,
+      "loss": 0.0863,
+      "step": 3859
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.07e-05,
+      "loss": 0.2558,
+      "step": 3860
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.0695e-05,
+      "loss": 0.0886,
+      "step": 3861
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.0690000000000005e-05,
+      "loss": 0.3157,
+      "step": 3862
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.0685e-05,
+      "loss": 0.0781,
+      "step": 3863
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.0680000000000004e-05,
+      "loss": 0.0769,
+      "step": 3864
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.0675e-05,
+      "loss": 0.0853,
+      "step": 3865
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.0669999999999995e-05,
+      "loss": 0.263,
+      "step": 3866
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.0665e-05,
+      "loss": 0.153,
+      "step": 3867
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.066e-05,
+      "loss": 0.2545,
+      "step": 3868
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.0655e-05,
+      "loss": 0.1509,
+      "step": 3869
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.065e-05,
+      "loss": 0.2829,
+      "step": 3870
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.0645e-05,
+      "loss": 0.1178,
+      "step": 3871
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.0640000000000004e-05,
+      "loss": 0.0825,
+      "step": 3872
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.0635e-05,
+      "loss": 0.1734,
+      "step": 3873
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.063e-05,
+      "loss": 0.2921,
+      "step": 3874
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.209,
+      "step": 3875
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.062e-05,
+      "loss": 0.2104,
+      "step": 3876
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.0615e-05,
+      "loss": 0.1777,
+      "step": 3877
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.061e-05,
+      "loss": 0.1432,
+      "step": 3878
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.0605e-05,
+      "loss": 0.0625,
+      "step": 3879
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.06e-05,
+      "loss": 0.1069,
+      "step": 3880
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.0595e-05,
+      "loss": 0.1574,
+      "step": 3881
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.059e-05,
+      "loss": 0.2831,
+      "step": 3882
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.0585e-05,
+      "loss": 0.176,
+      "step": 3883
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.058e-05,
+      "loss": 0.2569,
+      "step": 3884
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.0575000000000004e-05,
+      "loss": 0.0951,
+      "step": 3885
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.057e-05,
+      "loss": 0.2105,
+      "step": 3886
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.0565e-05,
+      "loss": 0.3025,
+      "step": 3887
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.0560000000000005e-05,
+      "loss": 0.097,
+      "step": 3888
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.055500000000001e-05,
+      "loss": 0.1023,
+      "step": 3889
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.0549999999999996e-05,
+      "loss": 0.1526,
+      "step": 3890
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.0545e-05,
+      "loss": 0.1085,
+      "step": 3891
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.054e-05,
+      "loss": 0.0698,
+      "step": 3892
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.0535e-05,
+      "loss": 0.055,
+      "step": 3893
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.053e-05,
+      "loss": 0.14,
+      "step": 3894
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.0525e-05,
+      "loss": 0.0687,
+      "step": 3895
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.052e-05,
+      "loss": 0.2639,
+      "step": 3896
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.0515e-05,
+      "loss": 0.1935,
+      "step": 3897
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.0510000000000003e-05,
+      "loss": 0.0829,
+      "step": 3898
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.0505000000000006e-05,
+      "loss": 0.2527,
+      "step": 3899
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.05e-05,
+      "loss": 0.0867,
+      "step": 3900
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.0495000000000004e-05,
+      "loss": 0.1932,
+      "step": 3901
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.049e-05,
+      "loss": 0.1544,
+      "step": 3902
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.0484999999999996e-05,
+      "loss": 0.1428,
+      "step": 3903
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.048e-05,
+      "loss": 0.0818,
+      "step": 3904
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.0475e-05,
+      "loss": 0.1046,
+      "step": 3905
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.047e-05,
+      "loss": 0.1761,
+      "step": 3906
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.0465e-05,
+      "loss": 0.1925,
+      "step": 3907
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.046e-05,
+      "loss": 0.1894,
+      "step": 3908
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.0455000000000005e-05,
+      "loss": 0.2579,
+      "step": 3909
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.045e-05,
+      "loss": 0.1643,
+      "step": 3910
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.0445e-05,
+      "loss": 0.1944,
+      "step": 3911
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.0440000000000006e-05,
+      "loss": 0.09,
+      "step": 3912
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.0435e-05,
+      "loss": 0.2366,
+      "step": 3913
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.043e-05,
+      "loss": 0.3812,
+      "step": 3914
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.0425e-05,
+      "loss": 0.1463,
+      "step": 3915
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.0419999999999996e-05,
+      "loss": 0.073,
+      "step": 3916
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.0415e-05,
+      "loss": 0.2501,
+      "step": 3917
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.041e-05,
+      "loss": 0.1144,
+      "step": 3918
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.0405000000000004e-05,
+      "loss": 0.1885,
+      "step": 3919
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.04e-05,
+      "loss": 0.1042,
+      "step": 3920
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.0395e-05,
+      "loss": 0.2,
+      "step": 3921
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.0390000000000005e-05,
+      "loss": 0.2782,
+      "step": 3922
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.0385e-05,
+      "loss": 0.1513,
+      "step": 3923
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.038e-05,
+      "loss": 0.0516,
+      "step": 3924
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.0375000000000006e-05,
+      "loss": 0.2625,
+      "step": 3925
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.0369999999999995e-05,
+      "loss": 0.0568,
+      "step": 3926
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.0365e-05,
+      "loss": 0.0854,
+      "step": 3927
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.036e-05,
+      "loss": 0.0849,
+      "step": 3928
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.0355e-05,
+      "loss": 0.0807,
+      "step": 3929
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.035e-05,
+      "loss": 0.1076,
+      "step": 3930
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.0345e-05,
+      "loss": 0.1862,
+      "step": 3931
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.0340000000000003e-05,
+      "loss": 0.081,
+      "step": 3932
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.0335e-05,
+      "loss": 0.1489,
+      "step": 3933
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.033e-05,
+      "loss": 0.2935,
+      "step": 3934
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.0325000000000004e-05,
+      "loss": 0.1126,
+      "step": 3935
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.032e-05,
+      "loss": 0.138,
+      "step": 3936
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.0315e-05,
+      "loss": 0.0454,
+      "step": 3937
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.031e-05,
+      "loss": 0.0799,
+      "step": 3938
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.0305e-05,
+      "loss": 0.0562,
+      "step": 3939
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.03e-05,
+      "loss": 0.0863,
+      "step": 3940
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.0295e-05,
+      "loss": 0.2023,
+      "step": 3941
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.029e-05,
+      "loss": 0.2722,
+      "step": 3942
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.0285e-05,
+      "loss": 0.0704,
+      "step": 3943
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.028e-05,
+      "loss": 0.1388,
+      "step": 3944
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.0275e-05,
+      "loss": 0.146,
+      "step": 3945
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.027e-05,
+      "loss": 0.1815,
+      "step": 3946
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.0265e-05,
+      "loss": 0.1124,
+      "step": 3947
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.0260000000000004e-05,
+      "loss": 0.159,
+      "step": 3948
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.025500000000001e-05,
+      "loss": 0.1734,
+      "step": 3949
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.0249999999999996e-05,
+      "loss": 0.066,
+      "step": 3950
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.0245e-05,
+      "loss": 0.0668,
+      "step": 3951
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.024e-05,
+      "loss": 0.0562,
+      "step": 3952
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.0235e-05,
+      "loss": 0.072,
+      "step": 3953
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.023e-05,
+      "loss": 0.0804,
+      "step": 3954
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.0225e-05,
+      "loss": 0.1941,
+      "step": 3955
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.0220000000000005e-05,
+      "loss": 0.0884,
+      "step": 3956
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 4.0215e-05,
+      "loss": 0.2978,
+      "step": 3957
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 4.021e-05,
+      "loss": 0.1141,
+      "step": 3958
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 4.0205000000000006e-05,
+      "loss": 0.052,
+      "step": 3959
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 4.02e-05,
+      "loss": 0.2635,
+      "step": 3960
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 4.0195000000000004e-05,
+      "loss": 0.3205,
+      "step": 3961
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 4.019e-05,
+      "loss": 0.0624,
+      "step": 3962
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 4.0184999999999996e-05,
+      "loss": 0.1761,
+      "step": 3963
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 4.018e-05,
+      "loss": 0.2186,
+      "step": 3964
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 4.0175e-05,
+      "loss": 0.063,
+      "step": 3965
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 4.017e-05,
+      "loss": 0.0856,
+      "step": 3966
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 4.0165e-05,
+      "loss": 0.073,
+      "step": 3967
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 4.016e-05,
+      "loss": 0.2166,
+      "step": 3968
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 4.0155000000000004e-05,
+      "loss": 0.202,
+      "step": 3969
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 4.015e-05,
+      "loss": 0.0622,
+      "step": 3970
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 4.0145e-05,
+      "loss": 0.0879,
+      "step": 3971
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 4.0140000000000005e-05,
+      "loss": 0.0508,
+      "step": 3972
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 4.0135e-05,
+      "loss": 0.1494,
+      "step": 3973
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 4.013e-05,
+      "loss": 0.2006,
+      "step": 3974
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 4.0125e-05,
+      "loss": 0.1928,
+      "step": 3975
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 4.012e-05,
+      "loss": 0.1764,
+      "step": 3976
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 4.0115e-05,
+      "loss": 0.1951,
+      "step": 3977
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 4.011e-05,
+      "loss": 0.1619,
+      "step": 3978
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 4.0105e-05,
+      "loss": 0.0913,
+      "step": 3979
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 4.01e-05,
+      "loss": 0.1401,
+      "step": 3980
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 4.0095e-05,
+      "loss": 0.0748,
+      "step": 3981
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 4.0090000000000004e-05,
+      "loss": 0.123,
+      "step": 3982
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 4.0085e-05,
+      "loss": 0.2087,
+      "step": 3983
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 4.008e-05,
+      "loss": 0.1553,
+      "step": 3984
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 4.0075000000000005e-05,
+      "loss": 0.2225,
+      "step": 3985
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 4.007e-05,
+      "loss": 0.1559,
+      "step": 3986
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 4.0065e-05,
+      "loss": 0.07,
+      "step": 3987
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 4.006e-05,
+      "loss": 0.3,
+      "step": 3988
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 4.0055e-05,
+      "loss": 0.272,
+      "step": 3989
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 4.005e-05,
+      "loss": 0.1897,
+      "step": 3990
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 4.0045e-05,
+      "loss": 0.0919,
+      "step": 3991
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 4.004e-05,
+      "loss": 0.2099,
+      "step": 3992
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 4.0035e-05,
+      "loss": 0.0594,
+      "step": 3993
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 4.003e-05,
+      "loss": 0.152,
+      "step": 3994
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 4.0025000000000004e-05,
+      "loss": 0.0925,
+      "step": 3995
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 4.0020000000000006e-05,
+      "loss": 0.0978,
+      "step": 3996
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 4.0015e-05,
+      "loss": 0.2605,
+      "step": 3997
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 4.001e-05,
+      "loss": 0.2279,
+      "step": 3998
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 4.0005e-05,
+      "loss": 0.0854,
+      "step": 3999
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 0.2117,
+      "step": 4000
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.9995e-05,
+      "loss": 0.1036,
+      "step": 4001
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.999e-05,
+      "loss": 0.0588,
+      "step": 4002
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.9985e-05,
+      "loss": 0.1442,
+      "step": 4003
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.998e-05,
+      "loss": 0.3124,
+      "step": 4004
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.9975e-05,
+      "loss": 0.1273,
+      "step": 4005
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.9970000000000005e-05,
+      "loss": 0.1036,
+      "step": 4006
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.9965e-05,
+      "loss": 0.1685,
+      "step": 4007
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.9960000000000004e-05,
+      "loss": 0.161,
+      "step": 4008
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.9955000000000006e-05,
+      "loss": 0.2678,
+      "step": 4009
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.995e-05,
+      "loss": 0.3189,
+      "step": 4010
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.9945e-05,
+      "loss": 0.1193,
+      "step": 4011
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.994e-05,
+      "loss": 0.152,
+      "step": 4012
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.9934999999999996e-05,
+      "loss": 0.2065,
+      "step": 4013
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.993e-05,
+      "loss": 0.1847,
+      "step": 4014
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.9925e-05,
+      "loss": 0.0662,
+      "step": 4015
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.9920000000000004e-05,
+      "loss": 0.208,
+      "step": 4016
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.9915e-05,
+      "loss": 0.0585,
+      "step": 4017
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.991e-05,
+      "loss": 0.2308,
+      "step": 4018
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.9905000000000005e-05,
+      "loss": 0.1654,
+      "step": 4019
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.99e-05,
+      "loss": 0.0811,
+      "step": 4020
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.9895000000000003e-05,
+      "loss": 0.1372,
+      "step": 4021
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.9890000000000006e-05,
+      "loss": 0.2258,
+      "step": 4022
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.9884999999999995e-05,
+      "loss": 0.1981,
+      "step": 4023
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.988e-05,
+      "loss": 0.0776,
+      "step": 4024
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.9875e-05,
+      "loss": 0.1602,
+      "step": 4025
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.987e-05,
+      "loss": 0.0754,
+      "step": 4026
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.9865e-05,
+      "loss": 0.0935,
+      "step": 4027
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.986e-05,
+      "loss": 0.1002,
+      "step": 4028
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.9855000000000004e-05,
+      "loss": 0.108,
+      "step": 4029
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.985e-05,
+      "loss": 0.1004,
+      "step": 4030
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.9845e-05,
+      "loss": 0.2782,
+      "step": 4031
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.9840000000000005e-05,
+      "loss": 0.1431,
+      "step": 4032
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.9835e-05,
+      "loss": 0.089,
+      "step": 4033
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.983e-05,
+      "loss": 0.1227,
+      "step": 4034
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.9825e-05,
+      "loss": 0.1982,
+      "step": 4035
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.982e-05,
+      "loss": 0.179,
+      "step": 4036
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.9815e-05,
+      "loss": 0.1684,
+      "step": 4037
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.981e-05,
+      "loss": 0.0913,
+      "step": 4038
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.9805e-05,
+      "loss": 0.2427,
+      "step": 4039
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.98e-05,
+      "loss": 0.1646,
+      "step": 4040
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.9795e-05,
+      "loss": 0.2968,
+      "step": 4041
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.9790000000000004e-05,
+      "loss": 0.199,
+      "step": 4042
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.9785e-05,
+      "loss": 0.0689,
+      "step": 4043
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.978e-05,
+      "loss": 0.0851,
+      "step": 4044
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.9775000000000005e-05,
+      "loss": 0.2196,
+      "step": 4045
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.977000000000001e-05,
+      "loss": 0.2091,
+      "step": 4046
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.9764999999999996e-05,
+      "loss": 0.2108,
+      "step": 4047
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.976e-05,
+      "loss": 0.1042,
+      "step": 4048
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.9755e-05,
+      "loss": 0.0707,
+      "step": 4049
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.975e-05,
+      "loss": 0.1641,
+      "step": 4050
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.9745e-05,
+      "loss": 0.2429,
+      "step": 4051
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.974e-05,
+      "loss": 0.097,
+      "step": 4052
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.9735e-05,
+      "loss": 0.0626,
+      "step": 4053
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.973e-05,
+      "loss": 0.0642,
+      "step": 4054
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.9725e-05,
+      "loss": 0.3002,
+      "step": 4055
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.9720000000000006e-05,
+      "loss": 0.4256,
+      "step": 4056
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.9715e-05,
+      "loss": 0.2042,
+      "step": 4057
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.9710000000000004e-05,
+      "loss": 0.0827,
+      "step": 4058
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.9705e-05,
+      "loss": 0.1064,
+      "step": 4059
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.9699999999999996e-05,
+      "loss": 0.0897,
+      "step": 4060
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.9695e-05,
+      "loss": 0.1934,
+      "step": 4061
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.969e-05,
+      "loss": 0.0665,
+      "step": 4062
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.9685e-05,
+      "loss": 0.0618,
+      "step": 4063
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.968e-05,
+      "loss": 0.1105,
+      "step": 4064
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.9675e-05,
+      "loss": 0.0515,
+      "step": 4065
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.9670000000000005e-05,
+      "loss": 0.0768,
+      "step": 4066
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.9665e-05,
+      "loss": 0.2367,
+      "step": 4067
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.966e-05,
+      "loss": 0.0941,
+      "step": 4068
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.9655000000000006e-05,
+      "loss": 0.2029,
+      "step": 4069
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.965e-05,
+      "loss": 0.2227,
+      "step": 4070
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.9645e-05,
+      "loss": 0.2566,
+      "step": 4071
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.964e-05,
+      "loss": 0.2458,
+      "step": 4072
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.9634999999999996e-05,
+      "loss": 0.0832,
+      "step": 4073
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.963e-05,
+      "loss": 0.0715,
+      "step": 4074
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.9625e-05,
+      "loss": 0.0672,
+      "step": 4075
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.9620000000000004e-05,
+      "loss": 0.156,
+      "step": 4076
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.9615e-05,
+      "loss": 0.2296,
+      "step": 4077
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.961e-05,
+      "loss": 0.2118,
+      "step": 4078
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.9605000000000005e-05,
+      "loss": 0.0528,
+      "step": 4079
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.96e-05,
+      "loss": 0.1134,
+      "step": 4080
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.9595e-05,
+      "loss": 0.1809,
+      "step": 4081
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.9590000000000006e-05,
+      "loss": 0.2987,
+      "step": 4082
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.9584999999999995e-05,
+      "loss": 0.0949,
+      "step": 4083
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.958e-05,
+      "loss": 0.2262,
+      "step": 4084
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.9575e-05,
+      "loss": 0.137,
+      "step": 4085
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.957e-05,
+      "loss": 0.174,
+      "step": 4086
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.9565e-05,
+      "loss": 0.0476,
+      "step": 4087
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.956e-05,
+      "loss": 0.0983,
+      "step": 4088
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.9555e-05,
+      "loss": 0.0579,
+      "step": 4089
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.955e-05,
+      "loss": 0.171,
+      "step": 4090
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.9545e-05,
+      "loss": 0.2617,
+      "step": 4091
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.9540000000000004e-05,
+      "loss": 0.1938,
+      "step": 4092
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.9535e-05,
+      "loss": 0.1673,
+      "step": 4093
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.953e-05,
+      "loss": 0.1973,
+      "step": 4094
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.9525e-05,
+      "loss": 0.363,
+      "step": 4095
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.952e-05,
+      "loss": 0.0902,
+      "step": 4096
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.9515e-05,
+      "loss": 0.0476,
+      "step": 4097
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.951e-05,
+      "loss": 0.0693,
+      "step": 4098
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.9505e-05,
+      "loss": 0.0618,
+      "step": 4099
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.95e-05,
+      "loss": 0.1104,
+      "step": 4100
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.9495e-05,
+      "loss": 0.1876,
+      "step": 4101
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.949e-05,
+      "loss": 0.1426,
+      "step": 4102
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.9485e-05,
+      "loss": 0.0585,
+      "step": 4103
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.948e-05,
+      "loss": 0.2776,
+      "step": 4104
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.9475000000000004e-05,
+      "loss": 0.1868,
+      "step": 4105
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.947000000000001e-05,
+      "loss": 0.1692,
+      "step": 4106
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.9464999999999996e-05,
+      "loss": 0.0889,
+      "step": 4107
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.946e-05,
+      "loss": 0.1472,
+      "step": 4108
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.9455e-05,
+      "loss": 0.0843,
+      "step": 4109
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.945e-05,
+      "loss": 0.2839,
+      "step": 4110
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.9445e-05,
+      "loss": 0.2318,
+      "step": 4111
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.944e-05,
+      "loss": 0.0595,
+      "step": 4112
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.9435e-05,
+      "loss": 0.1663,
+      "step": 4113
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.943e-05,
+      "loss": 0.2155,
+      "step": 4114
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.9425e-05,
+      "loss": 0.1154,
+      "step": 4115
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.9420000000000005e-05,
+      "loss": 0.0563,
+      "step": 4116
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.9415e-05,
+      "loss": 0.2112,
+      "step": 4117
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.9410000000000004e-05,
+      "loss": 0.1441,
+      "step": 4118
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.9405e-05,
+      "loss": 0.2266,
+      "step": 4119
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.9399999999999995e-05,
+      "loss": 0.0711,
+      "step": 4120
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.9395e-05,
+      "loss": 0.0633,
+      "step": 4121
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.939e-05,
+      "loss": 0.0677,
+      "step": 4122
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.9385e-05,
+      "loss": 0.0841,
+      "step": 4123
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.938e-05,
+      "loss": 0.2023,
+      "step": 4124
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.9375e-05,
+      "loss": 0.153,
+      "step": 4125
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.9370000000000004e-05,
+      "loss": 0.2206,
+      "step": 4126
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.9365e-05,
+      "loss": 0.2167,
+      "step": 4127
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.936e-05,
+      "loss": 0.128,
+      "step": 4128
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.9355000000000005e-05,
+      "loss": 0.1726,
+      "step": 4129
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.935e-05,
+      "loss": 0.0876,
+      "step": 4130
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.9345000000000004e-05,
+      "loss": 0.2994,
+      "step": 4131
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.934e-05,
+      "loss": 0.1112,
+      "step": 4132
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.9335e-05,
+      "loss": 0.1474,
+      "step": 4133
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.933e-05,
+      "loss": 0.0501,
+      "step": 4134
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.9325e-05,
+      "loss": 0.1981,
+      "step": 4135
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.932e-05,
+      "loss": 0.1735,
+      "step": 4136
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.9315e-05,
+      "loss": 0.2261,
+      "step": 4137
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.931e-05,
+      "loss": 0.2479,
+      "step": 4138
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.9305000000000004e-05,
+      "loss": 0.1778,
+      "step": 4139
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.93e-05,
+      "loss": 0.0666,
+      "step": 4140
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.9295e-05,
+      "loss": 0.1033,
+      "step": 4141
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.9290000000000005e-05,
+      "loss": 0.1021,
+      "step": 4142
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.928500000000001e-05,
+      "loss": 0.0699,
+      "step": 4143
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.928e-05,
+      "loss": 0.0834,
+      "step": 4144
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.9275e-05,
+      "loss": 0.0902,
+      "step": 4145
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.927e-05,
+      "loss": 0.0826,
+      "step": 4146
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.9265e-05,
+      "loss": 0.2397,
+      "step": 4147
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.926e-05,
+      "loss": 0.0891,
+      "step": 4148
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.9255e-05,
+      "loss": 0.2271,
+      "step": 4149
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.925e-05,
+      "loss": 0.2071,
+      "step": 4150
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.9245e-05,
+      "loss": 0.1639,
+      "step": 4151
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.9240000000000004e-05,
+      "loss": 0.1901,
+      "step": 4152
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.9235000000000006e-05,
+      "loss": 0.1358,
+      "step": 4153
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.923e-05,
+      "loss": 0.1463,
+      "step": 4154
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.9225000000000005e-05,
+      "loss": 0.1649,
+      "step": 4155
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.922e-05,
+      "loss": 0.2971,
+      "step": 4156
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.9214999999999996e-05,
+      "loss": 0.2134,
+      "step": 4157
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.921e-05,
+      "loss": 0.1539,
+      "step": 4158
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.9205e-05,
+      "loss": 0.1116,
+      "step": 4159
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.92e-05,
+      "loss": 0.3053,
+      "step": 4160
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.9195e-05,
+      "loss": 0.1761,
+      "step": 4161
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.919e-05,
+      "loss": 0.0948,
+      "step": 4162
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.9185000000000005e-05,
+      "loss": 0.2282,
+      "step": 4163
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.918e-05,
+      "loss": 0.1062,
+      "step": 4164
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.9175000000000004e-05,
+      "loss": 0.1285,
+      "step": 4165
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.9170000000000006e-05,
+      "loss": 0.0813,
+      "step": 4166
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.9165e-05,
+      "loss": 0.1625,
+      "step": 4167
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.916e-05,
+      "loss": 0.168,
+      "step": 4168
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.9155e-05,
+      "loss": 0.2134,
+      "step": 4169
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.9149999999999996e-05,
+      "loss": 0.1244,
+      "step": 4170
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.9145e-05,
+      "loss": 0.1358,
+      "step": 4171
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.914e-05,
+      "loss": 0.076,
+      "step": 4172
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.9135000000000004e-05,
+      "loss": 0.2822,
+      "step": 4173
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.913e-05,
+      "loss": 0.1823,
+      "step": 4174
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.9125e-05,
+      "loss": 0.0857,
+      "step": 4175
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.9120000000000005e-05,
+      "loss": 0.0715,
+      "step": 4176
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.9115e-05,
+      "loss": 0.1663,
+      "step": 4177
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.911e-05,
+      "loss": 0.1963,
+      "step": 4178
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.9105000000000006e-05,
+      "loss": 0.1828,
+      "step": 4179
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.9099999999999995e-05,
+      "loss": 0.1594,
+      "step": 4180
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.9095e-05,
+      "loss": 0.0506,
+      "step": 4181
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.909e-05,
+      "loss": 0.2018,
+      "step": 4182
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.9085e-05,
+      "loss": 0.2017,
+      "step": 4183
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.908e-05,
+      "loss": 0.1786,
+      "step": 4184
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.9075e-05,
+      "loss": 0.217,
+      "step": 4185
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.9070000000000004e-05,
+      "loss": 0.0746,
+      "step": 4186
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.9065e-05,
+      "loss": 0.274,
+      "step": 4187
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.906e-05,
+      "loss": 0.2091,
+      "step": 4188
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.9055000000000005e-05,
+      "loss": 0.2646,
+      "step": 4189
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.905e-05,
+      "loss": 0.0833,
+      "step": 4190
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.9045e-05,
+      "loss": 0.084,
+      "step": 4191
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.904e-05,
+      "loss": 0.0902,
+      "step": 4192
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.9035e-05,
+      "loss": 0.1816,
+      "step": 4193
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.903e-05,
+      "loss": 0.0819,
+      "step": 4194
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.9025e-05,
+      "loss": 0.2811,
+      "step": 4195
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.902e-05,
+      "loss": 0.0558,
+      "step": 4196
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.9015e-05,
+      "loss": 0.183,
+      "step": 4197
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.901e-05,
+      "loss": 0.0694,
+      "step": 4198
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.9005000000000003e-05,
+      "loss": 0.1268,
+      "step": 4199
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.9e-05,
+      "loss": 0.1048,
+      "step": 4200
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.8995e-05,
+      "loss": 0.2514,
+      "step": 4201
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.8990000000000004e-05,
+      "loss": 0.0668,
+      "step": 4202
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.898500000000001e-05,
+      "loss": 0.1501,
+      "step": 4203
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.8979999999999996e-05,
+      "loss": 0.1583,
+      "step": 4204
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.8975e-05,
+      "loss": 0.2693,
+      "step": 4205
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.897e-05,
+      "loss": 0.2028,
+      "step": 4206
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.8965e-05,
+      "loss": 0.1593,
+      "step": 4207
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.896e-05,
+      "loss": 0.0725,
+      "step": 4208
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.8955e-05,
+      "loss": 0.0769,
+      "step": 4209
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.895e-05,
+      "loss": 0.1161,
+      "step": 4210
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.8945e-05,
+      "loss": 0.1377,
+      "step": 4211
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.894e-05,
+      "loss": 0.0802,
+      "step": 4212
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.8935000000000006e-05,
+      "loss": 0.2476,
+      "step": 4213
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.893e-05,
+      "loss": 0.2026,
+      "step": 4214
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.8925000000000004e-05,
+      "loss": 0.0878,
+      "step": 4215
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.892e-05,
+      "loss": 0.0664,
+      "step": 4216
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.8914999999999996e-05,
+      "loss": 0.1525,
+      "step": 4217
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.891e-05,
+      "loss": 0.0649,
+      "step": 4218
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.8905e-05,
+      "loss": 0.0841,
+      "step": 4219
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.89e-05,
+      "loss": 0.0814,
+      "step": 4220
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.8895e-05,
+      "loss": 0.2451,
+      "step": 4221
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.889e-05,
+      "loss": 0.1744,
+      "step": 4222
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.8885000000000005e-05,
+      "loss": 0.0788,
+      "step": 4223
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.888e-05,
+      "loss": 0.3091,
+      "step": 4224
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.8875e-05,
+      "loss": 0.1637,
+      "step": 4225
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.8870000000000006e-05,
+      "loss": 0.1759,
+      "step": 4226
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.8865e-05,
+      "loss": 0.2092,
+      "step": 4227
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.886e-05,
+      "loss": 0.062,
+      "step": 4228
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.8855e-05,
+      "loss": 0.0731,
+      "step": 4229
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.8849999999999996e-05,
+      "loss": 0.0828,
+      "step": 4230
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.8845e-05,
+      "loss": 0.1353,
+      "step": 4231
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.884e-05,
+      "loss": 0.1155,
+      "step": 4232
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.8835000000000003e-05,
+      "loss": 0.1796,
+      "step": 4233
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.883e-05,
+      "loss": 0.2798,
+      "step": 4234
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.8825e-05,
+      "loss": 0.2292,
+      "step": 4235
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.8820000000000004e-05,
+      "loss": 0.1455,
+      "step": 4236
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.8815e-05,
+      "loss": 0.1644,
+      "step": 4237
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.881e-05,
+      "loss": 0.0704,
+      "step": 4238
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.8805000000000005e-05,
+      "loss": 0.1651,
+      "step": 4239
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.8799999999999994e-05,
+      "loss": 0.1824,
+      "step": 4240
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.8795e-05,
+      "loss": 0.0672,
+      "step": 4241
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.879e-05,
+      "loss": 0.1577,
+      "step": 4242
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.8785e-05,
+      "loss": 0.2018,
+      "step": 4243
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.878e-05,
+      "loss": 0.1472,
+      "step": 4244
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.8775e-05,
+      "loss": 0.0623,
+      "step": 4245
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.877e-05,
+      "loss": 0.2529,
+      "step": 4246
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.8765e-05,
+      "loss": 0.124,
+      "step": 4247
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.876e-05,
+      "loss": 0.3216,
+      "step": 4248
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.8755000000000004e-05,
+      "loss": 0.3152,
+      "step": 4249
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.875e-05,
+      "loss": 0.147,
+      "step": 4250
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.8745e-05,
+      "loss": 0.15,
+      "step": 4251
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.8740000000000005e-05,
+      "loss": 0.2605,
+      "step": 4252
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.8735e-05,
+      "loss": 0.1354,
+      "step": 4253
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.873e-05,
+      "loss": 0.1284,
+      "step": 4254
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.8725e-05,
+      "loss": 0.1179,
+      "step": 4255
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.872e-05,
+      "loss": 0.2803,
+      "step": 4256
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.8715e-05,
+      "loss": 0.1888,
+      "step": 4257
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.871e-05,
+      "loss": 0.106,
+      "step": 4258
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.8705e-05,
+      "loss": 0.1032,
+      "step": 4259
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.87e-05,
+      "loss": 0.1549,
+      "step": 4260
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.8695e-05,
+      "loss": 0.0733,
+      "step": 4261
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.8690000000000004e-05,
+      "loss": 0.0703,
+      "step": 4262
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.8685000000000007e-05,
+      "loss": 0.2245,
+      "step": 4263
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.868e-05,
+      "loss": 0.1018,
+      "step": 4264
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.8675e-05,
+      "loss": 0.111,
+      "step": 4265
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.867e-05,
+      "loss": 0.0553,
+      "step": 4266
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.8664999999999997e-05,
+      "loss": 0.1771,
+      "step": 4267
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.866e-05,
+      "loss": 0.2378,
+      "step": 4268
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.8655e-05,
+      "loss": 0.0689,
+      "step": 4269
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.865e-05,
+      "loss": 0.3576,
+      "step": 4270
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.8645e-05,
+      "loss": 0.2156,
+      "step": 4271
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.864e-05,
+      "loss": 0.1976,
+      "step": 4272
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.8635000000000005e-05,
+      "loss": 0.0961,
+      "step": 4273
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.863e-05,
+      "loss": 0.0546,
+      "step": 4274
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.8625000000000004e-05,
+      "loss": 0.0626,
+      "step": 4275
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.8620000000000006e-05,
+      "loss": 0.2261,
+      "step": 4276
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.8614999999999995e-05,
+      "loss": 0.2992,
+      "step": 4277
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.861e-05,
+      "loss": 0.1982,
+      "step": 4278
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.8605e-05,
+      "loss": 0.206,
+      "step": 4279
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.8599999999999996e-05,
+      "loss": 0.0671,
+      "step": 4280
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.8595e-05,
+      "loss": 0.0876,
+      "step": 4281
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.859e-05,
+      "loss": 0.0698,
+      "step": 4282
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.8585000000000004e-05,
+      "loss": 0.1967,
+      "step": 4283
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.858e-05,
+      "loss": 0.2641,
+      "step": 4284
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.8575e-05,
+      "loss": 0.0797,
+      "step": 4285
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.8570000000000005e-05,
+      "loss": 0.0818,
+      "step": 4286
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.8565e-05,
+      "loss": 0.1254,
+      "step": 4287
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.8560000000000004e-05,
+      "loss": 0.279,
+      "step": 4288
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.8555e-05,
+      "loss": 0.1514,
+      "step": 4289
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.855e-05,
+      "loss": 0.1109,
+      "step": 4290
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.8545e-05,
+      "loss": 0.097,
+      "step": 4291
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.854e-05,
+      "loss": 0.1378,
+      "step": 4292
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.8535e-05,
+      "loss": 0.2644,
+      "step": 4293
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.853e-05,
+      "loss": 0.2045,
+      "step": 4294
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.8525e-05,
+      "loss": 0.1817,
+      "step": 4295
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.8520000000000004e-05,
+      "loss": 0.2531,
+      "step": 4296
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.8515e-05,
+      "loss": 0.1141,
+      "step": 4297
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.851e-05,
+      "loss": 0.1708,
+      "step": 4298
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.8505000000000005e-05,
+      "loss": 0.0856,
+      "step": 4299
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.850000000000001e-05,
+      "loss": 0.1204,
+      "step": 4300
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.8494999999999997e-05,
+      "loss": 0.1183,
+      "step": 4301
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.849e-05,
+      "loss": 0.0747,
+      "step": 4302
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.8485e-05,
+      "loss": 0.0686,
+      "step": 4303
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.848e-05,
+      "loss": 0.1673,
+      "step": 4304
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.8475e-05,
+      "loss": 0.0982,
+      "step": 4305
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.847e-05,
+      "loss": 0.1718,
+      "step": 4306
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.8465e-05,
+      "loss": 0.1067,
+      "step": 4307
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.846e-05,
+      "loss": 0.0555,
+      "step": 4308
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.8455000000000004e-05,
+      "loss": 0.0577,
+      "step": 4309
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.8450000000000006e-05,
+      "loss": 0.2183,
+      "step": 4310
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.8445e-05,
+      "loss": 0.1523,
+      "step": 4311
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.8440000000000005e-05,
+      "loss": 0.2564,
+      "step": 4312
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.8435e-05,
+      "loss": 0.2602,
+      "step": 4313
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.8429999999999996e-05,
+      "loss": 0.0967,
+      "step": 4314
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.8425e-05,
+      "loss": 0.1457,
+      "step": 4315
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.842e-05,
+      "loss": 0.0856,
+      "step": 4316
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.8415e-05,
+      "loss": 0.084,
+      "step": 4317
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.841e-05,
+      "loss": 0.0814,
+      "step": 4318
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.8405e-05,
+      "loss": 0.2276,
+      "step": 4319
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 0.2642,
+      "step": 4320
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.8395e-05,
+      "loss": 0.0469,
+      "step": 4321
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.8390000000000003e-05,
+      "loss": 0.0462,
+      "step": 4322
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.8385000000000006e-05,
+      "loss": 0.2579,
+      "step": 4323
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 3.838e-05,
+      "loss": 0.0799,
+      "step": 4324
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 3.8375e-05,
+      "loss": 0.0731,
+      "step": 4325
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 3.837e-05,
+      "loss": 0.1851,
+      "step": 4326
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 3.8364999999999996e-05,
+      "loss": 0.3406,
+      "step": 4327
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 3.836e-05,
+      "loss": 0.1115,
+      "step": 4328
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 3.8355e-05,
+      "loss": 0.0766,
+      "step": 4329
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 3.8350000000000004e-05,
+      "loss": 0.2182,
+      "step": 4330
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 3.8345e-05,
+      "loss": 0.211,
+      "step": 4331
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 3.834e-05,
+      "loss": 0.3301,
+      "step": 4332
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 3.8335000000000005e-05,
+      "loss": 0.1731,
+      "step": 4333
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 3.833e-05,
+      "loss": 0.2134,
+      "step": 4334
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 3.8325e-05,
+      "loss": 0.0994,
+      "step": 4335
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 3.8320000000000006e-05,
+      "loss": 0.1205,
+      "step": 4336
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 3.8314999999999995e-05,
+      "loss": 0.1043,
+      "step": 4337
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 3.831e-05,
+      "loss": 0.1787,
+      "step": 4338
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 3.8305e-05,
+      "loss": 0.2178,
+      "step": 4339
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 3.83e-05,
+      "loss": 0.0666,
+      "step": 4340
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 3.8295e-05,
+      "loss": 0.1883,
+      "step": 4341
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 3.829e-05,
+      "loss": 0.2615,
+      "step": 4342
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 3.8285000000000004e-05,
+      "loss": 0.3605,
+      "step": 4343
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 3.828e-05,
+      "loss": 0.073,
+      "step": 4344
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 3.8275e-05,
+      "loss": 0.076,
+      "step": 4345
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 3.8270000000000005e-05,
+      "loss": 0.0606,
+      "step": 4346
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 3.8265e-05,
+      "loss": 0.1571,
+      "step": 4347
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 3.826e-05,
+      "loss": 0.0814,
+      "step": 4348
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 3.8255e-05,
+      "loss": 0.1217,
+      "step": 4349
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 3.825e-05,
+      "loss": 0.0789,
+      "step": 4350
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 3.8245e-05,
+      "loss": 0.1824,
+      "step": 4351
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 3.824e-05,
+      "loss": 0.2005,
+      "step": 4352
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 3.8235e-05,
+      "loss": 0.1472,
+      "step": 4353
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 3.823e-05,
+      "loss": 0.162,
+      "step": 4354
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 3.8225e-05,
+      "loss": 0.1465,
+      "step": 4355
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 3.822e-05,
+      "loss": 0.0852,
+      "step": 4356
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 3.8215e-05,
+      "loss": 0.2588,
+      "step": 4357
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 3.821e-05,
+      "loss": 0.2857,
+      "step": 4358
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 3.8205000000000004e-05,
+      "loss": 0.2338,
+      "step": 4359
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 3.820000000000001e-05,
+      "loss": 0.0785,
+      "step": 4360
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 3.8194999999999996e-05,
+      "loss": 0.188,
+      "step": 4361
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 3.819e-05,
+      "loss": 0.1005,
+      "step": 4362
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 3.8185e-05,
+      "loss": 0.0928,
+      "step": 4363
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 3.818e-05,
+      "loss": 0.0655,
+      "step": 4364
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 3.8175e-05,
+      "loss": 0.087,
+      "step": 4365
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 3.817e-05,
+      "loss": 0.0645,
+      "step": 4366
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 3.8165e-05,
+      "loss": 0.266,
+      "step": 4367
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 3.816e-05,
+      "loss": 0.3258,
+      "step": 4368
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 3.8155e-05,
+      "loss": 0.0766,
+      "step": 4369
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 3.8150000000000006e-05,
+      "loss": 0.1246,
+      "step": 4370
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 3.8145e-05,
+      "loss": 0.1492,
+      "step": 4371
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 3.8140000000000004e-05,
+      "loss": 0.2495,
+      "step": 4372
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 3.8135e-05,
+      "loss": 0.1506,
+      "step": 4373
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 3.8129999999999996e-05,
+      "loss": 0.0663,
+      "step": 4374
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 3.8125e-05,
+      "loss": 0.133,
+      "step": 4375
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 3.812e-05,
+      "loss": 0.2215,
+      "step": 4376
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 3.8115e-05,
+      "loss": 0.1975,
+      "step": 4377
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 3.811e-05,
+      "loss": 0.1847,
+      "step": 4378
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 3.8105e-05,
+      "loss": 0.1477,
+      "step": 4379
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 3.8100000000000005e-05,
+      "loss": 0.2874,
+      "step": 4380
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 3.8095e-05,
+      "loss": 0.1012,
+      "step": 4381
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 3.809e-05,
+      "loss": 0.0639,
+      "step": 4382
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 3.8085000000000006e-05,
+      "loss": 0.1804,
+      "step": 4383
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 3.808e-05,
+      "loss": 0.1693,
+      "step": 4384
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 3.8075000000000004e-05,
+      "loss": 0.1488,
+      "step": 4385
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 3.807e-05,
+      "loss": 0.2428,
+      "step": 4386
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 3.8064999999999996e-05,
+      "loss": 0.1553,
+      "step": 4387
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 3.806e-05,
+      "loss": 0.0574,
+      "step": 4388
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 3.8055e-05,
+      "loss": 0.0787,
+      "step": 4389
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 3.805e-05,
+      "loss": 0.1651,
+      "step": 4390
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 3.8045e-05,
+      "loss": 0.2163,
+      "step": 4391
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 3.804e-05,
+      "loss": 0.0688,
+      "step": 4392
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 3.8035000000000004e-05,
+      "loss": 0.1442,
+      "step": 4393
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 3.803e-05,
+      "loss": 0.1113,
+      "step": 4394
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 3.8025e-05,
+      "loss": 0.0682,
+      "step": 4395
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 3.8020000000000005e-05,
+      "loss": 0.3033,
+      "step": 4396
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 3.8015e-05,
+      "loss": 0.0886,
+      "step": 4397
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 3.801e-05,
+      "loss": 0.2071,
+      "step": 4398
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 3.8005e-05,
+      "loss": 0.2763,
+      "step": 4399
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 3.8e-05,
+      "loss": 0.2333,
+      "step": 4400
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 3.7995e-05,
+      "loss": 0.0829,
+      "step": 4401
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 3.799e-05,
+      "loss": 0.0645,
+      "step": 4402
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 3.7985e-05,
+      "loss": 0.2729,
+      "step": 4403
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 3.798e-05,
+      "loss": 0.0708,
+      "step": 4404
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 3.7975e-05,
+      "loss": 0.1709,
+      "step": 4405
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 3.7970000000000004e-05,
+      "loss": 0.1189,
+      "step": 4406
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 3.7965e-05,
+      "loss": 0.1314,
+      "step": 4407
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 3.796e-05,
+      "loss": 0.0725,
+      "step": 4408
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 3.7955000000000005e-05,
+      "loss": 0.0719,
+      "step": 4409
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 3.795e-05,
+      "loss": 0.2885,
+      "step": 4410
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 3.7945e-05,
+      "loss": 0.1245,
+      "step": 4411
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 3.794e-05,
+      "loss": 0.0529,
+      "step": 4412
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 3.7935e-05,
+      "loss": 0.0778,
+      "step": 4413
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 3.793e-05,
+      "loss": 0.1721,
+      "step": 4414
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 3.7925e-05,
+      "loss": 0.1351,
+      "step": 4415
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 3.792e-05,
+      "loss": 0.0977,
+      "step": 4416
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 3.7915e-05,
+      "loss": 0.0813,
+      "step": 4417
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 3.791e-05,
+      "loss": 0.2583,
+      "step": 4418
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 3.7905000000000004e-05,
+      "loss": 0.1772,
+      "step": 4419
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.7900000000000006e-05,
+      "loss": 0.1658,
+      "step": 4420
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.7895e-05,
+      "loss": 0.1322,
+      "step": 4421
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3.789e-05,
+      "loss": 0.1743,
+      "step": 4422
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3.7885e-05,
+      "loss": 0.1887,
+      "step": 4423
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3.7879999999999996e-05,
+      "loss": 0.0566,
+      "step": 4424
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3.7875e-05,
+      "loss": 0.2149,
+      "step": 4425
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3.787e-05,
+      "loss": 0.1075,
+      "step": 4426
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.7865e-05,
+      "loss": 0.0807,
+      "step": 4427
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.786e-05,
+      "loss": 0.1866,
+      "step": 4428
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.7855e-05,
+      "loss": 0.152,
+      "step": 4429
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3.7850000000000005e-05,
+      "loss": 0.1398,
+      "step": 4430
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3.7845e-05,
+      "loss": 0.0758,
+      "step": 4431
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3.7840000000000004e-05,
+      "loss": 0.1049,
+      "step": 4432
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3.7835000000000006e-05,
+      "loss": 0.2446,
+      "step": 4433
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.7829999999999995e-05,
+      "loss": 0.1447,
+      "step": 4434
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.7825e-05,
+      "loss": 0.1461,
+      "step": 4435
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.782e-05,
+      "loss": 0.1235,
+      "step": 4436
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.7814999999999996e-05,
+      "loss": 0.083,
+      "step": 4437
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.781e-05,
+      "loss": 0.233,
+      "step": 4438
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3.7805e-05,
+      "loss": 0.27,
+      "step": 4439
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 0.0817,
+      "step": 4440
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3.7795e-05,
+      "loss": 0.1137,
+      "step": 4441
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 3.779e-05,
+      "loss": 0.1061,
+      "step": 4442
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 3.7785000000000005e-05,
+      "loss": 0.2322,
+      "step": 4443
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3.778e-05,
+      "loss": 0.1066,
+      "step": 4444
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3.7775000000000003e-05,
+      "loss": 0.078,
+      "step": 4445
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3.777e-05,
+      "loss": 0.1428,
+      "step": 4446
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 3.7765e-05,
+      "loss": 0.0622,
+      "step": 4447
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 3.776e-05,
+      "loss": 0.0455,
+      "step": 4448
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 3.7755e-05,
+      "loss": 0.2424,
+      "step": 4449
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 3.775e-05,
+      "loss": 0.1779,
+      "step": 4450
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3.7745e-05,
+      "loss": 0.1393,
+      "step": 4451
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3.774e-05,
+      "loss": 0.2044,
+      "step": 4452
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3.7735000000000004e-05,
+      "loss": 0.0618,
+      "step": 4453
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 3.773e-05,
+      "loss": 0.0852,
+      "step": 4454
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 3.7725e-05,
+      "loss": 0.1604,
+      "step": 4455
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.7720000000000005e-05,
+      "loss": 0.1689,
+      "step": 4456
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.7715e-05,
+      "loss": 0.2368,
+      "step": 4457
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.7709999999999996e-05,
+      "loss": 0.0866,
+      "step": 4458
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3.7705e-05,
+      "loss": 0.1379,
+      "step": 4459
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3.77e-05,
+      "loss": 0.1761,
+      "step": 4460
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 3.7695e-05,
+      "loss": 0.1759,
+      "step": 4461
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 3.769e-05,
+      "loss": 0.0613,
+      "step": 4462
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.7685e-05,
+      "loss": 0.1595,
+      "step": 4463
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.768e-05,
+      "loss": 0.0557,
+      "step": 4464
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.7675e-05,
+      "loss": 0.0672,
+      "step": 4465
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 3.7670000000000004e-05,
+      "loss": 0.2023,
+      "step": 4466
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 3.7665000000000006e-05,
+      "loss": 0.2721,
+      "step": 4467
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.766e-05,
+      "loss": 0.205,
+      "step": 4468
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.7655000000000005e-05,
+      "loss": 0.0658,
+      "step": 4469
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.765e-05,
+      "loss": 0.1491,
+      "step": 4470
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 3.7644999999999996e-05,
+      "loss": 0.1286,
+      "step": 4471
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 3.764e-05,
+      "loss": 0.1633,
+      "step": 4472
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.7635e-05,
+      "loss": 0.0779,
+      "step": 4473
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.763e-05,
+      "loss": 0.2556,
+      "step": 4474
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 3.7625e-05,
+      "loss": 0.171,
+      "step": 4475
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 3.762e-05,
+      "loss": 0.2532,
+      "step": 4476
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 3.7615000000000005e-05,
+      "loss": 0.0566,
+      "step": 4477
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3.761e-05,
+      "loss": 0.0956,
+      "step": 4478
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3.7605e-05,
+      "loss": 0.194,
+      "step": 4479
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3.7600000000000006e-05,
+      "loss": 0.1745,
+      "step": 4480
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3.7595e-05,
+      "loss": 0.0494,
+      "step": 4481
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3.759e-05,
+      "loss": 0.126,
+      "step": 4482
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3.7585e-05,
+      "loss": 0.2363,
+      "step": 4483
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3.7579999999999996e-05,
+      "loss": 0.0779,
+      "step": 4484
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 3.7575e-05,
+      "loss": 0.1775,
+      "step": 4485
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 3.757e-05,
+      "loss": 0.2416,
+      "step": 4486
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3.7565000000000004e-05,
+      "loss": 0.0772,
+      "step": 4487
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3.756e-05,
+      "loss": 0.1288,
+      "step": 4488
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3.7555e-05,
+      "loss": 0.2366,
+      "step": 4489
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3.7550000000000005e-05,
+      "loss": 0.0685,
+      "step": 4490
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3.7545e-05,
+      "loss": 0.095,
+      "step": 4491
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 3.754e-05,
+      "loss": 0.1379,
+      "step": 4492
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 3.7535000000000006e-05,
+      "loss": 0.1521,
+      "step": 4493
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.7529999999999995e-05,
+      "loss": 0.1477,
+      "step": 4494
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.7525e-05,
+      "loss": 0.069,
+      "step": 4495
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.752e-05,
+      "loss": 0.185,
+      "step": 4496
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 3.7515e-05,
+      "loss": 0.1975,
+      "step": 4497
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 3.751e-05,
+      "loss": 0.0966,
+      "step": 4498
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.7505e-05,
+      "loss": 0.0888,
+      "step": 4499
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.157,
+      "step": 4500
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.7495e-05,
+      "loss": 0.0912,
+      "step": 4501
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 3.749e-05,
+      "loss": 0.0854,
+      "step": 4502
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 3.7485000000000004e-05,
+      "loss": 0.0713,
+      "step": 4503
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.748e-05,
+      "loss": 0.053,
+      "step": 4504
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.7475e-05,
+      "loss": 0.1646,
+      "step": 4505
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.7470000000000005e-05,
+      "loss": 0.1568,
+      "step": 4506
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3.7465e-05,
+      "loss": 0.057,
+      "step": 4507
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3.746e-05,
+      "loss": 0.0911,
+      "step": 4508
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 3.7455e-05,
+      "loss": 0.07,
+      "step": 4509
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 3.745e-05,
+      "loss": 0.1828,
+      "step": 4510
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3.7445e-05,
+      "loss": 0.3202,
+      "step": 4511
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3.744e-05,
+      "loss": 0.0831,
+      "step": 4512
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3.7435e-05,
+      "loss": 0.073,
+      "step": 4513
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 3.743e-05,
+      "loss": 0.0711,
+      "step": 4514
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 3.7425e-05,
+      "loss": 0.0852,
+      "step": 4515
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 3.7420000000000004e-05,
+      "loss": 0.1703,
+      "step": 4516
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 3.741500000000001e-05,
+      "loss": 0.2407,
+      "step": 4517
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 3.741e-05,
+      "loss": 0.074,
+      "step": 4518
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 3.7405e-05,
+      "loss": 0.1983,
+      "step": 4519
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 3.74e-05,
+      "loss": 0.2517,
+      "step": 4520
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 3.7395e-05,
+      "loss": 0.2354,
+      "step": 4521
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 3.739e-05,
+      "loss": 0.1298,
+      "step": 4522
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 3.7385e-05,
+      "loss": 0.0541,
+      "step": 4523
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 3.738e-05,
+      "loss": 0.1629,
+      "step": 4524
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 3.7375e-05,
+      "loss": 0.0594,
+      "step": 4525
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 3.737e-05,
+      "loss": 0.2905,
+      "step": 4526
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 3.7365000000000006e-05,
+      "loss": 0.1433,
+      "step": 4527
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 3.736e-05,
+      "loss": 0.1231,
+      "step": 4528
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 3.7355000000000004e-05,
+      "loss": 0.1974,
+      "step": 4529
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 3.735000000000001e-05,
+      "loss": 0.1409,
+      "step": 4530
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 3.7344999999999996e-05,
+      "loss": 0.1305,
+      "step": 4531
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 3.734e-05,
+      "loss": 0.2547,
+      "step": 4532
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 3.7335e-05,
+      "loss": 0.0717,
+      "step": 4533
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 3.733e-05,
+      "loss": 0.1588,
+      "step": 4534
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 3.7325e-05,
+      "loss": 0.176,
+      "step": 4535
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 3.732e-05,
+      "loss": 0.1459,
+      "step": 4536
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 3.7315000000000004e-05,
+      "loss": 0.4507,
+      "step": 4537
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 3.731e-05,
+      "loss": 0.0861,
+      "step": 4538
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 3.7305e-05,
+      "loss": 0.083,
+      "step": 4539
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 3.7300000000000005e-05,
+      "loss": 0.1657,
+      "step": 4540
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 3.7295e-05,
+      "loss": 0.1861,
+      "step": 4541
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3.7290000000000004e-05,
+      "loss": 0.2665,
+      "step": 4542
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3.7285e-05,
+      "loss": 0.1005,
+      "step": 4543
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3.7279999999999995e-05,
+      "loss": 0.2558,
+      "step": 4544
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 3.7275e-05,
+      "loss": 0.1426,
+      "step": 4545
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 3.727e-05,
+      "loss": 0.2298,
+      "step": 4546
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 3.7265e-05,
+      "loss": 0.0555,
+      "step": 4547
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 3.726e-05,
+      "loss": 0.2133,
+      "step": 4548
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 3.7255e-05,
+      "loss": 0.1151,
+      "step": 4549
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 3.7250000000000004e-05,
+      "loss": 0.1678,
+      "step": 4550
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 3.7245e-05,
+      "loss": 0.3258,
+      "step": 4551
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 3.724e-05,
+      "loss": 0.1308,
+      "step": 4552
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 3.7235000000000005e-05,
+      "loss": 0.1109,
+      "step": 4553
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3.723e-05,
+      "loss": 0.1699,
+      "step": 4554
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3.7225e-05,
+      "loss": 0.1424,
+      "step": 4555
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3.722e-05,
+      "loss": 0.102,
+      "step": 4556
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 3.7215e-05,
+      "loss": 0.1037,
+      "step": 4557
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 3.721e-05,
+      "loss": 0.1958,
+      "step": 4558
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 3.7205e-05,
+      "loss": 0.0747,
+      "step": 4559
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 3.72e-05,
+      "loss": 0.0619,
+      "step": 4560
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 3.7195e-05,
+      "loss": 0.0933,
+      "step": 4561
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 3.719e-05,
+      "loss": 0.064,
+      "step": 4562
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 3.7185000000000004e-05,
+      "loss": 0.269,
+      "step": 4563
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3.718e-05,
+      "loss": 0.1936,
+      "step": 4564
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3.7175e-05,
+      "loss": 0.1159,
+      "step": 4565
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3.7170000000000005e-05,
+      "loss": 0.1147,
+      "step": 4566
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 3.7165e-05,
+      "loss": 0.1045,
+      "step": 4567
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 3.7159999999999997e-05,
+      "loss": 0.0596,
+      "step": 4568
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 3.7155e-05,
+      "loss": 0.0893,
+      "step": 4569
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 3.715e-05,
+      "loss": 0.1506,
+      "step": 4570
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 3.7145e-05,
+      "loss": 0.1672,
+      "step": 4571
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 3.714e-05,
+      "loss": 0.1049,
+      "step": 4572
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 3.7135e-05,
+      "loss": 0.2106,
+      "step": 4573
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 3.713e-05,
+      "loss": 0.217,
+      "step": 4574
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 3.7125e-05,
+      "loss": 0.147,
+      "step": 4575
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 3.7120000000000004e-05,
+      "loss": 0.3559,
+      "step": 4576
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 3.7115000000000006e-05,
+      "loss": 0.0865,
+      "step": 4577
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 3.711e-05,
+      "loss": 0.2003,
+      "step": 4578
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 3.7105e-05,
+      "loss": 0.0593,
+      "step": 4579
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 3.71e-05,
+      "loss": 0.2576,
+      "step": 4580
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 3.7094999999999996e-05,
+      "loss": 0.1577,
+      "step": 4581
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 3.709e-05,
+      "loss": 0.1423,
+      "step": 4582
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 3.7085e-05,
+      "loss": 0.1705,
+      "step": 4583
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 3.708e-05,
+      "loss": 0.0717,
+      "step": 4584
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 3.7075e-05,
+      "loss": 0.1542,
+      "step": 4585
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 3.707e-05,
+      "loss": 0.2715,
+      "step": 4586
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 3.7065000000000005e-05,
+      "loss": 0.0683,
+      "step": 4587
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 3.706e-05,
+      "loss": 0.2029,
+      "step": 4588
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 3.7055000000000004e-05,
+      "loss": 0.1513,
+      "step": 4589
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 3.7050000000000006e-05,
+      "loss": 0.0491,
+      "step": 4590
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 3.7044999999999995e-05,
+      "loss": 0.2537,
+      "step": 4591
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 3.704e-05,
+      "loss": 0.139,
+      "step": 4592
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 3.7035e-05,
+      "loss": 0.1215,
+      "step": 4593
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 3.7029999999999996e-05,
+      "loss": 0.1369,
+      "step": 4594
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 3.7025e-05,
+      "loss": 0.0994,
+      "step": 4595
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 3.702e-05,
+      "loss": 0.1554,
+      "step": 4596
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 3.7015000000000004e-05,
+      "loss": 0.1663,
+      "step": 4597
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 3.701e-05,
+      "loss": 0.0566,
+      "step": 4598
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 3.7005e-05,
+      "loss": 0.0389,
+      "step": 4599
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 3.7000000000000005e-05,
+      "loss": 0.0588,
+      "step": 4600
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 3.6995e-05,
+      "loss": 0.1024,
+      "step": 4601
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 3.699e-05,
+      "loss": 0.0989,
+      "step": 4602
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 3.6985e-05,
+      "loss": 0.101,
+      "step": 4603
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 3.6979999999999995e-05,
+      "loss": 0.0576,
+      "step": 4604
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 3.6975e-05,
+      "loss": 0.1108,
+      "step": 4605
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 3.697e-05,
+      "loss": 0.2275,
+      "step": 4606
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 3.6965e-05,
+      "loss": 0.0528,
+      "step": 4607
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 3.696e-05,
+      "loss": 0.2219,
+      "step": 4608
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 3.6955e-05,
+      "loss": 0.0859,
+      "step": 4609
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 3.6950000000000004e-05,
+      "loss": 0.0928,
+      "step": 4610
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 3.6945e-05,
+      "loss": 0.1604,
+      "step": 4611
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 3.694e-05,
+      "loss": 0.1958,
+      "step": 4612
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 3.6935000000000005e-05,
+      "loss": 0.1357,
+      "step": 4613
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 3.693e-05,
+      "loss": 0.1509,
+      "step": 4614
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 3.6924999999999996e-05,
+      "loss": 0.1928,
+      "step": 4615
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 3.692e-05,
+      "loss": 0.1139,
+      "step": 4616
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 3.6915e-05,
+      "loss": 0.1236,
+      "step": 4617
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 3.691e-05,
+      "loss": 0.0791,
+      "step": 4618
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3.6905e-05,
+      "loss": 0.0597,
+      "step": 4619
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3.69e-05,
+      "loss": 0.054,
+      "step": 4620
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3.6895e-05,
+      "loss": 0.0652,
+      "step": 4621
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.689e-05,
+      "loss": 0.0694,
+      "step": 4622
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.6885000000000003e-05,
+      "loss": 0.3312,
+      "step": 4623
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.688e-05,
+      "loss": 0.0736,
+      "step": 4624
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.6875e-05,
+      "loss": 0.1625,
+      "step": 4625
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.6870000000000004e-05,
+      "loss": 0.2497,
+      "step": 4626
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.686500000000001e-05,
+      "loss": 0.2678,
+      "step": 4627
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.6859999999999996e-05,
+      "loss": 0.1781,
+      "step": 4628
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.6855e-05,
+      "loss": 0.1595,
+      "step": 4629
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.685e-05,
+      "loss": 0.09,
+      "step": 4630
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.6845e-05,
+      "loss": 0.1676,
+      "step": 4631
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.684e-05,
+      "loss": 0.2133,
+      "step": 4632
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.6835e-05,
+      "loss": 0.1453,
+      "step": 4633
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.6830000000000005e-05,
+      "loss": 0.0768,
+      "step": 4634
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.6825e-05,
+      "loss": 0.141,
+      "step": 4635
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.682e-05,
+      "loss": 0.0486,
+      "step": 4636
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.6815000000000006e-05,
+      "loss": 0.073,
+      "step": 4637
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.681e-05,
+      "loss": 0.2246,
+      "step": 4638
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.6805000000000004e-05,
+      "loss": 0.084,
+      "step": 4639
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.68e-05,
+      "loss": 0.0844,
+      "step": 4640
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.6794999999999996e-05,
+      "loss": 0.3821,
+      "step": 4641
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.679e-05,
+      "loss": 0.203,
+      "step": 4642
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.6785e-05,
+      "loss": 0.186,
+      "step": 4643
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.6780000000000004e-05,
+      "loss": 0.1456,
+      "step": 4644
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.6775e-05,
+      "loss": 0.0406,
+      "step": 4645
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.677e-05,
+      "loss": 0.1349,
+      "step": 4646
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.6765000000000005e-05,
+      "loss": 0.2465,
+      "step": 4647
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.676e-05,
+      "loss": 0.2391,
+      "step": 4648
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.6755e-05,
+      "loss": 0.1993,
+      "step": 4649
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.6750000000000006e-05,
+      "loss": 0.1305,
+      "step": 4650
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.6745e-05,
+      "loss": 0.2899,
+      "step": 4651
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.674e-05,
+      "loss": 0.0476,
+      "step": 4652
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.6735e-05,
+      "loss": 0.0767,
+      "step": 4653
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.673e-05,
+      "loss": 0.0564,
+      "step": 4654
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.6725e-05,
+      "loss": 0.0827,
+      "step": 4655
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.672e-05,
+      "loss": 0.205,
+      "step": 4656
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.6715e-05,
+      "loss": 0.1443,
+      "step": 4657
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.671e-05,
+      "loss": 0.1659,
+      "step": 4658
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.6705e-05,
+      "loss": 0.1417,
+      "step": 4659
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.6700000000000004e-05,
+      "loss": 0.2258,
+      "step": 4660
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.6695e-05,
+      "loss": 0.3034,
+      "step": 4661
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.669e-05,
+      "loss": 0.1315,
+      "step": 4662
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.6685000000000005e-05,
+      "loss": 0.0942,
+      "step": 4663
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.668e-05,
+      "loss": 0.1251,
+      "step": 4664
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 3.6675e-05,
+      "loss": 0.1302,
+      "step": 4665
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 3.667e-05,
+      "loss": 0.1515,
+      "step": 4666
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 3.6665e-05,
+      "loss": 0.1247,
+      "step": 4667
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 3.666e-05,
+      "loss": 0.1311,
+      "step": 4668
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 3.6655e-05,
+      "loss": 0.3963,
+      "step": 4669
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 3.665e-05,
+      "loss": 0.0867,
+      "step": 4670
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 3.6645e-05,
+      "loss": 0.0703,
+      "step": 4671
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 3.664e-05,
+      "loss": 0.0578,
+      "step": 4672
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 3.6635000000000004e-05,
+      "loss": 0.0766,
+      "step": 4673
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 3.663000000000001e-05,
+      "loss": 0.151,
+      "step": 4674
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 3.6625e-05,
+      "loss": 0.229,
+      "step": 4675
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 3.662e-05,
+      "loss": 0.1539,
+      "step": 4676
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 3.6615e-05,
+      "loss": 0.2401,
+      "step": 4677
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 3.661e-05,
+      "loss": 0.075,
+      "step": 4678
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 3.6605e-05,
+      "loss": 0.1498,
+      "step": 4679
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 3.66e-05,
+      "loss": 0.2323,
+      "step": 4680
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 3.6595e-05,
+      "loss": 0.1669,
+      "step": 4681
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 3.659e-05,
+      "loss": 0.3137,
+      "step": 4682
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 3.6585e-05,
+      "loss": 0.2011,
+      "step": 4683
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 3.6580000000000006e-05,
+      "loss": 0.0791,
+      "step": 4684
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 3.6575e-05,
+      "loss": 0.1864,
+      "step": 4685
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 3.6570000000000004e-05,
+      "loss": 0.1511,
+      "step": 4686
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 3.6565000000000007e-05,
+      "loss": 0.0552,
+      "step": 4687
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 3.6559999999999996e-05,
+      "loss": 0.0687,
+      "step": 4688
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 3.6555e-05,
+      "loss": 0.1023,
+      "step": 4689
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 3.655e-05,
+      "loss": 0.2065,
+      "step": 4690
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 3.6544999999999997e-05,
+      "loss": 0.2419,
+      "step": 4691
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 3.654e-05,
+      "loss": 0.0599,
+      "step": 4692
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 3.6535e-05,
+      "loss": 0.1734,
+      "step": 4693
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 3.6530000000000004e-05,
+      "loss": 0.2408,
+      "step": 4694
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 3.6525e-05,
+      "loss": 0.1795,
+      "step": 4695
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 3.652e-05,
+      "loss": 0.0423,
+      "step": 4696
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 3.6515000000000005e-05,
+      "loss": 0.297,
+      "step": 4697
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 3.651e-05,
+      "loss": 0.2139,
+      "step": 4698
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 3.6505000000000004e-05,
+      "loss": 0.2226,
+      "step": 4699
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 3.65e-05,
+      "loss": 0.108,
+      "step": 4700
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 3.6494999999999995e-05,
+      "loss": 0.2778,
+      "step": 4701
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 3.649e-05,
+      "loss": 0.3045,
+      "step": 4702
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 3.6485e-05,
+      "loss": 0.0697,
+      "step": 4703
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 3.648e-05,
+      "loss": 0.2703,
+      "step": 4704
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 3.6475e-05,
+      "loss": 0.0617,
+      "step": 4705
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 3.647e-05,
+      "loss": 0.1278,
+      "step": 4706
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 3.6465000000000004e-05,
+      "loss": 0.1132,
+      "step": 4707
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 3.646e-05,
+      "loss": 0.1142,
+      "step": 4708
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 3.6455e-05,
+      "loss": 0.0602,
+      "step": 4709
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 3.6450000000000005e-05,
+      "loss": 0.1807,
+      "step": 4710
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 3.6445e-05,
+      "loss": 0.0915,
+      "step": 4711
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 3.644e-05,
+      "loss": 0.0945,
+      "step": 4712
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 3.6435e-05,
+      "loss": 0.0663,
+      "step": 4713
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 3.643e-05,
+      "loss": 0.0843,
+      "step": 4714
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 3.6425e-05,
+      "loss": 0.3199,
+      "step": 4715
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 3.642e-05,
+      "loss": 0.0796,
+      "step": 4716
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 3.6415e-05,
+      "loss": 0.1508,
+      "step": 4717
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 3.641e-05,
+      "loss": 0.1314,
+      "step": 4718
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 3.6405e-05,
+      "loss": 0.2608,
+      "step": 4719
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 0.1093,
+      "step": 4720
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 3.6395e-05,
+      "loss": 0.2188,
+      "step": 4721
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 3.639e-05,
+      "loss": 0.223,
+      "step": 4722
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 3.6385000000000005e-05,
+      "loss": 0.0727,
+      "step": 4723
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 3.638e-05,
+      "loss": 0.0836,
+      "step": 4724
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 3.6374999999999996e-05,
+      "loss": 0.1404,
+      "step": 4725
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 3.637e-05,
+      "loss": 0.1103,
+      "step": 4726
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3.6365e-05,
+      "loss": 0.0827,
+      "step": 4727
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3.636e-05,
+      "loss": 0.055,
+      "step": 4728
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3.6355e-05,
+      "loss": 0.1522,
+      "step": 4729
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 3.635e-05,
+      "loss": 0.1948,
+      "step": 4730
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 3.6345e-05,
+      "loss": 0.1891,
+      "step": 4731
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 3.634e-05,
+      "loss": 0.1153,
+      "step": 4732
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 3.6335000000000004e-05,
+      "loss": 0.0806,
+      "step": 4733
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 3.6330000000000006e-05,
+      "loss": 0.2467,
+      "step": 4734
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 3.6325e-05,
+      "loss": 0.2687,
+      "step": 4735
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 3.632e-05,
+      "loss": 0.2092,
+      "step": 4736
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 3.6315e-05,
+      "loss": 0.1285,
+      "step": 4737
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 3.6309999999999996e-05,
+      "loss": 0.1876,
+      "step": 4738
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 3.6305e-05,
+      "loss": 0.2526,
+      "step": 4739
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 3.63e-05,
+      "loss": 0.2384,
+      "step": 4740
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 3.6295e-05,
+      "loss": 0.202,
+      "step": 4741
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 3.629e-05,
+      "loss": 0.165,
+      "step": 4742
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 3.6285e-05,
+      "loss": 0.0528,
+      "step": 4743
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 3.6280000000000005e-05,
+      "loss": 0.1685,
+      "step": 4744
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 3.6275e-05,
+      "loss": 0.1506,
+      "step": 4745
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 3.6270000000000003e-05,
+      "loss": 0.1502,
+      "step": 4746
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 3.6265000000000006e-05,
+      "loss": 0.1578,
+      "step": 4747
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 3.6259999999999995e-05,
+      "loss": 0.0934,
+      "step": 4748
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 3.6255e-05,
+      "loss": 0.078,
+      "step": 4749
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 3.625e-05,
+      "loss": 0.196,
+      "step": 4750
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 3.6244999999999996e-05,
+      "loss": 0.1123,
+      "step": 4751
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 3.624e-05,
+      "loss": 0.0794,
+      "step": 4752
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 3.6235e-05,
+      "loss": 0.1557,
+      "step": 4753
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 3.6230000000000004e-05,
+      "loss": 0.0596,
+      "step": 4754
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 3.6225e-05,
+      "loss": 0.0854,
+      "step": 4755
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 3.622e-05,
+      "loss": 0.0695,
+      "step": 4756
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 3.6215000000000005e-05,
+      "loss": 0.2223,
+      "step": 4757
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 3.621e-05,
+      "loss": 0.0603,
+      "step": 4758
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 3.6205e-05,
+      "loss": 0.0642,
+      "step": 4759
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 3.6200000000000006e-05,
+      "loss": 0.0487,
+      "step": 4760
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 3.6194999999999995e-05,
+      "loss": 0.1849,
+      "step": 4761
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 3.619e-05,
+      "loss": 0.0621,
+      "step": 4762
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 3.6185e-05,
+      "loss": 0.187,
+      "step": 4763
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 3.618e-05,
+      "loss": 0.0578,
+      "step": 4764
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 3.6175e-05,
+      "loss": 0.07,
+      "step": 4765
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 3.617e-05,
+      "loss": 0.2435,
+      "step": 4766
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 3.6165000000000004e-05,
+      "loss": 0.0888,
+      "step": 4767
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 3.616e-05,
+      "loss": 0.2308,
+      "step": 4768
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 3.6155e-05,
+      "loss": 0.1799,
+      "step": 4769
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 3.6150000000000005e-05,
+      "loss": 0.0862,
+      "step": 4770
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 3.6145e-05,
+      "loss": 0.105,
+      "step": 4771
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 3.614e-05,
+      "loss": 0.0525,
+      "step": 4772
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 3.6135e-05,
+      "loss": 0.081,
+      "step": 4773
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 3.613e-05,
+      "loss": 0.0749,
+      "step": 4774
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 3.6125e-05,
+      "loss": 0.1712,
+      "step": 4775
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 3.612e-05,
+      "loss": 0.2893,
+      "step": 4776
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 3.6115e-05,
+      "loss": 0.1527,
+      "step": 4777
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 3.611e-05,
+      "loss": 0.174,
+      "step": 4778
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 3.6105e-05,
+      "loss": 0.0765,
+      "step": 4779
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 3.61e-05,
+      "loss": 0.2204,
+      "step": 4780
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 3.6095e-05,
+      "loss": 0.151,
+      "step": 4781
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.609e-05,
+      "loss": 0.1324,
+      "step": 4782
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.6085000000000004e-05,
+      "loss": 0.0774,
+      "step": 4783
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.608000000000001e-05,
+      "loss": 0.2295,
+      "step": 4784
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.6074999999999996e-05,
+      "loss": 0.2473,
+      "step": 4785
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.607e-05,
+      "loss": 0.1582,
+      "step": 4786
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.6065e-05,
+      "loss": 0.2977,
+      "step": 4787
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.606e-05,
+      "loss": 0.3334,
+      "step": 4788
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.6055e-05,
+      "loss": 0.1651,
+      "step": 4789
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 3.605e-05,
+      "loss": 0.2578,
+      "step": 4790
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 3.6045e-05,
+      "loss": 0.1289,
+      "step": 4791
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 3.604e-05,
+      "loss": 0.1565,
+      "step": 4792
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 3.6035e-05,
+      "loss": 0.1113,
+      "step": 4793
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 3.6030000000000006e-05,
+      "loss": 0.097,
+      "step": 4794
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 3.6025e-05,
+      "loss": 0.0751,
+      "step": 4795
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 3.6020000000000004e-05,
+      "loss": 0.154,
+      "step": 4796
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 3.6015e-05,
+      "loss": 0.0575,
+      "step": 4797
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 3.6009999999999996e-05,
+      "loss": 0.1326,
+      "step": 4798
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 3.6005e-05,
+      "loss": 0.0913,
+      "step": 4799
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 3.6e-05,
+      "loss": 0.0565,
+      "step": 4800
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 3.5995000000000004e-05,
+      "loss": 0.1317,
+      "step": 4801
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 3.599e-05,
+      "loss": 0.1484,
+      "step": 4802
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 3.5985e-05,
+      "loss": 0.2345,
+      "step": 4803
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 3.5980000000000004e-05,
+      "loss": 0.1306,
+      "step": 4804
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 3.5975e-05,
+      "loss": 0.2032,
+      "step": 4805
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 3.597e-05,
+      "loss": 0.1513,
+      "step": 4806
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 3.5965000000000005e-05,
+      "loss": 0.2473,
+      "step": 4807
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 3.596e-05,
+      "loss": 0.1704,
+      "step": 4808
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 3.5955e-05,
+      "loss": 0.0658,
+      "step": 4809
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 3.595e-05,
+      "loss": 0.1342,
+      "step": 4810
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 3.5945e-05,
+      "loss": 0.0924,
+      "step": 4811
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 3.594e-05,
+      "loss": 0.1262,
+      "step": 4812
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 3.5935e-05,
+      "loss": 0.0518,
+      "step": 4813
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 3.593e-05,
+      "loss": 0.0676,
+      "step": 4814
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 3.5925e-05,
+      "loss": 0.0659,
+      "step": 4815
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 3.592e-05,
+      "loss": 0.1589,
+      "step": 4816
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 3.5915000000000004e-05,
+      "loss": 0.0921,
+      "step": 4817
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 3.591e-05,
+      "loss": 0.1689,
+      "step": 4818
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 3.5905e-05,
+      "loss": 0.0624,
+      "step": 4819
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 3.5900000000000005e-05,
+      "loss": 0.0834,
+      "step": 4820
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 3.5895e-05,
+      "loss": 0.1435,
+      "step": 4821
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 3.589e-05,
+      "loss": 0.0711,
+      "step": 4822
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 3.5885e-05,
+      "loss": 0.2164,
+      "step": 4823
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 3.588e-05,
+      "loss": 0.1768,
+      "step": 4824
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 3.5875e-05,
+      "loss": 0.1486,
+      "step": 4825
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 3.587e-05,
+      "loss": 0.1907,
+      "step": 4826
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 3.5865e-05,
+      "loss": 0.1828,
+      "step": 4827
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 3.586e-05,
+      "loss": 0.2126,
+      "step": 4828
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 3.5855e-05,
+      "loss": 0.2036,
+      "step": 4829
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 3.5850000000000004e-05,
+      "loss": 0.1704,
+      "step": 4830
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 3.584500000000001e-05,
+      "loss": 0.0711,
+      "step": 4831
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 3.584e-05,
+      "loss": 0.1104,
+      "step": 4832
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 3.5835e-05,
+      "loss": 0.1399,
+      "step": 4833
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 3.583e-05,
+      "loss": 0.0904,
+      "step": 4834
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 3.5825e-05,
+      "loss": 0.1347,
+      "step": 4835
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 3.582e-05,
+      "loss": 0.1415,
+      "step": 4836
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 3.5815e-05,
+      "loss": 0.2826,
+      "step": 4837
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 3.581e-05,
+      "loss": 0.0721,
+      "step": 4838
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 3.5805e-05,
+      "loss": 0.1317,
+      "step": 4839
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 3.58e-05,
+      "loss": 0.0956,
+      "step": 4840
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 3.5795000000000005e-05,
+      "loss": 0.225,
+      "step": 4841
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 3.579e-05,
+      "loss": 0.2207,
+      "step": 4842
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 3.5785000000000004e-05,
+      "loss": 0.2022,
+      "step": 4843
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 3.5780000000000006e-05,
+      "loss": 0.2203,
+      "step": 4844
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 3.5774999999999995e-05,
+      "loss": 0.0932,
+      "step": 4845
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 3.577e-05,
+      "loss": 0.1498,
+      "step": 4846
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 3.5765e-05,
+      "loss": 0.2107,
+      "step": 4847
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 3.5759999999999996e-05,
+      "loss": 0.1523,
+      "step": 4848
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 3.5755e-05,
+      "loss": 0.1128,
+      "step": 4849
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 3.575e-05,
+      "loss": 0.0891,
+      "step": 4850
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 3.5745000000000004e-05,
+      "loss": 0.0807,
+      "step": 4851
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 3.574e-05,
+      "loss": 0.1911,
+      "step": 4852
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 3.5735e-05,
+      "loss": 0.076,
+      "step": 4853
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 3.5730000000000005e-05,
+      "loss": 0.2007,
+      "step": 4854
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 3.5725e-05,
+      "loss": 0.0903,
+      "step": 4855
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 3.5720000000000004e-05,
+      "loss": 0.1658,
+      "step": 4856
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 3.5715e-05,
+      "loss": 0.2422,
+      "step": 4857
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 3.5709999999999995e-05,
+      "loss": 0.06,
+      "step": 4858
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 3.5705e-05,
+      "loss": 0.0858,
+      "step": 4859
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 3.57e-05,
+      "loss": 0.161,
+      "step": 4860
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 3.5695e-05,
+      "loss": 0.0575,
+      "step": 4861
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 3.569e-05,
+      "loss": 0.4558,
+      "step": 4862
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 3.5685e-05,
+      "loss": 0.2009,
+      "step": 4863
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 3.5680000000000004e-05,
+      "loss": 0.0598,
+      "step": 4864
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 3.5675e-05,
+      "loss": 0.1334,
+      "step": 4865
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 3.567e-05,
+      "loss": 0.2511,
+      "step": 4866
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 3.5665000000000005e-05,
+      "loss": 0.0939,
+      "step": 4867
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 3.566e-05,
+      "loss": 0.1511,
+      "step": 4868
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 3.5654999999999997e-05,
+      "loss": 0.0688,
+      "step": 4869
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 3.565e-05,
+      "loss": 0.0609,
+      "step": 4870
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 3.5645e-05,
+      "loss": 0.2109,
+      "step": 4871
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 3.564e-05,
+      "loss": 0.1011,
+      "step": 4872
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 3.5635e-05,
+      "loss": 0.1957,
+      "step": 4873
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 3.563e-05,
+      "loss": 0.1302,
+      "step": 4874
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 3.5625e-05,
+      "loss": 0.0692,
+      "step": 4875
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 3.562e-05,
+      "loss": 0.1563,
+      "step": 4876
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 3.5615000000000004e-05,
+      "loss": 0.1753,
+      "step": 4877
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 3.561e-05,
+      "loss": 0.1359,
+      "step": 4878
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 3.5605e-05,
+      "loss": 0.2519,
+      "step": 4879
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 3.5600000000000005e-05,
+      "loss": 0.0835,
+      "step": 4880
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 3.559500000000001e-05,
+      "loss": 0.0701,
+      "step": 4881
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 3.5589999999999996e-05,
+      "loss": 0.1294,
+      "step": 4882
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 3.5585e-05,
+      "loss": 0.1723,
+      "step": 4883
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 3.558e-05,
+      "loss": 0.1255,
+      "step": 4884
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 3.5575e-05,
+      "loss": 0.1723,
+      "step": 4885
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 3.557e-05,
+      "loss": 0.1545,
+      "step": 4886
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 3.5565e-05,
+      "loss": 0.3234,
+      "step": 4887
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 3.556e-05,
+      "loss": 0.3007,
+      "step": 4888
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 3.5555e-05,
+      "loss": 0.0649,
+      "step": 4889
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 3.5550000000000004e-05,
+      "loss": 0.0611,
+      "step": 4890
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 3.5545000000000006e-05,
+      "loss": 0.0672,
+      "step": 4891
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 3.554e-05,
+      "loss": 0.2912,
+      "step": 4892
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 3.5535000000000005e-05,
+      "loss": 0.0831,
+      "step": 4893
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 3.553e-05,
+      "loss": 0.2184,
+      "step": 4894
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 3.5524999999999996e-05,
+      "loss": 0.0604,
+      "step": 4895
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 3.552e-05,
+      "loss": 0.0569,
+      "step": 4896
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 3.5515e-05,
+      "loss": 0.1697,
+      "step": 4897
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 3.551e-05,
+      "loss": 0.1578,
+      "step": 4898
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 3.5505e-05,
+      "loss": 0.0631,
+      "step": 4899
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 3.55e-05,
+      "loss": 0.0745,
+      "step": 4900
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 3.5495000000000005e-05,
+      "loss": 0.0477,
+      "step": 4901
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 3.549e-05,
+      "loss": 0.2537,
+      "step": 4902
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 3.5485e-05,
+      "loss": 0.1405,
+      "step": 4903
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 3.5480000000000006e-05,
+      "loss": 0.1238,
+      "step": 4904
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 3.5475e-05,
+      "loss": 0.1035,
+      "step": 4905
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 3.547e-05,
+      "loss": 0.2113,
+      "step": 4906
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 3.5465e-05,
+      "loss": 0.0719,
+      "step": 4907
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 3.5459999999999996e-05,
+      "loss": 0.0928,
+      "step": 4908
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 3.5455e-05,
+      "loss": 0.0774,
+      "step": 4909
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 3.545e-05,
+      "loss": 0.165,
+      "step": 4910
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 3.5445000000000004e-05,
+      "loss": 0.1953,
+      "step": 4911
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 3.544e-05,
+      "loss": 0.2315,
+      "step": 4912
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 3.5435e-05,
+      "loss": 0.181,
+      "step": 4913
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 3.5430000000000005e-05,
+      "loss": 0.0406,
+      "step": 4914
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 3.5425e-05,
+      "loss": 0.0936,
+      "step": 4915
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 3.542e-05,
+      "loss": 0.1478,
+      "step": 4916
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 3.5415000000000006e-05,
+      "loss": 0.2129,
+      "step": 4917
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 3.5409999999999995e-05,
+      "loss": 0.1662,
+      "step": 4918
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 3.5405e-05,
+      "loss": 0.2379,
+      "step": 4919
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 3.54e-05,
+      "loss": 0.2076,
+      "step": 4920
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 3.5395e-05,
+      "loss": 0.2824,
+      "step": 4921
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 3.539e-05,
+      "loss": 0.0985,
+      "step": 4922
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 3.5385e-05,
+      "loss": 0.2349,
+      "step": 4923
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 3.5380000000000003e-05,
+      "loss": 0.1415,
+      "step": 4924
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 3.5375e-05,
+      "loss": 0.1705,
+      "step": 4925
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 3.537e-05,
+      "loss": 0.2088,
+      "step": 4926
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 3.5365000000000004e-05,
+      "loss": 0.0586,
+      "step": 4927
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 3.536e-05,
+      "loss": 0.2939,
+      "step": 4928
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 3.5355e-05,
+      "loss": 0.0786,
+      "step": 4929
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 3.535e-05,
+      "loss": 0.1601,
+      "step": 4930
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 3.5345e-05,
+      "loss": 0.2348,
+      "step": 4931
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 3.534e-05,
+      "loss": 0.1875,
+      "step": 4932
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 3.5335e-05,
+      "loss": 0.1338,
+      "step": 4933
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 3.533e-05,
+      "loss": 0.1286,
+      "step": 4934
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 3.5325e-05,
+      "loss": 0.1689,
+      "step": 4935
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 3.532e-05,
+      "loss": 0.2057,
+      "step": 4936
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 3.5315e-05,
+      "loss": 0.247,
+      "step": 4937
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 3.531e-05,
+      "loss": 0.0814,
+      "step": 4938
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 3.5305e-05,
+      "loss": 0.0971,
+      "step": 4939
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 3.5300000000000004e-05,
+      "loss": 0.0758,
+      "step": 4940
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 3.529500000000001e-05,
+      "loss": 0.2542,
+      "step": 4941
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 3.5289999999999996e-05,
+      "loss": 0.1464,
+      "step": 4942
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 3.5285e-05,
+      "loss": 0.1525,
+      "step": 4943
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 3.528e-05,
+      "loss": 0.2185,
+      "step": 4944
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 3.5275e-05,
+      "loss": 0.0545,
+      "step": 4945
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 3.527e-05,
+      "loss": 0.0628,
+      "step": 4946
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 3.5265e-05,
+      "loss": 0.1692,
+      "step": 4947
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 3.526e-05,
+      "loss": 0.1514,
+      "step": 4948
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 3.5255e-05,
+      "loss": 0.0627,
+      "step": 4949
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 3.525e-05,
+      "loss": 0.1569,
+      "step": 4950
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 3.5245000000000006e-05,
+      "loss": 0.0789,
+      "step": 4951
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 3.524e-05,
+      "loss": 0.0496,
+      "step": 4952
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 3.5235000000000004e-05,
+      "loss": 0.1323,
+      "step": 4953
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 3.523e-05,
+      "loss": 0.0695,
+      "step": 4954
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 3.5224999999999996e-05,
+      "loss": 0.1398,
+      "step": 4955
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 3.522e-05,
+      "loss": 0.1624,
+      "step": 4956
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 3.5215e-05,
+      "loss": 0.0855,
+      "step": 4957
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 3.521e-05,
+      "loss": 0.1854,
+      "step": 4958
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 3.5205e-05,
+      "loss": 0.3268,
+      "step": 4959
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 3.52e-05,
+      "loss": 0.093,
+      "step": 4960
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 3.5195000000000004e-05,
+      "loss": 0.0735,
+      "step": 4961
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 3.519e-05,
+      "loss": 0.0619,
+      "step": 4962
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 3.5185e-05,
+      "loss": 0.1788,
+      "step": 4963
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 3.5180000000000005e-05,
+      "loss": 0.1161,
+      "step": 4964
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 3.5175e-05,
+      "loss": 0.064,
+      "step": 4965
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 3.517e-05,
+      "loss": 0.083,
+      "step": 4966
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 3.5165e-05,
+      "loss": 0.0468,
+      "step": 4967
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 3.516e-05,
+      "loss": 0.1926,
+      "step": 4968
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 3.5155e-05,
+      "loss": 0.1426,
+      "step": 4969
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 3.515e-05,
+      "loss": 0.1736,
+      "step": 4970
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 3.5145e-05,
+      "loss": 0.211,
+      "step": 4971
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 3.514e-05,
+      "loss": 0.0657,
+      "step": 4972
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 3.5135e-05,
+      "loss": 0.0541,
+      "step": 4973
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 3.5130000000000004e-05,
+      "loss": 0.109,
+      "step": 4974
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 3.5125e-05,
+      "loss": 0.1196,
+      "step": 4975
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 3.512e-05,
+      "loss": 0.1076,
+      "step": 4976
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 3.5115000000000005e-05,
+      "loss": 0.0685,
+      "step": 4977
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 3.511e-05,
+      "loss": 0.1153,
+      "step": 4978
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 3.5105e-05,
+      "loss": 0.067,
+      "step": 4979
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 3.51e-05,
+      "loss": 0.066,
+      "step": 4980
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 3.5095e-05,
+      "loss": 0.1564,
+      "step": 4981
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 3.509e-05,
+      "loss": 0.0646,
+      "step": 4982
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 3.5085e-05,
+      "loss": 0.0506,
+      "step": 4983
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 3.508e-05,
+      "loss": 0.2029,
+      "step": 4984
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 3.5075e-05,
+      "loss": 0.0716,
+      "step": 4985
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 3.507e-05,
+      "loss": 0.1577,
+      "step": 4986
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 3.5065000000000004e-05,
+      "loss": 0.0612,
+      "step": 4987
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 3.5060000000000007e-05,
+      "loss": 0.224,
+      "step": 4988
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 3.5055e-05,
+      "loss": 0.0423,
+      "step": 4989
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 3.505e-05,
+      "loss": 0.1175,
+      "step": 4990
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 3.5045e-05,
+      "loss": 0.208,
+      "step": 4991
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 3.5039999999999997e-05,
+      "loss": 0.1239,
+      "step": 4992
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 3.5035e-05,
+      "loss": 0.0755,
+      "step": 4993
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 3.503e-05,
+      "loss": 0.0683,
+      "step": 4994
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 3.5025e-05,
+      "loss": 0.1771,
+      "step": 4995
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 3.502e-05,
+      "loss": 0.1098,
+      "step": 4996
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 3.5015e-05,
+      "loss": 0.1336,
+      "step": 4997
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 3.5010000000000005e-05,
+      "loss": 0.1785,
+      "step": 4998
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 3.5005e-05,
+      "loss": 0.1497,
+      "step": 4999
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 0.1308,
+      "step": 5000
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 3.4995000000000006e-05,
+      "loss": 0.2278,
+      "step": 5001
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 3.499e-05,
+      "loss": 0.2756,
+      "step": 5002
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 3.4985e-05,
+      "loss": 0.1633,
+      "step": 5003
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 3.498e-05,
+      "loss": 0.1457,
+      "step": 5004
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 3.4974999999999996e-05,
+      "loss": 0.1015,
+      "step": 5005
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 3.497e-05,
+      "loss": 0.1011,
+      "step": 5006
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 3.4965e-05,
+      "loss": 0.1621,
+      "step": 5007
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 3.4960000000000004e-05,
+      "loss": 0.2009,
+      "step": 5008
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 3.4955e-05,
+      "loss": 0.2508,
+      "step": 5009
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 3.495e-05,
+      "loss": 0.1461,
+      "step": 5010
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 3.4945000000000005e-05,
+      "loss": 0.0651,
+      "step": 5011
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 3.494e-05,
+      "loss": 0.2144,
+      "step": 5012
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 3.4935000000000003e-05,
+      "loss": 0.2016,
+      "step": 5013
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 3.4930000000000006e-05,
+      "loss": 0.2163,
+      "step": 5014
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 3.4924999999999995e-05,
+      "loss": 0.1413,
+      "step": 5015
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 3.492e-05,
+      "loss": 0.0639,
+      "step": 5016
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 3.4915e-05,
+      "loss": 0.0612,
+      "step": 5017
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 3.491e-05,
+      "loss": 0.2451,
+      "step": 5018
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 3.4905e-05,
+      "loss": 0.2621,
+      "step": 5019
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 3.49e-05,
+      "loss": 0.0542,
+      "step": 5020
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 3.4895000000000004e-05,
+      "loss": 0.1731,
+      "step": 5021
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 3.489e-05,
+      "loss": 0.1563,
+      "step": 5022
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 3.4885e-05,
+      "loss": 0.2477,
+      "step": 5023
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 3.4880000000000005e-05,
+      "loss": 0.1177,
+      "step": 5024
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 3.4875e-05,
+      "loss": 0.2773,
+      "step": 5025
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 3.487e-05,
+      "loss": 0.1103,
+      "step": 5026
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 3.4865e-05,
+      "loss": 0.1681,
+      "step": 5027
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 3.486e-05,
+      "loss": 0.2423,
+      "step": 5028
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 3.4855e-05,
+      "loss": 0.1384,
+      "step": 5029
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 3.485e-05,
+      "loss": 0.1097,
+      "step": 5030
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 3.4845e-05,
+      "loss": 0.1218,
+      "step": 5031
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 3.484e-05,
+      "loss": 0.1626,
+      "step": 5032
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 3.4835e-05,
+      "loss": 0.0733,
+      "step": 5033
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 3.4830000000000004e-05,
+      "loss": 0.1342,
+      "step": 5034
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 3.4825e-05,
+      "loss": 0.2042,
+      "step": 5035
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 3.482e-05,
+      "loss": 0.1207,
+      "step": 5036
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 3.4815000000000005e-05,
+      "loss": 0.0492,
+      "step": 5037
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 3.481000000000001e-05,
+      "loss": 0.0786,
+      "step": 5038
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 3.4804999999999996e-05,
+      "loss": 0.0441,
+      "step": 5039
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 3.48e-05,
+      "loss": 0.216,
+      "step": 5040
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 3.4795e-05,
+      "loss": 0.072,
+      "step": 5041
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 3.479e-05,
+      "loss": 0.1623,
+      "step": 5042
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 3.4785e-05,
+      "loss": 0.1423,
+      "step": 5043
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 3.478e-05,
+      "loss": 0.0732,
+      "step": 5044
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 3.4775e-05,
+      "loss": 0.34,
+      "step": 5045
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 3.477e-05,
+      "loss": 0.1848,
+      "step": 5046
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 3.4765000000000003e-05,
+      "loss": 0.0631,
+      "step": 5047
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 3.4760000000000006e-05,
+      "loss": 0.1839,
+      "step": 5048
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 3.4755e-05,
+      "loss": 0.0722,
+      "step": 5049
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 3.4750000000000004e-05,
+      "loss": 0.1075,
+      "step": 5050
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 3.4745e-05,
+      "loss": 0.3058,
+      "step": 5051
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 3.4739999999999996e-05,
+      "loss": 0.0527,
+      "step": 5052
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 3.4735e-05,
+      "loss": 0.3208,
+      "step": 5053
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 3.473e-05,
+      "loss": 0.0999,
+      "step": 5054
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 3.4725e-05,
+      "loss": 0.1605,
+      "step": 5055
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 3.472e-05,
+      "loss": 0.1374,
+      "step": 5056
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 3.4715e-05,
+      "loss": 0.1566,
+      "step": 5057
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 3.4710000000000005e-05,
+      "loss": 0.0455,
+      "step": 5058
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 3.4705e-05,
+      "loss": 0.2457,
+      "step": 5059
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 3.47e-05,
+      "loss": 0.0551,
+      "step": 5060
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 3.4695000000000006e-05,
+      "loss": 0.1887,
+      "step": 5061
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 3.469e-05,
+      "loss": 0.2227,
+      "step": 5062
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 3.4685e-05,
+      "loss": 0.1351,
+      "step": 5063
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 3.468e-05,
+      "loss": 0.1528,
+      "step": 5064
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 3.4674999999999996e-05,
+      "loss": 0.0453,
+      "step": 5065
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 3.467e-05,
+      "loss": 0.2431,
+      "step": 5066
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 3.4665e-05,
+      "loss": 0.1543,
+      "step": 5067
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 3.4660000000000004e-05,
+      "loss": 0.1898,
+      "step": 5068
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 3.4655e-05,
+      "loss": 0.1407,
+      "step": 5069
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 3.465e-05,
+      "loss": 0.1409,
+      "step": 5070
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 3.4645000000000005e-05,
+      "loss": 0.2098,
+      "step": 5071
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 3.464e-05,
+      "loss": 0.0584,
+      "step": 5072
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 3.4635e-05,
+      "loss": 0.0437,
+      "step": 5073
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 3.4630000000000006e-05,
+      "loss": 0.1143,
+      "step": 5074
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 3.4624999999999995e-05,
+      "loss": 0.1323,
+      "step": 5075
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 3.462e-05,
+      "loss": 0.1612,
+      "step": 5076
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 3.4615e-05,
+      "loss": 0.1399,
+      "step": 5077
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 3.461e-05,
+      "loss": 0.0817,
+      "step": 5078
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 3.4605e-05,
+      "loss": 0.1409,
+      "step": 5079
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 3.46e-05,
+      "loss": 0.1641,
+      "step": 5080
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 3.4595e-05,
+      "loss": 0.1471,
+      "step": 5081
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 3.459e-05,
+      "loss": 0.0797,
+      "step": 5082
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 3.4585e-05,
+      "loss": 0.0962,
+      "step": 5083
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 3.4580000000000004e-05,
+      "loss": 0.0784,
+      "step": 5084
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 3.4575e-05,
+      "loss": 0.2789,
+      "step": 5085
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 3.457e-05,
+      "loss": 0.2909,
+      "step": 5086
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 3.4565e-05,
+      "loss": 0.067,
+      "step": 5087
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 3.456e-05,
+      "loss": 0.2547,
+      "step": 5088
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 3.4555e-05,
+      "loss": 0.1695,
+      "step": 5089
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 3.455e-05,
+      "loss": 0.0602,
+      "step": 5090
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 3.4545e-05,
+      "loss": 0.1436,
+      "step": 5091
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 3.454e-05,
+      "loss": 0.0629,
+      "step": 5092
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 3.4535e-05,
+      "loss": 0.0522,
+      "step": 5093
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 3.453e-05,
+      "loss": 0.2543,
+      "step": 5094
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 3.4525e-05,
+      "loss": 0.1886,
+      "step": 5095
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 3.452e-05,
+      "loss": 0.3713,
+      "step": 5096
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 3.4515000000000004e-05,
+      "loss": 0.1025,
+      "step": 5097
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 3.451000000000001e-05,
+      "loss": 0.1417,
+      "step": 5098
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 3.4504999999999996e-05,
+      "loss": 0.079,
+      "step": 5099
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 3.45e-05,
+      "loss": 0.0672,
+      "step": 5100
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 3.4495e-05,
+      "loss": 0.1,
+      "step": 5101
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 3.449e-05,
+      "loss": 0.1217,
+      "step": 5102
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 3.4485e-05,
+      "loss": 0.2848,
+      "step": 5103
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 3.448e-05,
+      "loss": 0.2771,
+      "step": 5104
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 3.4475e-05,
+      "loss": 0.0666,
+      "step": 5105
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 3.447e-05,
+      "loss": 0.1519,
+      "step": 5106
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 3.4465e-05,
+      "loss": 0.163,
+      "step": 5107
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 3.4460000000000005e-05,
+      "loss": 0.1955,
+      "step": 5108
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 3.4455e-05,
+      "loss": 0.062,
+      "step": 5109
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 3.4450000000000004e-05,
+      "loss": 0.2435,
+      "step": 5110
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 3.4445e-05,
+      "loss": 0.2461,
+      "step": 5111
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 3.4439999999999996e-05,
+      "loss": 0.1557,
+      "step": 5112
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 3.4435e-05,
+      "loss": 0.138,
+      "step": 5113
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 3.443e-05,
+      "loss": 0.2045,
+      "step": 5114
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 3.4424999999999997e-05,
+      "loss": 0.1524,
+      "step": 5115
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 3.442e-05,
+      "loss": 0.0498,
+      "step": 5116
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 3.4415e-05,
+      "loss": 0.2565,
+      "step": 5117
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 3.4410000000000004e-05,
+      "loss": 0.1406,
+      "step": 5118
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 3.4405e-05,
+      "loss": 0.0562,
+      "step": 5119
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 3.44e-05,
+      "loss": 0.0766,
+      "step": 5120
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 3.4395000000000005e-05,
+      "loss": 0.1499,
+      "step": 5121
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 3.439e-05,
+      "loss": 0.1767,
+      "step": 5122
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 3.4385e-05,
+      "loss": 0.1482,
+      "step": 5123
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 3.438e-05,
+      "loss": 0.1231,
+      "step": 5124
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 3.4374999999999995e-05,
+      "loss": 0.2049,
+      "step": 5125
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 3.437e-05,
+      "loss": 0.1163,
+      "step": 5126
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 3.4365e-05,
+      "loss": 0.0555,
+      "step": 5127
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 3.436e-05,
+      "loss": 0.1688,
+      "step": 5128
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 3.4355e-05,
+      "loss": 0.0611,
+      "step": 5129
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 3.435e-05,
+      "loss": 0.1406,
+      "step": 5130
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 3.4345000000000004e-05,
+      "loss": 0.1513,
+      "step": 5131
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 3.434e-05,
+      "loss": 0.2469,
+      "step": 5132
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 3.4335e-05,
+      "loss": 0.0765,
+      "step": 5133
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 3.4330000000000005e-05,
+      "loss": 0.1695,
+      "step": 5134
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 3.4325e-05,
+      "loss": 0.0699,
+      "step": 5135
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 3.432e-05,
+      "loss": 0.1337,
+      "step": 5136
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 3.4315e-05,
+      "loss": 0.0637,
+      "step": 5137
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 3.431e-05,
+      "loss": 0.075,
+      "step": 5138
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 3.4305e-05,
+      "loss": 0.1392,
+      "step": 5139
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 3.43e-05,
+      "loss": 0.1272,
+      "step": 5140
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 3.4295e-05,
+      "loss": 0.1389,
+      "step": 5141
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 3.429e-05,
+      "loss": 0.1604,
+      "step": 5142
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 3.4285e-05,
+      "loss": 0.0829,
+      "step": 5143
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 3.4280000000000004e-05,
+      "loss": 0.0797,
+      "step": 5144
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 3.4275e-05,
+      "loss": 0.1391,
+      "step": 5145
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 3.427e-05,
+      "loss": 0.1617,
+      "step": 5146
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 3.4265000000000005e-05,
+      "loss": 0.1414,
+      "step": 5147
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 3.426e-05,
+      "loss": 0.2039,
+      "step": 5148
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 3.4254999999999996e-05,
+      "loss": 0.1026,
+      "step": 5149
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 3.425e-05,
+      "loss": 0.151,
+      "step": 5150
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 3.4245e-05,
+      "loss": 0.1469,
+      "step": 5151
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 3.424e-05,
+      "loss": 0.0674,
+      "step": 5152
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 3.4235e-05,
+      "loss": 0.0576,
+      "step": 5153
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 3.423e-05,
+      "loss": 0.1765,
+      "step": 5154
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 3.4225000000000005e-05,
+      "loss": 0.0852,
+      "step": 5155
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 3.422e-05,
+      "loss": 0.1784,
+      "step": 5156
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 3.4215000000000004e-05,
+      "loss": 0.0755,
+      "step": 5157
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 3.4210000000000006e-05,
+      "loss": 0.0719,
+      "step": 5158
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 3.4205e-05,
+      "loss": 0.2927,
+      "step": 5159
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 3.42e-05,
+      "loss": 0.1955,
+      "step": 5160
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 3.4195e-05,
+      "loss": 0.2108,
+      "step": 5161
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 3.4189999999999996e-05,
+      "loss": 0.1738,
+      "step": 5162
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 3.4185e-05,
+      "loss": 0.1178,
+      "step": 5163
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 3.418e-05,
+      "loss": 0.2146,
+      "step": 5164
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 3.4175000000000004e-05,
+      "loss": 0.1692,
+      "step": 5165
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 3.417e-05,
+      "loss": 0.1232,
+      "step": 5166
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 3.4165e-05,
+      "loss": 0.0839,
+      "step": 5167
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 3.4160000000000005e-05,
+      "loss": 0.1117,
+      "step": 5168
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 3.4155e-05,
+      "loss": 0.0969,
+      "step": 5169
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 3.415e-05,
+      "loss": 0.0947,
+      "step": 5170
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 3.4145000000000006e-05,
+      "loss": 0.1788,
+      "step": 5171
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 3.4139999999999995e-05,
+      "loss": 0.2355,
+      "step": 5172
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 3.4135e-05,
+      "loss": 0.1358,
+      "step": 5173
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 3.413e-05,
+      "loss": 0.087,
+      "step": 5174
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 3.4125e-05,
+      "loss": 0.185,
+      "step": 5175
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 3.412e-05,
+      "loss": 0.0463,
+      "step": 5176
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 3.4115e-05,
+      "loss": 0.1142,
+      "step": 5177
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 3.4110000000000004e-05,
+      "loss": 0.0838,
+      "step": 5178
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 3.4105e-05,
+      "loss": 0.1197,
+      "step": 5179
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 3.41e-05,
+      "loss": 0.0696,
+      "step": 5180
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 3.4095000000000005e-05,
+      "loss": 0.1786,
+      "step": 5181
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 3.409e-05,
+      "loss": 0.1405,
+      "step": 5182
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 3.4085e-05,
+      "loss": 0.1921,
+      "step": 5183
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 3.408e-05,
+      "loss": 0.1357,
+      "step": 5184
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 3.4075e-05,
+      "loss": 0.1034,
+      "step": 5185
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 3.407e-05,
+      "loss": 0.0762,
+      "step": 5186
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 3.4065e-05,
+      "loss": 0.0957,
+      "step": 5187
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 3.406e-05,
+      "loss": 0.0718,
+      "step": 5188
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 3.4055e-05,
+      "loss": 0.2451,
+      "step": 5189
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 3.405e-05,
+      "loss": 0.1706,
+      "step": 5190
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 3.4045000000000004e-05,
+      "loss": 0.0724,
+      "step": 5191
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 3.404e-05,
+      "loss": 0.2395,
+      "step": 5192
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 3.4035e-05,
+      "loss": 0.2421,
+      "step": 5193
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 3.4030000000000005e-05,
+      "loss": 0.1143,
+      "step": 5194
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 3.402500000000001e-05,
+      "loss": 0.0684,
+      "step": 5195
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 3.4019999999999996e-05,
+      "loss": 0.0963,
+      "step": 5196
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 3.4015e-05,
+      "loss": 0.0542,
+      "step": 5197
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 3.401e-05,
+      "loss": 0.1389,
+      "step": 5198
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 3.4005e-05,
+      "loss": 0.1688,
+      "step": 5199
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 3.4e-05,
+      "loss": 0.1502,
+      "step": 5200
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 3.3995e-05,
+      "loss": 0.1679,
+      "step": 5201
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 3.399e-05,
+      "loss": 0.1617,
+      "step": 5202
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 3.3985e-05,
+      "loss": 0.1468,
+      "step": 5203
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 3.398e-05,
+      "loss": 0.0724,
+      "step": 5204
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 3.3975000000000006e-05,
+      "loss": 0.1157,
+      "step": 5205
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 3.397e-05,
+      "loss": 0.2103,
+      "step": 5206
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 3.3965000000000004e-05,
+      "loss": 0.151,
+      "step": 5207
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 3.396e-05,
+      "loss": 0.1937,
+      "step": 5208
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 3.3954999999999996e-05,
+      "loss": 0.2363,
+      "step": 5209
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 3.395e-05,
+      "loss": 0.1868,
+      "step": 5210
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 3.3945e-05,
+      "loss": 0.1757,
+      "step": 5211
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 3.394e-05,
+      "loss": 0.1401,
+      "step": 5212
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 3.3935e-05,
+      "loss": 0.1773,
+      "step": 5213
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 3.393e-05,
+      "loss": 0.0794,
+      "step": 5214
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 3.3925000000000005e-05,
+      "loss": 0.0575,
+      "step": 5215
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 3.392e-05,
+      "loss": 0.1124,
+      "step": 5216
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 3.3915e-05,
+      "loss": 0.1413,
+      "step": 5217
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 3.3910000000000006e-05,
+      "loss": 0.1485,
+      "step": 5218
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 3.3905e-05,
+      "loss": 0.2869,
+      "step": 5219
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 3.39e-05,
+      "loss": 0.2388,
+      "step": 5220
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 3.3895e-05,
+      "loss": 0.2375,
+      "step": 5221
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 3.3889999999999996e-05,
+      "loss": 0.1349,
+      "step": 5222
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 3.3885e-05,
+      "loss": 0.1518,
+      "step": 5223
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 3.388e-05,
+      "loss": 0.1694,
+      "step": 5224
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 3.3875000000000003e-05,
+      "loss": 0.2129,
+      "step": 5225
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 3.387e-05,
+      "loss": 0.2492,
+      "step": 5226
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 3.3865e-05,
+      "loss": 0.0417,
+      "step": 5227
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 3.3860000000000004e-05,
+      "loss": 0.0524,
+      "step": 5228
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 3.3855e-05,
+      "loss": 0.0927,
+      "step": 5229
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 3.385e-05,
+      "loss": 0.0655,
+      "step": 5230
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 3.3845000000000005e-05,
+      "loss": 0.1048,
+      "step": 5231
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 3.3839999999999994e-05,
+      "loss": 0.2143,
+      "step": 5232
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 3.3835e-05,
+      "loss": 0.2683,
+      "step": 5233
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 3.383e-05,
+      "loss": 0.1278,
+      "step": 5234
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 3.3825e-05,
+      "loss": 0.1662,
+      "step": 5235
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 3.382e-05,
+      "loss": 0.1723,
+      "step": 5236
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 3.3815e-05,
+      "loss": 0.0925,
+      "step": 5237
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 3.381e-05,
+      "loss": 0.1158,
+      "step": 5238
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 3.3805e-05,
+      "loss": 0.0491,
+      "step": 5239
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 3.38e-05,
+      "loss": 0.186,
+      "step": 5240
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 3.3795000000000004e-05,
+      "loss": 0.268,
+      "step": 5241
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 3.379e-05,
+      "loss": 0.1594,
+      "step": 5242
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 3.3785e-05,
+      "loss": 0.1385,
+      "step": 5243
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 3.378e-05,
+      "loss": 0.1148,
+      "step": 5244
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 3.3775e-05,
+      "loss": 0.0682,
+      "step": 5245
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 3.377e-05,
+      "loss": 0.0649,
+      "step": 5246
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 3.3765e-05,
+      "loss": 0.272,
+      "step": 5247
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 3.376e-05,
+      "loss": 0.0716,
+      "step": 5248
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 3.3755e-05,
+      "loss": 0.0717,
+      "step": 5249
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 3.375e-05,
+      "loss": 0.1067,
+      "step": 5250
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 3.3745e-05,
+      "loss": 0.1785,
+      "step": 5251
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 3.374e-05,
+      "loss": 0.0679,
+      "step": 5252
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 3.3735e-05,
+      "loss": 0.0624,
+      "step": 5253
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 3.3730000000000004e-05,
+      "loss": 0.1351,
+      "step": 5254
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 3.3725000000000007e-05,
+      "loss": 0.1424,
+      "step": 5255
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 3.372e-05,
+      "loss": 0.1003,
+      "step": 5256
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 3.3715e-05,
+      "loss": 0.171,
+      "step": 5257
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 3.371e-05,
+      "loss": 0.1522,
+      "step": 5258
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 3.3705e-05,
+      "loss": 0.1397,
+      "step": 5259
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 3.37e-05,
+      "loss": 0.145,
+      "step": 5260
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 3.3695e-05,
+      "loss": 0.2107,
+      "step": 5261
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 3.369e-05,
+      "loss": 0.0552,
+      "step": 5262
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 3.3685e-05,
+      "loss": 0.1658,
+      "step": 5263
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 3.368e-05,
+      "loss": 0.2166,
+      "step": 5264
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 3.3675000000000005e-05,
+      "loss": 0.064,
+      "step": 5265
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 3.367e-05,
+      "loss": 0.0915,
+      "step": 5266
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 3.3665000000000004e-05,
+      "loss": 0.0698,
+      "step": 5267
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 3.3660000000000006e-05,
+      "loss": 0.1895,
+      "step": 5268
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 3.3654999999999995e-05,
+      "loss": 0.0841,
+      "step": 5269
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 3.365e-05,
+      "loss": 0.1636,
+      "step": 5270
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 3.3645e-05,
+      "loss": 0.1425,
+      "step": 5271
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 3.3639999999999996e-05,
+      "loss": 0.0984,
+      "step": 5272
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 3.3635e-05,
+      "loss": 0.071,
+      "step": 5273
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 3.363e-05,
+      "loss": 0.1463,
+      "step": 5274
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 3.3625000000000004e-05,
+      "loss": 0.1584,
+      "step": 5275
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 3.362e-05,
+      "loss": 0.2633,
+      "step": 5276
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 3.3615e-05,
+      "loss": 0.055,
+      "step": 5277
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 3.3610000000000005e-05,
+      "loss": 0.17,
+      "step": 5278
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 3.3605e-05,
+      "loss": 0.1463,
+      "step": 5279
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 0.0668,
+      "step": 5280
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 3.3595e-05,
+      "loss": 0.0691,
+      "step": 5281
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 3.3589999999999995e-05,
+      "loss": 0.085,
+      "step": 5282
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 3.3585e-05,
+      "loss": 0.1089,
+      "step": 5283
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 3.358e-05,
+      "loss": 0.1712,
+      "step": 5284
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 3.3575e-05,
+      "loss": 0.155,
+      "step": 5285
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 3.357e-05,
+      "loss": 0.2272,
+      "step": 5286
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.3565e-05,
+      "loss": 0.1951,
+      "step": 5287
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.3560000000000004e-05,
+      "loss": 0.046,
+      "step": 5288
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.3555e-05,
+      "loss": 0.2515,
+      "step": 5289
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.355e-05,
+      "loss": 0.0843,
+      "step": 5290
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 3.3545000000000005e-05,
+      "loss": 0.1093,
+      "step": 5291
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 3.354e-05,
+      "loss": 0.0667,
+      "step": 5292
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 3.3534999999999997e-05,
+      "loss": 0.0661,
+      "step": 5293
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 3.353e-05,
+      "loss": 0.0854,
+      "step": 5294
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 3.3525e-05,
+      "loss": 0.1301,
+      "step": 5295
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 3.352e-05,
+      "loss": 0.3033,
+      "step": 5296
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 3.3515e-05,
+      "loss": 0.1757,
+      "step": 5297
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 3.351e-05,
+      "loss": 0.0681,
+      "step": 5298
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 3.3505e-05,
+      "loss": 0.22,
+      "step": 5299
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 3.35e-05,
+      "loss": 0.2691,
+      "step": 5300
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 3.3495000000000004e-05,
+      "loss": 0.126,
+      "step": 5301
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 3.349e-05,
+      "loss": 0.1202,
+      "step": 5302
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 3.3485e-05,
+      "loss": 0.0567,
+      "step": 5303
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 3.3480000000000005e-05,
+      "loss": 0.1206,
+      "step": 5304
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 3.3475e-05,
+      "loss": 0.1861,
+      "step": 5305
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 3.3469999999999996e-05,
+      "loss": 0.1688,
+      "step": 5306
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 3.3465e-05,
+      "loss": 0.2615,
+      "step": 5307
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 3.346e-05,
+      "loss": 0.1604,
+      "step": 5308
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 3.3455e-05,
+      "loss": 0.0657,
+      "step": 5309
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 3.345e-05,
+      "loss": 0.1637,
+      "step": 5310
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 3.3445e-05,
+      "loss": 0.0453,
+      "step": 5311
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 3.344e-05,
+      "loss": 0.1821,
+      "step": 5312
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 3.3435e-05,
+      "loss": 0.1813,
+      "step": 5313
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 3.3430000000000003e-05,
+      "loss": 0.0949,
+      "step": 5314
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 3.3425000000000006e-05,
+      "loss": 0.1053,
+      "step": 5315
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 3.342e-05,
+      "loss": 0.0587,
+      "step": 5316
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 3.3415e-05,
+      "loss": 0.0376,
+      "step": 5317
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 3.341e-05,
+      "loss": 0.1589,
+      "step": 5318
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 3.3404999999999996e-05,
+      "loss": 0.1231,
+      "step": 5319
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 3.34e-05,
+      "loss": 0.1946,
+      "step": 5320
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 3.3395e-05,
+      "loss": 0.0631,
+      "step": 5321
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 3.3390000000000004e-05,
+      "loss": 0.1478,
+      "step": 5322
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 3.3385e-05,
+      "loss": 0.2204,
+      "step": 5323
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 3.338e-05,
+      "loss": 0.1399,
+      "step": 5324
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 3.3375000000000005e-05,
+      "loss": 0.0689,
+      "step": 5325
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 3.337e-05,
+      "loss": 0.2607,
+      "step": 5326
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 3.3365e-05,
+      "loss": 0.1205,
+      "step": 5327
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 3.3360000000000006e-05,
+      "loss": 0.2064,
+      "step": 5328
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 3.3354999999999995e-05,
+      "loss": 0.2044,
+      "step": 5329
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 3.335e-05,
+      "loss": 0.1579,
+      "step": 5330
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 3.3345e-05,
+      "loss": 0.0853,
+      "step": 5331
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 3.334e-05,
+      "loss": 0.1569,
+      "step": 5332
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 3.3335e-05,
+      "loss": 0.0697,
+      "step": 5333
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 3.333e-05,
+      "loss": 0.2438,
+      "step": 5334
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 3.3325000000000004e-05,
+      "loss": 0.0973,
+      "step": 5335
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 3.332e-05,
+      "loss": 0.0949,
+      "step": 5336
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 3.3315e-05,
+      "loss": 0.0777,
+      "step": 5337
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 3.3310000000000005e-05,
+      "loss": 0.0548,
+      "step": 5338
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 3.3305e-05,
+      "loss": 0.0501,
+      "step": 5339
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 3.33e-05,
+      "loss": 0.0996,
+      "step": 5340
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 3.3295e-05,
+      "loss": 0.068,
+      "step": 5341
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 3.329e-05,
+      "loss": 0.0472,
+      "step": 5342
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 3.3285e-05,
+      "loss": 0.1007,
+      "step": 5343
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 3.328e-05,
+      "loss": 0.0599,
+      "step": 5344
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 3.3275e-05,
+      "loss": 0.0609,
+      "step": 5345
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 3.327e-05,
+      "loss": 0.0662,
+      "step": 5346
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 3.3265e-05,
+      "loss": 0.2764,
+      "step": 5347
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 3.3260000000000003e-05,
+      "loss": 0.0703,
+      "step": 5348
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 3.3255e-05,
+      "loss": 0.1868,
+      "step": 5349
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 3.325e-05,
+      "loss": 0.1251,
+      "step": 5350
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 3.3245000000000004e-05,
+      "loss": 0.0894,
+      "step": 5351
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 3.324000000000001e-05,
+      "loss": 0.1455,
+      "step": 5352
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 3.3234999999999996e-05,
+      "loss": 0.2289,
+      "step": 5353
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 3.323e-05,
+      "loss": 0.172,
+      "step": 5354
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 3.3225e-05,
+      "loss": 0.0526,
+      "step": 5355
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 3.322e-05,
+      "loss": 0.0714,
+      "step": 5356
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 3.3215e-05,
+      "loss": 0.181,
+      "step": 5357
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 3.321e-05,
+      "loss": 0.1216,
+      "step": 5358
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 3.3205e-05,
+      "loss": 0.0603,
+      "step": 5359
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 3.32e-05,
+      "loss": 0.1388,
+      "step": 5360
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 3.3195e-05,
+      "loss": 0.1302,
+      "step": 5361
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 3.3190000000000006e-05,
+      "loss": 0.0654,
+      "step": 5362
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 3.3185e-05,
+      "loss": 0.0896,
+      "step": 5363
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 3.3180000000000004e-05,
+      "loss": 0.1508,
+      "step": 5364
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 3.3175e-05,
+      "loss": 0.3067,
+      "step": 5365
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 3.3169999999999996e-05,
+      "loss": 0.0562,
+      "step": 5366
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 3.3165e-05,
+      "loss": 0.1365,
+      "step": 5367
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 3.316e-05,
+      "loss": 0.0477,
+      "step": 5368
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 3.3155e-05,
+      "loss": 0.3753,
+      "step": 5369
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 3.315e-05,
+      "loss": 0.0745,
+      "step": 5370
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 3.3145e-05,
+      "loss": 0.122,
+      "step": 5371
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 3.3140000000000005e-05,
+      "loss": 0.129,
+      "step": 5372
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 3.3135e-05,
+      "loss": 0.0377,
+      "step": 5373
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 3.313e-05,
+      "loss": 0.1876,
+      "step": 5374
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 3.3125000000000006e-05,
+      "loss": 0.0413,
+      "step": 5375
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 3.312e-05,
+      "loss": 0.0861,
+      "step": 5376
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 3.3115000000000004e-05,
+      "loss": 0.0973,
+      "step": 5377
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 3.311e-05,
+      "loss": 0.1653,
+      "step": 5378
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 3.3104999999999996e-05,
+      "loss": 0.2865,
+      "step": 5379
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 3.31e-05,
+      "loss": 0.1404,
+      "step": 5380
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 3.3095e-05,
+      "loss": 0.1304,
+      "step": 5381
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 3.309e-05,
+      "loss": 0.1375,
+      "step": 5382
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 3.3085e-05,
+      "loss": 0.0751,
+      "step": 5383
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 3.308e-05,
+      "loss": 0.0583,
+      "step": 5384
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 3.3075000000000004e-05,
+      "loss": 0.2322,
+      "step": 5385
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 3.307e-05,
+      "loss": 0.2051,
+      "step": 5386
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 3.3065e-05,
+      "loss": 0.1958,
+      "step": 5387
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 3.3060000000000005e-05,
+      "loss": 0.2532,
+      "step": 5388
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 3.3055e-05,
+      "loss": 0.0499,
+      "step": 5389
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 3.305e-05,
+      "loss": 0.1128,
+      "step": 5390
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 3.3045e-05,
+      "loss": 0.1666,
+      "step": 5391
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 3.304e-05,
+      "loss": 0.1727,
+      "step": 5392
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 3.3035e-05,
+      "loss": 0.23,
+      "step": 5393
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 3.303e-05,
+      "loss": 0.3563,
+      "step": 5394
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 3.3025e-05,
+      "loss": 0.1929,
+      "step": 5395
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 3.302e-05,
+      "loss": 0.0488,
+      "step": 5396
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 3.3015e-05,
+      "loss": 0.1694,
+      "step": 5397
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 3.3010000000000004e-05,
+      "loss": 0.0518,
+      "step": 5398
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 3.3005e-05,
+      "loss": 0.1332,
+      "step": 5399
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 3.3e-05,
+      "loss": 0.1354,
+      "step": 5400
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 3.2995000000000005e-05,
+      "loss": 0.0559,
+      "step": 5401
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 3.299e-05,
+      "loss": 0.172,
+      "step": 5402
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 3.2985e-05,
+      "loss": 0.0454,
+      "step": 5403
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 3.298e-05,
+      "loss": 0.0661,
+      "step": 5404
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 3.2975e-05,
+      "loss": 0.2924,
+      "step": 5405
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 3.297e-05,
+      "loss": 0.0546,
+      "step": 5406
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 3.2965e-05,
+      "loss": 0.184,
+      "step": 5407
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 3.296e-05,
+      "loss": 0.166,
+      "step": 5408
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 3.2955e-05,
+      "loss": 0.2629,
+      "step": 5409
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 3.295e-05,
+      "loss": 0.1153,
+      "step": 5410
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 3.2945000000000004e-05,
+      "loss": 0.0684,
+      "step": 5411
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 3.2940000000000006e-05,
+      "loss": 0.1519,
+      "step": 5412
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 3.2935e-05,
+      "loss": 0.1326,
+      "step": 5413
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 3.293e-05,
+      "loss": 0.0575,
+      "step": 5414
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 3.2925e-05,
+      "loss": 0.1438,
+      "step": 5415
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 3.2919999999999997e-05,
+      "loss": 0.2207,
+      "step": 5416
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 3.2915e-05,
+      "loss": 0.1836,
+      "step": 5417
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 3.291e-05,
+      "loss": 0.1631,
+      "step": 5418
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 3.2905e-05,
+      "loss": 0.0738,
+      "step": 5419
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 3.29e-05,
+      "loss": 0.3049,
+      "step": 5420
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 3.2895e-05,
+      "loss": 0.1531,
+      "step": 5421
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 3.2890000000000005e-05,
+      "loss": 0.3007,
+      "step": 5422
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 3.2885e-05,
+      "loss": 0.2039,
+      "step": 5423
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 3.2880000000000004e-05,
+      "loss": 0.0787,
+      "step": 5424
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 3.2875000000000006e-05,
+      "loss": 0.0828,
+      "step": 5425
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 3.2869999999999995e-05,
+      "loss": 0.0843,
+      "step": 5426
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 3.2865e-05,
+      "loss": 0.1417,
+      "step": 5427
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 3.286e-05,
+      "loss": 0.171,
+      "step": 5428
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 3.2854999999999996e-05,
+      "loss": 0.0836,
+      "step": 5429
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 3.285e-05,
+      "loss": 0.0702,
+      "step": 5430
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 3.2845e-05,
+      "loss": 0.1242,
+      "step": 5431
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 3.2840000000000004e-05,
+      "loss": 0.1091,
+      "step": 5432
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 3.2835e-05,
+      "loss": 0.2567,
+      "step": 5433
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 3.283e-05,
+      "loss": 0.1478,
+      "step": 5434
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 3.2825000000000005e-05,
+      "loss": 0.2411,
+      "step": 5435
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 3.282e-05,
+      "loss": 0.1205,
+      "step": 5436
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 3.2815000000000003e-05,
+      "loss": 0.1446,
+      "step": 5437
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 3.281e-05,
+      "loss": 0.0781,
+      "step": 5438
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 3.2804999999999995e-05,
+      "loss": 0.2082,
+      "step": 5439
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 3.28e-05,
+      "loss": 0.0746,
+      "step": 5440
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 3.2795e-05,
+      "loss": 0.0902,
+      "step": 5441
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 3.279e-05,
+      "loss": 0.1617,
+      "step": 5442
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 3.2785e-05,
+      "loss": 0.085,
+      "step": 5443
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 3.278e-05,
+      "loss": 0.0503,
+      "step": 5444
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 3.2775000000000004e-05,
+      "loss": 0.1866,
+      "step": 5445
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 3.277e-05,
+      "loss": 0.2858,
+      "step": 5446
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 3.2765e-05,
+      "loss": 0.2055,
+      "step": 5447
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 3.2760000000000005e-05,
+      "loss": 0.2202,
+      "step": 5448
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 3.2755e-05,
+      "loss": 0.0789,
+      "step": 5449
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 3.2749999999999996e-05,
+      "loss": 0.057,
+      "step": 5450
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 3.2745e-05,
+      "loss": 0.1657,
+      "step": 5451
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 3.274e-05,
+      "loss": 0.0626,
+      "step": 5452
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 3.2735e-05,
+      "loss": 0.214,
+      "step": 5453
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 3.273e-05,
+      "loss": 0.2274,
+      "step": 5454
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 3.2725e-05,
+      "loss": 0.1593,
+      "step": 5455
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 3.272e-05,
+      "loss": 0.0808,
+      "step": 5456
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 3.2715e-05,
+      "loss": 0.2144,
+      "step": 5457
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 3.2710000000000004e-05,
+      "loss": 0.1519,
+      "step": 5458
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 3.2705e-05,
+      "loss": 0.0691,
+      "step": 5459
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 3.27e-05,
+      "loss": 0.0547,
+      "step": 5460
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 3.2695000000000005e-05,
+      "loss": 0.0397,
+      "step": 5461
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 3.269e-05,
+      "loss": 0.1451,
+      "step": 5462
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 3.2684999999999996e-05,
+      "loss": 0.0948,
+      "step": 5463
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 3.268e-05,
+      "loss": 0.0416,
+      "step": 5464
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 3.2675e-05,
+      "loss": 0.1647,
+      "step": 5465
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 3.267e-05,
+      "loss": 0.2726,
+      "step": 5466
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 3.2665e-05,
+      "loss": 0.1014,
+      "step": 5467
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 3.266e-05,
+      "loss": 0.1421,
+      "step": 5468
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 3.2655e-05,
+      "loss": 0.089,
+      "step": 5469
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 3.265e-05,
+      "loss": 0.1842,
+      "step": 5470
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 3.2645e-05,
+      "loss": 0.1726,
+      "step": 5471
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 3.2640000000000006e-05,
+      "loss": 0.2739,
+      "step": 5472
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 3.2635e-05,
+      "loss": 0.0933,
+      "step": 5473
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 3.263e-05,
+      "loss": 0.0712,
+      "step": 5474
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 3.2625e-05,
+      "loss": 0.2935,
+      "step": 5475
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 3.2619999999999996e-05,
+      "loss": 0.2184,
+      "step": 5476
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 3.2615e-05,
+      "loss": 0.0663,
+      "step": 5477
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 3.261e-05,
+      "loss": 0.1438,
+      "step": 5478
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 3.2605e-05,
+      "loss": 0.1118,
+      "step": 5479
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 3.26e-05,
+      "loss": 0.0643,
+      "step": 5480
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 3.2595e-05,
+      "loss": 0.1191,
+      "step": 5481
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 3.2590000000000005e-05,
+      "loss": 0.1629,
+      "step": 5482
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 3.2585e-05,
+      "loss": 0.2012,
+      "step": 5483
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 3.258e-05,
+      "loss": 0.0596,
+      "step": 5484
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 3.2575000000000006e-05,
+      "loss": 0.0552,
+      "step": 5485
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 3.2569999999999995e-05,
+      "loss": 0.0701,
+      "step": 5486
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 3.2565e-05,
+      "loss": 0.0455,
+      "step": 5487
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 3.256e-05,
+      "loss": 0.2313,
+      "step": 5488
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 3.2555e-05,
+      "loss": 0.1789,
+      "step": 5489
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 3.255e-05,
+      "loss": 0.1754,
+      "step": 5490
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 3.2545e-05,
+      "loss": 0.119,
+      "step": 5491
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 3.2540000000000004e-05,
+      "loss": 0.1845,
+      "step": 5492
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 3.2535e-05,
+      "loss": 0.1687,
+      "step": 5493
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 3.253e-05,
+      "loss": 0.1694,
+      "step": 5494
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 3.2525000000000005e-05,
+      "loss": 0.0558,
+      "step": 5495
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 3.252e-05,
+      "loss": 0.0777,
+      "step": 5496
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 3.2515e-05,
+      "loss": 0.0436,
+      "step": 5497
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 3.251e-05,
+      "loss": 0.0995,
+      "step": 5498
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 3.2505e-05,
+      "loss": 0.067,
+      "step": 5499
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 3.25e-05,
+      "loss": 0.1787,
+      "step": 5500
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 3.2495e-05,
+      "loss": 0.1053,
+      "step": 5501
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 3.249e-05,
+      "loss": 0.146,
+      "step": 5502
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 3.2485e-05,
+      "loss": 0.1717,
+      "step": 5503
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 3.248e-05,
+      "loss": 0.1389,
+      "step": 5504
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 3.2475e-05,
+      "loss": 0.2789,
+      "step": 5505
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 3.247e-05,
+      "loss": 0.1339,
+      "step": 5506
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 3.2465e-05,
+      "loss": 0.1275,
+      "step": 5507
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 3.2460000000000004e-05,
+      "loss": 0.1737,
+      "step": 5508
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 3.245500000000001e-05,
+      "loss": 0.2218,
+      "step": 5509
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 3.245e-05,
+      "loss": 0.1246,
+      "step": 5510
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 3.2445e-05,
+      "loss": 0.0618,
+      "step": 5511
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 3.244e-05,
+      "loss": 0.1403,
+      "step": 5512
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 3.2435e-05,
+      "loss": 0.0586,
+      "step": 5513
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 3.243e-05,
+      "loss": 0.1996,
+      "step": 5514
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 3.2425e-05,
+      "loss": 0.1771,
+      "step": 5515
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 3.242e-05,
+      "loss": 0.0706,
+      "step": 5516
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 3.2415e-05,
+      "loss": 0.2743,
+      "step": 5517
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 3.241e-05,
+      "loss": 0.095,
+      "step": 5518
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 3.2405000000000006e-05,
+      "loss": 0.1427,
+      "step": 5519
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 3.24e-05,
+      "loss": 0.0707,
+      "step": 5520
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 3.2395000000000004e-05,
+      "loss": 0.091,
+      "step": 5521
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 3.239000000000001e-05,
+      "loss": 0.061,
+      "step": 5522
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 3.2384999999999996e-05,
+      "loss": 0.1394,
+      "step": 5523
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 3.238e-05,
+      "loss": 0.2245,
+      "step": 5524
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 3.2375e-05,
+      "loss": 0.0845,
+      "step": 5525
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 3.237e-05,
+      "loss": 0.0881,
+      "step": 5526
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 3.2365e-05,
+      "loss": 0.0759,
+      "step": 5527
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 3.236e-05,
+      "loss": 0.1205,
+      "step": 5528
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 3.2355000000000004e-05,
+      "loss": 0.1147,
+      "step": 5529
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 3.235e-05,
+      "loss": 0.0789,
+      "step": 5530
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 3.2345e-05,
+      "loss": 0.1236,
+      "step": 5531
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 3.2340000000000005e-05,
+      "loss": 0.0655,
+      "step": 5532
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 3.2335e-05,
+      "loss": 0.0403,
+      "step": 5533
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 3.2330000000000004e-05,
+      "loss": 0.0621,
+      "step": 5534
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 3.2325e-05,
+      "loss": 0.2922,
+      "step": 5535
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 3.2319999999999995e-05,
+      "loss": 0.1729,
+      "step": 5536
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 3.2315e-05,
+      "loss": 0.0822,
+      "step": 5537
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 3.231e-05,
+      "loss": 0.1595,
+      "step": 5538
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 3.2305e-05,
+      "loss": 0.0628,
+      "step": 5539
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 3.23e-05,
+      "loss": 0.0788,
+      "step": 5540
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 3.2295e-05,
+      "loss": 0.0853,
+      "step": 5541
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 3.2290000000000004e-05,
+      "loss": 0.1823,
+      "step": 5542
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 3.2285e-05,
+      "loss": 0.1604,
+      "step": 5543
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 3.228e-05,
+      "loss": 0.0898,
+      "step": 5544
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 3.2275000000000005e-05,
+      "loss": 0.0868,
+      "step": 5545
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 3.227e-05,
+      "loss": 0.1566,
+      "step": 5546
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 3.2265e-05,
+      "loss": 0.0793,
+      "step": 5547
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 3.226e-05,
+      "loss": 0.1814,
+      "step": 5548
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 3.2255e-05,
+      "loss": 0.0837,
+      "step": 5549
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 3.225e-05,
+      "loss": 0.1426,
+      "step": 5550
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 3.2245e-05,
+      "loss": 0.0456,
+      "step": 5551
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 3.224e-05,
+      "loss": 0.0563,
+      "step": 5552
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 3.2235e-05,
+      "loss": 0.2439,
+      "step": 5553
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 3.223e-05,
+      "loss": 0.0453,
+      "step": 5554
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 3.2225000000000004e-05,
+      "loss": 0.0546,
+      "step": 5555
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 3.222e-05,
+      "loss": 0.2518,
+      "step": 5556
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 3.2215e-05,
+      "loss": 0.0542,
+      "step": 5557
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 3.2210000000000005e-05,
+      "loss": 0.0846,
+      "step": 5558
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 3.2205e-05,
+      "loss": 0.1366,
+      "step": 5559
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 3.22e-05,
+      "loss": 0.2948,
+      "step": 5560
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 3.2195e-05,
+      "loss": 0.0684,
+      "step": 5561
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 3.219e-05,
+      "loss": 0.128,
+      "step": 5562
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 3.2185e-05,
+      "loss": 0.0592,
+      "step": 5563
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 3.218e-05,
+      "loss": 0.0285,
+      "step": 5564
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 3.2175e-05,
+      "loss": 0.1128,
+      "step": 5565
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 3.217e-05,
+      "loss": 0.0635,
+      "step": 5566
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 3.2165e-05,
+      "loss": 0.1057,
+      "step": 5567
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 3.2160000000000004e-05,
+      "loss": 0.2286,
+      "step": 5568
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 3.2155000000000006e-05,
+      "loss": 0.1172,
+      "step": 5569
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 3.215e-05,
+      "loss": 0.1182,
+      "step": 5570
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 3.2145e-05,
+      "loss": 0.0571,
+      "step": 5571
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 3.214e-05,
+      "loss": 0.1147,
+      "step": 5572
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 3.2134999999999996e-05,
+      "loss": 0.1328,
+      "step": 5573
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 3.213e-05,
+      "loss": 0.2233,
+      "step": 5574
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 3.2125e-05,
+      "loss": 0.2361,
+      "step": 5575
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 3.212e-05,
+      "loss": 0.1837,
+      "step": 5576
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 3.2115e-05,
+      "loss": 0.0487,
+      "step": 5577
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 3.211e-05,
+      "loss": 0.1326,
+      "step": 5578
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 3.2105000000000005e-05,
+      "loss": 0.0664,
+      "step": 5579
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 3.21e-05,
+      "loss": 0.0933,
+      "step": 5580
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 3.2095000000000004e-05,
+      "loss": 0.1727,
+      "step": 5581
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 3.2090000000000006e-05,
+      "loss": 0.1955,
+      "step": 5582
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 3.2084999999999995e-05,
+      "loss": 0.1563,
+      "step": 5583
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 3.208e-05,
+      "loss": 0.1733,
+      "step": 5584
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 3.2075e-05,
+      "loss": 0.3044,
+      "step": 5585
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 3.2069999999999996e-05,
+      "loss": 0.173,
+      "step": 5586
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 3.2065e-05,
+      "loss": 0.1284,
+      "step": 5587
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 3.206e-05,
+      "loss": 0.0513,
+      "step": 5588
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 3.2055000000000004e-05,
+      "loss": 0.1257,
+      "step": 5589
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 3.205e-05,
+      "loss": 0.0398,
+      "step": 5590
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 3.2045e-05,
+      "loss": 0.1443,
+      "step": 5591
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 3.2040000000000005e-05,
+      "loss": 0.1447,
+      "step": 5592
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 3.2035e-05,
+      "loss": 0.1229,
+      "step": 5593
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 3.203e-05,
+      "loss": 0.2248,
+      "step": 5594
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 3.2025e-05,
+      "loss": 0.2033,
+      "step": 5595
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 3.2019999999999995e-05,
+      "loss": 0.1444,
+      "step": 5596
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 3.2015e-05,
+      "loss": 0.1735,
+      "step": 5597
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 3.201e-05,
+      "loss": 0.089,
+      "step": 5598
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 3.2005e-05,
+      "loss": 0.1005,
+      "step": 5599
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 3.2e-05,
+      "loss": 0.1833,
+      "step": 5600
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 3.1995e-05,
+      "loss": 0.1699,
+      "step": 5601
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 3.1990000000000004e-05,
+      "loss": 0.1851,
+      "step": 5602
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 3.1985e-05,
+      "loss": 0.191,
+      "step": 5603
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 3.198e-05,
+      "loss": 0.0533,
+      "step": 5604
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 3.1975000000000005e-05,
+      "loss": 0.0874,
+      "step": 5605
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 3.197e-05,
+      "loss": 0.1005,
+      "step": 5606
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 3.1964999999999996e-05,
+      "loss": 0.0661,
+      "step": 5607
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 3.196e-05,
+      "loss": 0.2291,
+      "step": 5608
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 3.1955e-05,
+      "loss": 0.1663,
+      "step": 5609
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 3.195e-05,
+      "loss": 0.1172,
+      "step": 5610
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 3.1945e-05,
+      "loss": 0.1146,
+      "step": 5611
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 3.194e-05,
+      "loss": 0.2475,
+      "step": 5612
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 3.1935e-05,
+      "loss": 0.1387,
+      "step": 5613
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 3.193e-05,
+      "loss": 0.0926,
+      "step": 5614
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 3.1925000000000004e-05,
+      "loss": 0.1763,
+      "step": 5615
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 3.192e-05,
+      "loss": 0.1664,
+      "step": 5616
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 3.1915e-05,
+      "loss": 0.1206,
+      "step": 5617
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 3.1910000000000004e-05,
+      "loss": 0.0856,
+      "step": 5618
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 3.1905e-05,
+      "loss": 0.07,
+      "step": 5619
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 3.1899999999999996e-05,
+      "loss": 0.1678,
+      "step": 5620
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 3.1895e-05,
+      "loss": 0.0482,
+      "step": 5621
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 3.189e-05,
+      "loss": 0.1812,
+      "step": 5622
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 3.1885e-05,
+      "loss": 0.158,
+      "step": 5623
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 3.188e-05,
+      "loss": 0.1066,
+      "step": 5624
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 3.1875e-05,
+      "loss": 0.1413,
+      "step": 5625
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 3.187e-05,
+      "loss": 0.072,
+      "step": 5626
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 3.1865e-05,
+      "loss": 0.1881,
+      "step": 5627
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 3.186e-05,
+      "loss": 0.1554,
+      "step": 5628
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 3.1855000000000006e-05,
+      "loss": 0.1414,
+      "step": 5629
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 3.185e-05,
+      "loss": 0.0741,
+      "step": 5630
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 3.1845000000000004e-05,
+      "loss": 0.24,
+      "step": 5631
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 3.184e-05,
+      "loss": 0.0601,
+      "step": 5632
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 3.1834999999999996e-05,
+      "loss": 0.0422,
+      "step": 5633
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 3.183e-05,
+      "loss": 0.084,
+      "step": 5634
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 3.1825e-05,
+      "loss": 0.1347,
+      "step": 5635
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 3.182e-05,
+      "loss": 0.1767,
+      "step": 5636
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 3.1815e-05,
+      "loss": 0.1244,
+      "step": 5637
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 3.181e-05,
+      "loss": 0.1158,
+      "step": 5638
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 3.1805000000000005e-05,
+      "loss": 0.147,
+      "step": 5639
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 3.18e-05,
+      "loss": 0.0538,
+      "step": 5640
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 3.1795e-05,
+      "loss": 0.1836,
+      "step": 5641
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 3.1790000000000006e-05,
+      "loss": 0.1781,
+      "step": 5642
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 3.1785e-05,
+      "loss": 0.1501,
+      "step": 5643
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 3.178e-05,
+      "loss": 0.1077,
+      "step": 5644
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 3.1775e-05,
+      "loss": 0.0974,
+      "step": 5645
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 3.1769999999999996e-05,
+      "loss": 0.1009,
+      "step": 5646
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 3.1765e-05,
+      "loss": 0.0527,
+      "step": 5647
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 3.176e-05,
+      "loss": 0.0575,
+      "step": 5648
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 3.1755000000000003e-05,
+      "loss": 0.1549,
+      "step": 5649
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 3.175e-05,
+      "loss": 0.1402,
+      "step": 5650
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 3.1745e-05,
+      "loss": 0.1972,
+      "step": 5651
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 3.1740000000000004e-05,
+      "loss": 0.0925,
+      "step": 5652
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 3.1735e-05,
+      "loss": 0.1166,
+      "step": 5653
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 3.173e-05,
+      "loss": 0.1653,
+      "step": 5654
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 3.1725000000000005e-05,
+      "loss": 0.0831,
+      "step": 5655
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 3.172e-05,
+      "loss": 0.1521,
+      "step": 5656
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 3.1715e-05,
+      "loss": 0.2116,
+      "step": 5657
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 3.171e-05,
+      "loss": 0.1864,
+      "step": 5658
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 3.1705e-05,
+      "loss": 0.1773,
+      "step": 5659
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 3.17e-05,
+      "loss": 0.1391,
+      "step": 5660
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 3.1695e-05,
+      "loss": 0.128,
+      "step": 5661
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 3.169e-05,
+      "loss": 0.1142,
+      "step": 5662
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 3.1685e-05,
+      "loss": 0.0981,
+      "step": 5663
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 3.168e-05,
+      "loss": 0.061,
+      "step": 5664
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 3.1675000000000004e-05,
+      "loss": 0.1175,
+      "step": 5665
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 3.167e-05,
+      "loss": 0.1535,
+      "step": 5666
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 3.1665e-05,
+      "loss": 0.1845,
+      "step": 5667
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 3.166e-05,
+      "loss": 0.0645,
+      "step": 5668
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 3.1655e-05,
+      "loss": 0.2822,
+      "step": 5669
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 3.165e-05,
+      "loss": 0.2162,
+      "step": 5670
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 3.1645e-05,
+      "loss": 0.0464,
+      "step": 5671
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 3.164e-05,
+      "loss": 0.1814,
+      "step": 5672
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 3.1635e-05,
+      "loss": 0.346,
+      "step": 5673
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 3.163e-05,
+      "loss": 0.1957,
+      "step": 5674
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 3.1625e-05,
+      "loss": 0.2622,
+      "step": 5675
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 3.1620000000000006e-05,
+      "loss": 0.2962,
+      "step": 5676
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 3.1615e-05,
+      "loss": 0.1914,
+      "step": 5677
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 3.1610000000000004e-05,
+      "loss": 0.0583,
+      "step": 5678
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 3.1605000000000007e-05,
+      "loss": 0.0703,
+      "step": 5679
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 3.1599999999999996e-05,
+      "loss": 0.0803,
+      "step": 5680
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 3.1595e-05,
+      "loss": 0.1155,
+      "step": 5681
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 3.159e-05,
+      "loss": 0.1456,
+      "step": 5682
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 3.1584999999999997e-05,
+      "loss": 0.1634,
+      "step": 5683
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 3.158e-05,
+      "loss": 0.0562,
+      "step": 5684
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 3.1575e-05,
+      "loss": 0.1887,
+      "step": 5685
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 3.1570000000000004e-05,
+      "loss": 0.0872,
+      "step": 5686
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 3.1565e-05,
+      "loss": 0.0653,
+      "step": 5687
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 3.156e-05,
+      "loss": 0.059,
+      "step": 5688
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 3.1555000000000005e-05,
+      "loss": 0.0834,
+      "step": 5689
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 3.155e-05,
+      "loss": 0.0731,
+      "step": 5690
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 3.1545000000000004e-05,
+      "loss": 0.1288,
+      "step": 5691
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 3.154e-05,
+      "loss": 0.0581,
+      "step": 5692
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 3.1534999999999995e-05,
+      "loss": 0.1272,
+      "step": 5693
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 3.153e-05,
+      "loss": 0.0842,
+      "step": 5694
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 3.1525e-05,
+      "loss": 0.0784,
+      "step": 5695
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 3.152e-05,
+      "loss": 0.1383,
+      "step": 5696
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 3.1515e-05,
+      "loss": 0.1615,
+      "step": 5697
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 3.151e-05,
+      "loss": 0.2939,
+      "step": 5698
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 3.1505000000000004e-05,
+      "loss": 0.056,
+      "step": 5699
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 3.15e-05,
+      "loss": 0.2806,
+      "step": 5700
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 3.1495e-05,
+      "loss": 0.2228,
+      "step": 5701
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 3.1490000000000005e-05,
+      "loss": 0.1844,
+      "step": 5702
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 3.1485e-05,
+      "loss": 0.1889,
+      "step": 5703
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 3.148e-05,
+      "loss": 0.145,
+      "step": 5704
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 3.1475e-05,
+      "loss": 0.129,
+      "step": 5705
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 3.147e-05,
+      "loss": 0.1515,
+      "step": 5706
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 3.1465e-05,
+      "loss": 0.0947,
+      "step": 5707
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 3.146e-05,
+      "loss": 0.0598,
+      "step": 5708
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 3.1455e-05,
+      "loss": 0.0812,
+      "step": 5709
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 3.145e-05,
+      "loss": 0.2104,
+      "step": 5710
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 3.1445e-05,
+      "loss": 0.0635,
+      "step": 5711
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 3.1440000000000004e-05,
+      "loss": 0.144,
+      "step": 5712
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 3.1435e-05,
+      "loss": 0.0885,
+      "step": 5713
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 3.143e-05,
+      "loss": 0.1148,
+      "step": 5714
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 3.1425000000000005e-05,
+      "loss": 0.1621,
+      "step": 5715
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 3.142e-05,
+      "loss": 0.2226,
+      "step": 5716
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 3.1414999999999997e-05,
+      "loss": 0.3135,
+      "step": 5717
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 3.141e-05,
+      "loss": 0.1516,
+      "step": 5718
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 3.1405e-05,
+      "loss": 0.121,
+      "step": 5719
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 3.14e-05,
+      "loss": 0.064,
+      "step": 5720
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 3.1395e-05,
+      "loss": 0.1984,
+      "step": 5721
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 3.139e-05,
+      "loss": 0.2152,
+      "step": 5722
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 3.1385e-05,
+      "loss": 0.0797,
+      "step": 5723
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 3.138e-05,
+      "loss": 0.1254,
+      "step": 5724
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 3.1375000000000004e-05,
+      "loss": 0.252,
+      "step": 5725
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 3.1370000000000006e-05,
+      "loss": 0.0563,
+      "step": 5726
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 3.1365e-05,
+      "loss": 0.1351,
+      "step": 5727
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 3.136e-05,
+      "loss": 0.2043,
+      "step": 5728
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 3.1355e-05,
+      "loss": 0.2038,
+      "step": 5729
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 3.1349999999999996e-05,
+      "loss": 0.169,
+      "step": 5730
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 3.1345e-05,
+      "loss": 0.0791,
+      "step": 5731
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 3.134e-05,
+      "loss": 0.1533,
+      "step": 5732
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 3.1335e-05,
+      "loss": 0.3012,
+      "step": 5733
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 3.133e-05,
+      "loss": 0.1559,
+      "step": 5734
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 3.1325e-05,
+      "loss": 0.0968,
+      "step": 5735
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 3.1320000000000005e-05,
+      "loss": 0.1476,
+      "step": 5736
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 3.1315e-05,
+      "loss": 0.1615,
+      "step": 5737
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 3.1310000000000003e-05,
+      "loss": 0.1908,
+      "step": 5738
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 3.1305000000000006e-05,
+      "loss": 0.1395,
+      "step": 5739
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 3.1299999999999995e-05,
+      "loss": 0.1119,
+      "step": 5740
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 3.1295e-05,
+      "loss": 0.1767,
+      "step": 5741
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 3.129e-05,
+      "loss": 0.0584,
+      "step": 5742
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 3.1284999999999996e-05,
+      "loss": 0.1382,
+      "step": 5743
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 3.128e-05,
+      "loss": 0.1674,
+      "step": 5744
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 3.1275e-05,
+      "loss": 0.1612,
+      "step": 5745
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 3.1270000000000004e-05,
+      "loss": 0.1925,
+      "step": 5746
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 3.1265e-05,
+      "loss": 0.2936,
+      "step": 5747
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 3.126e-05,
+      "loss": 0.0839,
+      "step": 5748
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 3.1255000000000005e-05,
+      "loss": 0.0747,
+      "step": 5749
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 3.125e-05,
+      "loss": 0.069,
+      "step": 5750
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 3.1245e-05,
+      "loss": 0.1178,
+      "step": 5751
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 3.1240000000000006e-05,
+      "loss": 0.0711,
+      "step": 5752
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 3.1234999999999995e-05,
+      "loss": 0.1343,
+      "step": 5753
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 3.123e-05,
+      "loss": 0.0582,
+      "step": 5754
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 3.1225e-05,
+      "loss": 0.1481,
+      "step": 5755
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 3.122e-05,
+      "loss": 0.0544,
+      "step": 5756
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 3.1215e-05,
+      "loss": 0.1115,
+      "step": 5757
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 3.121e-05,
+      "loss": 0.1746,
+      "step": 5758
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 3.1205000000000004e-05,
+      "loss": 0.056,
+      "step": 5759
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 3.12e-05,
+      "loss": 0.2219,
+      "step": 5760
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 3.1195e-05,
+      "loss": 0.2736,
+      "step": 5761
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.1190000000000005e-05,
+      "loss": 0.0604,
+      "step": 5762
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.1185e-05,
+      "loss": 0.0677,
+      "step": 5763
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 3.118e-05,
+      "loss": 0.1852,
+      "step": 5764
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 3.1175e-05,
+      "loss": 0.075,
+      "step": 5765
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 3.117e-05,
+      "loss": 0.142,
+      "step": 5766
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3.1165e-05,
+      "loss": 0.1374,
+      "step": 5767
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3.116e-05,
+      "loss": 0.1611,
+      "step": 5768
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 3.1155e-05,
+      "loss": 0.0467,
+      "step": 5769
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 3.115e-05,
+      "loss": 0.1367,
+      "step": 5770
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 3.1145e-05,
+      "loss": 0.0497,
+      "step": 5771
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 3.1140000000000003e-05,
+      "loss": 0.1414,
+      "step": 5772
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 3.1135e-05,
+      "loss": 0.0757,
+      "step": 5773
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 3.113e-05,
+      "loss": 0.0601,
+      "step": 5774
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 3.1125000000000004e-05,
+      "loss": 0.234,
+      "step": 5775
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 3.112000000000001e-05,
+      "loss": 0.0642,
+      "step": 5776
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 3.1114999999999996e-05,
+      "loss": 0.1382,
+      "step": 5777
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 3.111e-05,
+      "loss": 0.1572,
+      "step": 5778
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 3.1105e-05,
+      "loss": 0.0617,
+      "step": 5779
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 3.11e-05,
+      "loss": 0.152,
+      "step": 5780
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 3.1095e-05,
+      "loss": 0.0756,
+      "step": 5781
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 3.109e-05,
+      "loss": 0.0589,
+      "step": 5782
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 3.1085e-05,
+      "loss": 0.2457,
+      "step": 5783
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 3.108e-05,
+      "loss": 0.0892,
+      "step": 5784
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 3.1075e-05,
+      "loss": 0.1367,
+      "step": 5785
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 3.1070000000000006e-05,
+      "loss": 0.0705,
+      "step": 5786
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 3.1065e-05,
+      "loss": 0.1133,
+      "step": 5787
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 3.1060000000000004e-05,
+      "loss": 0.1733,
+      "step": 5788
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 3.1055e-05,
+      "loss": 0.2055,
+      "step": 5789
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 3.1049999999999996e-05,
+      "loss": 0.1501,
+      "step": 5790
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 3.1045e-05,
+      "loss": 0.1901,
+      "step": 5791
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 3.104e-05,
+      "loss": 0.0756,
+      "step": 5792
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 3.1035e-05,
+      "loss": 0.1533,
+      "step": 5793
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 3.103e-05,
+      "loss": 0.1629,
+      "step": 5794
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 3.1025e-05,
+      "loss": 0.1441,
+      "step": 5795
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 3.1020000000000005e-05,
+      "loss": 0.2182,
+      "step": 5796
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 3.1015e-05,
+      "loss": 0.1274,
+      "step": 5797
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 3.101e-05,
+      "loss": 0.2099,
+      "step": 5798
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 3.1005000000000006e-05,
+      "loss": 0.1374,
+      "step": 5799
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 3.1e-05,
+      "loss": 0.1453,
+      "step": 5800
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 3.0995e-05,
+      "loss": 0.0526,
+      "step": 5801
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 3.099e-05,
+      "loss": 0.0862,
+      "step": 5802
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 3.0984999999999996e-05,
+      "loss": 0.0639,
+      "step": 5803
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 3.098e-05,
+      "loss": 0.1902,
+      "step": 5804
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 3.0975e-05,
+      "loss": 0.1603,
+      "step": 5805
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 3.097e-05,
+      "loss": 0.0539,
+      "step": 5806
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 3.0965e-05,
+      "loss": 0.1707,
+      "step": 5807
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 3.096e-05,
+      "loss": 0.1582,
+      "step": 5808
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 3.0955000000000004e-05,
+      "loss": 0.1099,
+      "step": 5809
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 3.095e-05,
+      "loss": 0.2607,
+      "step": 5810
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 3.0945e-05,
+      "loss": 0.1825,
+      "step": 5811
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 3.0940000000000005e-05,
+      "loss": 0.155,
+      "step": 5812
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 3.0934999999999994e-05,
+      "loss": 0.1803,
+      "step": 5813
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 3.093e-05,
+      "loss": 0.1037,
+      "step": 5814
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 3.0925e-05,
+      "loss": 0.0474,
+      "step": 5815
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 3.092e-05,
+      "loss": 0.0659,
+      "step": 5816
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 3.0915e-05,
+      "loss": 0.152,
+      "step": 5817
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 3.091e-05,
+      "loss": 0.0581,
+      "step": 5818
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 3.0905e-05,
+      "loss": 0.2584,
+      "step": 5819
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 3.09e-05,
+      "loss": 0.0833,
+      "step": 5820
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 3.0895e-05,
+      "loss": 0.139,
+      "step": 5821
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 3.0890000000000004e-05,
+      "loss": 0.1591,
+      "step": 5822
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 3.0885e-05,
+      "loss": 0.2732,
+      "step": 5823
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 3.088e-05,
+      "loss": 0.0879,
+      "step": 5824
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 3.0875e-05,
+      "loss": 0.0512,
+      "step": 5825
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 3.087e-05,
+      "loss": 0.0838,
+      "step": 5826
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 3.0865e-05,
+      "loss": 0.0385,
+      "step": 5827
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 3.086e-05,
+      "loss": 0.1695,
+      "step": 5828
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 3.0855e-05,
+      "loss": 0.0826,
+      "step": 5829
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 3.085e-05,
+      "loss": 0.0642,
+      "step": 5830
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 3.0845e-05,
+      "loss": 0.1467,
+      "step": 5831
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 3.084e-05,
+      "loss": 0.2188,
+      "step": 5832
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 3.0835e-05,
+      "loss": 0.0628,
+      "step": 5833
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 3.083e-05,
+      "loss": 0.1012,
+      "step": 5834
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 3.0825000000000004e-05,
+      "loss": 0.0656,
+      "step": 5835
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 3.0820000000000006e-05,
+      "loss": 0.047,
+      "step": 5836
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 3.0814999999999995e-05,
+      "loss": 0.1683,
+      "step": 5837
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 3.081e-05,
+      "loss": 0.1632,
+      "step": 5838
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 3.0805e-05,
+      "loss": 0.1089,
+      "step": 5839
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 3.0799999999999996e-05,
+      "loss": 0.1687,
+      "step": 5840
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 3.0795e-05,
+      "loss": 0.1091,
+      "step": 5841
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 3.079e-05,
+      "loss": 0.2485,
+      "step": 5842
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 3.0785000000000004e-05,
+      "loss": 0.0561,
+      "step": 5843
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 3.078e-05,
+      "loss": 0.1691,
+      "step": 5844
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 3.0775e-05,
+      "loss": 0.1866,
+      "step": 5845
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 3.0770000000000005e-05,
+      "loss": 0.09,
+      "step": 5846
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 3.0765e-05,
+      "loss": 0.2328,
+      "step": 5847
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 3.0760000000000004e-05,
+      "loss": 0.2019,
+      "step": 5848
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 3.0755e-05,
+      "loss": 0.1929,
+      "step": 5849
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 3.0749999999999995e-05,
+      "loss": 0.0522,
+      "step": 5850
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 3.0745e-05,
+      "loss": 0.0672,
+      "step": 5851
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 3.074e-05,
+      "loss": 0.1756,
+      "step": 5852
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 3.0735e-05,
+      "loss": 0.1484,
+      "step": 5853
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 3.073e-05,
+      "loss": 0.0909,
+      "step": 5854
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 3.0725e-05,
+      "loss": 0.162,
+      "step": 5855
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 3.0720000000000004e-05,
+      "loss": 0.1004,
+      "step": 5856
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 3.0715e-05,
+      "loss": 0.2469,
+      "step": 5857
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 3.071e-05,
+      "loss": 0.147,
+      "step": 5858
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 3.0705000000000005e-05,
+      "loss": 0.1763,
+      "step": 5859
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 3.07e-05,
+      "loss": 0.1864,
+      "step": 5860
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 3.0695e-05,
+      "loss": 0.1183,
+      "step": 5861
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 3.069e-05,
+      "loss": 0.0819,
+      "step": 5862
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 3.0685e-05,
+      "loss": 0.1368,
+      "step": 5863
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 3.068e-05,
+      "loss": 0.0608,
+      "step": 5864
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 3.0675e-05,
+      "loss": 0.0737,
+      "step": 5865
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 3.067e-05,
+      "loss": 0.171,
+      "step": 5866
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 3.0665e-05,
+      "loss": 0.0495,
+      "step": 5867
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 3.066e-05,
+      "loss": 0.2475,
+      "step": 5868
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 3.0655000000000004e-05,
+      "loss": 0.1187,
+      "step": 5869
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 3.065e-05,
+      "loss": 0.125,
+      "step": 5870
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 3.0645e-05,
+      "loss": 0.1987,
+      "step": 5871
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 3.0640000000000005e-05,
+      "loss": 0.0568,
+      "step": 5872
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 3.0635e-05,
+      "loss": 0.1793,
+      "step": 5873
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 3.0629999999999996e-05,
+      "loss": 0.1501,
+      "step": 5874
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 3.0625e-05,
+      "loss": 0.0408,
+      "step": 5875
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 3.062e-05,
+      "loss": 0.1547,
+      "step": 5876
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 3.0615e-05,
+      "loss": 0.0819,
+      "step": 5877
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 3.061e-05,
+      "loss": 0.0535,
+      "step": 5878
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 3.0605e-05,
+      "loss": 0.2257,
+      "step": 5879
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 3.06e-05,
+      "loss": 0.2153,
+      "step": 5880
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 3.0595e-05,
+      "loss": 0.0827,
+      "step": 5881
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 3.0590000000000004e-05,
+      "loss": 0.2399,
+      "step": 5882
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 3.0585000000000006e-05,
+      "loss": 0.1947,
+      "step": 5883
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 3.058e-05,
+      "loss": 0.07,
+      "step": 5884
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 3.0575000000000005e-05,
+      "loss": 0.188,
+      "step": 5885
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 3.057e-05,
+      "loss": 0.1514,
+      "step": 5886
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 3.0564999999999996e-05,
+      "loss": 0.1283,
+      "step": 5887
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 3.056e-05,
+      "loss": 0.046,
+      "step": 5888
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 3.0555e-05,
+      "loss": 0.1797,
+      "step": 5889
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 3.055e-05,
+      "loss": 0.1095,
+      "step": 5890
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 3.0545e-05,
+      "loss": 0.1612,
+      "step": 5891
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 3.054e-05,
+      "loss": 0.3997,
+      "step": 5892
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 3.0535000000000005e-05,
+      "loss": 0.1957,
+      "step": 5893
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 3.053e-05,
+      "loss": 0.0382,
+      "step": 5894
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 3.0525e-05,
+      "loss": 0.0876,
+      "step": 5895
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 3.0520000000000006e-05,
+      "loss": 0.0784,
+      "step": 5896
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 3.0515000000000005e-05,
+      "loss": 0.1152,
+      "step": 5897
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 3.0509999999999998e-05,
+      "loss": 0.1348,
+      "step": 5898
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 3.0504999999999997e-05,
+      "loss": 0.2806,
+      "step": 5899
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 3.05e-05,
+      "loss": 0.0489,
+      "step": 5900
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 3.0495e-05,
+      "loss": 0.1934,
+      "step": 5901
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 3.049e-05,
+      "loss": 0.2267,
+      "step": 5902
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 3.0485e-05,
+      "loss": 0.1501,
+      "step": 5903
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 3.048e-05,
+      "loss": 0.0696,
+      "step": 5904
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 3.0475000000000002e-05,
+      "loss": 0.182,
+      "step": 5905
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 3.047e-05,
+      "loss": 0.1212,
+      "step": 5906
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 3.0465000000000004e-05,
+      "loss": 0.1233,
+      "step": 5907
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 3.0460000000000003e-05,
+      "loss": 0.1783,
+      "step": 5908
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 3.0455000000000002e-05,
+      "loss": 0.1544,
+      "step": 5909
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 3.0449999999999998e-05,
+      "loss": 0.0895,
+      "step": 5910
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 3.0444999999999997e-05,
+      "loss": 0.0663,
+      "step": 5911
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 3.044e-05,
+      "loss": 0.0884,
+      "step": 5912
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 3.0435e-05,
+      "loss": 0.2151,
+      "step": 5913
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 3.0430000000000002e-05,
+      "loss": 0.1623,
+      "step": 5914
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 3.0425e-05,
+      "loss": 0.116,
+      "step": 5915
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 3.042e-05,
+      "loss": 0.1181,
+      "step": 5916
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 3.0415000000000003e-05,
+      "loss": 0.2171,
+      "step": 5917
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 3.0410000000000002e-05,
+      "loss": 0.1691,
+      "step": 5918
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 3.0405e-05,
+      "loss": 0.134,
+      "step": 5919
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 3.0400000000000004e-05,
+      "loss": 0.0546,
+      "step": 5920
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 3.0395000000000003e-05,
+      "loss": 0.172,
+      "step": 5921
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 3.039e-05,
+      "loss": 0.1651,
+      "step": 5922
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 3.0384999999999998e-05,
+      "loss": 0.158,
+      "step": 5923
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 3.038e-05,
+      "loss": 0.2006,
+      "step": 5924
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 3.0375e-05,
+      "loss": 0.1019,
+      "step": 5925
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 3.037e-05,
+      "loss": 0.1423,
+      "step": 5926
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 3.0365e-05,
+      "loss": 0.0887,
+      "step": 5927
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 3.036e-05,
+      "loss": 0.0672,
+      "step": 5928
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 3.0355000000000003e-05,
+      "loss": 0.1984,
+      "step": 5929
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 3.0350000000000002e-05,
+      "loss": 0.1122,
+      "step": 5930
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 3.0345e-05,
+      "loss": 0.153,
+      "step": 5931
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 3.0340000000000004e-05,
+      "loss": 0.1643,
+      "step": 5932
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 3.0335000000000003e-05,
+      "loss": 0.0691,
+      "step": 5933
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 3.033e-05,
+      "loss": 0.1891,
+      "step": 5934
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 3.0325e-05,
+      "loss": 0.1292,
+      "step": 5935
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 3.0319999999999998e-05,
+      "loss": 0.192,
+      "step": 5936
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 3.0315e-05,
+      "loss": 0.1052,
+      "step": 5937
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 3.031e-05,
+      "loss": 0.4067,
+      "step": 5938
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 3.0305000000000002e-05,
+      "loss": 0.1589,
+      "step": 5939
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 3.03e-05,
+      "loss": 0.0922,
+      "step": 5940
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 3.0295e-05,
+      "loss": 0.1228,
+      "step": 5941
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 3.0290000000000003e-05,
+      "loss": 0.0512,
+      "step": 5942
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 3.0285000000000002e-05,
+      "loss": 0.0821,
+      "step": 5943
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 3.0280000000000005e-05,
+      "loss": 0.09,
+      "step": 5944
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 3.0275000000000004e-05,
+      "loss": 0.0831,
+      "step": 5945
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 3.0269999999999996e-05,
+      "loss": 0.0361,
+      "step": 5946
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 3.0265e-05,
+      "loss": 0.123,
+      "step": 5947
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 3.0259999999999998e-05,
+      "loss": 0.1369,
+      "step": 5948
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 3.0255e-05,
+      "loss": 0.2372,
+      "step": 5949
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 3.025e-05,
+      "loss": 0.0687,
+      "step": 5950
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 3.0245e-05,
+      "loss": 0.1299,
+      "step": 5951
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 3.0240000000000002e-05,
+      "loss": 0.2018,
+      "step": 5952
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 3.0235e-05,
+      "loss": 0.1423,
+      "step": 5953
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 3.0230000000000004e-05,
+      "loss": 0.1625,
+      "step": 5954
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 3.0225000000000003e-05,
+      "loss": 0.07,
+      "step": 5955
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 3.0220000000000002e-05,
+      "loss": 0.0855,
+      "step": 5956
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 3.0215000000000005e-05,
+      "loss": 0.1307,
+      "step": 5957
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 3.0209999999999997e-05,
+      "loss": 0.14,
+      "step": 5958
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 3.0205e-05,
+      "loss": 0.0731,
+      "step": 5959
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 3.02e-05,
+      "loss": 0.1278,
+      "step": 5960
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 3.0194999999999998e-05,
+      "loss": 0.0763,
+      "step": 5961
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 3.019e-05,
+      "loss": 0.1684,
+      "step": 5962
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 3.0185e-05,
+      "loss": 0.1274,
+      "step": 5963
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 3.0180000000000002e-05,
+      "loss": 0.0876,
+      "step": 5964
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 3.0175e-05,
+      "loss": 0.1229,
+      "step": 5965
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 3.017e-05,
+      "loss": 0.1341,
+      "step": 5966
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 3.0165000000000003e-05,
+      "loss": 0.0492,
+      "step": 5967
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 3.0160000000000003e-05,
+      "loss": 0.0875,
+      "step": 5968
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 3.0155000000000005e-05,
+      "loss": 0.0634,
+      "step": 5969
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 3.0149999999999998e-05,
+      "loss": 0.0924,
+      "step": 5970
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 3.0144999999999997e-05,
+      "loss": 0.2144,
+      "step": 5971
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 3.014e-05,
+      "loss": 0.1514,
+      "step": 5972
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 3.0135e-05,
+      "loss": 0.139,
+      "step": 5973
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 3.013e-05,
+      "loss": 0.0674,
+      "step": 5974
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 3.0125e-05,
+      "loss": 0.144,
+      "step": 5975
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 3.012e-05,
+      "loss": 0.0616,
+      "step": 5976
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 3.0115000000000002e-05,
+      "loss": 0.0505,
+      "step": 5977
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 3.011e-05,
+      "loss": 0.2254,
+      "step": 5978
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 3.0105000000000004e-05,
+      "loss": 0.1275,
+      "step": 5979
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 3.0100000000000003e-05,
+      "loss": 0.1145,
+      "step": 5980
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 3.0095000000000002e-05,
+      "loss": 0.1528,
+      "step": 5981
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 3.0089999999999998e-05,
+      "loss": 0.1464,
+      "step": 5982
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 3.0084999999999997e-05,
+      "loss": 0.2369,
+      "step": 5983
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 3.008e-05,
+      "loss": 0.2286,
+      "step": 5984
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 3.0075e-05,
+      "loss": 0.0607,
+      "step": 5985
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 3.007e-05,
+      "loss": 0.0911,
+      "step": 5986
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 3.0065e-05,
+      "loss": 0.0625,
+      "step": 5987
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 3.006e-05,
+      "loss": 0.0753,
+      "step": 5988
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 3.0055000000000003e-05,
+      "loss": 0.0563,
+      "step": 5989
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 3.0050000000000002e-05,
+      "loss": 0.1268,
+      "step": 5990
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 3.0045e-05,
+      "loss": 0.2739,
+      "step": 5991
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 3.0040000000000004e-05,
+      "loss": 0.0556,
+      "step": 5992
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 3.0035000000000003e-05,
+      "loss": 0.0622,
+      "step": 5993
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 3.003e-05,
+      "loss": 0.0435,
+      "step": 5994
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 3.0024999999999998e-05,
+      "loss": 0.2574,
+      "step": 5995
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 3.0019999999999997e-05,
+      "loss": 0.111,
+      "step": 5996
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 3.0015e-05,
+      "loss": 0.3503,
+      "step": 5997
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 3.001e-05,
+      "loss": 0.0982,
+      "step": 5998
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 3.0005e-05,
+      "loss": 0.1556,
+      "step": 5999
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 3e-05,
+      "loss": 0.0636,
+      "step": 6000
+    },
+    {
+      "epoch": 25.0,
+      "eval_loss": 0.4488964080810547,
+      "eval_mean_accuracy": 0.7688703095990338,
+      "eval_mean_iou": 0.6634750323993962,
+      "eval_overall_accuracy": 0.8661323547363281,
+      "eval_per_category_accuracy": [
+        0.8948000727084287,
+        0.9051120589389424,
+        0.5066987971497301
+      ],
+      "eval_per_category_iou": [
+        0.8050725526330943,
+        0.80482903564041,
+        0.3805235089246844
+      ],
+      "eval_runtime": 159.3184,
+      "eval_samples_per_second": 0.753,
+      "eval_steps_per_second": 0.377,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 12000,
+  "num_train_epochs": 50,
+  "save_steps": 3000,
+  "total_flos": 3.623614282727424e+18,
+  "trial_name": null,
+  "trial_params": null
+}