diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,28525 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 4750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 2.4297,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 2.3998,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 2.4045,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 2.463,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.3666,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 2.4288,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 2.4251,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 2.4803,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 2.4075,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 2.3769,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 2.4314,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 2.4275,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 2.2976,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 2.3859,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 2.4325,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 2.3814,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 2.3957,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 2.4433,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 2.4051,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.441,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 2.4263,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 2.3631,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 2.356,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 2.4343,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.3514,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-07,
+      "loss": 2.3838,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 2.4551,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6e-07,
+      "loss": 2.3974,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 2.42,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.3417,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 2.3386,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4e-07,
+      "loss": 2.3946,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6e-07,
+      "loss": 2.3426,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 2.3515,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 2.3762,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-07,
+      "loss": 2.3918,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4e-07,
+      "loss": 2.3511,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6e-07,
+      "loss": 2.3745,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.8e-07,
+      "loss": 2.4072,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 2.3796,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 2.4053,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 2.3779,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 2.3364,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 2.3931,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 2.4008,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 2.3857,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 2.3786,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 2.4144,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 2.3599,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.3386,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 2.302,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 2.3927,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 2.397,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 2.4071,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 2.2871,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 2.3563,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 2.3115,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 2.3033,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 2.4122,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.3899,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 2.3562,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 2.3298,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 2.3283,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 2.3635,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 2.3346,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 2.3955,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 2.3561,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 2.3594,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 2.3455,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.3766,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 2.3199,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.44e-06,
+      "loss": 2.3718,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.46e-06,
+      "loss": 2.2958,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.48e-06,
+      "loss": 2.3729,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 2.306,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.52e-06,
+      "loss": 2.3088,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.54e-06,
+      "loss": 2.3136,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.56e-06,
+      "loss": 2.3787,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 2.3865,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.3478,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 2.3092,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 2.3089,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 2.2677,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 2.3223,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.2566,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.72e-06,
+      "loss": 2.3574,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.74e-06,
+      "loss": 2.2636,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.76e-06,
+      "loss": 2.2462,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 2.3132,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.3108,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 2.3002,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 2.2583,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 2.2776,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 2.3276,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 2.3006,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 2.3155,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.94e-06,
+      "loss": 2.2945,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 2.3316,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98e-06,
+      "loss": 2.2689,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.2502,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.02e-06,
+      "loss": 2.2731,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.04e-06,
+      "loss": 2.3311,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.06e-06,
+      "loss": 2.2969,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.08e-06,
+      "loss": 2.2813,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.2813,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.12e-06,
+      "loss": 2.2637,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 2.3322,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.16e-06,
+      "loss": 2.3053,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 2.2352,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2e-06,
+      "loss": 2.2798,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 2.3529,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.24e-06,
+      "loss": 2.281,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 2.214,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.28e-06,
+      "loss": 2.3061,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 2.2771,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 2.2761,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 2.2297,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 2.2779,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.38e-06,
+      "loss": 2.2204,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.2745,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.42e-06,
+      "loss": 2.2289,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 2.246,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.46e-06,
+      "loss": 2.2318,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 2.2939,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5e-06,
+      "loss": 2.2354,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.52e-06,
+      "loss": 2.2607,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 2.2516,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.56e-06,
+      "loss": 2.3537,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 2.2989,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6e-06,
+      "loss": 2.286,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 2.2812,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.64e-06,
+      "loss": 2.2314,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 2.3007,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.68e-06,
+      "loss": 2.2702,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 2.2682,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 2.2499,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 2.3094,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 2.259,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 2.2673,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.3074,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.82e-06,
+      "loss": 2.178,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.84e-06,
+      "loss": 2.234,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.86e-06,
+      "loss": 2.1925,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.88e-06,
+      "loss": 2.2439,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9e-06,
+      "loss": 2.2741,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.92e-06,
+      "loss": 2.2212,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 2.2689,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.96e-06,
+      "loss": 2.2101,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 2.1976,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3e-06,
+      "loss": 2.2739,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 2.1931,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.04e-06,
+      "loss": 2.2415,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 2.2128,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.08e-06,
+      "loss": 2.3045,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 2.2146,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.12e-06,
+      "loss": 2.2223,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 2.2488,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 2.2285,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 2.2455,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.2266,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 2.2259,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 2.2256,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 2.243,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 2.2133,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 2.2105,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 2.2472,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 2.192,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 2.2025,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 2.1808,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.2221,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 2.2041,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.44e-06,
+      "loss": 2.2352,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.46e-06,
+      "loss": 2.2183,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.48e-06,
+      "loss": 2.2081,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5e-06,
+      "loss": 2.2135,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.52e-06,
+      "loss": 2.1838,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.54e-06,
+      "loss": 2.2103,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 2.1972,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.58e-06,
+      "loss": 2.1864,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.2045,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.62e-06,
+      "loss": 2.2037,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 2.1621,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.66e-06,
+      "loss": 2.2479,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 2.1534,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7e-06,
+      "loss": 2.2509,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 2.1431,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.74e-06,
+      "loss": 2.2582,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 2.1786,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 2.1511,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.2128,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 2.2329,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 2.2446,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.86e-06,
+      "loss": 2.2367,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.88e-06,
+      "loss": 2.2109,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 2.245,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 2.1475,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.94e-06,
+      "loss": 2.2136,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.96e-06,
+      "loss": 2.1769,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 2.1853,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.1926,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 2.2072,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.04e-06,
+      "loss": 2.1358,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 2.2077,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.08e-06,
+      "loss": 2.1996,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1e-06,
+      "loss": 2.1911,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.12e-06,
+      "loss": 2.2074,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.14e-06,
+      "loss": 2.2137,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.16e-06,
+      "loss": 2.2004,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.18e-06,
+      "loss": 2.1245,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.1873,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.22e-06,
+      "loss": 2.1469,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.24e-06,
+      "loss": 2.1944,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.26e-06,
+      "loss": 2.1885,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 2.2611,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.3e-06,
+      "loss": 2.2101,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.32e-06,
+      "loss": 2.1665,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.34e-06,
+      "loss": 2.1715,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 2.1942,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.38e-06,
+      "loss": 2.1685,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4e-06,
+      "loss": 2.1805,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.42e-06,
+      "loss": 2.1973,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 2.1743,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 2.1102,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.48e-06,
+      "loss": 2.159,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5e-06,
+      "loss": 2.1208,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 2.1746,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 2.0865,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.56e-06,
+      "loss": 2.1479,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.58e-06,
+      "loss": 2.1494,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.2117,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 2.1934,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 2.1598,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.66e-06,
+      "loss": 2.1795,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 2.1219,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7e-06,
+      "loss": 2.1903,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 2.178,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74e-06,
+      "loss": 2.1803,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76e-06,
+      "loss": 2.14,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.78e-06,
+      "loss": 2.1068,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.2054,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 2.1719,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.84e-06,
+      "loss": 2.1219,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.86e-06,
+      "loss": 2.1386,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 2.1634,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.2024,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92e-06,
+      "loss": 2.2108,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.94e-06,
+      "loss": 2.1571,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 2.2098,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 2.0911,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5e-06,
+      "loss": 2.16,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.02e-06,
+      "loss": 2.1382,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.04e-06,
+      "loss": 2.1547,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 2.1671,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 2.1645,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1e-06,
+      "loss": 2.1023,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.12e-06,
+      "loss": 2.1519,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 2.1554,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 2.1464,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.18e-06,
+      "loss": 2.1145,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2e-06,
+      "loss": 2.1364,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 2.1366,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 2.2169,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 2.1534,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.28e-06,
+      "loss": 2.1054,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 2.1245,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 2.1839,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 2.088,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.36e-06,
+      "loss": 2.2276,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 2.1575,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.1937,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 2.1604,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 2.1389,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 2.0966,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 2.1428,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.0956,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 2.153,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 2.1615,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 2.0671,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 2.1442,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.1649,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 2.1464,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.64e-06,
+      "loss": 2.1415,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.66e-06,
+      "loss": 2.0962,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.68e-06,
+      "loss": 2.1675,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.7e-06,
+      "loss": 2.1487,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.72e-06,
+      "loss": 2.1434,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.74e-06,
+      "loss": 2.1281,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.76e-06,
+      "loss": 2.1304,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.78e-06,
+      "loss": 2.1338,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.8e-06,
+      "loss": 2.1143,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.82e-06,
+      "loss": 2.1047,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.84e-06,
+      "loss": 2.0695,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.86e-06,
+      "loss": 2.0493,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 2.1104,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9e-06,
+      "loss": 2.1247,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.92e-06,
+      "loss": 2.0872,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.94e-06,
+      "loss": 2.0791,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 2.137,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.98e-06,
+      "loss": 2.1143,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6e-06,
+      "loss": 2.1071,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.02e-06,
+      "loss": 2.1147,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 2.109,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 2.1336,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.08e-06,
+      "loss": 2.1766,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1e-06,
+      "loss": 2.1253,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 2.1203,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 2.0683,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.16e-06,
+      "loss": 2.1012,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.18e-06,
+      "loss": 2.1294,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 2.1388,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 2.1204,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.24e-06,
+      "loss": 2.0995,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.26e-06,
+      "loss": 2.1132,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 2.1821,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 2.1686,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 2.1454,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.34e-06,
+      "loss": 2.1214,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 2.0956,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 2.1438,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.1409,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.42e-06,
+      "loss": 2.178,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 2.1678,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 2.1068,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 2.0314,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.1014,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 2.1324,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 1.9781,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 2.0717,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 2.0929,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 2.1325,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 2.1189,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 2.0761,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 2.0821,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 2.106,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 2.1459,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 2.1263,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 2.1051,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 2.161,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 2.0873,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.1235,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 2.1174,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 2.0807,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 2.0947,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.88e-06,
+      "loss": 2.0682,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9e-06,
+      "loss": 2.113,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.92e-06,
+      "loss": 2.1043,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 2.0799,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.96e-06,
+      "loss": 2.0958,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.98e-06,
+      "loss": 2.1264,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7e-06,
+      "loss": 2.1102,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 2.1133,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.04e-06,
+      "loss": 2.089,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.06e-06,
+      "loss": 2.0793,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.08e-06,
+      "loss": 2.0669,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 2.1005,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 2.1483,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.14e-06,
+      "loss": 2.1394,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.16e-06,
+      "loss": 2.0503,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 2.1166,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 2.0876,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.22e-06,
+      "loss": 2.1126,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.24e-06,
+      "loss": 2.0517,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 2.0544,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 2.0982,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.3e-06,
+      "loss": 2.1314,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.32e-06,
+      "loss": 2.0793,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 2.0969,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 2.0851,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 2.1063,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.4e-06,
+      "loss": 2.1229,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 2.0818,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 2.0713,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 2.089,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.48e-06,
+      "loss": 2.0832,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.0265,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 2.1032,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 2.1003,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 2.0314,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.58e-06,
+      "loss": 2.0596,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 2.1133,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 2.1042,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 2.126,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 2.1003,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 2.1337,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.7e-06,
+      "loss": 2.0588,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.72e-06,
+      "loss": 2.1079,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.74e-06,
+      "loss": 2.0663,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.76e-06,
+      "loss": 1.9847,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.78e-06,
+      "loss": 2.1238,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.0657,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 2.0993,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 2.0504,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 2.1016,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.88e-06,
+      "loss": 2.0768,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.9e-06,
+      "loss": 2.0269,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.92e-06,
+      "loss": 2.0757,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.94e-06,
+      "loss": 2.0987,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 2.0688,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 2.0565,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.0868,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 2.0593,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 2.0884,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.06e-06,
+      "loss": 2.0597,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.08e-06,
+      "loss": 2.0515,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.1e-06,
+      "loss": 2.024,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 2.0569,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.14e-06,
+      "loss": 2.0612,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.16e-06,
+      "loss": 2.0523,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.18e-06,
+      "loss": 2.0799,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.2e-06,
+      "loss": 2.023,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 2.0376,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.24e-06,
+      "loss": 2.0284,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.26e-06,
+      "loss": 2.0587,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.28e-06,
+      "loss": 2.0511,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.3e-06,
+      "loss": 2.0146,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.32e-06,
+      "loss": 1.9799,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.34e-06,
+      "loss": 2.0923,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.36e-06,
+      "loss": 2.0695,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 2.0305,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.9944,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.42e-06,
+      "loss": 2.0465,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.44e-06,
+      "loss": 2.0507,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.46e-06,
+      "loss": 2.0974,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.48e-06,
+      "loss": 2.0776,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.5e-06,
+      "loss": 2.0426,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.52e-06,
+      "loss": 2.0376,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 2.0439,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 2.0538,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 2.0593,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.6e-06,
+      "loss": 2.0516,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.62e-06,
+      "loss": 2.0356,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.64e-06,
+      "loss": 2.0675,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.66e-06,
+      "loss": 2.0728,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.68e-06,
+      "loss": 2.0298,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 2.0285,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 2.0548,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 2.0298,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.76e-06,
+      "loss": 2.05,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.78e-06,
+      "loss": 2.0684,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.8e-06,
+      "loss": 2.0483,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.82e-06,
+      "loss": 2.071,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.84e-06,
+      "loss": 2.0114,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 1.9854,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 2.0758,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 1.9694,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 2.0452,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.94e-06,
+      "loss": 2.0176,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.96e-06,
+      "loss": 2.0725,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.98e-06,
+      "loss": 2.0416,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9e-06,
+      "loss": 1.9993,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 2.0587,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 2.0711,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 2.0027,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 2.0642,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.1002,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.12e-06,
+      "loss": 2.0219,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.14e-06,
+      "loss": 2.0083,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.16e-06,
+      "loss": 1.9901,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 2.0165,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 2.0257,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 2.0583,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 2.0275,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 1.9874,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 2.0249,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.3e-06,
+      "loss": 2.0662,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.32e-06,
+      "loss": 1.9838,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 2.0019,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 2.0455,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.38e-06,
+      "loss": 2.0329,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.4e-06,
+      "loss": 2.0305,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.42e-06,
+      "loss": 2.035,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 2.0233,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 2.0155,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.48e-06,
+      "loss": 2.0285,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.5e-06,
+      "loss": 2.0274,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.52e-06,
+      "loss": 2.0047,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.54e-06,
+      "loss": 1.9921,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.56e-06,
+      "loss": 2.0408,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.58e-06,
+      "loss": 2.0477,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.0446,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 2.0255,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 2.0061,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.66e-06,
+      "loss": 1.9435,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.68e-06,
+      "loss": 1.9942,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7e-06,
+      "loss": 2.0197,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.72e-06,
+      "loss": 1.9987,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74e-06,
+      "loss": 2.0144,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 2.0122,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 2.0707,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.9903,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 2.0328,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.84e-06,
+      "loss": 1.9926,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86e-06,
+      "loss": 2.026,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88e-06,
+      "loss": 1.9945,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9e-06,
+      "loss": 2.0078,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 1.984,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 2.0054,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 1.9845,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 1.991,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 1.9933,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99999863396485e-06,
+      "loss": 2.0023,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999994535860152e-06,
+      "loss": 2.0278,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99998770568814e-06,
+      "loss": 2.0456,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999978143452545e-06,
+      "loss": 2.0214,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999965849158597e-06,
+      "loss": 2.0401,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99995082281301e-06,
+      "loss": 1.98,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999933064423997e-06,
+      "loss": 1.953,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99991257400126e-06,
+      "loss": 2.0016,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999889351555997e-06,
+      "loss": 2.0149,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999863397100895e-06,
+      "loss": 1.9647,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999834710650136e-06,
+      "loss": 2.0101,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999803292219396e-06,
+      "loss": 2.0089,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999769141825842e-06,
+      "loss": 2.0112,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999732259488134e-06,
+      "loss": 2.0166,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999692645226426e-06,
+      "loss": 2.0395,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999650299062362e-06,
+      "loss": 1.9675,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999605221019082e-06,
+      "loss": 1.9849,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999557411121216e-06,
+      "loss": 1.9701,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999506869394888e-06,
+      "loss": 1.9557,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999453595867717e-06,
+      "loss": 1.9935,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999397590568811e-06,
+      "loss": 2.0018,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99933885352877e-06,
+      "loss": 1.9937,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999277384779693e-06,
+      "loss": 1.9397,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999213184355164e-06,
+      "loss": 2.0341,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999146252290264e-06,
+      "loss": 1.9732,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999076588621565e-06,
+      "loss": 1.9881,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999004193387135e-06,
+      "loss": 1.9777,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998929066626529e-06,
+      "loss": 2.0081,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998851208380797e-06,
+      "loss": 1.9975,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998770618692483e-06,
+      "loss": 1.9646,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998687297605623e-06,
+      "loss": 1.9942,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998601245165744e-06,
+      "loss": 2.0118,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998512461419865e-06,
+      "loss": 2.0125,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9984209464165e-06,
+      "loss": 1.9335,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998326700205656e-06,
+      "loss": 1.9887,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998229722838826e-06,
+      "loss": 1.9622,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998130014369005e-06,
+      "loss": 2.0034,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99802757485067e-06,
+      "loss": 1.9807,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9979224043398e-06,
+      "loss": 1.9805,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997814502893857e-06,
+      "loss": 1.9464,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997703870571804e-06,
+      "loss": 1.9768,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99759050743409e-06,
+      "loss": 1.9689,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997474413542658e-06,
+      "loss": 2.0163,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997355588960945e-06,
+      "loss": 1.943,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997234033753876e-06,
+      "loss": 1.9665,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997109747987873e-06,
+      "loss": 1.9253,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996982731730845e-06,
+      "loss": 2.0027,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996852985052197e-06,
+      "loss": 2.025,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996720508022826e-06,
+      "loss": 1.9192,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996585300715117e-06,
+      "loss": 1.973,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996447363202947e-06,
+      "loss": 1.9786,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996306695561692e-06,
+      "loss": 1.9571,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99616329786821e-06,
+      "loss": 1.954,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996017170200861e-06,
+      "loss": 1.9707,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995868312639488e-06,
+      "loss": 1.9443,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995716725265427e-06,
+      "loss": 1.9426,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99556240816151e-06,
+      "loss": 1.9473,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995405361412058e-06,
+      "loss": 1.9187,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995245585102884e-06,
+      "loss": 1.9688,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99508307932129e-06,
+      "loss": 1.9549,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994917844156074e-06,
+      "loss": 1.983,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994749879697518e-06,
+      "loss": 1.9266,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994579186037407e-06,
+      "loss": 1.9374,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994405763269004e-06,
+      "loss": 1.9755,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994229611487075e-06,
+      "loss": 1.9967,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994050730787868e-06,
+      "loss": 1.9294,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993869121269129e-06,
+      "loss": 1.9984,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99368478303009e-06,
+      "loss": 1.923,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993497716171475e-06,
+      "loss": 1.9212,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.9933079207955e-06,
+      "loss": 1.9833,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993115397005874e-06,
+      "loss": 1.946,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992920144907796e-06,
+      "loss": 1.9262,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992722164607952e-06,
+      "loss": 1.9194,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992521456214519e-06,
+      "loss": 1.9771,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992318019837171e-06,
+      "loss": 1.9573,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992111855587067e-06,
+      "loss": 1.9363,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991902963576856e-06,
+      "loss": 1.9799,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991691343920684e-06,
+      "loss": 2.0101,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99147699673418e-06,
+      "loss": 1.9282,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991259922134466e-06,
+      "loss": 1.9218,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991040120240156e-06,
+      "loss": 1.9885,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990817591171352e-06,
+      "loss": 1.959,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990592335049648e-06,
+      "loss": 1.9397,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990364351998126e-06,
+      "loss": 1.9976,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990133642141359e-06,
+      "loss": 1.9065,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98990020560541e-06,
+      "loss": 1.9464,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989664042517834e-06,
+      "loss": 1.8935,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989425153007672e-06,
+      "loss": 1.9307,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989183537205455e-06,
+      "loss": 1.9388,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98893919524321e-06,
+      "loss": 1.9318,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988692127254447e-06,
+      "loss": 1.9183,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988442333374164e-06,
+      "loss": 1.9559,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988189813738856e-06,
+      "loss": 1.9098,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.9879345684865e-06,
+      "loss": 1.8982,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.987676597756569e-06,
+      "loss": 1.9529,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.987415901690021e-06,
+      "loss": 1.9435,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.9871524804293e-06,
+      "loss": 1.8581,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986886334118349e-06,
+      "loss": 1.9454,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986617462902591e-06,
+      "loss": 1.9689,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98634586692894e-06,
+      "loss": 1.9454,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986071546345804e-06,
+      "loss": 1.922,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98579450130307e-06,
+      "loss": 1.9485,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985514731952123e-06,
+      "loss": 1.9494,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985232238445834e-06,
+      "loss": 1.9275,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984947020938556e-06,
+      "loss": 1.9284,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984659079586142e-06,
+      "loss": 1.8922,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984368414545923e-06,
+      "loss": 1.9058,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984075025976727e-06,
+      "loss": 1.9274,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98377891403886e-06,
+      "loss": 1.9434,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983480078894125e-06,
+      "loss": 1.8698,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983178520705808e-06,
+      "loss": 1.9599,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.982874239638687e-06,
+      "loss": 1.9065,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.982567235859024e-06,
+      "loss": 1.9031,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98225750953457e-06,
+      "loss": 1.9676,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981945060834565e-06,
+      "loss": 1.8881,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981629889929734e-06,
+      "loss": 1.9332,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981311996992291e-06,
+      "loss": 1.9205,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980991382195938e-06,
+      "loss": 1.9184,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980668045715864e-06,
+      "loss": 1.9621,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98034198772874e-06,
+      "loss": 1.8632,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980013208412736e-06,
+      "loss": 1.9222,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979681707947498e-06,
+      "loss": 1.9181,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979347486514161e-06,
+      "loss": 1.9045,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979010544295348e-06,
+      "loss": 1.8787,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978670881475173e-06,
+      "loss": 1.9012,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978328498239229e-06,
+      "loss": 1.9361,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977983394774601e-06,
+      "loss": 1.8778,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977635571269858e-06,
+      "loss": 1.9253,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977285027915056e-06,
+      "loss": 1.9303,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976931764901733e-06,
+      "loss": 1.8999,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976575782422923e-06,
+      "loss": 1.934,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976217080673135e-06,
+      "loss": 1.8842,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97585565984837e-06,
+      "loss": 1.9329,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.975491520146116e-06,
+      "loss": 1.9433,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97512466176534e-06,
+      "loss": 1.9238,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974755084906503e-06,
+      "loss": 1.8809,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974382789771541e-06,
+      "loss": 1.9284,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974007776563888e-06,
+      "loss": 1.9397,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97363004548845e-06,
+      "loss": 1.8867,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97324959675163e-06,
+      "loss": 1.9121,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.972866430561307e-06,
+      "loss": 1.9191,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.972480547126853e-06,
+      "loss": 1.9092,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.972091946659114e-06,
+      "loss": 1.9463,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971700629370431e-06,
+      "loss": 1.8923,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971306595474623e-06,
+      "loss": 1.8936,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.970909845186997e-06,
+      "loss": 1.8935,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.970510378724343e-06,
+      "loss": 1.9519,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.970108196304934e-06,
+      "loss": 1.8829,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.969703298148528e-06,
+      "loss": 1.8961,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96929568447637e-06,
+      "loss": 1.9369,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.968885355511183e-06,
+      "loss": 1.879,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.968472311477176e-06,
+      "loss": 1.8588,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.968056552600043e-06,
+      "loss": 1.9068,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.967638079106961e-06,
+      "loss": 1.894,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96721689122659e-06,
+      "loss": 1.8738,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.966792989189073e-06,
+      "loss": 1.8856,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.966366373226036e-06,
+      "loss": 1.931,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.965937043570585e-06,
+      "loss": 1.9099,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.965505000457317e-06,
+      "loss": 1.9352,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.965070244122302e-06,
+      "loss": 1.8911,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.964632774803099e-06,
+      "loss": 1.9041,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.964192592738747e-06,
+      "loss": 1.8452,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96374969816977e-06,
+      "loss": 1.8887,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.963304091338166e-06,
+      "loss": 1.8872,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.962855772487425e-06,
+      "loss": 1.8924,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.962404741862517e-06,
+      "loss": 1.8844,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.961950999709886e-06,
+      "loss": 1.9072,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.961494546277467e-06,
+      "loss": 1.9154,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96103538181467e-06,
+      "loss": 1.8801,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.960573506572391e-06,
+      "loss": 1.9024,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.960108920803002e-06,
+      "loss": 1.8648,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959641624760364e-06,
+      "loss": 1.8808,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959171618699812e-06,
+      "loss": 1.941,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958698902878163e-06,
+      "loss": 1.8572,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958223477553715e-06,
+      "loss": 1.8757,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.957745342986249e-06,
+      "loss": 1.8656,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.957264499437023e-06,
+      "loss": 1.8728,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.956780947168779e-06,
+      "loss": 1.8689,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.956294686445733e-06,
+      "loss": 1.909,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.955805717533586e-06,
+      "loss": 1.8802,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.955314040699522e-06,
+      "loss": 1.8443,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.954819656212193e-06,
+      "loss": 1.8576,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.954322564341743e-06,
+      "loss": 1.8809,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.953822765359785e-06,
+      "loss": 1.8968,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.953320259539423e-06,
+      "loss": 1.8576,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.952815047155229e-06,
+      "loss": 1.9088,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.952307128483257e-06,
+      "loss": 1.8406,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.951796503801044e-06,
+      "loss": 1.9066,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.9512831733876e-06,
+      "loss": 1.8915,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.950767137523417e-06,
+      "loss": 1.868,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.950248396490462e-06,
+      "loss": 1.8759,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.949726950572186e-06,
+      "loss": 1.9059,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.949202800053515e-06,
+      "loss": 1.8898,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.948675945220849e-06,
+      "loss": 1.8611,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.948146386362071e-06,
+      "loss": 1.8554,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.947614123766538e-06,
+      "loss": 1.8758,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.947079157725086e-06,
+      "loss": 1.8655,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.946541488530028e-06,
+      "loss": 1.9132,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.946001116475154e-06,
+      "loss": 1.8573,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.945458041855732e-06,
+      "loss": 1.8518,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.944912264968502e-06,
+      "loss": 1.8733,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.944363786111691e-06,
+      "loss": 1.8824,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.943812605584988e-06,
+      "loss": 1.8347,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94325872368957e-06,
+      "loss": 1.849,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.942702140728086e-06,
+      "loss": 1.8998,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94214285700466e-06,
+      "loss": 1.8315,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.941580872824892e-06,
+      "loss": 1.8687,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.941016188495858e-06,
+      "loss": 1.8563,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94044880432611e-06,
+      "loss": 1.8898,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.939878720625675e-06,
+      "loss": 1.8544,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.939305937706054e-06,
+      "loss": 1.8722,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.938730455880224e-06,
+      "loss": 1.8658,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.938152275462637e-06,
+      "loss": 1.8575,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.937571396769216e-06,
+      "loss": 1.9082,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.936987820117365e-06,
+      "loss": 1.8541,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.936401545825957e-06,
+      "loss": 1.8953,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.935812574215341e-06,
+      "loss": 1.827,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93522090560734e-06,
+      "loss": 1.8287,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.934626540325245e-06,
+      "loss": 1.8456,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.934029478693832e-06,
+      "loss": 1.8167,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93342972103934e-06,
+      "loss": 1.8638,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.932827267689489e-06,
+      "loss": 1.8698,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.932222118973463e-06,
+      "loss": 1.8576,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.931614275221926e-06,
+      "loss": 1.8544,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.931003736767013e-06,
+      "loss": 1.8625,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.930390503942332e-06,
+      "loss": 1.8358,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.929774577082957e-06,
+      "loss": 1.8382,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.929155956525446e-06,
+      "loss": 1.8017,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.928534642607815e-06,
+      "loss": 1.8687,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.927910635669563e-06,
+      "loss": 1.8245,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.927283936051655e-06,
+      "loss": 1.8247,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.92665454409653e-06,
+      "loss": 1.7977,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.926022460148095e-06,
+      "loss": 1.8367,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.925387684551729e-06,
+      "loss": 1.856,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924750217654283e-06,
+      "loss": 1.8477,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924110059804078e-06,
+      "loss": 1.8279,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.923467211350906e-06,
+      "loss": 1.8367,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.922821672646028e-06,
+      "loss": 1.815,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.922173444042173e-06,
+      "loss": 1.8252,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.921522525893548e-06,
+      "loss": 1.8535,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.920868918555817e-06,
+      "loss": 1.8151,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.920212622386125e-06,
+      "loss": 1.8173,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.919553637743078e-06,
+      "loss": 1.8348,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.918891964986759e-06,
+      "loss": 1.8491,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91822760447871e-06,
+      "loss": 1.8473,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.917560556581952e-06,
+      "loss": 1.8419,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.916890821660966e-06,
+      "loss": 1.8587,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.916218400081706e-06,
+      "loss": 1.8379,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.915543292211589e-06,
+      "loss": 1.8173,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91486549841951e-06,
+      "loss": 1.8241,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.914185019075819e-06,
+      "loss": 1.8399,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.913501854552345e-06,
+      "loss": 1.7914,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.912816005222375e-06,
+      "loss": 1.8078,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.912127471460666e-06,
+      "loss": 1.8392,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.911436253643445e-06,
+      "loss": 1.8196,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.910742352148401e-06,
+      "loss": 1.8259,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.910045767354694e-06,
+      "loss": 1.824,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.909346499642946e-06,
+      "loss": 1.8265,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.908644549395247e-06,
+      "loss": 1.7853,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.907939916995154e-06,
+      "loss": 1.7966,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.907232602827683e-06,
+      "loss": 1.8247,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.906522607279327e-06,
+      "loss": 1.8532,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.905809930738033e-06,
+      "loss": 1.8243,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90509457359322e-06,
+      "loss": 1.8682,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90437653623577e-06,
+      "loss": 1.818,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.903655819058024e-06,
+      "loss": 1.8049,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.902932422453795e-06,
+      "loss": 1.8288,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90220634681836e-06,
+      "loss": 1.8262,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.901477592548451e-06,
+      "loss": 1.7863,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.900746160042273e-06,
+      "loss": 1.8211,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.900012049699492e-06,
+      "loss": 1.8061,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.899275261921236e-06,
+      "loss": 1.8116,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.898535797110093e-06,
+      "loss": 1.8383,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.897793655670118e-06,
+      "loss": 1.788,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.89704883800683e-06,
+      "loss": 1.8119,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.896301344527205e-06,
+      "loss": 1.7946,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.895551175639688e-06,
+      "loss": 1.7913,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.894798331754176e-06,
+      "loss": 1.8442,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.894042813282039e-06,
+      "loss": 1.8192,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.8932846206361e-06,
+      "loss": 1.7724,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.892523754230646e-06,
+      "loss": 1.7978,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.891760214481425e-06,
+      "loss": 1.8064,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.890994001805646e-06,
+      "loss": 1.8268,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.89022511662198e-06,
+      "loss": 1.7726,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.889453559350557e-06,
+      "loss": 1.7836,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.888679330412965e-06,
+      "loss": 1.8418,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.887902430232252e-06,
+      "loss": 1.8108,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.887122859232931e-06,
+      "loss": 1.8015,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.886340617840968e-06,
+      "loss": 1.8097,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.885555706483792e-06,
+      "loss": 1.7988,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.884768125590288e-06,
+      "loss": 1.7667,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.883977875590803e-06,
+      "loss": 1.7711,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.883184956917139e-06,
+      "loss": 1.8113,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88238937000256e-06,
+      "loss": 1.8374,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.881591115281783e-06,
+      "loss": 1.7806,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88079019319099e-06,
+      "loss": 1.7975,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87998660416781e-06,
+      "loss": 1.7898,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.879180348651343e-06,
+      "loss": 1.8208,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87837142708213e-06,
+      "loss": 1.8104,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.877559839902185e-06,
+      "loss": 1.8218,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.876745587554963e-06,
+      "loss": 1.8144,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87592867048539e-06,
+      "loss": 1.8213,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.875109089139836e-06,
+      "loss": 1.8226,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.874286843966134e-06,
+      "loss": 1.8354,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87346193541357e-06,
+      "loss": 1.7959,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.872634363932887e-06,
+      "loss": 1.8179,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87180412997628e-06,
+      "loss": 1.7997,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.870971233997397e-06,
+      "loss": 1.7826,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.870135676451352e-06,
+      "loss": 1.7845,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.869297457794698e-06,
+      "loss": 1.8512,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.868456578485454e-06,
+      "loss": 1.7878,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.867613038983088e-06,
+      "loss": 1.8127,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.86676683974852e-06,
+      "loss": 1.7951,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.865917981244127e-06,
+      "loss": 1.7908,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.865066463933735e-06,
+      "loss": 1.8139,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.864212288282624e-06,
+      "loss": 1.8106,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.863355454757532e-06,
+      "loss": 1.7552,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.862495963826643e-06,
+      "loss": 1.7906,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.861633815959595e-06,
+      "loss": 1.7679,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.860769011627476e-06,
+      "loss": 1.7959,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.859901551302828e-06,
+      "loss": 1.796,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.859031435459646e-06,
+      "loss": 1.7913,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85815866457337e-06,
+      "loss": 1.7774,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.857283239120896e-06,
+      "loss": 1.782,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85640515958057e-06,
+      "loss": 1.7694,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.855524426432184e-06,
+      "loss": 1.7521,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.854641040156986e-06,
+      "loss": 1.7582,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.853755001237668e-06,
+      "loss": 1.7396,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.852866310158378e-06,
+      "loss": 1.8146,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.851974967404703e-06,
+      "loss": 1.7274,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.851080973463692e-06,
+      "loss": 1.8011,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.85018432882383e-06,
+      "loss": 1.8011,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.849285033975062e-06,
+      "loss": 1.7428,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.84838308940877e-06,
+      "loss": 1.7852,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.847478495617793e-06,
+      "loss": 1.7329,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.846571253096412e-06,
+      "loss": 1.8113,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.845661362340357e-06,
+      "loss": 1.8088,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.844748823846806e-06,
+      "loss": 1.7779,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.843833638114382e-06,
+      "loss": 1.7812,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.842915805643156e-06,
+      "loss": 1.7879,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.841995326934644e-06,
+      "loss": 1.7648,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.84107220249181e-06,
+      "loss": 1.768,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.84014643281906e-06,
+      "loss": 1.7637,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.839218018422248e-06,
+      "loss": 1.7132,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.838286959808674e-06,
+      "loss": 1.7675,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.83735325748708e-06,
+      "loss": 1.7935,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.836416911967657e-06,
+      "loss": 1.7993,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.835477923762033e-06,
+      "loss": 1.7658,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.834536293383287e-06,
+      "loss": 1.769,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.833592021345938e-06,
+      "loss": 1.7541,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.832645108165951e-06,
+      "loss": 1.7038,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.831695554360732e-06,
+      "loss": 1.7348,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.83074336044913e-06,
+      "loss": 1.7594,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.829788526951435e-06,
+      "loss": 1.7499,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.828831054389388e-06,
+      "loss": 1.7454,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.82787094328616e-06,
+      "loss": 1.7757,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.82690819416637e-06,
+      "loss": 1.7729,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.825942807556079e-06,
+      "loss": 1.74,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.824974783982787e-06,
+      "loss": 1.8088,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.824004123975434e-06,
+      "loss": 1.777,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.823030828064406e-06,
+      "loss": 1.77,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.822054896781523e-06,
+      "loss": 1.7669,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.821076330660049e-06,
+      "loss": 1.8021,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.820095130234684e-06,
+      "loss": 1.7565,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.819111296041573e-06,
+      "loss": 1.7727,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.818124828618294e-06,
+      "loss": 1.7307,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.817135728503868e-06,
+      "loss": 1.7444,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.816143996238751e-06,
+      "loss": 1.7697,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.815149632364845e-06,
+      "loss": 1.7847,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.814152637425478e-06,
+      "loss": 1.7653,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.813153011965426e-06,
+      "loss": 1.8015,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.812150756530896e-06,
+      "loss": 1.7577,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.811145871669536e-06,
+      "loss": 1.7415,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.81013835793043e-06,
+      "loss": 1.7759,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.809128215864096e-06,
+      "loss": 1.7482,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.808115446022491e-06,
+      "loss": 1.7075,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.807100048959004e-06,
+      "loss": 1.7583,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.806082025228468e-06,
+      "loss": 1.7377,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.805061375387139e-06,
+      "loss": 1.7502,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.804038099992716e-06,
+      "loss": 1.7168,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.803012199604334e-06,
+      "loss": 1.7412,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.801983674782559e-06,
+      "loss": 1.7324,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.800952526089387e-06,
+      "loss": 1.7261,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.799918754088257e-06,
+      "loss": 1.737,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.798882359344036e-06,
+      "loss": 1.7401,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.79784334242302e-06,
+      "loss": 1.7521,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.79680170389295e-06,
+      "loss": 1.7368,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.795757444322987e-06,
+      "loss": 1.7115,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.794710564283727e-06,
+      "loss": 1.7358,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.793661064347205e-06,
+      "loss": 1.7793,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.79260894508688e-06,
+      "loss": 1.7271,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.791554207077647e-06,
+      "loss": 1.7479,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.790496850895825e-06,
+      "loss": 1.6916,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.789436877119171e-06,
+      "loss": 1.7046,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.788374286326872e-06,
+      "loss": 1.7358,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.78730907909954e-06,
+      "loss": 1.7158,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.786241256019218e-06,
+      "loss": 1.7064,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.785170817669383e-06,
+      "loss": 1.6805,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.784097764634934e-06,
+      "loss": 1.6968,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.783022097502204e-06,
+      "loss": 1.7251,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.781943816858955e-06,
+      "loss": 1.7161,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.780862923294369e-06,
+      "loss": 1.7238,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.779779417399066e-06,
+      "loss": 1.7239,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.778693299765088e-06,
+      "loss": 1.6942,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.777604570985903e-06,
+      "loss": 1.727,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77651323165641e-06,
+      "loss": 1.7247,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77541928237293e-06,
+      "loss": 1.7052,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.774322723733216e-06,
+      "loss": 1.7342,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77322355633644e-06,
+      "loss": 1.692,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.772121780783202e-06,
+      "loss": 1.7097,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.771017397675528e-06,
+      "loss": 1.7497,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.76991040761687e-06,
+      "loss": 1.73,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.768800811212104e-06,
+      "loss": 1.6739,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.767688609067526e-06,
+      "loss": 1.7559,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.766573801790861e-06,
+      "loss": 1.743,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.765456389991255e-06,
+      "loss": 1.7328,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.764336374279276e-06,
+      "loss": 1.7138,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.763213755266916e-06,
+      "loss": 1.7131,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.762088533567593e-06,
+      "loss": 1.7108,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.760960709796141e-06,
+      "loss": 1.765,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.759830284568821e-06,
+      "loss": 1.6892,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.75869725850331e-06,
+      "loss": 1.7071,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.757561632218715e-06,
+      "loss": 1.7373,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.756423406335554e-06,
+      "loss": 1.6702,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 1.6932,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.754139158262725e-06,
+      "loss": 1.7019,
+      "step": 926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.752993137321202e-06,
+      "loss": 1.697,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.751844519277405e-06,
+      "loss": 1.6825,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.750693304758953e-06,
+      "loss": 1.67,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.749539494394887e-06,
+      "loss": 1.6587,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.748383088815663e-06,
+      "loss": 1.7098,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74722408865316e-06,
+      "loss": 1.6785,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.746062494540672e-06,
+      "loss": 1.6851,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.744898307112906e-06,
+      "loss": 1.7144,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.743731527005998e-06,
+      "loss": 1.6784,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.742562154857484e-06,
+      "loss": 1.6867,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.741390191306334e-06,
+      "loss": 1.7131,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74021563699292e-06,
+      "loss": 1.674,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.739038492559034e-06,
+      "loss": 1.6945,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.737858758647889e-06,
+      "loss": 1.6849,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.736676435904107e-06,
+      "loss": 1.7004,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.735491524973723e-06,
+      "loss": 1.7029,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.73430402650419e-06,
+      "loss": 1.7066,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.733113941144375e-06,
+      "loss": 1.6908,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.731921269544557e-06,
+      "loss": 1.7194,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.730726012356428e-06,
+      "loss": 1.6919,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.729528170233094e-06,
+      "loss": 1.6609,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.728327743829074e-06,
+      "loss": 1.6636,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.727124733800293e-06,
+      "loss": 1.6679,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.7259191408041e-06,
+      "loss": 1.6771,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.72471096549924e-06,
+      "loss": 1.6702,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.723500208545882e-06,
+      "loss": 1.6518,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.7222868706056e-06,
+      "loss": 1.6887,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.72107095234138e-06,
+      "loss": 1.6767,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.719852454417613e-06,
+      "loss": 1.703,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.718631377500105e-06,
+      "loss": 1.6776,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.717407722256072e-06,
+      "loss": 1.6849,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.716181489354135e-06,
+      "loss": 1.6956,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.714952679464324e-06,
+      "loss": 1.7368,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.713721293258079e-06,
+      "loss": 1.6679,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.712487331408244e-06,
+      "loss": 1.6565,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.711250794589076e-06,
+      "loss": 1.675,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.710011683476237e-06,
+      "loss": 1.6794,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.708769998746792e-06,
+      "loss": 1.6687,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.707525741079219e-06,
+      "loss": 1.6573,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.706278911153391e-06,
+      "loss": 1.68,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.7050295096506e-06,
+      "loss": 1.6724,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.703777537253534e-06,
+      "loss": 1.645,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.702522994646288e-06,
+      "loss": 1.6867,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.701265882514362e-06,
+      "loss": 1.7107,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.70000620154466e-06,
+      "loss": 1.6807,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.69874395242549e-06,
+      "loss": 1.6436,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.697479135846559e-06,
+      "loss": 1.6857,
+      "step": 973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.696211752498985e-06,
+      "loss": 1.6905,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.694941803075285e-06,
+      "loss": 1.6342,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.693669288269371e-06,
+      "loss": 1.6567,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.692394208776568e-06,
+      "loss": 1.6365,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.691116565293597e-06,
+      "loss": 1.6316,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.689836358518579e-06,
+      "loss": 1.6508,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.688553589151038e-06,
+      "loss": 1.6569,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.687268257891895e-06,
+      "loss": 1.6586,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.685980365443477e-06,
+      "loss": 1.6888,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.684689912509503e-06,
+      "loss": 1.6512,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.683396899795096e-06,
+      "loss": 1.6711,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.682101328006776e-06,
+      "loss": 1.6148,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.680803197852462e-06,
+      "loss": 1.656,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.67950251004147e-06,
+      "loss": 1.6344,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.678199265284516e-06,
+      "loss": 1.6453,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.676893464293708e-06,
+      "loss": 1.6617,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.675585107782557e-06,
+      "loss": 1.6029,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.674274196465965e-06,
+      "loss": 1.627,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.672960731060235e-06,
+      "loss": 1.628,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.671644712283061e-06,
+      "loss": 1.6761,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.670326140853535e-06,
+      "loss": 1.6385,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.669005017492142e-06,
+      "loss": 1.7022,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.667681342920764e-06,
+      "loss": 1.6294,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.666355117862674e-06,
+      "loss": 1.6326,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.66502634304254e-06,
+      "loss": 1.6364,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.663695019186423e-06,
+      "loss": 1.6294,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.66236114702178e-06,
+      "loss": 1.6423,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.661024727277454e-06,
+      "loss": 1.6104,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.659685760683685e-06,
+      "loss": 1.6377,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.658344247972103e-06,
+      "loss": 1.6473,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.65700018987573e-06,
+      "loss": 1.6246,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.655653587128978e-06,
+      "loss": 1.6012,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.654304440467647e-06,
+      "loss": 1.599,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.652952750628933e-06,
+      "loss": 1.6158,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.651598518351418e-06,
+      "loss": 1.6494,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.650241744375073e-06,
+      "loss": 1.63,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.648882429441258e-06,
+      "loss": 1.6283,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.647520574292721e-06,
+      "loss": 1.6125,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.646156179673602e-06,
+      "loss": 1.6161,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.644789246329422e-06,
+      "loss": 1.6352,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.643419775007095e-06,
+      "loss": 1.6196,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.642047766454917e-06,
+      "loss": 1.6428,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.640673221422575e-06,
+      "loss": 1.5941,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.639296140661137e-06,
+      "loss": 1.6379,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.637916524923063e-06,
+      "loss": 1.6115,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.636534374962192e-06,
+      "loss": 1.6141,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.635149691533749e-06,
+      "loss": 1.6098,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.633762475394348e-06,
+      "loss": 1.6265,
+      "step": 1021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.632372727301982e-06,
+      "loss": 1.6104,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.630980448016028e-06,
+      "loss": 1.6382,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.629585638297247e-06,
+      "loss": 1.6244,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.628188298907782e-06,
+      "loss": 1.6081,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.626788430611162e-06,
+      "loss": 1.6067,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.62538603417229e-06,
+      "loss": 1.6166,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.62398111035746e-06,
+      "loss": 1.5886,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.62257365993434e-06,
+      "loss": 1.5845,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.62116368367198e-06,
+      "loss": 1.5939,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.619751182340811e-06,
+      "loss": 1.5994,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.618336156712646e-06,
+      "loss": 1.6032,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.616918607560672e-06,
+      "loss": 1.6328,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.615498535659457e-06,
+      "loss": 1.5841,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.614075941784953e-06,
+      "loss": 1.6315,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.61265082671448e-06,
+      "loss": 1.5775,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.611223191226744e-06,
+      "loss": 1.6215,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.609793036101825e-06,
+      "loss": 1.5926,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.608360362121178e-06,
+      "loss": 1.6009,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.606925170067637e-06,
+      "loss": 1.6003,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.605487460725412e-06,
+      "loss": 1.6253,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.604047234880087e-06,
+      "loss": 1.6038,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.602604493318624e-06,
+      "loss": 1.5572,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.601159236829353e-06,
+      "loss": 1.5816,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.599711466201985e-06,
+      "loss": 1.6122,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.598261182227601e-06,
+      "loss": 1.5947,
+      "step": 1046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.596808385698658e-06,
+      "loss": 1.5797,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.595353077408984e-06,
+      "loss": 1.6088,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.593895258153778e-06,
+      "loss": 1.5518,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.592434928729617e-06,
+      "loss": 1.588,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.59097208993444e-06,
+      "loss": 1.5704,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.589506742567567e-06,
+      "loss": 1.5929,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.588038887429684e-06,
+      "loss": 1.5706,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.586568525322847e-06,
+      "loss": 1.6184,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.585095657050483e-06,
+      "loss": 1.5958,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.583620283417386e-06,
+      "loss": 1.5942,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.582142405229723e-06,
+      "loss": 1.6014,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.580662023295026e-06,
+      "loss": 1.5897,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.579179138422196e-06,
+      "loss": 1.6266,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.577693751421507e-06,
+      "loss": 1.6143,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.576205863104588e-06,
+      "loss": 1.5936,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.574715474284448e-06,
+      "loss": 1.5885,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.573222585775452e-06,
+      "loss": 1.5518,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.571727198393338e-06,
+      "loss": 1.6002,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.570229312955207e-06,
+      "loss": 1.5612,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.568728930279525e-06,
+      "loss": 1.5684,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.56722605118612e-06,
+      "loss": 1.5705,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.565720676496186e-06,
+      "loss": 1.5969,
+      "step": 1068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.564212807032282e-06,
+      "loss": 1.5725,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.562702443618332e-06,
+      "loss": 1.5839,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.561189587079614e-06,
+      "loss": 1.5646,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.559674238242778e-06,
+      "loss": 1.5804,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.558156397935832e-06,
+      "loss": 1.5512,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.556636066988144e-06,
+      "loss": 1.5728,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.555113246230443e-06,
+      "loss": 1.5667,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.553587936494821e-06,
+      "loss": 1.5861,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.552060138614731e-06,
+      "loss": 1.5857,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.550529853424979e-06,
+      "loss": 1.5651,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.548997081761737e-06,
+      "loss": 1.6347,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.547461824462534e-06,
+      "loss": 1.6315,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.545924082366253e-06,
+      "loss": 1.6307,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.54438385631314e-06,
+      "loss": 1.5805,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.542841147144795e-06,
+      "loss": 1.5523,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.541295955704177e-06,
+      "loss": 1.558,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.5397482828356e-06,
+      "loss": 1.6018,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.538198129384734e-06,
+      "loss": 1.5709,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.536645496198605e-06,
+      "loss": 1.5633,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.535090384125593e-06,
+      "loss": 1.5916,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.533532794015434e-06,
+      "loss": 1.5784,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.531972726719216e-06,
+      "loss": 1.5699,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.530410183089385e-06,
+      "loss": 1.5747,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.528845163979731e-06,
+      "loss": 1.5907,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.527277670245407e-06,
+      "loss": 1.5575,
+      "step": 1093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.525707702742913e-06,
+      "loss": 1.5419,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.524135262330098e-06,
+      "loss": 1.54,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.52256034986617e-06,
+      "loss": 1.5799,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.52098296621168e-06,
+      "loss": 1.5559,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.519403112228536e-06,
+      "loss": 1.5635,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.51782078877999e-06,
+      "loss": 1.5603,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.516235996730645e-06,
+      "loss": 1.5735,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.514648736946455e-06,
+      "loss": 1.5956,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.513059010294723e-06,
+      "loss": 1.5296,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.511466817644095e-06,
+      "loss": 1.569,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.509872159864565e-06,
+      "loss": 1.5628,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.508275037827483e-06,
+      "loss": 1.4957,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.506675452405537e-06,
+      "loss": 1.5536,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.50507340447276e-06,
+      "loss": 1.5444,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.503468894904534e-06,
+      "loss": 1.5413,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.501861924577588e-06,
+      "loss": 1.5209,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.500252494369992e-06,
+      "loss": 1.5749,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.498640605161163e-06,
+      "loss": 1.5476,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.497026257831856e-06,
+      "loss": 1.5414,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.495409453264176e-06,
+      "loss": 1.559,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.493790192341568e-06,
+      "loss": 1.5695,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.492168475948815e-06,
+      "loss": 1.5592,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.490544304972051e-06,
+      "loss": 1.5318,
+      "step": 1116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.488917680298743e-06,
+      "loss": 1.5531,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.487288602817702e-06,
+      "loss": 1.5538,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.485657073419078e-06,
+      "loss": 1.5768,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.484023092994366e-06,
+      "loss": 1.5606,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.48238666243639e-06,
+      "loss": 1.555,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.48074778263932e-06,
+      "loss": 1.5395,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.479106454498664e-06,
+      "loss": 1.5478,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.477462678911268e-06,
+      "loss": 1.551,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.475816456775313e-06,
+      "loss": 1.5286,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.474167788990316e-06,
+      "loss": 1.5142,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.472516676457137e-06,
+      "loss": 1.5632,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.470863120077962e-06,
+      "loss": 1.5092,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.46920712075632e-06,
+      "loss": 1.5361,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.467548679397072e-06,
+      "loss": 1.542,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.465887796906413e-06,
+      "loss": 1.5361,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.464224474191874e-06,
+      "loss": 1.5032,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.462558712162317e-06,
+      "loss": 1.5203,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.460890511727939e-06,
+      "loss": 1.5504,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.459219873800265e-06,
+      "loss": 1.5263,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.457546799292156e-06,
+      "loss": 1.5349,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.455871289117808e-06,
+      "loss": 1.5712,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.454193344192737e-06,
+      "loss": 1.5436,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4525129654338e-06,
+      "loss": 1.5408,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.450830153759177e-06,
+      "loss": 1.5412,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.449144910088381e-06,
+      "loss": 1.5176,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.447457235342254e-06,
+      "loss": 1.5148,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.445767130442964e-06,
+      "loss": 1.5394,
+      "step": 1143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.444074596314007e-06,
+      "loss": 1.539,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.442379633880212e-06,
+      "loss": 1.5351,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.440682244067724e-06,
+      "loss": 1.5229,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.438982427804024e-06,
+      "loss": 1.5371,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.437280186017914e-06,
+      "loss": 1.5335,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.435575519639526e-06,
+      "loss": 1.4822,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.43386842960031e-06,
+      "loss": 1.525,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.432158916833047e-06,
+      "loss": 1.5199,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.430446982271837e-06,
+      "loss": 1.4936,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.428732626852104e-06,
+      "loss": 1.5513,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4270158515106e-06,
+      "loss": 1.5081,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.42529665718539e-06,
+      "loss": 1.5171,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.423575044815872e-06,
+      "loss": 1.4968,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.421851015342753e-06,
+      "loss": 1.5342,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.420124569708069e-06,
+      "loss": 1.5044,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.418395708855178e-06,
+      "loss": 1.5102,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.416664433728749e-06,
+      "loss": 1.5086,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.414930745274775e-06,
+      "loss": 1.5378,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.413194644440573e-06,
+      "loss": 1.5362,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.411456132174768e-06,
+      "loss": 1.5031,
+      "step": 1163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.409715209427307e-06,
+      "loss": 1.514,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.40797187714946e-06,
+      "loss": 1.4741,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.406226136293804e-06,
+      "loss": 1.5109,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.404477987814236e-06,
+      "loss": 1.4898,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.402727432665972e-06,
+      "loss": 1.5309,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.400974471805538e-06,
+      "loss": 1.5496,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.399219106190776e-06,
+      "loss": 1.476,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.397461336780841e-06,
+      "loss": 1.5332,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.39570116453621e-06,
+      "loss": 1.5155,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.393938590418657e-06,
+      "loss": 1.5069,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.392173615391282e-06,
+      "loss": 1.5372,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.39040624041849e-06,
+      "loss": 1.4934,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.388636466466003e-06,
+      "loss": 1.5125,
+      "step": 1176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.386864294500846e-06,
+      "loss": 1.4975,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.385089725491364e-06,
+      "loss": 1.4861,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.383312760407201e-06,
+      "loss": 1.5158,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.381533400219319e-06,
+      "loss": 1.5051,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.379751645899984e-06,
+      "loss": 1.496,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.377967498422772e-06,
+      "loss": 1.4889,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.376180958762567e-06,
+      "loss": 1.4758,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.374392027895557e-06,
+      "loss": 1.5194,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.372600706799244e-06,
+      "loss": 1.5071,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.370806996452425e-06,
+      "loss": 1.4768,
+      "step": 1186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.369010897835212e-06,
+      "loss": 1.4894,
+      "step": 1187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.367212411929019e-06,
+      "loss": 1.5191,
+      "step": 1188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.365411539716562e-06,
+      "loss": 1.4862,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.363608282181862e-06,
+      "loss": 1.4817,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.361802640310248e-06,
+      "loss": 1.4875,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.359994615088345e-06,
+      "loss": 1.4846,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.358184207504086e-06,
+      "loss": 1.4907,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.3563714185467e-06,
+      "loss": 1.525,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.354556249206724e-06,
+      "loss": 1.5014,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.352738700475988e-06,
+      "loss": 1.4978,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.35091877334763e-06,
+      "loss": 1.4614,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.349096468816081e-06,
+      "loss": 1.4805,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.347271787877074e-06,
+      "loss": 1.518,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.345444731527642e-06,
+      "loss": 1.4613,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.343615300766113e-06,
+      "loss": 1.4993,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.341783496592116e-06,
+      "loss": 1.4546,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.339949320006572e-06,
+      "loss": 1.4669,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.3381127720117e-06,
+      "loss": 1.4647,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.336273853611017e-06,
+      "loss": 1.5118,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.334432565809338e-06,
+      "loss": 1.4636,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.332588909612761e-06,
+      "loss": 1.476,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.33074288602869e-06,
+      "loss": 1.4416,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.328894496065816e-06,
+      "loss": 1.5026,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.327043740734129e-06,
+      "loss": 1.4638,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.325190621044905e-06,
+      "loss": 1.5239,
+      "step": 1211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.323335138010717e-06,
+      "loss": 1.5185,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.321477292645422e-06,
+      "loss": 1.4983,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.319617085964177e-06,
+      "loss": 1.4943,
+      "step": 1214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.317754518983424e-06,
+      "loss": 1.4925,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.315889592720897e-06,
+      "loss": 1.4804,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.314022308195616e-06,
+      "loss": 1.4721,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.312152666427894e-06,
+      "loss": 1.5023,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.310280668439325e-06,
+      "loss": 1.4444,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.308406315252799e-06,
+      "loss": 1.5029,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.306529607892489e-06,
+      "loss": 1.5082,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.304650547383852e-06,
+      "loss": 1.4728,
+      "step": 1222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.302769134753635e-06,
+      "loss": 1.4869,
+      "step": 1223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.300885371029867e-06,
+      "loss": 1.4787,
+      "step": 1224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.298999257241862e-06,
+      "loss": 1.5056,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.297110794420223e-06,
+      "loss": 1.4769,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.29521998359683e-06,
+      "loss": 1.4794,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.293326825804847e-06,
+      "loss": 1.5131,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.291431322078725e-06,
+      "loss": 1.4707,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.289533473454194e-06,
+      "loss": 1.473,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.287633280968263e-06,
+      "loss": 1.481,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.285730745659224e-06,
+      "loss": 1.4782,
+      "step": 1232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.28382586856665e-06,
+      "loss": 1.4452,
+      "step": 1233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.281918650731392e-06,
+      "loss": 1.4797,
+      "step": 1234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.280009093195583e-06,
+      "loss": 1.4573,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.278097197002628e-06,
+      "loss": 1.4766,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.27618296319722e-06,
+      "loss": 1.4882,
+      "step": 1237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.274266392825316e-06,
+      "loss": 1.4625,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.272347486934162e-06,
+      "loss": 1.4734,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.270426246572273e-06,
+      "loss": 1.487,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.26850267278944e-06,
+      "loss": 1.4582,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.266576766636737e-06,
+      "loss": 1.4692,
+      "step": 1242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.2646485291665e-06,
+      "loss": 1.4706,
+      "step": 1243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.26271796143235e-06,
+      "loss": 1.4487,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.26078506448917e-06,
+      "loss": 1.5028,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.25884983939313e-06,
+      "loss": 1.4465,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.256912287201658e-06,
+      "loss": 1.4938,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.25497240897346e-06,
+      "loss": 1.458,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.253030205768516e-06,
+      "loss": 1.4507,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.251085678648072e-06,
+      "loss": 1.4493,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.249138828674645e-06,
+      "loss": 1.4255,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.24718965691202e-06,
+      "loss": 1.4654,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.245238164425251e-06,
+      "loss": 1.4593,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.243284352280664e-06,
+      "loss": 1.4524,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.241328221545849e-06,
+      "loss": 1.4451,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.239369773289661e-06,
+      "loss": 1.4457,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.237409008582226e-06,
+      "loss": 1.4568,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.235445928494931e-06,
+      "loss": 1.4018,
+      "step": 1258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.233480534100432e-06,
+      "loss": 1.4764,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.231512826472651e-06,
+      "loss": 1.4559,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.229542806686764e-06,
+      "loss": 1.4475,
+      "step": 1261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.227570475819226e-06,
+      "loss": 1.4276,
+      "step": 1262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.225595834947739e-06,
+      "loss": 1.4564,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.223618885151278e-06,
+      "loss": 1.4213,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.221639627510076e-06,
+      "loss": 1.4259,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.219658063105628e-06,
+      "loss": 1.4689,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.217674193020686e-06,
+      "loss": 1.4659,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.215688018339266e-06,
+      "loss": 1.4783,
+      "step": 1268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.213699540146643e-06,
+      "loss": 1.4425,
+      "step": 1269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.211708759529347e-06,
+      "loss": 1.4418,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.209715677575172e-06,
+      "loss": 1.4607,
+      "step": 1271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.207720295373162e-06,
+      "loss": 1.4417,
+      "step": 1272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.205722614013625e-06,
+      "loss": 1.4566,
+      "step": 1273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.203722634588119e-06,
+      "loss": 1.4376,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.201720358189464e-06,
+      "loss": 1.4611,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.199715785911731e-06,
+      "loss": 1.4652,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.197708918850245e-06,
+      "loss": 1.439,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.195699758101589e-06,
+      "loss": 1.4151,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.193688304763595e-06,
+      "loss": 1.4344,
+      "step": 1279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.191674559935349e-06,
+      "loss": 1.4489,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.189658524717191e-06,
+      "loss": 1.453,
+      "step": 1281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.18764020021071e-06,
+      "loss": 1.428,
+      "step": 1282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.185619587518746e-06,
+      "loss": 1.4282,
+      "step": 1283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.183596687745392e-06,
+      "loss": 1.4694,
+      "step": 1284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.181571501995986e-06,
+      "loss": 1.4487,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.179544031377124e-06,
+      "loss": 1.4767,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.177514276996638e-06,
+      "loss": 1.4493,
+      "step": 1287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.175482239963619e-06,
+      "loss": 1.4235,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.173447921388396e-06,
+      "loss": 1.4253,
+      "step": 1289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.171411322382552e-06,
+      "loss": 1.4435,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.169372444058914e-06,
+      "loss": 1.466,
+      "step": 1291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.16733128753155e-06,
+      "loss": 1.4192,
+      "step": 1292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.165287853915782e-06,
+      "loss": 1.4369,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.163242144328169e-06,
+      "loss": 1.4393,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.161194159886515e-06,
+      "loss": 1.4047,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.159143901709865e-06,
+      "loss": 1.4391,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.157091370918512e-06,
+      "loss": 1.4119,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.155036568633988e-06,
+      "loss": 1.4415,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.152979495979064e-06,
+      "loss": 1.4263,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.150920154077753e-06,
+      "loss": 1.4394,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.148858544055313e-06,
+      "loss": 1.4004,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.14679466703823e-06,
+      "loss": 1.4043,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.144728524154238e-06,
+      "loss": 1.4148,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.14266011653231e-06,
+      "loss": 1.4401,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.140589445302647e-06,
+      "loss": 1.4403,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.138516511596698e-06,
+      "loss": 1.4076,
+      "step": 1306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.13644131654714e-06,
+      "loss": 1.4521,
+      "step": 1307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.134363861287892e-06,
+      "loss": 1.3745,
+      "step": 1308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.132284146954101e-06,
+      "loss": 1.4026,
+      "step": 1309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.130202174682154e-06,
+      "loss": 1.4244,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.12811794560967e-06,
+      "loss": 1.4243,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.1260314608755e-06,
+      "loss": 1.4016,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.12394272161973e-06,
+      "loss": 1.3993,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.121851728983677e-06,
+      "loss": 1.4289,
+      "step": 1314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.119758484109886e-06,
+      "loss": 1.4303,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.117662988142138e-06,
+      "loss": 1.3877,
+      "step": 1316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.11556524222544e-06,
+      "loss": 1.4532,
+      "step": 1317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.11346524750603e-06,
+      "loss": 1.4502,
+      "step": 1318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.111363005131377e-06,
+      "loss": 1.4519,
+      "step": 1319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.109258516250172e-06,
+      "loss": 1.4075,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.107151782012341e-06,
+      "loss": 1.4199,
+      "step": 1321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.105042803569029e-06,
+      "loss": 1.4084,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.102931582072615e-06,
+      "loss": 1.4229,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.100818118676697e-06,
+      "loss": 1.4467,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.098702414536107e-06,
+      "loss": 1.3929,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.096584470806889e-06,
+      "loss": 1.4202,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.09446428864632e-06,
+      "loss": 1.4251,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.092341869212898e-06,
+      "loss": 1.4317,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.09021721366634e-06,
+      "loss": 1.4252,
+      "step": 1329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.08809032316759e-06,
+      "loss": 1.4171,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.085961198878812e-06,
+      "loss": 1.4236,
+      "step": 1331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.083829841963388e-06,
+      "loss": 1.4258,
+      "step": 1332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.08169625358592e-06,
+      "loss": 1.4161,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.079560434912234e-06,
+      "loss": 1.4268,
+      "step": 1334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.077422387109367e-06,
+      "loss": 1.4327,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.075282111345583e-06,
+      "loss": 1.3991,
+      "step": 1336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.073139608790353e-06,
+      "loss": 1.4041,
+      "step": 1337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.070994880614377e-06,
+      "loss": 1.4091,
+      "step": 1338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.06884792798956e-06,
+      "loss": 1.404,
+      "step": 1339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.06669875208903e-06,
+      "loss": 1.3754,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.064547354087126e-06,
+      "loss": 1.4151,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.062393735159401e-06,
+      "loss": 1.4326,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.060237896482622e-06,
+      "loss": 1.4128,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.058079839234772e-06,
+      "loss": 1.4413,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.055919564595043e-06,
+      "loss": 1.4279,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.05375707374384e-06,
+      "loss": 1.4285,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.051592367862778e-06,
+      "loss": 1.4435,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.049425448134682e-06,
+      "loss": 1.4157,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.047256315743586e-06,
+      "loss": 1.4341,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 1.4066,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.042911417714589e-06,
+      "loss": 1.3865,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.040735654450798e-06,
+      "loss": 1.4194,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.038557683272235e-06,
+      "loss": 1.3981,
+      "step": 1353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.036377505368975e-06,
+      "loss": 1.376,
+      "step": 1354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.034195121932294e-06,
+      "loss": 1.4221,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.03201053415468e-06,
+      "loss": 1.4169,
+      "step": 1356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.029823743229823e-06,
+      "loss": 1.3914,
+      "step": 1357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.027634750352615e-06,
+      "loss": 1.3851,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.02544355671915e-06,
+      "loss": 1.39,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.023250163526731e-06,
+      "loss": 1.4221,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.021054571973856e-06,
+      "loss": 1.4169,
+      "step": 1361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.01885678326023e-06,
+      "loss": 1.3909,
+      "step": 1362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.016656798586753e-06,
+      "loss": 1.4103,
+      "step": 1363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.01445461915553e-06,
+      "loss": 1.4193,
+      "step": 1364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.01225024616986e-06,
+      "loss": 1.4088,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.010043680834245e-06,
+      "loss": 1.4,
+      "step": 1366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.007834924354384e-06,
+      "loss": 1.3964,
+      "step": 1367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.00562397793717e-06,
+      "loss": 1.405,
+      "step": 1368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.003410842790698e-06,
+      "loss": 1.3883,
+      "step": 1369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.001195520124257e-06,
+      "loss": 1.3759,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.998978011148325e-06,
+      "loss": 1.3906,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.996758317074587e-06,
+      "loss": 1.4255,
+      "step": 1372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.99453643911591e-06,
+      "loss": 1.4163,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.99231237848636e-06,
+      "loss": 1.38,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.990086136401199e-06,
+      "loss": 1.3647,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.987857714076872e-06,
+      "loss": 1.3585,
+      "step": 1376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.985627112731022e-06,
+      "loss": 1.4374,
+      "step": 1377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.983394333582482e-06,
+      "loss": 1.4174,
+      "step": 1378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.981159377851273e-06,
+      "loss": 1.3758,
+      "step": 1379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.978922246758607e-06,
+      "loss": 1.3787,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.976682941526881e-06,
+      "loss": 1.3837,
+      "step": 1381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.974441463379688e-06,
+      "loss": 1.3889,
+      "step": 1382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.972197813541799e-06,
+      "loss": 1.3994,
+      "step": 1383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.969951993239177e-06,
+      "loss": 1.4395,
+      "step": 1384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.967704003698972e-06,
+      "loss": 1.355,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.965453846149513e-06,
+      "loss": 1.4011,
+      "step": 1386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.963201521820319e-06,
+      "loss": 1.3689,
+      "step": 1387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.960947031942093e-06,
+      "loss": 1.3894,
+      "step": 1388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.95869037774672e-06,
+      "loss": 1.3734,
+      "step": 1389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.956431560467267e-06,
+      "loss": 1.3548,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.954170581337983e-06,
+      "loss": 1.4175,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.9519074415943e-06,
+      "loss": 1.4082,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.949642142472827e-06,
+      "loss": 1.4019,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.947374685211359e-06,
+      "loss": 1.3628,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.945105071048863e-06,
+      "loss": 1.4093,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.942833301225491e-06,
+      "loss": 1.3693,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.940559376982567e-06,
+      "loss": 1.3888,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.938283299562596e-06,
+      "loss": 1.4025,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.93600507020926e-06,
+      "loss": 1.3727,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.933724690167417e-06,
+      "loss": 1.3896,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.931442160683094e-06,
+      "loss": 1.4167,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.929157483003503e-06,
+      "loss": 1.3846,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.926870658377018e-06,
+      "loss": 1.3542,
+      "step": 1403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.924581688053195e-06,
+      "loss": 1.3953,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.922290573282761e-06,
+      "loss": 1.4095,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.919997315317613e-06,
+      "loss": 1.3512,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.917701915410817e-06,
+      "loss": 1.3526,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.915404374816611e-06,
+      "loss": 1.4174,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.913104694790409e-06,
+      "loss": 1.3613,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.910802876588783e-06,
+      "loss": 1.3715,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.90849892146948e-06,
+      "loss": 1.4041,
+      "step": 1411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.906192830691416e-06,
+      "loss": 1.3609,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.90388460551467e-06,
+      "loss": 1.3754,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.901574247200488e-06,
+      "loss": 1.3945,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.899261757011281e-06,
+      "loss": 1.3672,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.896947136210631e-06,
+      "loss": 1.3825,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.894630386063274e-06,
+      "loss": 1.4042,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.892311507835118e-06,
+      "loss": 1.3615,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.889990502793229e-06,
+      "loss": 1.3775,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.88766737220584e-06,
+      "loss": 1.3797,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.885342117342336e-06,
+      "loss": 1.3724,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.883014739473274e-06,
+      "loss": 1.3805,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.880685239870366e-06,
+      "loss": 1.3779,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.878353619806482e-06,
+      "loss": 1.3955,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.87601988055565e-06,
+      "loss": 1.353,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.873684023393062e-06,
+      "loss": 1.3498,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.87134604959506e-06,
+      "loss": 1.3645,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.869005960439147e-06,
+      "loss": 1.3386,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.86666375720398e-06,
+      "loss": 1.3606,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.864319441169373e-06,
+      "loss": 1.3687,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.861973013616292e-06,
+      "loss": 1.3352,
+      "step": 1431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.85962447582686e-06,
+      "loss": 1.3492,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.857273829084348e-06,
+      "loss": 1.3549,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.854921074673184e-06,
+      "loss": 1.3936,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.852566213878947e-06,
+      "loss": 1.3911,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.850209247988365e-06,
+      "loss": 1.3779,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.847850178289316e-06,
+      "loss": 1.3623,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.845489006070833e-06,
+      "loss": 1.3866,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.84312573262309e-06,
+      "loss": 1.3358,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.840760359237412e-06,
+      "loss": 1.3564,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.838392887206276e-06,
+      "loss": 1.3696,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.8360233178233e-06,
+      "loss": 1.3565,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83365165238325e-06,
+      "loss": 1.3545,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.831277892182038e-06,
+      "loss": 1.3847,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.828902038516717e-06,
+      "loss": 1.3504,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.826524092685493e-06,
+      "loss": 1.3945,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.824144055987703e-06,
+      "loss": 1.36,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.821761929723835e-06,
+      "loss": 1.3962,
+      "step": 1448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.819377715195516e-06,
+      "loss": 1.3674,
+      "step": 1449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.816991413705515e-06,
+      "loss": 1.3124,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.81460302655774e-06,
+      "loss": 1.3554,
+      "step": 1451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.81221255505724e-06,
+      "loss": 1.3448,
+      "step": 1452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.809820000510201e-06,
+      "loss": 1.3669,
+      "step": 1453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.807425364223949e-06,
+      "loss": 1.4075,
+      "step": 1454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.805028647506948e-06,
+      "loss": 1.3479,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.802629851668796e-06,
+      "loss": 1.3602,
+      "step": 1456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.80022897802023e-06,
+      "loss": 1.3689,
+      "step": 1457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.79782602787312e-06,
+      "loss": 1.3665,
+      "step": 1458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.795421002540473e-06,
+      "loss": 1.305,
+      "step": 1459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.793013903336428e-06,
+      "loss": 1.3538,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.790604731576257e-06,
+      "loss": 1.3731,
+      "step": 1461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.788193488576369e-06,
+      "loss": 1.3526,
+      "step": 1462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.785780175654296e-06,
+      "loss": 1.31,
+      "step": 1463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.783364794128709e-06,
+      "loss": 1.3511,
+      "step": 1464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.780947345319406e-06,
+      "loss": 1.3611,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.778527830547314e-06,
+      "loss": 1.3855,
+      "step": 1466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.776106251134492e-06,
+      "loss": 1.3172,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.773682608404123e-06,
+      "loss": 1.3284,
+      "step": 1468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.77125690368052e-06,
+      "loss": 1.3928,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.768829138289124e-06,
+      "loss": 1.3385,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.766399313556496e-06,
+      "loss": 1.3524,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.763967430810331e-06,
+      "loss": 1.3511,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.761533491379441e-06,
+      "loss": 1.3491,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.759097496593766e-06,
+      "loss": 1.3867,
+      "step": 1474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.756659447784367e-06,
+      "loss": 1.3459,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.75421934628343e-06,
+      "loss": 1.3624,
+      "step": 1476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.751777193424258e-06,
+      "loss": 1.3292,
+      "step": 1477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.74933299054128e-06,
+      "loss": 1.3466,
+      "step": 1478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.74688673897004e-06,
+      "loss": 1.3168,
+      "step": 1479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.744438440047207e-06,
+      "loss": 1.3413,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.741988095110566e-06,
+      "loss": 1.3464,
+      "step": 1481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.739535705499018e-06,
+      "loss": 1.3159,
+      "step": 1482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.737081272552583e-06,
+      "loss": 1.3406,
+      "step": 1483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.734624797612402e-06,
+      "loss": 1.3252,
+      "step": 1484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.73216628202072e-06,
+      "loss": 1.3501,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.729705727120911e-06,
+      "loss": 1.3357,
+      "step": 1486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.727243134257454e-06,
+      "loss": 1.3329,
+      "step": 1487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.724778504775946e-06,
+      "loss": 1.3277,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.722311840023094e-06,
+      "loss": 1.3448,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.719843141346717e-06,
+      "loss": 1.3254,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.717372410095749e-06,
+      "loss": 1.3598,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.714899647620232e-06,
+      "loss": 1.3811,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.712424855271316e-06,
+      "loss": 1.3594,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.709948034401265e-06,
+      "loss": 1.354,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.707469186363446e-06,
+      "loss": 1.3556,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.70498831251234e-06,
+      "loss": 1.3234,
+      "step": 1496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.702505414203528e-06,
+      "loss": 1.3484,
+      "step": 1497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.700020492793702e-06,
+      "loss": 1.3607,
+      "step": 1498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.697533549640658e-06,
+      "loss": 1.339,
+      "step": 1499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.695044586103297e-06,
+      "loss": 1.3611,
+      "step": 1500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.692553603541622e-06,
+      "loss": 1.3293,
+      "step": 1501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.690060603316744e-06,
+      "loss": 1.312,
+      "step": 1502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.68756558679087e-06,
+      "loss": 1.3016,
+      "step": 1503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.685068555327314e-06,
+      "loss": 1.3637,
+      "step": 1504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.68256951029049e-06,
+      "loss": 1.3427,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.680068453045908e-06,
+      "loss": 1.3144,
+      "step": 1506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.677565384960185e-06,
+      "loss": 1.3691,
+      "step": 1507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.67506030740103e-06,
+      "loss": 1.36,
+      "step": 1508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.672553221737253e-06,
+      "loss": 1.3637,
+      "step": 1509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.67004412933876e-06,
+      "loss": 1.3238,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.667533031576556e-06,
+      "loss": 1.3162,
+      "step": 1511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.66501992982274e-06,
+      "loss": 1.3167,
+      "step": 1512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.662504825450505e-06,
+      "loss": 1.324,
+      "step": 1513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.65998771983414e-06,
+      "loss": 1.3445,
+      "step": 1514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.657468614349028e-06,
+      "loss": 1.3493,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.654947510371641e-06,
+      "loss": 1.3305,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.652424409279547e-06,
+      "loss": 1.2844,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.649899312451407e-06,
+      "loss": 1.3479,
+      "step": 1518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.647372221266962e-06,
+      "loss": 1.3098,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.644843137107058e-06,
+      "loss": 1.3525,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.64231206135362e-06,
+      "loss": 1.3428,
+      "step": 1521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.639778995389658e-06,
+      "loss": 1.3385,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.637243940599282e-06,
+      "loss": 1.3678,
+      "step": 1523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.634706898367677e-06,
+      "loss": 1.3576,
+      "step": 1524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.632167870081122e-06,
+      "loss": 1.3326,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.629626857126975e-06,
+      "loss": 1.3464,
+      "step": 1526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.627083860893682e-06,
+      "loss": 1.3341,
+      "step": 1527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.624538882770772e-06,
+      "loss": 1.341,
+      "step": 1528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.621991924148856e-06,
+      "loss": 1.317,
+      "step": 1529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.619442986419631e-06,
+      "loss": 1.2973,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.616892070975869e-06,
+      "loss": 1.3193,
+      "step": 1531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.614339179211426e-06,
+      "loss": 1.3334,
+      "step": 1532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.611784312521241e-06,
+      "loss": 1.3316,
+      "step": 1533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.609227472301327e-06,
+      "loss": 1.3098,
+      "step": 1534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.606668659948778e-06,
+      "loss": 1.2965,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.604107876861765e-06,
+      "loss": 1.3372,
+      "step": 1536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.601545124439535e-06,
+      "loss": 1.3168,
+      "step": 1537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.598980404082413e-06,
+      "loss": 1.2885,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.596413717191799e-06,
+      "loss": 1.3348,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.593845065170165e-06,
+      "loss": 1.297,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.59127444942106e-06,
+      "loss": 1.3355,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.5887018713491e-06,
+      "loss": 1.3281,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.586127332359985e-06,
+      "loss": 1.3356,
+      "step": 1543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.583550833860477e-06,
+      "loss": 1.3106,
+      "step": 1544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.58097237725841e-06,
+      "loss": 1.3078,
+      "step": 1545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.578391963962687e-06,
+      "loss": 1.2985,
+      "step": 1546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.575809595383285e-06,
+      "loss": 1.323,
+      "step": 1547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.573225272931246e-06,
+      "loss": 1.279,
+      "step": 1548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.570638998018679e-06,
+      "loss": 1.3107,
+      "step": 1549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.568050772058763e-06,
+      "loss": 1.317,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.565460596465738e-06,
+      "loss": 1.2831,
+      "step": 1551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.562868472654915e-06,
+      "loss": 1.3125,
+      "step": 1552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.560274402042666e-06,
+      "loss": 1.3449,
+      "step": 1553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.557678386046429e-06,
+      "loss": 1.3171,
+      "step": 1554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.555080426084699e-06,
+      "loss": 1.3173,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.552480523577043e-06,
+      "loss": 1.3349,
+      "step": 1556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.549878679944083e-06,
+      "loss": 1.3244,
+      "step": 1557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.547274896607501e-06,
+      "loss": 1.3641,
+      "step": 1558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.544669174990044e-06,
+      "loss": 1.3374,
+      "step": 1559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.542061516515512e-06,
+      "loss": 1.3298,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.539451922608769e-06,
+      "loss": 1.3337,
+      "step": 1561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.536840394695729e-06,
+      "loss": 1.3114,
+      "step": 1562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.534226934203373e-06,
+      "loss": 1.289,
+      "step": 1563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.531611542559729e-06,
+      "loss": 1.3241,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.528994221193885e-06,
+      "loss": 1.3315,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.526374971535985e-06,
+      "loss": 1.3205,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.523753795017217e-06,
+      "loss": 1.3143,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.521130693069836e-06,
+      "loss": 1.3269,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.518505667127139e-06,
+      "loss": 1.2853,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.515878718623473e-06,
+      "loss": 1.3461,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.513249848994248e-06,
+      "loss": 1.3008,
+      "step": 1571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.510619059675907e-06,
+      "loss": 1.3089,
+      "step": 1572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.507986352105956e-06,
+      "loss": 1.2938,
+      "step": 1573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.50535172772294e-06,
+      "loss": 1.3037,
+      "step": 1574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.502715187966455e-06,
+      "loss": 1.3364,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.500076734277147e-06,
+      "loss": 1.3189,
+      "step": 1576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.497436368096701e-06,
+      "loss": 1.2775,
+      "step": 1577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.494794090867849e-06,
+      "loss": 1.3176,
+      "step": 1578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.492149904034372e-06,
+      "loss": 1.3057,
+      "step": 1579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.489503809041089e-06,
+      "loss": 1.3238,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.486855807333862e-06,
+      "loss": 1.2901,
+      "step": 1581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.484205900359602e-06,
+      "loss": 1.2786,
+      "step": 1582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.48155408956625e-06,
+      "loss": 1.2969,
+      "step": 1583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.478900376402793e-06,
+      "loss": 1.3271,
+      "step": 1584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.476244762319257e-06,
+      "loss": 1.3288,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.47358724876671e-06,
+      "loss": 1.2993,
+      "step": 1586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.470927837197253e-06,
+      "loss": 1.2807,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.468266529064025e-06,
+      "loss": 1.2972,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.465603325821205e-06,
+      "loss": 1.3032,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.462938228924e-06,
+      "loss": 1.3107,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.460271239828661e-06,
+      "loss": 1.3252,
+      "step": 1591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.457602359992468e-06,
+      "loss": 1.2833,
+      "step": 1592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.45493159087373e-06,
+      "loss": 1.2573,
+      "step": 1593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.452258933931798e-06,
+      "loss": 1.2951,
+      "step": 1594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.449584390627046e-06,
+      "loss": 1.3001,
+      "step": 1595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.446907962420882e-06,
+      "loss": 1.2553,
+      "step": 1596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.444229650775746e-06,
+      "loss": 1.2994,
+      "step": 1597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.441549457155104e-06,
+      "loss": 1.3022,
+      "step": 1598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.438867383023453e-06,
+      "loss": 1.3086,
+      "step": 1599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.436183429846314e-06,
+      "loss": 1.2915,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.433497599090239e-06,
+      "loss": 1.264,
+      "step": 1601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.4308098922228e-06,
+      "loss": 1.3163,
+      "step": 1602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.428120310712601e-06,
+      "loss": 1.2874,
+      "step": 1603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.425428856029264e-06,
+      "loss": 1.2958,
+      "step": 1604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.422735529643445e-06,
+      "loss": 1.2868,
+      "step": 1605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.420040333026807e-06,
+      "loss": 1.2876,
+      "step": 1606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.417343267652046e-06,
+      "loss": 1.2915,
+      "step": 1607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.414644334992878e-06,
+      "loss": 1.2782,
+      "step": 1608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.411943536524036e-06,
+      "loss": 1.2606,
+      "step": 1609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.409240873721277e-06,
+      "loss": 1.3243,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.406536348061371e-06,
+      "loss": 1.2852,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.40382996102211e-06,
+      "loss": 1.2918,
+      "step": 1612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.4011217140823e-06,
+      "loss": 1.3232,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.39841160872177e-06,
+      "loss": 1.3071,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.395699646421355e-06,
+      "loss": 1.3075,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.392985828662911e-06,
+      "loss": 1.2812,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.390270156929306e-06,
+      "loss": 1.2801,
+      "step": 1617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.38755263270442e-06,
+      "loss": 1.3144,
+      "step": 1618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.38483325747315e-06,
+      "loss": 1.2815,
+      "step": 1619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.3821120327214e-06,
+      "loss": 1.2779,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.379388959936082e-06,
+      "loss": 1.2887,
+      "step": 1621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.376664040605122e-06,
+      "loss": 1.2909,
+      "step": 1622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.373937276217458e-06,
+      "loss": 1.3147,
+      "step": 1623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.37120866826303e-06,
+      "loss": 1.2921,
+      "step": 1624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.368478218232787e-06,
+      "loss": 1.3162,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.365745927618688e-06,
+      "loss": 1.2812,
+      "step": 1626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.363011797913689e-06,
+      "loss": 1.2952,
+      "step": 1627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.360275830611764e-06,
+      "loss": 1.2835,
+      "step": 1628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.35753802720788e-06,
+      "loss": 1.2975,
+      "step": 1629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.354798389198013e-06,
+      "loss": 1.285,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.352056918079139e-06,
+      "loss": 1.2815,
+      "step": 1631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.349313615349235e-06,
+      "loss": 1.2914,
+      "step": 1632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.346568482507282e-06,
+      "loss": 1.2676,
+      "step": 1633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.34382152105326e-06,
+      "loss": 1.2911,
+      "step": 1634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.341072732488145e-06,
+      "loss": 1.2966,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.338322118313916e-06,
+      "loss": 1.316,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.335569680033545e-06,
+      "loss": 1.2608,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.332815419151002e-06,
+      "loss": 1.2735,
+      "step": 1638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.33005933717126e-06,
+      "loss": 1.2676,
+      "step": 1639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.327301435600273e-06,
+      "loss": 1.3404,
+      "step": 1640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.324541715945004e-06,
+      "loss": 1.3041,
+      "step": 1641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.321780179713397e-06,
+      "loss": 1.2859,
+      "step": 1642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.319016828414399e-06,
+      "loss": 1.2256,
+      "step": 1643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.316251663557941e-06,
+      "loss": 1.2823,
+      "step": 1644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.313484686654949e-06,
+      "loss": 1.3166,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.310715899217339e-06,
+      "loss": 1.3222,
+      "step": 1646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.307945302758014e-06,
+      "loss": 1.2793,
+      "step": 1647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.305172898790865e-06,
+      "loss": 1.2888,
+      "step": 1648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.302398688830777e-06,
+      "loss": 1.2703,
+      "step": 1649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.299622674393615e-06,
+      "loss": 1.295,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.296844856996232e-06,
+      "loss": 1.2681,
+      "step": 1651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.294065238156464e-06,
+      "loss": 1.2847,
+      "step": 1652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.29128381939314e-06,
+      "loss": 1.3056,
+      "step": 1653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.28850060222606e-06,
+      "loss": 1.2765,
+      "step": 1654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.285715588176018e-06,
+      "loss": 1.2701,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.282928778764783e-06,
+      "loss": 1.2759,
+      "step": 1656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.280140175515106e-06,
+      "loss": 1.3074,
+      "step": 1657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.27734977995072e-06,
+      "loss": 1.2859,
+      "step": 1658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.274557593596335e-06,
+      "loss": 1.2432,
+      "step": 1659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.271763617977642e-06,
+      "loss": 1.283,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.268967854621308e-06,
+      "loss": 1.2534,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.26617030505498e-06,
+      "loss": 1.2788,
+      "step": 1662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.263370970807276e-06,
+      "loss": 1.277,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.260569853407789e-06,
+      "loss": 1.2959,
+      "step": 1664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.257766954387093e-06,
+      "loss": 1.2874,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.254962275276732e-06,
+      "loss": 1.2755,
+      "step": 1666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.25215581760922e-06,
+      "loss": 1.2716,
+      "step": 1667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.249347582918043e-06,
+      "loss": 1.297,
+      "step": 1668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.246537572737664e-06,
+      "loss": 1.3019,
+      "step": 1669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.243725788603509e-06,
+      "loss": 1.3099,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.240912232051978e-06,
+      "loss": 1.2667,
+      "step": 1671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.238096904620438e-06,
+      "loss": 1.2566,
+      "step": 1672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.235279807847223e-06,
+      "loss": 1.2821,
+      "step": 1673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.232460943271633e-06,
+      "loss": 1.2734,
+      "step": 1674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.229640312433938e-06,
+      "loss": 1.2706,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.226817916875368e-06,
+      "loss": 1.2851,
+      "step": 1676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.22399375813812e-06,
+      "loss": 1.2723,
+      "step": 1677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.221167837765355e-06,
+      "loss": 1.2882,
+      "step": 1678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.218340157301194e-06,
+      "loss": 1.3023,
+      "step": 1679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.215510718290724e-06,
+      "loss": 1.2754,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.212679522279988e-06,
+      "loss": 1.2649,
+      "step": 1681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.209846570815992e-06,
+      "loss": 1.239,
+      "step": 1682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.207011865446701e-06,
+      "loss": 1.2715,
+      "step": 1683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.204175407721037e-06,
+      "loss": 1.2765,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.201337199188881e-06,
+      "loss": 1.2704,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.19849724140107e-06,
+      "loss": 1.2572,
+      "step": 1686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.195655535909398e-06,
+      "loss": 1.2596,
+      "step": 1687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.192812084266609e-06,
+      "loss": 1.2517,
+      "step": 1688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.189966888026408e-06,
+      "loss": 1.2728,
+      "step": 1689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.18711994874345e-06,
+      "loss": 1.2358,
+      "step": 1690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.18427126797334e-06,
+      "loss": 1.2517,
+      "step": 1691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.181420847272643e-06,
+      "loss": 1.3009,
+      "step": 1692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.178568688198865e-06,
+      "loss": 1.2595,
+      "step": 1693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.175714792310465e-06,
+      "loss": 1.2523,
+      "step": 1694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.172859161166855e-06,
+      "loss": 1.2845,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.170001796328389e-06,
+      "loss": 1.2708,
+      "step": 1696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.167142699356371e-06,
+      "loss": 1.2719,
+      "step": 1697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.164281871813058e-06,
+      "loss": 1.2994,
+      "step": 1698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.161419315261638e-06,
+      "loss": 1.241,
+      "step": 1699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.158555031266255e-06,
+      "loss": 1.2616,
+      "step": 1700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.155689021391997e-06,
+      "loss": 1.2607,
+      "step": 1701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.152821287204888e-06,
+      "loss": 1.2399,
+      "step": 1702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.149951830271903e-06,
+      "loss": 1.2905,
+      "step": 1703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.14708065216095e-06,
+      "loss": 1.2884,
+      "step": 1704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.14420775444088e-06,
+      "loss": 1.2372,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.141333138681489e-06,
+      "loss": 1.2652,
+      "step": 1706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.138456806453503e-06,
+      "loss": 1.2842,
+      "step": 1707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.135578759328593e-06,
+      "loss": 1.3131,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.132698998879365e-06,
+      "loss": 1.2846,
+      "step": 1709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.129817526679358e-06,
+      "loss": 1.2655,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.126934344303052e-06,
+      "loss": 1.2845,
+      "step": 1711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.124049453325856e-06,
+      "loss": 1.2717,
+      "step": 1712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.121162855324116e-06,
+      "loss": 1.2683,
+      "step": 1713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.11827455187511e-06,
+      "loss": 1.2569,
+      "step": 1714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.115384544557046e-06,
+      "loss": 1.253,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.112492834949066e-06,
+      "loss": 1.2629,
+      "step": 1716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.109599424631242e-06,
+      "loss": 1.2429,
+      "step": 1717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.10670431518457e-06,
+      "loss": 1.2453,
+      "step": 1718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.103807508190984e-06,
+      "loss": 1.2538,
+      "step": 1719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.100909005233336e-06,
+      "loss": 1.247,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.09800880789541e-06,
+      "loss": 1.2304,
+      "step": 1721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.095106917761914e-06,
+      "loss": 1.2742,
+      "step": 1722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.092203336418483e-06,
+      "loss": 1.2394,
+      "step": 1723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.089298065451673e-06,
+      "loss": 1.2426,
+      "step": 1724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.086391106448965e-06,
+      "loss": 1.2556,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.083482460998765e-06,
+      "loss": 1.2674,
+      "step": 1726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.080572130690394e-06,
+      "loss": 1.2738,
+      "step": 1727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.0776601171141e-06,
+      "loss": 1.2579,
+      "step": 1728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.074746421861048e-06,
+      "loss": 1.2365,
+      "step": 1729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.07183104652332e-06,
+      "loss": 1.2481,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.06891399269392e-06,
+      "loss": 1.2492,
+      "step": 1731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.065995261966767e-06,
+      "loss": 1.2572,
+      "step": 1732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.063074855936696e-06,
+      "loss": 1.261,
+      "step": 1733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.060152776199458e-06,
+      "loss": 1.2412,
+      "step": 1734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.057229024351719e-06,
+      "loss": 1.2946,
+      "step": 1735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.054303601991056e-06,
+      "loss": 1.288,
+      "step": 1736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.051376510715964e-06,
+      "loss": 1.2606,
+      "step": 1737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.048447752125846e-06,
+      "loss": 1.2555,
+      "step": 1738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.045517327821015e-06,
+      "loss": 1.2433,
+      "step": 1739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.042585239402697e-06,
+      "loss": 1.2182,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.039651488473028e-06,
+      "loss": 1.2841,
+      "step": 1741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.03671607663505e-06,
+      "loss": 1.2645,
+      "step": 1742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.03377900549271e-06,
+      "loss": 1.2752,
+      "step": 1743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.030840276650868e-06,
+      "loss": 1.2636,
+      "step": 1744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.027899891715288e-06,
+      "loss": 1.2213,
+      "step": 1745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.024957852292637e-06,
+      "loss": 1.2364,
+      "step": 1746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.022014159990484e-06,
+      "loss": 1.2583,
+      "step": 1747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.019068816417304e-06,
+      "loss": 1.2443,
+      "step": 1748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.016121823182478e-06,
+      "loss": 1.2492,
+      "step": 1749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.013173181896283e-06,
+      "loss": 1.2735,
+      "step": 1750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.010222894169897e-06,
+      "loss": 1.2653,
+      "step": 1751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.007270961615397e-06,
+      "loss": 1.2577,
+      "step": 1752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.004317385845765e-06,
+      "loss": 1.2447,
+      "step": 1753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.001362168474874e-06,
+      "loss": 1.2525,
+      "step": 1754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.998405311117493e-06,
+      "loss": 1.2493,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.995446815389296e-06,
+      "loss": 1.2898,
+      "step": 1756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.992486682906844e-06,
+      "loss": 1.2553,
+      "step": 1757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.989524915287595e-06,
+      "loss": 1.2578,
+      "step": 1758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.986561514149901e-06,
+      "loss": 1.2271,
+      "step": 1759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.983596481113005e-06,
+      "loss": 1.2628,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.980629817797045e-06,
+      "loss": 1.2674,
+      "step": 1761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.977661525823046e-06,
+      "loss": 1.2714,
+      "step": 1762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.974691606812923e-06,
+      "loss": 1.231,
+      "step": 1763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.971720062389482e-06,
+      "loss": 1.2631,
+      "step": 1764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.96874689417642e-06,
+      "loss": 1.2673,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.965772103798315e-06,
+      "loss": 1.2223,
+      "step": 1766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.962795692880636e-06,
+      "loss": 1.268,
+      "step": 1767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.959817663049733e-06,
+      "loss": 1.272,
+      "step": 1768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.956838015932846e-06,
+      "loss": 1.2909,
+      "step": 1769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.953856753158096e-06,
+      "loss": 1.2371,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.950873876354485e-06,
+      "loss": 1.2358,
+      "step": 1771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.9478893871519e-06,
+      "loss": 1.267,
+      "step": 1772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.944903287181106e-06,
+      "loss": 1.2113,
+      "step": 1773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.941915578073753e-06,
+      "loss": 1.2312,
+      "step": 1774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.938926261462366e-06,
+      "loss": 1.2457,
+      "step": 1775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.93593533898035e-06,
+      "loss": 1.2658,
+      "step": 1776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.932942812261987e-06,
+      "loss": 1.2707,
+      "step": 1777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.929948682942435e-06,
+      "loss": 1.2292,
+      "step": 1778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.926952952657728e-06,
+      "loss": 1.2435,
+      "step": 1779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.923955623044775e-06,
+      "loss": 1.2268,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.92095669574136e-06,
+      "loss": 1.2413,
+      "step": 1781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.91795617238614e-06,
+      "loss": 1.2404,
+      "step": 1782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.914954054618643e-06,
+      "loss": 1.2753,
+      "step": 1783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.911950344079267e-06,
+      "loss": 1.2474,
+      "step": 1784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.908945042409282e-06,
+      "loss": 1.2747,
+      "step": 1785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.905938151250827e-06,
+      "loss": 1.2368,
+      "step": 1786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.902929672246909e-06,
+      "loss": 1.2297,
+      "step": 1787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.899919607041404e-06,
+      "loss": 1.2548,
+      "step": 1788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.896907957279056e-06,
+      "loss": 1.2596,
+      "step": 1789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.89389472460547e-06,
+      "loss": 1.2336,
+      "step": 1790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.890879910667118e-06,
+      "loss": 1.2107,
+      "step": 1791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.887863517111337e-06,
+      "loss": 1.2194,
+      "step": 1792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.884845545586333e-06,
+      "loss": 1.2287,
+      "step": 1793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.881825997741159e-06,
+      "loss": 1.2611,
+      "step": 1794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.878804875225743e-06,
+      "loss": 1.2182,
+      "step": 1795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.87578217969087e-06,
+      "loss": 1.2698,
+      "step": 1796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.87275791278818e-06,
+      "loss": 1.2251,
+      "step": 1797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.869732076170178e-06,
+      "loss": 1.2271,
+      "step": 1798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.866704671490219e-06,
+      "loss": 1.2327,
+      "step": 1799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.863675700402527e-06,
+      "loss": 1.2193,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.860645164562168e-06,
+      "loss": 1.2129,
+      "step": 1801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.857613065625072e-06,
+      "loss": 1.2463,
+      "step": 1802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.85457940524802e-06,
+      "loss": 1.2377,
+      "step": 1803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.851544185088647e-06,
+      "loss": 1.2323,
+      "step": 1804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.848507406805439e-06,
+      "loss": 1.2531,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.845469072057735e-06,
+      "loss": 1.2144,
+      "step": 1806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.842429182505724e-06,
+      "loss": 1.2514,
+      "step": 1807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.839387739810444e-06,
+      "loss": 1.22,
+      "step": 1808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.836344745633785e-06,
+      "loss": 1.265,
+      "step": 1809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.833300201638475e-06,
+      "loss": 1.2244,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.830254109488102e-06,
+      "loss": 1.2172,
+      "step": 1811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.82720647084709e-06,
+      "loss": 1.2438,
+      "step": 1812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.824157287380714e-06,
+      "loss": 1.2651,
+      "step": 1813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.82110656075509e-06,
+      "loss": 1.2224,
+      "step": 1814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.818054292637176e-06,
+      "loss": 1.2514,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.815000484694777e-06,
+      "loss": 1.2493,
+      "step": 1816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.811945138596535e-06,
+      "loss": 1.206,
+      "step": 1817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.808888256011933e-06,
+      "loss": 1.2536,
+      "step": 1818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.805829838611298e-06,
+      "loss": 1.2511,
+      "step": 1819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.80276988806579e-06,
+      "loss": 1.2323,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.79970840604741e-06,
+      "loss": 1.2208,
+      "step": 1821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.796645394228992e-06,
+      "loss": 1.227,
+      "step": 1822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.793580854284213e-06,
+      "loss": 1.2677,
+      "step": 1823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.790514787887577e-06,
+      "loss": 1.2007,
+      "step": 1824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.787447196714428e-06,
+      "loss": 1.2225,
+      "step": 1825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.78437808244094e-06,
+      "loss": 1.271,
+      "step": 1826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.781307446744121e-06,
+      "loss": 1.2391,
+      "step": 1827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.77823529130181e-06,
+      "loss": 1.2462,
+      "step": 1828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.775161617792673e-06,
+      "loss": 1.2277,
+      "step": 1829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.772086427896211e-06,
+      "loss": 1.2053,
+      "step": 1830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.769009723292751e-06,
+      "loss": 1.2483,
+      "step": 1831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.765931505663446e-06,
+      "loss": 1.2238,
+      "step": 1832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.762851776690279e-06,
+      "loss": 1.2366,
+      "step": 1833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.759770538056058e-06,
+      "loss": 1.2335,
+      "step": 1834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.75668779144441e-06,
+      "loss": 1.2401,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.753603538539797e-06,
+      "loss": 1.1825,
+      "step": 1836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.750517781027496e-06,
+      "loss": 1.2622,
+      "step": 1837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.747430520593609e-06,
+      "loss": 1.2226,
+      "step": 1838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.744341758925056e-06,
+      "loss": 1.2522,
+      "step": 1839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.741251497709583e-06,
+      "loss": 1.2217,
+      "step": 1840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.73815973863575e-06,
+      "loss": 1.2338,
+      "step": 1841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.73506648339294e-06,
+      "loss": 1.2076,
+      "step": 1842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.731971733671347e-06,
+      "loss": 1.2521,
+      "step": 1843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.728875491161988e-06,
+      "loss": 1.2365,
+      "step": 1844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.725777757556697e-06,
+      "loss": 1.217,
+      "step": 1845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.722678534548112e-06,
+      "loss": 1.2214,
+      "step": 1846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.719577823829698e-06,
+      "loss": 1.2222,
+      "step": 1847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.716475627095726e-06,
+      "loss": 1.2574,
+      "step": 1848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.713371946041277e-06,
+      "loss": 1.2276,
+      "step": 1849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.710266782362248e-06,
+      "loss": 1.2236,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.707160137755344e-06,
+      "loss": 1.224,
+      "step": 1851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.70405201391808e-06,
+      "loss": 1.2251,
+      "step": 1852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.700942412548775e-06,
+      "loss": 1.2426,
+      "step": 1853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.697831335346564e-06,
+      "loss": 1.2207,
+      "step": 1854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.69471878401138e-06,
+      "loss": 1.2231,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.691604760243965e-06,
+      "loss": 1.2488,
+      "step": 1856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.688489265745868e-06,
+      "loss": 1.2597,
+      "step": 1857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.685372302219436e-06,
+      "loss": 1.2318,
+      "step": 1858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.682253871367823e-06,
+      "loss": 1.2384,
+      "step": 1859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.679133974894984e-06,
+      "loss": 1.2111,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.676012614505673e-06,
+      "loss": 1.2129,
+      "step": 1861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.672889791905445e-06,
+      "loss": 1.2251,
+      "step": 1862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.669765508800655e-06,
+      "loss": 1.2186,
+      "step": 1863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.666639766898456e-06,
+      "loss": 1.2322,
+      "step": 1864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.663512567906797e-06,
+      "loss": 1.2496,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.660383913534421e-06,
+      "loss": 1.2559,
+      "step": 1866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.657253805490872e-06,
+      "loss": 1.245,
+      "step": 1867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.654122245486484e-06,
+      "loss": 1.2684,
+      "step": 1868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.650989235232384e-06,
+      "loss": 1.2204,
+      "step": 1869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.647854776440496e-06,
+      "loss": 1.2787,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.644718870823529e-06,
+      "loss": 1.2031,
+      "step": 1871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.641581520094985e-06,
+      "loss": 1.2355,
+      "step": 1872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.638442725969162e-06,
+      "loss": 1.2106,
+      "step": 1873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.635302490161136e-06,
+      "loss": 1.2249,
+      "step": 1874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.63216081438678e-06,
+      "loss": 1.2521,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.6290177003627465e-06,
+      "loss": 1.2148,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.6258731498064796e-06,
+      "loss": 1.2118,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.622727164436205e-06,
+      "loss": 1.2137,
+      "step": 1878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.619579745970932e-06,
+      "loss": 1.2343,
+      "step": 1879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.616430896130456e-06,
+      "loss": 1.1996,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.613280616635352e-06,
+      "loss": 1.242,
+      "step": 1881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.610128909206978e-06,
+      "loss": 1.2503,
+      "step": 1882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.60697577556747e-06,
+      "loss": 1.2253,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.603821217439747e-06,
+      "loss": 1.1821,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.6006652365475e-06,
+      "loss": 1.2031,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.597507834615203e-06,
+      "loss": 1.2903,
+      "step": 1886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.594349013368107e-06,
+      "loss": 1.1985,
+      "step": 1887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.591188774532234e-06,
+      "loss": 1.201,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.588027119834383e-06,
+      "loss": 1.2142,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.584864051002128e-06,
+      "loss": 1.2398,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.581699569763812e-06,
+      "loss": 1.2231,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.578533677848554e-06,
+      "loss": 1.2021,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5753663769862415e-06,
+      "loss": 1.2061,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.572197668907533e-06,
+      "loss": 1.2355,
+      "step": 1894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.569027555343853e-06,
+      "loss": 1.2101,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.565856038027397e-06,
+      "loss": 1.2381,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.562683118691128e-06,
+      "loss": 1.2041,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.559508799068771e-06,
+      "loss": 1.2362,
+      "step": 1898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.556333080894822e-06,
+      "loss": 1.1969,
+      "step": 1899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.553155965904535e-06,
+      "loss": 1.2532,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.549977455833933e-06,
+      "loss": 1.2226,
+      "step": 1901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.546797552419798e-06,
+      "loss": 1.2302,
+      "step": 1902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5436162573996726e-06,
+      "loss": 1.2174,
+      "step": 1903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.540433572511861e-06,
+      "loss": 1.2274,
+      "step": 1904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.537249499495429e-06,
+      "loss": 1.2248,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.534064040090198e-06,
+      "loss": 1.2011,
+      "step": 1906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.530877196036748e-06,
+      "loss": 1.1907,
+      "step": 1907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.527688969076412e-06,
+      "loss": 1.2006,
+      "step": 1908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.524499360951287e-06,
+      "loss": 1.2213,
+      "step": 1909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.521308373404218e-06,
+      "loss": 1.2316,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.518116008178805e-06,
+      "loss": 1.234,
+      "step": 1911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.514922267019402e-06,
+      "loss": 1.2299,
+      "step": 1912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.511727151671112e-06,
+      "loss": 1.2553,
+      "step": 1913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.508530663879792e-06,
+      "loss": 1.2453,
+      "step": 1914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5053328053920495e-06,
+      "loss": 1.191,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.502133577955239e-06,
+      "loss": 1.1991,
+      "step": 1916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.498932983317461e-06,
+      "loss": 1.2364,
+      "step": 1917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.495731023227569e-06,
+      "loss": 1.205,
+      "step": 1918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.492527699435155e-06,
+      "loss": 1.251,
+      "step": 1919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.489323013690561e-06,
+      "loss": 1.2422,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.486116967744875e-06,
+      "loss": 1.2104,
+      "step": 1921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.482909563349925e-06,
+      "loss": 1.244,
+      "step": 1922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.47970080225828e-06,
+      "loss": 1.1963,
+      "step": 1923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.476490686223252e-06,
+      "loss": 1.2286,
+      "step": 1924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.473279216998896e-06,
+      "loss": 1.2117,
+      "step": 1925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.47006639634e-06,
+      "loss": 1.2034,
+      "step": 1926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4668522260020995e-06,
+      "loss": 1.2014,
+      "step": 1927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.463636707741458e-06,
+      "loss": 1.2291,
+      "step": 1928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.460419843315083e-06,
+      "loss": 1.2631,
+      "step": 1929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.457201634480713e-06,
+      "loss": 1.1973,
+      "step": 1930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.453982082996822e-06,
+      "loss": 1.2056,
+      "step": 1931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.450761190622618e-06,
+      "loss": 1.2059,
+      "step": 1932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.447538959118045e-06,
+      "loss": 1.2135,
+      "step": 1933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.444315390243771e-06,
+      "loss": 1.2164,
+      "step": 1934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.441090485761203e-06,
+      "loss": 1.2125,
+      "step": 1935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.437864247432474e-06,
+      "loss": 1.222,
+      "step": 1936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.434636677020442e-06,
+      "loss": 1.1948,
+      "step": 1937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.431407776288702e-06,
+      "loss": 1.1876,
+      "step": 1938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.428177547001567e-06,
+      "loss": 1.1966,
+      "step": 1939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.42494599092408e-06,
+      "loss": 1.2253,
+      "step": 1940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.42171310982201e-06,
+      "loss": 1.242,
+      "step": 1941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.418478905461851e-06,
+      "loss": 1.2429,
+      "step": 1942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.415243379610814e-06,
+      "loss": 1.2154,
+      "step": 1943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.412006534036836e-06,
+      "loss": 1.2369,
+      "step": 1944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.408768370508577e-06,
+      "loss": 1.2191,
+      "step": 1945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.405528890795412e-06,
+      "loss": 1.2014,
+      "step": 1946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402288096667443e-06,
+      "loss": 1.2362,
+      "step": 1947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.399045989895481e-06,
+      "loss": 1.2169,
+      "step": 1948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.395802572251063e-06,
+      "loss": 1.181,
+      "step": 1949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.392557845506433e-06,
+      "loss": 1.2377,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.389311811434559e-06,
+      "loss": 1.1754,
+      "step": 1951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.386064471809118e-06,
+      "loss": 1.1939,
+      "step": 1952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3828158284045015e-06,
+      "loss": 1.2008,
+      "step": 1953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.379565882995816e-06,
+      "loss": 1.2083,
+      "step": 1954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3763146373588765e-06,
+      "loss": 1.2252,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.373062093270209e-06,
+      "loss": 1.1961,
+      "step": 1956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.369808252507049e-06,
+      "loss": 1.2143,
+      "step": 1957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.366553116847342e-06,
+      "loss": 1.2168,
+      "step": 1958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.363296688069739e-06,
+      "loss": 1.2267,
+      "step": 1959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.360038967953598e-06,
+      "loss": 1.233,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.356779958278984e-06,
+      "loss": 1.1855,
+      "step": 1961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.353519660826665e-06,
+      "loss": 1.1978,
+      "step": 1962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.350258077378112e-06,
+      "loss": 1.1712,
+      "step": 1963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.346995209715504e-06,
+      "loss": 1.1639,
+      "step": 1964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.343731059621715e-06,
+      "loss": 1.1854,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.340465628880324e-06,
+      "loss": 1.2028,
+      "step": 1966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3371989192756065e-06,
+      "loss": 1.1843,
+      "step": 1967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.33393093259254e-06,
+      "loss": 1.2308,
+      "step": 1968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.330661670616798e-06,
+      "loss": 1.2277,
+      "step": 1969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3273911351347495e-06,
+      "loss": 1.2348,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3241193279334636e-06,
+      "loss": 1.2034,
+      "step": 1971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.3208462508006995e-06,
+      "loss": 1.2146,
+      "step": 1972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.317571905524916e-06,
+      "loss": 1.1975,
+      "step": 1973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.3142962938952575e-06,
+      "loss": 1.241,
+      "step": 1974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.311019417701567e-06,
+      "loss": 1.1799,
+      "step": 1975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.307741278734374e-06,
+      "loss": 1.2176,
+      "step": 1976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.304461878784901e-06,
+      "loss": 1.1689,
+      "step": 1977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.301181219645056e-06,
+      "loss": 1.2369,
+      "step": 1978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.297899303107441e-06,
+      "loss": 1.1786,
+      "step": 1979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2946161309653375e-06,
+      "loss": 1.206,
+      "step": 1980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.291331705012718e-06,
+      "loss": 1.2062,
+      "step": 1981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.288046027044241e-06,
+      "loss": 1.2065,
+      "step": 1982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.284759098855246e-06,
+      "loss": 1.2346,
+      "step": 1983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.281470922241757e-06,
+      "loss": 1.1998,
+      "step": 1984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2781814990004785e-06,
+      "loss": 1.2009,
+      "step": 1985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.274890830928799e-06,
+      "loss": 1.2183,
+      "step": 1986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.271598919824785e-06,
+      "loss": 1.2046,
+      "step": 1987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2683057674871835e-06,
+      "loss": 1.1966,
+      "step": 1988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2650113757154195e-06,
+      "loss": 1.2117,
+      "step": 1989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.261715746309594e-06,
+      "loss": 1.2163,
+      "step": 1990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.258418881070487e-06,
+      "loss": 1.2017,
+      "step": 1991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.255120781799551e-06,
+      "loss": 1.2069,
+      "step": 1992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.251821450298912e-06,
+      "loss": 1.1675,
+      "step": 1993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2485208883713745e-06,
+      "loss": 1.1945,
+      "step": 1994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.24521909782041e-06,
+      "loss": 1.2188,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.241916080450163e-06,
+      "loss": 1.1803,
+      "step": 1996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.238611838065448e-06,
+      "loss": 1.1894,
+      "step": 1997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.235306372471752e-06,
+      "loss": 1.1979,
+      "step": 1998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.231999685475227e-06,
+      "loss": 1.1954,
+      "step": 1999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2286917788826926e-06,
+      "loss": 1.2051,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.225382654501634e-06,
+      "loss": 1.2019,
+      "step": 2001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.222072314140207e-06,
+      "loss": 1.1855,
+      "step": 2002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.218760759607225e-06,
+      "loss": 1.1894,
+      "step": 2003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.21544799271217e-06,
+      "loss": 1.2538,
+      "step": 2004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.212134015265183e-06,
+      "loss": 1.1978,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.208818829077068e-06,
+      "loss": 1.2258,
+      "step": 2006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2055024359592906e-06,
+      "loss": 1.2321,
+      "step": 2007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.202184837723972e-06,
+      "loss": 1.2108,
+      "step": 2008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.198866036183898e-06,
+      "loss": 1.1891,
+      "step": 2009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.195546033152506e-06,
+      "loss": 1.1889,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.192224830443893e-06,
+      "loss": 1.1981,
+      "step": 2011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.188902429872811e-06,
+      "loss": 1.2023,
+      "step": 2012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.185578833254665e-06,
+      "loss": 1.2119,
+      "step": 2013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.182254042405518e-06,
+      "loss": 1.2219,
+      "step": 2014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.17892805914208e-06,
+      "loss": 1.198,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.175600885281715e-06,
+      "loss": 1.1647,
+      "step": 2016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.172272522642437e-06,
+      "loss": 1.1891,
+      "step": 2017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.168942973042912e-06,
+      "loss": 1.2028,
+      "step": 2018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.165612238302454e-06,
+      "loss": 1.1931,
+      "step": 2019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.16228032024102e-06,
+      "loss": 1.1676,
+      "step": 2020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.158947220679218e-06,
+      "loss": 1.1649,
+      "step": 2021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.155612941438301e-06,
+      "loss": 1.1876,
+      "step": 2022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.152277484340167e-06,
+      "loss": 1.234,
+      "step": 2023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.148940851207353e-06,
+      "loss": 1.2094,
+      "step": 2024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.145603043863045e-06,
+      "loss": 1.2037,
+      "step": 2025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.142264064131069e-06,
+      "loss": 1.2074,
+      "step": 2026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.138923913835889e-06,
+      "loss": 1.1922,
+      "step": 2027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.135582594802608e-06,
+      "loss": 1.1977,
+      "step": 2028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.132240108856972e-06,
+      "loss": 1.1909,
+      "step": 2029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.128896457825364e-06,
+      "loss": 1.1506,
+      "step": 2030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.125551643534798e-06,
+      "loss": 1.2041,
+      "step": 2031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.122205667812931e-06,
+      "loss": 1.1833,
+      "step": 2032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.11885853248805e-06,
+      "loss": 1.2259,
+      "step": 2033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.115510239389076e-06,
+      "loss": 1.1974,
+      "step": 2034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.112160790345563e-06,
+      "loss": 1.1857,
+      "step": 2035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.108810187187698e-06,
+      "loss": 1.179,
+      "step": 2036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.105458431746299e-06,
+      "loss": 1.1657,
+      "step": 2037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.10210552585281e-06,
+      "loss": 1.2359,
+      "step": 2038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.098751471339308e-06,
+      "loss": 1.1732,
+      "step": 2039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0953962700384935e-06,
+      "loss": 1.1756,
+      "step": 2040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.092039923783697e-06,
+      "loss": 1.1914,
+      "step": 2041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.088682434408873e-06,
+      "loss": 1.1999,
+      "step": 2042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.085323803748601e-06,
+      "loss": 1.18,
+      "step": 2043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.081964033638082e-06,
+      "loss": 1.2054,
+      "step": 2044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.078603125913146e-06,
+      "loss": 1.1988,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.075241082410236e-06,
+      "loss": 1.1997,
+      "step": 2046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.071877904966422e-06,
+      "loss": 1.2047,
+      "step": 2047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.068513595419391e-06,
+      "loss": 1.2439,
+      "step": 2048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.065148155607449e-06,
+      "loss": 1.1808,
+      "step": 2049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.061781587369518e-06,
+      "loss": 1.1981,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.058413892545142e-06,
+      "loss": 1.1733,
+      "step": 2051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0550450729744735e-06,
+      "loss": 1.2487,
+      "step": 2052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.051675130498283e-06,
+      "loss": 1.2064,
+      "step": 2053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.048304066957958e-06,
+      "loss": 1.1513,
+      "step": 2054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.044931884195492e-06,
+      "loss": 1.1982,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.041558584053492e-06,
+      "loss": 1.1981,
+      "step": 2056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.038184168375178e-06,
+      "loss": 1.1677,
+      "step": 2057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0348086390043805e-06,
+      "loss": 1.1793,
+      "step": 2058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.031431997785532e-06,
+      "loss": 1.1664,
+      "step": 2059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.028054246563679e-06,
+      "loss": 1.2213,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.024675387184472e-06,
+      "loss": 1.1446,
+      "step": 2061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0212954214941686e-06,
+      "loss": 1.2201,
+      "step": 2062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0179143513396274e-06,
+      "loss": 1.1732,
+      "step": 2063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.014532178568314e-06,
+      "loss": 1.1799,
+      "step": 2064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.011148905028295e-06,
+      "loss": 1.1884,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.007764532568238e-06,
+      "loss": 1.2222,
+      "step": 2066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.004379063037411e-06,
+      "loss": 1.2256,
+      "step": 2067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.000992498285684e-06,
+      "loss": 1.174,
+      "step": 2068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9976048401635234e-06,
+      "loss": 1.1625,
+      "step": 2069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.994216090521991e-06,
+      "loss": 1.194,
+      "step": 2070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.99082625121275e-06,
+      "loss": 1.182,
+      "step": 2071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.987435324088054e-06,
+      "loss": 1.1869,
+      "step": 2072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.984043311000754e-06,
+      "loss": 1.1379,
+      "step": 2073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.980650213804296e-06,
+      "loss": 1.1913,
+      "step": 2074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.977256034352713e-06,
+      "loss": 1.2082,
+      "step": 2075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.973860774500633e-06,
+      "loss": 1.1863,
+      "step": 2076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.970464436103276e-06,
+      "loss": 1.2138,
+      "step": 2077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.967067021016445e-06,
+      "loss": 1.1741,
+      "step": 2078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.963668531096539e-06,
+      "loss": 1.1718,
+      "step": 2079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.960268968200539e-06,
+      "loss": 1.1767,
+      "step": 2080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9568683341860135e-06,
+      "loss": 1.1587,
+      "step": 2081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.953466630911118e-06,
+      "loss": 1.1951,
+      "step": 2082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9500638602345895e-06,
+      "loss": 1.2034,
+      "step": 2083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.946660024015751e-06,
+      "loss": 1.1943,
+      "step": 2084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.943255124114505e-06,
+      "loss": 1.1787,
+      "step": 2085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.939849162391339e-06,
+      "loss": 1.1765,
+      "step": 2086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9364421407073176e-06,
+      "loss": 1.1947,
+      "step": 2087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.933034060924084e-06,
+      "loss": 1.1769,
+      "step": 2088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.929624924903863e-06,
+      "loss": 1.1604,
+      "step": 2089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.926214734509451e-06,
+      "loss": 1.1966,
+      "step": 2090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9228034916042285e-06,
+      "loss": 1.184,
+      "step": 2091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.919391198052143e-06,
+      "loss": 1.1934,
+      "step": 2092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.915977855717722e-06,
+      "loss": 1.2155,
+      "step": 2093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.912563466466062e-06,
+      "loss": 1.1833,
+      "step": 2094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.909148032162835e-06,
+      "loss": 1.2008,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9057315546742794e-06,
+      "loss": 1.1575,
+      "step": 2096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.90231403586721e-06,
+      "loss": 1.1822,
+      "step": 2097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.898895477609007e-06,
+      "loss": 1.1751,
+      "step": 2098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.895475881767615e-06,
+      "loss": 1.1863,
+      "step": 2099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.892055250211552e-06,
+      "loss": 1.1903,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.888633584809899e-06,
+      "loss": 1.176,
+      "step": 2101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.885210887432302e-06,
+      "loss": 1.2033,
+      "step": 2102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.88178715994897e-06,
+      "loss": 1.1735,
+      "step": 2103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.878362404230678e-06,
+      "loss": 1.1672,
+      "step": 2104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.874936622148758e-06,
+      "loss": 1.1732,
+      "step": 2105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.871509815575108e-06,
+      "loss": 1.1836,
+      "step": 2106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.868081986382181e-06,
+      "loss": 1.1779,
+      "step": 2107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.864653136442991e-06,
+      "loss": 1.1563,
+      "step": 2108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.861223267631113e-06,
+      "loss": 1.2255,
+      "step": 2109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.8577923818206724e-06,
+      "loss": 1.1714,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.854360480886355e-06,
+      "loss": 1.1818,
+      "step": 2111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.8509275667034e-06,
+      "loss": 1.1724,
+      "step": 2112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.847493641147599e-06,
+      "loss": 1.1769,
+      "step": 2113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.8440587060952965e-06,
+      "loss": 1.1993,
+      "step": 2114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.840622763423391e-06,
+      "loss": 1.175,
+      "step": 2115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.837185815009329e-06,
+      "loss": 1.1789,
+      "step": 2116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.833747862731106e-06,
+      "loss": 1.1578,
+      "step": 2117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.830308908467271e-06,
+      "loss": 1.1731,
+      "step": 2118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.826868954096915e-06,
+      "loss": 1.1766,
+      "step": 2119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.823428001499677e-06,
+      "loss": 1.1945,
+      "step": 2120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.8199860525557414e-06,
+      "loss": 1.1878,
+      "step": 2121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.816543109145839e-06,
+      "loss": 1.1534,
+      "step": 2122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.813099173151242e-06,
+      "loss": 1.1761,
+      "step": 2123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.809654246453765e-06,
+      "loss": 1.1609,
+      "step": 2124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.806208330935766e-06,
+      "loss": 1.2013,
+      "step": 2125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.802761428480138e-06,
+      "loss": 1.1857,
+      "step": 2126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.799313540970322e-06,
+      "loss": 1.1697,
+      "step": 2127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.795864670290288e-06,
+      "loss": 1.1601,
+      "step": 2128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.792414818324549e-06,
+      "loss": 1.1705,
+      "step": 2129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.788963986958153e-06,
+      "loss": 1.1788,
+      "step": 2130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.785512178076683e-06,
+      "loss": 1.2033,
+      "step": 2131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.782059393566254e-06,
+      "loss": 1.1541,
+      "step": 2132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.778605635313517e-06,
+      "loss": 1.1589,
+      "step": 2133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.775150905205656e-06,
+      "loss": 1.1618,
+      "step": 2134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.771695205130382e-06,
+      "loss": 1.152,
+      "step": 2135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.768238536975939e-06,
+      "loss": 1.1876,
+      "step": 2136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.764780902631098e-06,
+      "loss": 1.1676,
+      "step": 2137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.7613223039851595e-06,
+      "loss": 1.1822,
+      "step": 2138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.757862742927951e-06,
+      "loss": 1.1793,
+      "step": 2139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.754402221349825e-06,
+      "loss": 1.1369,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.750940741141658e-06,
+      "loss": 1.1546,
+      "step": 2141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.747478304194854e-06,
+      "loss": 1.1704,
+      "step": 2142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.744014912401334e-06,
+      "loss": 1.1814,
+      "step": 2143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.740550567653546e-06,
+      "loss": 1.1689,
+      "step": 2144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.737085271844458e-06,
+      "loss": 1.1863,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.7336190268675536e-06,
+      "loss": 1.1759,
+      "step": 2146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.730151834616838e-06,
+      "loss": 1.1824,
+      "step": 2147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.726683696986836e-06,
+      "loss": 1.1874,
+      "step": 2148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.723214615872585e-06,
+      "loss": 1.1526,
+      "step": 2149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.719744593169642e-06,
+      "loss": 1.1656,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.716273630774072e-06,
+      "loss": 1.1796,
+      "step": 2151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.712801730582461e-06,
+      "loss": 1.1568,
+      "step": 2152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.709328894491902e-06,
+      "loss": 1.2031,
+      "step": 2153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.705855124400004e-06,
+      "loss": 1.1854,
+      "step": 2154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.702380422204882e-06,
+      "loss": 1.2032,
+      "step": 2155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.698904789805163e-06,
+      "loss": 1.1614,
+      "step": 2156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.69542822909998e-06,
+      "loss": 1.166,
+      "step": 2157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.691950741988975e-06,
+      "loss": 1.1774,
+      "step": 2158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.688472330372298e-06,
+      "loss": 1.1361,
+      "step": 2159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.684992996150599e-06,
+      "loss": 1.1874,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.681512741225038e-06,
+      "loss": 1.1552,
+      "step": 2161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.678031567497272e-06,
+      "loss": 1.168,
+      "step": 2162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6745494768694655e-06,
+      "loss": 1.1796,
+      "step": 2163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.671066471244282e-06,
+      "loss": 1.1599,
+      "step": 2164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.667582552524884e-06,
+      "loss": 1.2083,
+      "step": 2165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.664097722614934e-06,
+      "loss": 1.1604,
+      "step": 2166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.660611983418591e-06,
+      "loss": 1.1529,
+      "step": 2167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.657125336840513e-06,
+      "loss": 1.1701,
+      "step": 2168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.653637784785853e-06,
+      "loss": 1.1687,
+      "step": 2169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.650149329160258e-06,
+      "loss": 1.1719,
+      "step": 2170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.646659971869869e-06,
+      "loss": 1.1794,
+      "step": 2171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.643169714821319e-06,
+      "loss": 1.1308,
+      "step": 2172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.639678559921735e-06,
+      "loss": 1.1747,
+      "step": 2173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.636186509078733e-06,
+      "loss": 1.1384,
+      "step": 2174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6326935642004165e-06,
+      "loss": 1.1449,
+      "step": 2175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.629199727195384e-06,
+      "loss": 1.1666,
+      "step": 2176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.625704999972713e-06,
+      "loss": 1.1565,
+      "step": 2177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.622209384441974e-06,
+      "loss": 1.1974,
+      "step": 2178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6187128825132196e-06,
+      "loss": 1.1341,
+      "step": 2179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.615215496096987e-06,
+      "loss": 1.1624,
+      "step": 2180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.611717227104298e-06,
+      "loss": 1.1946,
+      "step": 2181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.608218077446657e-06,
+      "loss": 1.1844,
+      "step": 2182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.604718049036047e-06,
+      "loss": 1.211,
+      "step": 2183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.601217143784935e-06,
+      "loss": 1.1662,
+      "step": 2184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.597715363606261e-06,
+      "loss": 1.1752,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.59421271041345e-06,
+      "loss": 1.1754,
+      "step": 2186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5907091861204e-06,
+      "loss": 1.151,
+      "step": 2187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.587204792641486e-06,
+      "loss": 1.1861,
+      "step": 2188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5836995318915575e-06,
+      "loss": 1.139,
+      "step": 2189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.580193405785939e-06,
+      "loss": 1.1789,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.576686416240425e-06,
+      "loss": 1.1301,
+      "step": 2191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.573178565171288e-06,
+      "loss": 1.1435,
+      "step": 2192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5696698544952645e-06,
+      "loss": 1.1843,
+      "step": 2193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.566160286129561e-06,
+      "loss": 1.2141,
+      "step": 2194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.56264986199186e-06,
+      "loss": 1.1934,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.559138584000304e-06,
+      "loss": 1.1585,
+      "step": 2196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.555626454073504e-06,
+      "loss": 1.1636,
+      "step": 2197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.552113474130538e-06,
+      "loss": 1.1559,
+      "step": 2198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.548599646090948e-06,
+      "loss": 1.1612,
+      "step": 2199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 1.1822,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5415694534023765e-06,
+      "loss": 1.1659,
+      "step": 2201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.538053092594791e-06,
+      "loss": 1.1798,
+      "step": 2202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5345358913733735e-06,
+      "loss": 1.1865,
+      "step": 2203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5310178516599686e-06,
+      "loss": 1.1863,
+      "step": 2204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.527498975376882e-06,
+      "loss": 1.196,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.523979264446883e-06,
+      "loss": 1.1482,
+      "step": 2206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.520458720793186e-06,
+      "loss": 1.1575,
+      "step": 2207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.516937346339466e-06,
+      "loss": 1.1577,
+      "step": 2208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.513415143009853e-06,
+      "loss": 1.1543,
+      "step": 2209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.509892112728928e-06,
+      "loss": 1.1823,
+      "step": 2210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.506368257421724e-06,
+      "loss": 1.1501,
+      "step": 2211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.502843579013725e-06,
+      "loss": 1.1779,
+      "step": 2212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.499318079430865e-06,
+      "loss": 1.1427,
+      "step": 2213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.495791760599527e-06,
+      "loss": 1.1817,
+      "step": 2214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.492264624446541e-06,
+      "loss": 1.1507,
+      "step": 2215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.488736672899183e-06,
+      "loss": 1.1608,
+      "step": 2216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.485207907885175e-06,
+      "loss": 1.1684,
+      "step": 2217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.481678331332686e-06,
+      "loss": 1.148,
+      "step": 2218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.478147945170324e-06,
+      "loss": 1.1603,
+      "step": 2219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.474616751327143e-06,
+      "loss": 1.1345,
+      "step": 2220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.471084751732637e-06,
+      "loss": 1.1728,
+      "step": 2221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.467551948316739e-06,
+      "loss": 1.1768,
+      "step": 2222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.464018343009824e-06,
+      "loss": 1.1104,
+      "step": 2223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.460483937742703e-06,
+      "loss": 1.1615,
+      "step": 2224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.456948734446624e-06,
+      "loss": 1.1456,
+      "step": 2225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4534127350532725e-06,
+      "loss": 1.18,
+      "step": 2226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.449875941494769e-06,
+      "loss": 1.1817,
+      "step": 2227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.446338355703665e-06,
+      "loss": 1.1485,
+      "step": 2228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.442799979612949e-06,
+      "loss": 1.1833,
+      "step": 2229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.439260815156039e-06,
+      "loss": 1.1725,
+      "step": 2230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4357208642667836e-06,
+      "loss": 1.1413,
+      "step": 2231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.432180128879462e-06,
+      "loss": 1.1358,
+      "step": 2232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.428638610928783e-06,
+      "loss": 1.1281,
+      "step": 2233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.425096312349881e-06,
+      "loss": 1.1529,
+      "step": 2234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.421553235078317e-06,
+      "loss": 1.1308,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4180093810500785e-06,
+      "loss": 1.1546,
+      "step": 2236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.414464752201579e-06,
+      "loss": 1.1323,
+      "step": 2237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.410919350469651e-06,
+      "loss": 1.1308,
+      "step": 2238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.407373177791553e-06,
+      "loss": 1.1416,
+      "step": 2239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4038262361049655e-06,
+      "loss": 1.1591,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.400278527347983e-06,
+      "loss": 1.1416,
+      "step": 2241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.396730053459127e-06,
+      "loss": 1.1821,
+      "step": 2242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.393180816377331e-06,
+      "loss": 1.1621,
+      "step": 2243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3896308180419505e-06,
+      "loss": 1.1456,
+      "step": 2244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.386080060392755e-06,
+      "loss": 1.1501,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.382528545369924e-06,
+      "loss": 1.1728,
+      "step": 2246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.378976274914058e-06,
+      "loss": 1.1464,
+      "step": 2247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.375423250966168e-06,
+      "loss": 1.173,
+      "step": 2248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.371869475467677e-06,
+      "loss": 1.165,
+      "step": 2249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.368314950360416e-06,
+      "loss": 1.1171,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.364759677586627e-06,
+      "loss": 1.1567,
+      "step": 2251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.361203659088964e-06,
+      "loss": 1.1714,
+      "step": 2252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.357646896810481e-06,
+      "loss": 1.1778,
+      "step": 2253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.354089392694648e-06,
+      "loss": 1.1426,
+      "step": 2254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.350531148685333e-06,
+      "loss": 1.1555,
+      "step": 2255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.346972166726809e-06,
+      "loss": 1.1718,
+      "step": 2256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.343412448763755e-06,
+      "loss": 1.1157,
+      "step": 2257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.339851996741253e-06,
+      "loss": 1.1457,
+      "step": 2258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.336290812604781e-06,
+      "loss": 1.1556,
+      "step": 2259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.3327288983002215e-06,
+      "loss": 1.1464,
+      "step": 2260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.329166255773854e-06,
+      "loss": 1.1259,
+      "step": 2261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.325602886972356e-06,
+      "loss": 1.1364,
+      "step": 2262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.322038793842804e-06,
+      "loss": 1.1425,
+      "step": 2263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.318473978332669e-06,
+      "loss": 1.1404,
+      "step": 2264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.314908442389813e-06,
+      "loss": 1.1618,
+      "step": 2265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.311342187962498e-06,
+      "loss": 1.1872,
+      "step": 2266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.307775216999374e-06,
+      "loss": 1.1656,
+      "step": 2267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.304207531449486e-06,
+      "loss": 1.1298,
+      "step": 2268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.300639133262265e-06,
+      "loss": 1.1422,
+      "step": 2269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.297070024387535e-06,
+      "loss": 1.1564,
+      "step": 2270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.293500206775507e-06,
+      "loss": 1.1221,
+      "step": 2271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.289929682376781e-06,
+      "loss": 1.1663,
+      "step": 2272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.286358453142339e-06,
+      "loss": 1.1553,
+      "step": 2273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.282786521023554e-06,
+      "loss": 1.1725,
+      "step": 2274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.279213887972179e-06,
+      "loss": 1.1668,
+      "step": 2275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.275640555940349e-06,
+      "loss": 1.168,
+      "step": 2276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.272066526880584e-06,
+      "loss": 1.1524,
+      "step": 2277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.268491802745785e-06,
+      "loss": 1.1581,
+      "step": 2278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2649163854892305e-06,
+      "loss": 1.1446,
+      "step": 2279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.261340277064578e-06,
+      "loss": 1.159,
+      "step": 2280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.257763479425866e-06,
+      "loss": 1.1379,
+      "step": 2281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.254185994527503e-06,
+      "loss": 1.1521,
+      "step": 2282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.250607824324279e-06,
+      "loss": 1.1737,
+      "step": 2283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.247028970771359e-06,
+      "loss": 1.1505,
+      "step": 2284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.243449435824276e-06,
+      "loss": 1.1718,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.239869221438938e-06,
+      "loss": 1.1691,
+      "step": 2286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.236288329571625e-06,
+      "loss": 1.1332,
+      "step": 2287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.232706762178986e-06,
+      "loss": 1.1564,
+      "step": 2288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.229124521218043e-06,
+      "loss": 1.1564,
+      "step": 2289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2255416086461795e-06,
+      "loss": 1.1516,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.221958026421151e-06,
+      "loss": 1.1615,
+      "step": 2291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.218373776501077e-06,
+      "loss": 1.1311,
+      "step": 2292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.214788860844439e-06,
+      "loss": 1.1808,
+      "step": 2293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.211203281410091e-06,
+      "loss": 1.1494,
+      "step": 2294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.20761704015724e-06,
+      "loss": 1.1221,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.204030139045459e-06,
+      "loss": 1.105,
+      "step": 2296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.200442580034683e-06,
+      "loss": 1.1703,
+      "step": 2297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.196854365085203e-06,
+      "loss": 1.1641,
+      "step": 2298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.193265496157671e-06,
+      "loss": 1.122,
+      "step": 2299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.189675975213094e-06,
+      "loss": 1.1639,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.186085804212839e-06,
+      "loss": 1.1584,
+      "step": 2301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.182494985118625e-06,
+      "loss": 1.1799,
+      "step": 2302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.178903519892524e-06,
+      "loss": 1.1753,
+      "step": 2303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.175311410496966e-06,
+      "loss": 1.1319,
+      "step": 2304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.171718658894729e-06,
+      "loss": 1.1557,
+      "step": 2305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.168125267048942e-06,
+      "loss": 1.1278,
+      "step": 2306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.164531236923086e-06,
+      "loss": 1.1522,
+      "step": 2307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1609365704809886e-06,
+      "loss": 1.1262,
+      "step": 2308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.157341269686828e-06,
+      "loss": 1.1669,
+      "step": 2309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.153745336505125e-06,
+      "loss": 1.1383,
+      "step": 2310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.150148772900749e-06,
+      "loss": 1.1885,
+      "step": 2311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.146551580838913e-06,
+      "loss": 1.1251,
+      "step": 2312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.142953762285174e-06,
+      "loss": 1.1357,
+      "step": 2313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.139355319205429e-06,
+      "loss": 1.1796,
+      "step": 2314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.135756253565919e-06,
+      "loss": 1.168,
+      "step": 2315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.132156567333224e-06,
+      "loss": 1.1137,
+      "step": 2316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.128556262474262e-06,
+      "loss": 1.1487,
+      "step": 2317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.124955340956293e-06,
+      "loss": 1.1877,
+      "step": 2318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.121353804746907e-06,
+      "loss": 1.1293,
+      "step": 2319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.117751655814037e-06,
+      "loss": 1.1431,
+      "step": 2320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.114148896125946e-06,
+      "loss": 1.0975,
+      "step": 2321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.110545527651236e-06,
+      "loss": 1.1719,
+      "step": 2322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.106941552358835e-06,
+      "loss": 1.1532,
+      "step": 2323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.103336972218007e-06,
+      "loss": 1.1516,
+      "step": 2324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.099731789198344e-06,
+      "loss": 1.1246,
+      "step": 2325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.096126005269769e-06,
+      "loss": 1.1411,
+      "step": 2326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.092519622402535e-06,
+      "loss": 1.1551,
+      "step": 2327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.088912642567218e-06,
+      "loss": 1.1499,
+      "step": 2328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.085305067734725e-06,
+      "loss": 1.1583,
+      "step": 2329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.081696899876282e-06,
+      "loss": 1.1268,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.078088140963446e-06,
+      "loss": 1.1478,
+      "step": 2331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.074478792968092e-06,
+      "loss": 1.1398,
+      "step": 2332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.070868857862419e-06,
+      "loss": 1.1211,
+      "step": 2333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.067258337618945e-06,
+      "loss": 1.128,
+      "step": 2334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.063647234210511e-06,
+      "loss": 1.1302,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.060035549610275e-06,
+      "loss": 1.1734,
+      "step": 2336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.056423285791708e-06,
+      "loss": 1.1605,
+      "step": 2337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.052810444728608e-06,
+      "loss": 1.2069,
+      "step": 2338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.049197028395076e-06,
+      "loss": 1.1616,
+      "step": 2339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.045583038765538e-06,
+      "loss": 1.1314,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.041968477814727e-06,
+      "loss": 1.1775,
+      "step": 2341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.03835334751769e-06,
+      "loss": 1.1384,
+      "step": 2342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.034737649849785e-06,
+      "loss": 1.1537,
+      "step": 2343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.031121386786679e-06,
+      "loss": 1.1281,
+      "step": 2344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0275045603043516e-06,
+      "loss": 1.1194,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.023887172379085e-06,
+      "loss": 1.1618,
+      "step": 2346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.020269224987473e-06,
+      "loss": 1.1558,
+      "step": 2347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.016650720106411e-06,
+      "loss": 1.1393,
+      "step": 2348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.013031659713102e-06,
+      "loss": 1.1581,
+      "step": 2349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.009412045785051e-06,
+      "loss": 1.1271,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.005791880300067e-06,
+      "loss": 1.1532,
+      "step": 2351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.002171165236259e-06,
+      "loss": 1.1377,
+      "step": 2352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9985499025720354e-06,
+      "loss": 1.1453,
+      "step": 2353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.994928094286106e-06,
+      "loss": 1.1436,
+      "step": 2354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.991305742357478e-06,
+      "loss": 1.1528,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.987682848765455e-06,
+      "loss": 1.1704,
+      "step": 2356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.984059415489636e-06,
+      "loss": 1.1408,
+      "step": 2357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.980435444509918e-06,
+      "loss": 1.1131,
+      "step": 2358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9768109378064875e-06,
+      "loss": 1.1504,
+      "step": 2359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.973185897359828e-06,
+      "loss": 1.1516,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.969560325150712e-06,
+      "loss": 1.1057,
+      "step": 2361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9659342231602e-06,
+      "loss": 1.1647,
+      "step": 2362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.96230759336965e-06,
+      "loss": 1.1359,
+      "step": 2363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.958680437760701e-06,
+      "loss": 1.1514,
+      "step": 2364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.955052758315282e-06,
+      "loss": 1.1981,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.951424557015608e-06,
+      "loss": 1.1108,
+      "step": 2366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9477958358441804e-06,
+      "loss": 1.2001,
+      "step": 2367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.944166596783782e-06,
+      "loss": 1.1369,
+      "step": 2368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.94053684181748e-06,
+      "loss": 1.1165,
+      "step": 2369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.936906572928625e-06,
+      "loss": 1.1419,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.933275792100844e-06,
+      "loss": 1.165,
+      "step": 2371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.92964450131805e-06,
+      "loss": 1.1443,
+      "step": 2372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.926012702564431e-06,
+      "loss": 1.1463,
+      "step": 2373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.922380397824451e-06,
+      "loss": 1.133,
+      "step": 2374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.918747589082853e-06,
+      "loss": 1.1684,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.915114278324654e-06,
+      "loss": 1.1024,
+      "step": 2376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.911480467535149e-06,
+      "loss": 1.1423,
+      "step": 2377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.907846158699901e-06,
+      "loss": 1.1206,
+      "step": 2378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.904211353804748e-06,
+      "loss": 1.1313,
+      "step": 2379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.900576054835798e-06,
+      "loss": 1.1142,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.896940263779431e-06,
+      "loss": 1.1527,
+      "step": 2381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.893303982622292e-06,
+      "loss": 1.1486,
+      "step": 2382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.889667213351298e-06,
+      "loss": 1.1508,
+      "step": 2383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.88602995795363e-06,
+      "loss": 1.1619,
+      "step": 2384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.882392218416736e-06,
+      "loss": 1.1444,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.878753996728328e-06,
+      "loss": 1.1654,
+      "step": 2386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8751152948763815e-06,
+      "loss": 1.1467,
+      "step": 2387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.871476114849133e-06,
+      "loss": 1.1385,
+      "step": 2388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.867836458635083e-06,
+      "loss": 1.1429,
+      "step": 2389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.864196328222991e-06,
+      "loss": 1.1243,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.860555725601875e-06,
+      "loss": 1.1245,
+      "step": 2391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.856914652761011e-06,
+      "loss": 1.1792,
+      "step": 2392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.853273111689934e-06,
+      "loss": 1.101,
+      "step": 2393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.849631104378431e-06,
+      "loss": 1.1679,
+      "step": 2394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.845988632816547e-06,
+      "loss": 1.1326,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8423456989945805e-06,
+      "loss": 1.1446,
+      "step": 2396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.83870230490308e-06,
+      "loss": 1.1203,
+      "step": 2397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.835058452532848e-06,
+      "loss": 1.1418,
+      "step": 2398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.831414143874938e-06,
+      "loss": 1.1373,
+      "step": 2399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.82776938092065e-06,
+      "loss": 1.1367,
+      "step": 2400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.8241241656615354e-06,
+      "loss": 1.1387,
+      "step": 2401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.820478500089388e-06,
+      "loss": 1.1268,
+      "step": 2402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.816832386196254e-06,
+      "loss": 1.1409,
+      "step": 2403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.813185825974419e-06,
+      "loss": 1.1476,
+      "step": 2404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.809538821416416e-06,
+      "loss": 1.1251,
+      "step": 2405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.805891374515019e-06,
+      "loss": 1.1432,
+      "step": 2406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.8022434872632434e-06,
+      "loss": 1.1363,
+      "step": 2407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.798595161654348e-06,
+      "loss": 1.1029,
+      "step": 2408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.794946399681829e-06,
+      "loss": 1.1422,
+      "step": 2409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.79129720333942e-06,
+      "loss": 1.145,
+      "step": 2410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.787647574621094e-06,
+      "loss": 1.1116,
+      "step": 2411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7839975155210595e-06,
+      "loss": 1.1209,
+      "step": 2412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.78034702803376e-06,
+      "loss": 1.1474,
+      "step": 2413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.776696114153871e-06,
+      "loss": 1.1222,
+      "step": 2414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.773044775876306e-06,
+      "loss": 1.1148,
+      "step": 2415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.769393015196208e-06,
+      "loss": 1.1496,
+      "step": 2416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765740834108948e-06,
+      "loss": 1.1386,
+      "step": 2417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76208823461013e-06,
+      "loss": 1.1458,
+      "step": 2418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.758435218695587e-06,
+      "loss": 1.172,
+      "step": 2419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.754781788361376e-06,
+      "loss": 1.1019,
+      "step": 2420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.751127945603786e-06,
+      "loss": 1.1121,
+      "step": 2421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.747473692419324e-06,
+      "loss": 1.18,
+      "step": 2422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7438190308047285e-06,
+      "loss": 1.1216,
+      "step": 2423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.740163962756956e-06,
+      "loss": 1.1328,
+      "step": 2424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.736508490273189e-06,
+      "loss": 1.1426,
+      "step": 2425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.732852615350827e-06,
+      "loss": 1.1347,
+      "step": 2426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.729196339987492e-06,
+      "loss": 1.1437,
+      "step": 2427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.725539666181025e-06,
+      "loss": 1.1511,
+      "step": 2428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.721882595929483e-06,
+      "loss": 1.1869,
+      "step": 2429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.718225131231142e-06,
+      "loss": 1.1428,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7145672740844915e-06,
+      "loss": 1.1149,
+      "step": 2431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.710909026488235e-06,
+      "loss": 1.1514,
+      "step": 2432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7072503904412925e-06,
+      "loss": 1.1301,
+      "step": 2433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.703591367942791e-06,
+      "loss": 1.1266,
+      "step": 2434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.699931960992076e-06,
+      "loss": 1.1243,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.696272171588695e-06,
+      "loss": 1.1283,
+      "step": 2436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.69261200173241e-06,
+      "loss": 1.1659,
+      "step": 2437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.68895145342319e-06,
+      "loss": 1.1383,
+      "step": 2438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.68529052866121e-06,
+      "loss": 1.1199,
+      "step": 2439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.68162922944685e-06,
+      "loss": 1.149,
+      "step": 2440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6779675577806955e-06,
+      "loss": 1.1499,
+      "step": 2441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.674305515663537e-06,
+      "loss": 1.1261,
+      "step": 2442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.670643105096361e-06,
+      "loss": 1.1333,
+      "step": 2443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.666980328080367e-06,
+      "loss": 1.1473,
+      "step": 2444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.663317186616942e-06,
+      "loss": 1.115,
+      "step": 2445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.659653682707681e-06,
+      "loss": 1.1353,
+      "step": 2446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.655989818354373e-06,
+      "loss": 1.1067,
+      "step": 2447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.652325595559005e-06,
+      "loss": 1.1357,
+      "step": 2448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.648661016323761e-06,
+      "loss": 1.1376,
+      "step": 2449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.644996082651018e-06,
+      "loss": 1.1117,
+      "step": 2450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.641330796543346e-06,
+      "loss": 1.1154,
+      "step": 2451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.63766516000351e-06,
+      "loss": 1.1346,
+      "step": 2452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.633999175034465e-06,
+      "loss": 1.153,
+      "step": 2453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.6303328436393565e-06,
+      "loss": 1.1357,
+      "step": 2454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.626666167821522e-06,
+      "loss": 1.1426,
+      "step": 2455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.622999149584482e-06,
+      "loss": 1.1533,
+      "step": 2456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.619331790931946e-06,
+      "loss": 1.1881,
+      "step": 2457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.6156640938678144e-06,
+      "loss": 1.137,
+      "step": 2458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.611996060396164e-06,
+      "loss": 1.1513,
+      "step": 2459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.608327692521263e-06,
+      "loss": 1.131,
+      "step": 2460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.604658992247557e-06,
+      "loss": 1.1311,
+      "step": 2461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.600989961579677e-06,
+      "loss": 1.1306,
+      "step": 2462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.597320602522433e-06,
+      "loss": 1.1459,
+      "step": 2463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.593650917080812e-06,
+      "loss": 1.114,
+      "step": 2464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.589980907259983e-06,
+      "loss": 1.1222,
+      "step": 2465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.586310575065293e-06,
+      "loss": 1.1428,
+      "step": 2466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.582639922502261e-06,
+      "loss": 1.105,
+      "step": 2467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.578968951576583e-06,
+      "loss": 1.1111,
+      "step": 2468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.57529766429413e-06,
+      "loss": 1.1527,
+      "step": 2469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.571626062660947e-06,
+      "loss": 1.1461,
+      "step": 2470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.567954148683242e-06,
+      "loss": 1.1478,
+      "step": 2471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5642819243674085e-06,
+      "loss": 1.1267,
+      "step": 2472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.560609391719996e-06,
+      "loss": 1.1819,
+      "step": 2473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5569365527477325e-06,
+      "loss": 1.1371,
+      "step": 2474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.553263409457504e-06,
+      "loss": 1.1234,
+      "step": 2475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.549589963856371e-06,
+      "loss": 1.1292,
+      "step": 2476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.545916217951555e-06,
+      "loss": 1.1334,
+      "step": 2477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.542242173750442e-06,
+      "loss": 1.1474,
+      "step": 2478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.538567833260581e-06,
+      "loss": 1.1075,
+      "step": 2479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5348931984896836e-06,
+      "loss": 1.1337,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.531218271445623e-06,
+      "loss": 1.1665,
+      "step": 2481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.52754305413643e-06,
+      "loss": 1.1224,
+      "step": 2482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.523867548570294e-06,
+      "loss": 1.1056,
+      "step": 2483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5201917567555655e-06,
+      "loss": 1.1115,
+      "step": 2484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.516515680700747e-06,
+      "loss": 1.1192,
+      "step": 2485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.512839322414498e-06,
+      "loss": 1.0815,
+      "step": 2486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.509162683905633e-06,
+      "loss": 1.1263,
+      "step": 2487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.50548576718312e-06,
+      "loss": 1.141,
+      "step": 2488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5018085742560745e-06,
+      "loss": 1.1219,
+      "step": 2489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.4981311071337704e-06,
+      "loss": 1.136,
+      "step": 2490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.494453367825626e-06,
+      "loss": 1.1467,
+      "step": 2491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.49077535834121e-06,
+      "loss": 1.1209,
+      "step": 2492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.487097080690238e-06,
+      "loss": 1.1406,
+      "step": 2493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.483418536882572e-06,
+      "loss": 1.0988,
+      "step": 2494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.47973972892822e-06,
+      "loss": 1.1119,
+      "step": 2495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.476060658837336e-06,
+      "loss": 1.1427,
+      "step": 2496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.472381328620216e-06,
+      "loss": 1.1216,
+      "step": 2497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.468701740287295e-06,
+      "loss": 1.099,
+      "step": 2498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4650218958491544e-06,
+      "loss": 1.1065,
+      "step": 2499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.46134179731651e-06,
+      "loss": 1.1514,
+      "step": 2500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.457661446700223e-06,
+      "loss": 1.1255,
+      "step": 2501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.453980846011287e-06,
+      "loss": 1.1223,
+      "step": 2502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.450299997260833e-06,
+      "loss": 1.1416,
+      "step": 2503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.44661890246013e-06,
+      "loss": 1.1347,
+      "step": 2504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4429375636205795e-06,
+      "loss": 1.1181,
+      "step": 2505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.439255982753717e-06,
+      "loss": 1.1374,
+      "step": 2506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.435574161871209e-06,
+      "loss": 1.153,
+      "step": 2507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4318921029848545e-06,
+      "loss": 1.1347,
+      "step": 2508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.428209808106584e-06,
+      "loss": 1.1336,
+      "step": 2509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.424527279248452e-06,
+      "loss": 1.1124,
+      "step": 2510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.420844518422649e-06,
+      "loss": 1.1363,
+      "step": 2511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.417161527641482e-06,
+      "loss": 1.1045,
+      "step": 2512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.413478308917391e-06,
+      "loss": 1.1605,
+      "step": 2513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4097948642629405e-06,
+      "loss": 1.1204,
+      "step": 2514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.406111195690813e-06,
+      "loss": 1.1158,
+      "step": 2515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4024273052138195e-06,
+      "loss": 1.0985,
+      "step": 2516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.398743194844888e-06,
+      "loss": 1.1377,
+      "step": 2517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.395058866597068e-06,
+      "loss": 1.1386,
+      "step": 2518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.39137432248353e-06,
+      "loss": 1.0953,
+      "step": 2519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.387689564517558e-06,
+      "loss": 1.1345,
+      "step": 2520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.38400459471256e-06,
+      "loss": 1.1521,
+      "step": 2521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.380319415082049e-06,
+      "loss": 1.1343,
+      "step": 2522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.376634027639664e-06,
+      "loss": 1.107,
+      "step": 2523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3729484343991516e-06,
+      "loss": 1.1308,
+      "step": 2524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.36926263737437e-06,
+      "loss": 1.1246,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.365576638579292e-06,
+      "loss": 1.1109,
+      "step": 2526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.361890440027999e-06,
+      "loss": 1.1305,
+      "step": 2527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.358204043734682e-06,
+      "loss": 1.0965,
+      "step": 2528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.354517451713639e-06,
+      "loss": 1.1097,
+      "step": 2529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.350830665979276e-06,
+      "loss": 1.115,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.347143688546105e-06,
+      "loss": 1.1502,
+      "step": 2531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.343456521428742e-06,
+      "loss": 1.1328,
+      "step": 2532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.339769166641906e-06,
+      "loss": 1.147,
+      "step": 2533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.33608162620042e-06,
+      "loss": 1.1228,
+      "step": 2534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.33239390211921e-06,
+      "loss": 1.1029,
+      "step": 2535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.328705996413297e-06,
+      "loss": 1.144,
+      "step": 2536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.325017911097806e-06,
+      "loss": 1.0941,
+      "step": 2537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.321329648187959e-06,
+      "loss": 1.1216,
+      "step": 2538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3176412096990745e-06,
+      "loss": 1.1074,
+      "step": 2539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3139525976465675e-06,
+      "loss": 1.1225,
+      "step": 2540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.310263814045947e-06,
+      "loss": 1.115,
+      "step": 2541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.306574860912816e-06,
+      "loss": 1.1156,
+      "step": 2542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.302885740262872e-06,
+      "loss": 1.0773,
+      "step": 2543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.299196454111899e-06,
+      "loss": 1.1497,
+      "step": 2544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2955070044757775e-06,
+      "loss": 1.1126,
+      "step": 2545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.291817393370474e-06,
+      "loss": 1.1295,
+      "step": 2546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.288127622812044e-06,
+      "loss": 1.1003,
+      "step": 2547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.284437694816631e-06,
+      "loss": 1.1303,
+      "step": 2548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.280747611400462e-06,
+      "loss": 1.1154,
+      "step": 2549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.27705737457985e-06,
+      "loss": 1.1378,
+      "step": 2550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.273366986371195e-06,
+      "loss": 1.1331,
+      "step": 2551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.269676448790974e-06,
+      "loss": 1.0961,
+      "step": 2552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.265985763855749e-06,
+      "loss": 1.0668,
+      "step": 2553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.262294933582164e-06,
+      "loss": 1.1242,
+      "step": 2554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.258603959986939e-06,
+      "loss": 1.1493,
+      "step": 2555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2549128450868744e-06,
+      "loss": 1.1179,
+      "step": 2556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.251221590898848e-06,
+      "loss": 1.1337,
+      "step": 2557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.247530199439812e-06,
+      "loss": 1.1016,
+      "step": 2558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.243838672726795e-06,
+      "loss": 1.0838,
+      "step": 2559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.240147012776899e-06,
+      "loss": 1.1332,
+      "step": 2560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.236455221607299e-06,
+      "loss": 1.1144,
+      "step": 2561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2327633012352395e-06,
+      "loss": 1.1215,
+      "step": 2562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.229071253678042e-06,
+      "loss": 1.1031,
+      "step": 2563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.225379080953089e-06,
+      "loss": 1.1308,
+      "step": 2564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.221686785077838e-06,
+      "loss": 1.0948,
+      "step": 2565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.217994368069812e-06,
+      "loss": 1.1165,
+      "step": 2566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.214301831946598e-06,
+      "loss": 1.1075,
+      "step": 2567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.210609178725849e-06,
+      "loss": 1.1197,
+      "step": 2568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.206916410425284e-06,
+      "loss": 1.115,
+      "step": 2569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.203223529062684e-06,
+      "loss": 1.1269,
+      "step": 2570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.19953053665589e-06,
+      "loss": 1.1354,
+      "step": 2571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.195837435222806e-06,
+      "loss": 1.1137,
+      "step": 2572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.192144226781394e-06,
+      "loss": 1.1266,
+      "step": 2573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.188450913349674e-06,
+      "loss": 1.1198,
+      "step": 2574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.184757496945726e-06,
+      "loss": 1.0971,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1810639795876835e-06,
+      "loss": 1.1508,
+      "step": 2576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.177370363293737e-06,
+      "loss": 1.1245,
+      "step": 2577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.173676650082131e-06,
+      "loss": 1.0953,
+      "step": 2578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.169982841971159e-06,
+      "loss": 1.1252,
+      "step": 2579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.166288940979175e-06,
+      "loss": 1.1028,
+      "step": 2580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.162594949124575e-06,
+      "loss": 1.1486,
+      "step": 2581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1589008684258094e-06,
+      "loss": 1.0861,
+      "step": 2582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.155206700901374e-06,
+      "loss": 1.1096,
+      "step": 2583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.151512448569815e-06,
+      "loss": 1.0808,
+      "step": 2584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.147818113449725e-06,
+      "loss": 1.1174,
+      "step": 2585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.144123697559738e-06,
+      "loss": 1.1514,
+      "step": 2586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.140429202918538e-06,
+      "loss": 1.0915,
+      "step": 2587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.136734631544847e-06,
+      "loss": 1.1131,
+      "step": 2588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.133039985457431e-06,
+      "loss": 1.0919,
+      "step": 2589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.129345266675095e-06,
+      "loss": 1.1017,
+      "step": 2590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.1256504772166885e-06,
+      "loss": 1.1227,
+      "step": 2591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.121955619101094e-06,
+      "loss": 1.12,
+      "step": 2592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.118260694347233e-06,
+      "loss": 1.1323,
+      "step": 2593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.114565704974069e-06,
+      "loss": 1.1326,
+      "step": 2594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.11087065300059e-06,
+      "loss": 1.0931,
+      "step": 2595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.107175540445829e-06,
+      "loss": 1.1165,
+      "step": 2596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.103480369328846e-06,
+      "loss": 1.111,
+      "step": 2597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0997851416687326e-06,
+      "loss": 1.1175,
+      "step": 2598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.096089859484616e-06,
+      "loss": 1.0916,
+      "step": 2599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.09239452479565e-06,
+      "loss": 1.1407,
+      "step": 2600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.088699139621016e-06,
+      "loss": 1.1088,
+      "step": 2601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0850037059799254e-06,
+      "loss": 1.0831,
+      "step": 2602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.081308225891614e-06,
+      "loss": 1.1094,
+      "step": 2603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.077612701375345e-06,
+      "loss": 1.1117,
+      "step": 2604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.073917134450406e-06,
+      "loss": 1.1388,
+      "step": 2605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.070221527136103e-06,
+      "loss": 1.0926,
+      "step": 2606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0665258814517725e-06,
+      "loss": 1.1101,
+      "step": 2607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.062830199416764e-06,
+      "loss": 1.0952,
+      "step": 2608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.059134483050452e-06,
+      "loss": 1.1063,
+      "step": 2609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.055438734372225e-06,
+      "loss": 1.1343,
+      "step": 2610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0517429554014965e-06,
+      "loss": 1.1253,
+      "step": 2611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.048047148157687e-06,
+      "loss": 1.1405,
+      "step": 2612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.044351314660243e-06,
+      "loss": 1.0907,
+      "step": 2613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.040655456928616e-06,
+      "loss": 1.1085,
+      "step": 2614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.036959576982276e-06,
+      "loss": 1.1092,
+      "step": 2615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.033263676840703e-06,
+      "loss": 1.1151,
+      "step": 2616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0295677585233895e-06,
+      "loss": 1.1408,
+      "step": 2617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.025871824049838e-06,
+      "loss": 1.1006,
+      "step": 2618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.022175875439556e-06,
+      "loss": 1.1354,
+      "step": 2619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.018479914712065e-06,
+      "loss": 1.1335,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.01478394388689e-06,
+      "loss": 1.132,
+      "step": 2621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.011087964983559e-06,
+      "loss": 1.1033,
+      "step": 2622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.007391980021608e-06,
+      "loss": 1.1437,
+      "step": 2623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.003695991020575e-06,
+      "loss": 1.1354,
+      "step": 2624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 1.0998,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.996304008979426e-06,
+      "loss": 1.1326,
+      "step": 2626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.992608019978394e-06,
+      "loss": 1.1255,
+      "step": 2627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.988912035016442e-06,
+      "loss": 1.088,
+      "step": 2628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.985216056113113e-06,
+      "loss": 1.147,
+      "step": 2629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.981520085287936e-06,
+      "loss": 1.1126,
+      "step": 2630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.977824124560445e-06,
+      "loss": 1.1176,
+      "step": 2631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.974128175950163e-06,
+      "loss": 1.0789,
+      "step": 2632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.970432241476613e-06,
+      "loss": 1.1295,
+      "step": 2633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.966736323159298e-06,
+      "loss": 1.1212,
+      "step": 2634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.963040423017726e-06,
+      "loss": 1.1137,
+      "step": 2635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.959344543071386e-06,
+      "loss": 1.0982,
+      "step": 2636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.955648685339758e-06,
+      "loss": 1.0849,
+      "step": 2637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.951952851842314e-06,
+      "loss": 1.0792,
+      "step": 2638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.948257044598506e-06,
+      "loss": 1.1156,
+      "step": 2639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.944561265627776e-06,
+      "loss": 1.1058,
+      "step": 2640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9408655169495505e-06,
+      "loss": 1.1129,
+      "step": 2641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.937169800583237e-06,
+      "loss": 1.1122,
+      "step": 2642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.93347411854823e-06,
+      "loss": 1.1191,
+      "step": 2643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.929778472863898e-06,
+      "loss": 1.1032,
+      "step": 2644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.926082865549597e-06,
+      "loss": 1.135,
+      "step": 2645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.922387298624656e-06,
+      "loss": 1.1274,
+      "step": 2646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.918691774108386e-06,
+      "loss": 1.0925,
+      "step": 2647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914996294020076e-06,
+      "loss": 1.1229,
+      "step": 2648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911300860378984e-06,
+      "loss": 1.1145,
+      "step": 2649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907605475204352e-06,
+      "loss": 1.1216,
+      "step": 2650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.903910140515384e-06,
+      "loss": 1.0999,
+      "step": 2651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.900214858331268e-06,
+      "loss": 1.1095,
+      "step": 2652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.896519630671156e-06,
+      "loss": 1.1055,
+      "step": 2653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.892824459554173e-06,
+      "loss": 1.1091,
+      "step": 2654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.889129346999411e-06,
+      "loss": 1.0769,
+      "step": 2655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.885434295025934e-06,
+      "loss": 1.1065,
+      "step": 2656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.881739305652768e-06,
+      "loss": 1.0811,
+      "step": 2657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.878044380898909e-06,
+      "loss": 1.0967,
+      "step": 2658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.874349522783313e-06,
+      "loss": 1.1033,
+      "step": 2659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.870654733324907e-06,
+      "loss": 1.0794,
+      "step": 2660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.866960014542571e-06,
+      "loss": 1.1312,
+      "step": 2661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.863265368455153e-06,
+      "loss": 1.1166,
+      "step": 2662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.859570797081463e-06,
+      "loss": 1.1064,
+      "step": 2663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.855876302440262e-06,
+      "loss": 1.1282,
+      "step": 2664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.852181886550277e-06,
+      "loss": 1.1484,
+      "step": 2665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.848487551430186e-06,
+      "loss": 1.1217,
+      "step": 2666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.844793299098627e-06,
+      "loss": 1.1067,
+      "step": 2667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.841099131574192e-06,
+      "loss": 1.1091,
+      "step": 2668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.837405050875426e-06,
+      "loss": 1.1172,
+      "step": 2669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8337110590208255e-06,
+      "loss": 1.0671,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.830017158028842e-06,
+      "loss": 1.1028,
+      "step": 2671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.82632334991787e-06,
+      "loss": 1.0914,
+      "step": 2672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.822629636706265e-06,
+      "loss": 1.1162,
+      "step": 2673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.818936020412317e-06,
+      "loss": 1.1221,
+      "step": 2674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.815242503054277e-06,
+      "loss": 1.1035,
+      "step": 2675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.811549086650327e-06,
+      "loss": 1.1317,
+      "step": 2676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.80785577321861e-06,
+      "loss": 1.0881,
+      "step": 2677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.804162564777196e-06,
+      "loss": 1.0964,
+      "step": 2678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.800469463344112e-06,
+      "loss": 1.1113,
+      "step": 2679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.796776470937317e-06,
+      "loss": 1.0807,
+      "step": 2680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.793083589574715e-06,
+      "loss": 1.0742,
+      "step": 2681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.789390821274152e-06,
+      "loss": 1.1083,
+      "step": 2682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.785698168053403e-06,
+      "loss": 1.0813,
+      "step": 2683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7820056319301895e-06,
+      "loss": 1.1214,
+      "step": 2684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.778313214922161e-06,
+      "loss": 1.1113,
+      "step": 2685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.774620919046911e-06,
+      "loss": 1.0983,
+      "step": 2686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.770928746321959e-06,
+      "loss": 1.101,
+      "step": 2687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.767236698764762e-06,
+      "loss": 1.0721,
+      "step": 2688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.763544778392702e-06,
+      "loss": 1.0987,
+      "step": 2689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.759852987223104e-06,
+      "loss": 1.1057,
+      "step": 2690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.756161327273206e-06,
+      "loss": 1.1156,
+      "step": 2691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.75246980056019e-06,
+      "loss": 1.123,
+      "step": 2692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.748778409101153e-06,
+      "loss": 1.0941,
+      "step": 2693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.745087154913126e-06,
+      "loss": 1.102,
+      "step": 2694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.741396040013062e-06,
+      "loss": 1.0914,
+      "step": 2695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.737705066417838e-06,
+      "loss": 1.1241,
+      "step": 2696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7340142361442526e-06,
+      "loss": 1.1112,
+      "step": 2697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.730323551209027e-06,
+      "loss": 1.1097,
+      "step": 2698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.726633013628807e-06,
+      "loss": 1.1445,
+      "step": 2699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7229426254201504e-06,
+      "loss": 1.0899,
+      "step": 2700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71925238859954e-06,
+      "loss": 1.1046,
+      "step": 2701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71556230518337e-06,
+      "loss": 1.1047,
+      "step": 2702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.711872377187956e-06,
+      "loss": 1.0858,
+      "step": 2703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.708182606629526e-06,
+      "loss": 1.1258,
+      "step": 2704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.704492995524223e-06,
+      "loss": 1.0528,
+      "step": 2705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.700803545888102e-06,
+      "loss": 1.0895,
+      "step": 2706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.697114259737131e-06,
+      "loss": 1.0654,
+      "step": 2707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.693425139087185e-06,
+      "loss": 1.1117,
+      "step": 2708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.689736185954055e-06,
+      "loss": 1.0704,
+      "step": 2709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.686047402353433e-06,
+      "loss": 1.0912,
+      "step": 2710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.682358790300927e-06,
+      "loss": 1.1052,
+      "step": 2711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6786703518120415e-06,
+      "loss": 1.0815,
+      "step": 2712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.674982088902196e-06,
+      "loss": 1.0706,
+      "step": 2713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.671294003586705e-06,
+      "loss": 1.0787,
+      "step": 2714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6676060978807936e-06,
+      "loss": 1.0634,
+      "step": 2715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.663918373799581e-06,
+      "loss": 1.1161,
+      "step": 2716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.660230833358094e-06,
+      "loss": 1.1163,
+      "step": 2717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.656543478571261e-06,
+      "loss": 1.1241,
+      "step": 2718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.652856311453896e-06,
+      "loss": 1.1058,
+      "step": 2719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6491693340207254e-06,
+      "loss": 1.1519,
+      "step": 2720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.645482548286361e-06,
+      "loss": 1.1083,
+      "step": 2721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.64179595626532e-06,
+      "loss": 1.1167,
+      "step": 2722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.638109559972001e-06,
+      "loss": 1.0838,
+      "step": 2723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.63442336142071e-06,
+      "loss": 1.1199,
+      "step": 2724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.630737362625631e-06,
+      "loss": 1.1132,
+      "step": 2725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.62705156560085e-06,
+      "loss": 1.1047,
+      "step": 2726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6233659723603374e-06,
+      "loss": 1.077,
+      "step": 2727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.619680584917952e-06,
+      "loss": 1.1224,
+      "step": 2728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.615995405287442e-06,
+      "loss": 1.132,
+      "step": 2729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.612310435482443e-06,
+      "loss": 1.1083,
+      "step": 2730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.608625677516472e-06,
+      "loss": 1.0853,
+      "step": 2731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.604941133402934e-06,
+      "loss": 1.0675,
+      "step": 2732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.601256805155114e-06,
+      "loss": 1.124,
+      "step": 2733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.597572694786181e-06,
+      "loss": 1.1173,
+      "step": 2734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.593888804309188e-06,
+      "loss": 1.116,
+      "step": 2735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.59020513573706e-06,
+      "loss": 1.1201,
+      "step": 2736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5865216910826095e-06,
+      "loss": 1.1182,
+      "step": 2737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.582838472358519e-06,
+      "loss": 1.0814,
+      "step": 2738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.579155481577353e-06,
+      "loss": 1.1135,
+      "step": 2739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.575472720751547e-06,
+      "loss": 1.1117,
+      "step": 2740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.571790191893418e-06,
+      "loss": 1.0945,
+      "step": 2741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5681078970151455e-06,
+      "loss": 1.1156,
+      "step": 2742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.564425838128793e-06,
+      "loss": 1.1073,
+      "step": 2743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.560744017246284e-06,
+      "loss": 1.0845,
+      "step": 2744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.557062436379422e-06,
+      "loss": 1.1276,
+      "step": 2745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5533810975398715e-06,
+      "loss": 1.0991,
+      "step": 2746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.549700002739169e-06,
+      "loss": 1.0842,
+      "step": 2747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.546019153988714e-06,
+      "loss": 1.0926,
+      "step": 2748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5423385532997795e-06,
+      "loss": 1.1062,
+      "step": 2749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.53865820268349e-06,
+      "loss": 1.0826,
+      "step": 2750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.534978104150848e-06,
+      "loss": 1.1087,
+      "step": 2751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.531298259712707e-06,
+      "loss": 1.0627,
+      "step": 2752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.527618671379785e-06,
+      "loss": 1.0964,
+      "step": 2753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.523939341162665e-06,
+      "loss": 1.0907,
+      "step": 2754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.52026027107178e-06,
+      "loss": 1.0976,
+      "step": 2755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5165814631174296e-06,
+      "loss": 1.0804,
+      "step": 2756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.512902919309764e-06,
+      "loss": 1.0932,
+      "step": 2757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.509224641658792e-06,
+      "loss": 1.1188,
+      "step": 2758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.505546632174375e-06,
+      "loss": 1.068,
+      "step": 2759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.50186889286623e-06,
+      "loss": 1.0677,
+      "step": 2760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4981914257439254e-06,
+      "loss": 1.0868,
+      "step": 2761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.494514232816883e-06,
+      "loss": 1.1435,
+      "step": 2762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.490837316094368e-06,
+      "loss": 1.1107,
+      "step": 2763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.487160677585504e-06,
+      "loss": 1.0886,
+      "step": 2764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.483484319299254e-06,
+      "loss": 1.1039,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.479808243244437e-06,
+      "loss": 1.0937,
+      "step": 2766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.476132451429707e-06,
+      "loss": 1.1161,
+      "step": 2767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.472456945863573e-06,
+      "loss": 1.101,
+      "step": 2768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.468781728554379e-06,
+      "loss": 1.1495,
+      "step": 2769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.465106801510319e-06,
+      "loss": 1.0943,
+      "step": 2770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.461432166739421e-06,
+      "loss": 1.077,
+      "step": 2771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.457757826249559e-06,
+      "loss": 1.0979,
+      "step": 2772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.454083782048447e-06,
+      "loss": 1.0869,
+      "step": 2773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.450410036143629e-06,
+      "loss": 1.1221,
+      "step": 2774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.446736590542497e-06,
+      "loss": 1.1083,
+      "step": 2775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.443063447252268e-06,
+      "loss": 1.0889,
+      "step": 2776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4393906082800046e-06,
+      "loss": 1.0984,
+      "step": 2777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4357180756325915e-06,
+      "loss": 1.0876,
+      "step": 2778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.432045851316759e-06,
+      "loss": 1.0763,
+      "step": 2779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.428373937339056e-06,
+      "loss": 1.104,
+      "step": 2780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.424702335705871e-06,
+      "loss": 1.0857,
+      "step": 2781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.421031048423418e-06,
+      "loss": 1.0912,
+      "step": 2782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.417360077497741e-06,
+      "loss": 1.1061,
+      "step": 2783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.413689424934709e-06,
+      "loss": 1.1071,
+      "step": 2784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.410019092740018e-06,
+      "loss": 1.0634,
+      "step": 2785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.40634908291919e-06,
+      "loss": 1.0928,
+      "step": 2786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.402679397477571e-06,
+      "loss": 1.1065,
+      "step": 2787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.399010038420325e-06,
+      "loss": 1.097,
+      "step": 2788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.395341007752444e-06,
+      "loss": 1.1007,
+      "step": 2789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.391672307478739e-06,
+      "loss": 1.0864,
+      "step": 2790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3880039396038356e-06,
+      "loss": 1.0814,
+      "step": 2791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.384335906132188e-06,
+      "loss": 1.0538,
+      "step": 2792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.380668209068054e-06,
+      "loss": 1.1463,
+      "step": 2793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.37700085041552e-06,
+      "loss": 1.1604,
+      "step": 2794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.373333832178478e-06,
+      "loss": 1.0793,
+      "step": 2795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.369667156360644e-06,
+      "loss": 1.108,
+      "step": 2796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.366000824965535e-06,
+      "loss": 1.1023,
+      "step": 2797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.362334839996492e-06,
+      "loss": 1.1148,
+      "step": 2798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3586692034566554e-06,
+      "loss": 1.0975,
+      "step": 2799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.355003917348985e-06,
+      "loss": 1.1486,
+      "step": 2800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.35133898367624e-06,
+      "loss": 1.0879,
+      "step": 2801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3476744044409965e-06,
+      "loss": 1.1032,
+      "step": 2802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.344010181645628e-06,
+      "loss": 1.0947,
+      "step": 2803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.340346317292321e-06,
+      "loss": 1.0732,
+      "step": 2804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.336682813383059e-06,
+      "loss": 1.0941,
+      "step": 2805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.333019671919636e-06,
+      "loss": 1.1167,
+      "step": 2806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3293568949036394e-06,
+      "loss": 1.0726,
+      "step": 2807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.325694484336464e-06,
+      "loss": 1.1003,
+      "step": 2808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.322032442219305e-06,
+      "loss": 1.1082,
+      "step": 2809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3183707705531494e-06,
+      "loss": 1.0939,
+      "step": 2810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.314709471338792e-06,
+      "loss": 1.099,
+      "step": 2811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.31104854657681e-06,
+      "loss": 1.0774,
+      "step": 2812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.307387998267591e-06,
+      "loss": 1.0781,
+      "step": 2813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.303727828411306e-06,
+      "loss": 1.109,
+      "step": 2814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.300068039007926e-06,
+      "loss": 1.0515,
+      "step": 2815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.29640863205721e-06,
+      "loss": 1.1226,
+      "step": 2816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.292749609558711e-06,
+      "loss": 1.0774,
+      "step": 2817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.289090973511766e-06,
+      "loss": 1.0906,
+      "step": 2818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.285432725915512e-06,
+      "loss": 1.1061,
+      "step": 2819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2817748687688596e-06,
+      "loss": 1.0608,
+      "step": 2820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2781174040705195e-06,
+      "loss": 1.0851,
+      "step": 2821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.274460333818977e-06,
+      "loss": 1.0884,
+      "step": 2822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.270803660012511e-06,
+      "loss": 1.1203,
+      "step": 2823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2671473846491755e-06,
+      "loss": 1.0864,
+      "step": 2824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.263491509726812e-06,
+      "loss": 1.0873,
+      "step": 2825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2598360372430455e-06,
+      "loss": 1.1244,
+      "step": 2826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2561809691952715e-06,
+      "loss": 1.0542,
+      "step": 2827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.252526307580677e-06,
+      "loss": 1.0806,
+      "step": 2828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.248872054396215e-06,
+      "loss": 1.1118,
+      "step": 2829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2452182116386244e-06,
+      "loss": 1.1004,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.241564781304413e-06,
+      "loss": 1.0865,
+      "step": 2831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.237911765389871e-06,
+      "loss": 1.0828,
+      "step": 2832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.234259165891053e-06,
+      "loss": 1.1047,
+      "step": 2833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.230606984803794e-06,
+      "loss": 1.0995,
+      "step": 2834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2269552241236945e-06,
+      "loss": 1.0691,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.223303885846131e-06,
+      "loss": 1.085,
+      "step": 2836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.219652971966243e-06,
+      "loss": 1.0777,
+      "step": 2837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.216002484478944e-06,
+      "loss": 1.0914,
+      "step": 2838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.212352425378908e-06,
+      "loss": 1.124,
+      "step": 2839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.208702796660581e-06,
+      "loss": 1.0916,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.205053600318173e-06,
+      "loss": 1.0989,
+      "step": 2841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.201404838345653e-06,
+      "loss": 1.0653,
+      "step": 2842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.197756512736758e-06,
+      "loss": 1.0932,
+      "step": 2843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.194108625484983e-06,
+      "loss": 1.0914,
+      "step": 2844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.190461178583586e-06,
+      "loss": 1.0948,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.186814174025582e-06,
+      "loss": 1.1515,
+      "step": 2846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.183167613803747e-06,
+      "loss": 1.0711,
+      "step": 2847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.179521499910612e-06,
+      "loss": 1.0883,
+      "step": 2848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.175875834338466e-06,
+      "loss": 1.1136,
+      "step": 2849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.17223061907935e-06,
+      "loss": 1.1257,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.168585856125063e-06,
+      "loss": 1.0703,
+      "step": 2851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.164941547467152e-06,
+      "loss": 1.1191,
+      "step": 2852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.161297695096922e-06,
+      "loss": 1.1215,
+      "step": 2853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.157654301005421e-06,
+      "loss": 1.0974,
+      "step": 2854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.154011367183455e-06,
+      "loss": 1.078,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.15036889562157e-06,
+      "loss": 1.1007,
+      "step": 2856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.146726888310069e-06,
+      "loss": 1.1203,
+      "step": 2857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.14308534723899e-06,
+      "loss": 1.0781,
+      "step": 2858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1394442743981275e-06,
+      "loss": 1.1188,
+      "step": 2859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.135803671777011e-06,
+      "loss": 1.0444,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.132163541364919e-06,
+      "loss": 1.079,
+      "step": 2861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.128523885150869e-06,
+      "loss": 1.1115,
+      "step": 2862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.124884705123619e-06,
+      "loss": 1.0753,
+      "step": 2863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.121246003271674e-06,
+      "loss": 1.0993,
+      "step": 2864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.117607781583265e-06,
+      "loss": 1.1016,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.113970042046371e-06,
+      "loss": 1.0797,
+      "step": 2866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.110332786648702e-06,
+      "loss": 1.1171,
+      "step": 2867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.10669601737771e-06,
+      "loss": 1.1105,
+      "step": 2868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.10305973622057e-06,
+      "loss": 1.0811,
+      "step": 2869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0994239451642024e-06,
+      "loss": 1.0946,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.095788646195253e-06,
+      "loss": 1.1261,
+      "step": 2871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0921538413001e-06,
+      "loss": 1.0921,
+      "step": 2872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0885195324648526e-06,
+      "loss": 1.1162,
+      "step": 2873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.084885721675347e-06,
+      "loss": 1.0697,
+      "step": 2874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.081252410917148e-06,
+      "loss": 1.1006,
+      "step": 2875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.077619602175552e-06,
+      "loss": 1.0854,
+      "step": 2876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.073987297435571e-06,
+      "loss": 1.0906,
+      "step": 2877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.070355498681952e-06,
+      "loss": 1.1189,
+      "step": 2878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.066724207899157e-06,
+      "loss": 1.0731,
+      "step": 2879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.063093427071376e-06,
+      "loss": 1.1198,
+      "step": 2880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0594631581825214e-06,
+      "loss": 1.1092,
+      "step": 2881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.055833403216219e-06,
+      "loss": 1.083,
+      "step": 2882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.052204164155821e-06,
+      "loss": 1.0716,
+      "step": 2883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.048575442984392e-06,
+      "loss": 1.0851,
+      "step": 2884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.04494724168472e-06,
+      "loss": 1.0967,
+      "step": 2885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0413195622393e-06,
+      "loss": 1.1008,
+      "step": 2886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.037692406630351e-06,
+      "loss": 1.0616,
+      "step": 2887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0340657768398006e-06,
+      "loss": 1.0954,
+      "step": 2888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.030439674849291e-06,
+      "loss": 1.1035,
+      "step": 2889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.026814102640174e-06,
+      "loss": 1.1178,
+      "step": 2890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.023189062193514e-06,
+      "loss": 1.0751,
+      "step": 2891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.019564555490084e-06,
+      "loss": 1.0861,
+      "step": 2892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.015940584510366e-06,
+      "loss": 1.0919,
+      "step": 2893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.012317151234547e-06,
+      "loss": 1.0801,
+      "step": 2894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0086942576425245e-06,
+      "loss": 1.103,
+      "step": 2895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0050719057138955e-06,
+      "loss": 1.0794,
+      "step": 2896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.001450097427965e-06,
+      "loss": 1.1025,
+      "step": 2897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.997828834763743e-06,
+      "loss": 1.0613,
+      "step": 2898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994208119699933e-06,
+      "loss": 1.0837,
+      "step": 2899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99058795421495e-06,
+      "loss": 1.0678,
+      "step": 2900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9869683402868986e-06,
+      "loss": 1.0677,
+      "step": 2901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98334927989359e-06,
+      "loss": 1.0653,
+      "step": 2902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979730775012528e-06,
+      "loss": 1.0773,
+      "step": 2903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.976112827620916e-06,
+      "loss": 1.0708,
+      "step": 2904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.972495439695651e-06,
+      "loss": 1.1351,
+      "step": 2905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.968878613213323e-06,
+      "loss": 1.1078,
+      "step": 2906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.965262350150216e-06,
+      "loss": 1.105,
+      "step": 2907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.961646652482313e-06,
+      "loss": 1.0931,
+      "step": 2908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.958031522185275e-06,
+      "loss": 1.0961,
+      "step": 2909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9544169612344644e-06,
+      "loss": 1.1109,
+      "step": 2910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.950802971604925e-06,
+      "loss": 1.0688,
+      "step": 2911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.947189555271395e-06,
+      "loss": 1.0902,
+      "step": 2912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.943576714208293e-06,
+      "loss": 1.1209,
+      "step": 2913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.939964450389728e-06,
+      "loss": 1.1086,
+      "step": 2914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.93635276578949e-06,
+      "loss": 1.0765,
+      "step": 2915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.932741662381055e-06,
+      "loss": 1.0648,
+      "step": 2916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.929131142137583e-06,
+      "loss": 1.0725,
+      "step": 2917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.925521207031908e-06,
+      "loss": 1.1081,
+      "step": 2918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.921911859036556e-06,
+      "loss": 1.0744,
+      "step": 2919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9183031001237185e-06,
+      "loss": 1.0996,
+      "step": 2920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.914694932265277e-06,
+      "loss": 1.1129,
+      "step": 2921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9110873574327825e-06,
+      "loss": 1.082,
+      "step": 2922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.907480377597467e-06,
+      "loss": 1.0817,
+      "step": 2923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.903873994730231e-06,
+      "loss": 1.0707,
+      "step": 2924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9002682108016585e-06,
+      "loss": 1.1146,
+      "step": 2925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.896663027781995e-06,
+      "loss": 1.0612,
+      "step": 2926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8930584476411664e-06,
+      "loss": 1.0616,
+      "step": 2927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.889454472348766e-06,
+      "loss": 1.1055,
+      "step": 2928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.885851103874055e-06,
+      "loss": 1.1287,
+      "step": 2929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.882248344185965e-06,
+      "loss": 1.06,
+      "step": 2930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8786461952530955e-06,
+      "loss": 1.121,
+      "step": 2931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.875044659043709e-06,
+      "loss": 1.067,
+      "step": 2932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8714437375257394e-06,
+      "loss": 1.1074,
+      "step": 2933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.867843432666777e-06,
+      "loss": 1.1016,
+      "step": 2934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.864243746434082e-06,
+      "loss": 1.1225,
+      "step": 2935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.860644680794572e-06,
+      "loss": 1.093,
+      "step": 2936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.857046237714827e-06,
+      "loss": 1.065,
+      "step": 2937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.853448419161088e-06,
+      "loss": 1.0714,
+      "step": 2938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8498512270992515e-06,
+      "loss": 1.0941,
+      "step": 2939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.846254663494877e-06,
+      "loss": 1.0878,
+      "step": 2940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.842658730313173e-06,
+      "loss": 1.0927,
+      "step": 2941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.839063429519013e-06,
+      "loss": 1.0898,
+      "step": 2942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.835468763076916e-06,
+      "loss": 1.0846,
+      "step": 2943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.831874732951061e-06,
+      "loss": 1.0962,
+      "step": 2944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.828281341105273e-06,
+      "loss": 1.1081,
+      "step": 2945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.824688589503037e-06,
+      "loss": 1.0815,
+      "step": 2946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.821096480107477e-06,
+      "loss": 1.0522,
+      "step": 2947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.817505014881378e-06,
+      "loss": 1.1102,
+      "step": 2948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.813914195787162e-06,
+      "loss": 1.0439,
+      "step": 2949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8103240247869077e-06,
+      "loss": 1.0739,
+      "step": 2950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8067345038423305e-06,
+      "loss": 1.0502,
+      "step": 2951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.803145634914797e-06,
+      "loss": 1.0882,
+      "step": 2952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7995574199653184e-06,
+      "loss": 1.081,
+      "step": 2953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.795969860954541e-06,
+      "loss": 1.0795,
+      "step": 2954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7923829598427614e-06,
+      "loss": 1.0803,
+      "step": 2955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7887967185899095e-06,
+      "loss": 1.0434,
+      "step": 2956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.785211139155562e-06,
+      "loss": 1.0753,
+      "step": 2957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7816262234989247e-06,
+      "loss": 1.1092,
+      "step": 2958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.77804197357885e-06,
+      "loss": 1.1155,
+      "step": 2959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7744583913538213e-06,
+      "loss": 1.0977,
+      "step": 2960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7708754787819583e-06,
+      "loss": 1.0977,
+      "step": 2961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.767293237821015e-06,
+      "loss": 1.1073,
+      "step": 2962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.763711670428377e-06,
+      "loss": 1.0969,
+      "step": 2963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7601307785610643e-06,
+      "loss": 1.1095,
+      "step": 2964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.756550564175727e-06,
+      "loss": 1.0738,
+      "step": 2965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7529710292286427e-06,
+      "loss": 1.0851,
+      "step": 2966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7493921756757224e-06,
+      "loss": 1.0699,
+      "step": 2967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.745814005472498e-06,
+      "loss": 1.1032,
+      "step": 2968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7422365205741374e-06,
+      "loss": 1.0346,
+      "step": 2969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7386597229354226e-06,
+      "loss": 1.0573,
+      "step": 2970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7350836145107694e-06,
+      "loss": 1.0431,
+      "step": 2971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7315081972542155e-06,
+      "loss": 1.1061,
+      "step": 2972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7279334731194157e-06,
+      "loss": 1.0968,
+      "step": 2973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.724359444059652e-06,
+      "loss": 1.0779,
+      "step": 2974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.720786112027822e-06,
+      "loss": 1.1,
+      "step": 2975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.717213478976447e-06,
+      "loss": 1.0899,
+      "step": 2976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7136415468576607e-06,
+      "loss": 1.1018,
+      "step": 2977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.710070317623221e-06,
+      "loss": 1.0806,
+      "step": 2978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.706499793224494e-06,
+      "loss": 1.0589,
+      "step": 2979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7029299756124672e-06,
+      "loss": 1.0813,
+      "step": 2980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6993608667377367e-06,
+      "loss": 1.0824,
+      "step": 2981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.695792468550517e-06,
+      "loss": 1.0983,
+      "step": 2982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.692224783000627e-06,
+      "loss": 1.0945,
+      "step": 2983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6886578120375043e-06,
+      "loss": 1.0622,
+      "step": 2984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.685091557610188e-06,
+      "loss": 1.066,
+      "step": 2985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6815260216673333e-06,
+      "loss": 1.086,
+      "step": 2986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.677961206157197e-06,
+      "loss": 1.0973,
+      "step": 2987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.674397113027645e-06,
+      "loss": 1.0422,
+      "step": 2988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6708337442261473e-06,
+      "loss": 1.0816,
+      "step": 2989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6672711016997798e-06,
+      "loss": 1.1019,
+      "step": 2990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6637091873952202e-06,
+      "loss": 1.0353,
+      "step": 2991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6601480032587487e-06,
+      "loss": 1.073,
+      "step": 2992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.656587551236246e-06,
+      "loss": 1.0652,
+      "step": 2993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6530278332731915e-06,
+      "loss": 1.0735,
+      "step": 2994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.649468851314669e-06,
+      "loss": 1.0869,
+      "step": 2995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.645910607305353e-06,
+      "loss": 1.0866,
+      "step": 2996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6423531031895197e-06,
+      "loss": 1.0985,
+      "step": 2997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6387963409110378e-06,
+      "loss": 1.098,
+      "step": 2998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.635240322413375e-06,
+      "loss": 1.0297,
+      "step": 2999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6316850496395863e-06,
+      "loss": 1.0589,
+      "step": 3000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6281305245323257e-06,
+      "loss": 1.1112,
+      "step": 3001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6245767490338324e-06,
+      "loss": 1.0931,
+      "step": 3002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6210237250859443e-06,
+      "loss": 1.0903,
+      "step": 3003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6174714546300783e-06,
+      "loss": 1.0751,
+      "step": 3004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6139199396072488e-06,
+      "loss": 1.1161,
+      "step": 3005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6103691819580507e-06,
+      "loss": 1.0723,
+      "step": 3006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6068191836226685e-06,
+      "loss": 1.0859,
+      "step": 3007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.603269946540875e-06,
+      "loss": 1.0678,
+      "step": 3008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5997214726520167e-06,
+      "loss": 1.1059,
+      "step": 3009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5961737638950366e-06,
+      "loss": 1.0833,
+      "step": 3010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.592626822208446e-06,
+      "loss": 1.0953,
+      "step": 3011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.58908064953035e-06,
+      "loss": 1.084,
+      "step": 3012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.585535247798421e-06,
+      "loss": 1.0804,
+      "step": 3013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.581990618949922e-06,
+      "loss": 1.064,
+      "step": 3014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.578446764921684e-06,
+      "loss": 1.0806,
+      "step": 3015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5749036876501196e-06,
+      "loss": 1.0846,
+      "step": 3016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.571361389071218e-06,
+      "loss": 1.089,
+      "step": 3017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5678198711205384e-06,
+      "loss": 1.0985,
+      "step": 3018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.564279135733218e-06,
+      "loss": 1.078,
+      "step": 3019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.560739184843963e-06,
+      "loss": 1.0768,
+      "step": 3020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.557200020387052e-06,
+      "loss": 1.0864,
+      "step": 3021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.553661644296337e-06,
+      "loss": 1.045,
+      "step": 3022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.550124058505233e-06,
+      "loss": 1.1039,
+      "step": 3023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5465872649467296e-06,
+      "loss": 1.0957,
+      "step": 3024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5430512655533774e-06,
+      "loss": 1.0824,
+      "step": 3025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.539516062257298e-06,
+      "loss": 1.0594,
+      "step": 3026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5359816569901773e-06,
+      "loss": 1.0692,
+      "step": 3027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5324480516832606e-06,
+      "loss": 1.091,
+      "step": 3028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5289152482673643e-06,
+      "loss": 1.0522,
+      "step": 3029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5253832486728567e-06,
+      "loss": 1.0793,
+      "step": 3030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5218520548296777e-06,
+      "loss": 1.079,
+      "step": 3031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.518321668667315e-06,
+      "loss": 1.0847,
+      "step": 3032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5147920921148267e-06,
+      "loss": 1.0661,
+      "step": 3033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5112633271008183e-06,
+      "loss": 1.0492,
+      "step": 3034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5077353755534615e-06,
+      "loss": 1.0475,
+      "step": 3035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.504208239400474e-06,
+      "loss": 1.0716,
+      "step": 3036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5006819205691374e-06,
+      "loss": 1.0429,
+      "step": 3037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4971564209862763e-06,
+      "loss": 1.08,
+      "step": 3038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4936317425782783e-06,
+      "loss": 1.0635,
+      "step": 3039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4901078872710736e-06,
+      "loss": 1.103,
+      "step": 3040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4865848569901497e-06,
+      "loss": 1.0894,
+      "step": 3041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4830626536605355e-06,
+      "loss": 1.0441,
+      "step": 3042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.479541279206815e-06,
+      "loss": 1.0483,
+      "step": 3043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.476020735553119e-06,
+      "loss": 1.0454,
+      "step": 3044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4725010246231174e-06,
+      "loss": 1.0844,
+      "step": 3045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.468982148340033e-06,
+      "loss": 1.0794,
+      "step": 3046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.465464108626628e-06,
+      "loss": 1.0925,
+      "step": 3047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4619469074052094e-06,
+      "loss": 1.0776,
+      "step": 3048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4584305465976248e-06,
+      "loss": 1.1022,
+      "step": 3049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 1.1109,
+      "step": 3050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.451400353909053e-06,
+      "loss": 1.0865,
+      "step": 3051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.447886525869464e-06,
+      "loss": 1.0712,
+      "step": 3052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.444373545926497e-06,
+      "loss": 1.0835,
+      "step": 3053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.440861415999699e-06,
+      "loss": 1.0746,
+      "step": 3054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.437350138008141e-06,
+      "loss": 1.0395,
+      "step": 3055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4338397138704403e-06,
+      "loss": 1.0921,
+      "step": 3056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4303301455047376e-06,
+      "loss": 1.0726,
+      "step": 3057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.426821434828714e-06,
+      "loss": 1.0642,
+      "step": 3058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4233135837595753e-06,
+      "loss": 1.0861,
+      "step": 3059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4198065942140633e-06,
+      "loss": 1.1003,
+      "step": 3060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4163004681084446e-06,
+      "loss": 1.0739,
+      "step": 3061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4127952073585146e-06,
+      "loss": 1.0809,
+      "step": 3062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4092908138796015e-06,
+      "loss": 1.1117,
+      "step": 3063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.405787289586551e-06,
+      "loss": 1.0668,
+      "step": 3064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.40228463639374e-06,
+      "loss": 1.053,
+      "step": 3065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.398782856215066e-06,
+      "loss": 1.0747,
+      "step": 3066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3952819509639534e-06,
+      "loss": 1.0377,
+      "step": 3067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3917819225533435e-06,
+      "loss": 1.0621,
+      "step": 3068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.388282772895703e-06,
+      "loss": 1.0675,
+      "step": 3069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3847845039030138e-06,
+      "loss": 1.093,
+      "step": 3070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.381287117486783e-06,
+      "loss": 1.1031,
+      "step": 3071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3777906155580275e-06,
+      "loss": 1.1094,
+      "step": 3072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3742950000272877e-06,
+      "loss": 1.0502,
+      "step": 3073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.370800272804618e-06,
+      "loss": 1.0627,
+      "step": 3074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3673064357995844e-06,
+      "loss": 1.1062,
+      "step": 3075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3638134909212694e-06,
+      "loss": 1.1027,
+      "step": 3076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.360321440078267e-06,
+      "loss": 1.0841,
+      "step": 3077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3568302851786823e-06,
+      "loss": 1.083,
+      "step": 3078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3533400281301344e-06,
+      "loss": 1.0725,
+      "step": 3079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3498506708397437e-06,
+      "loss": 1.0761,
+      "step": 3080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3463622152141482e-06,
+      "loss": 1.0724,
+      "step": 3081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3428746631594877e-06,
+      "loss": 1.0328,
+      "step": 3082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3393880165814084e-06,
+      "loss": 1.083,
+      "step": 3083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3359022773850673e-06,
+      "loss": 1.0723,
+      "step": 3084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.332417447475116e-06,
+      "loss": 1.095,
+      "step": 3085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.328933528755719e-06,
+      "loss": 1.0707,
+      "step": 3086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3254505231305345e-06,
+      "loss": 1.0331,
+      "step": 3087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3219684325027297e-06,
+      "loss": 1.0528,
+      "step": 3088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.318487258774964e-06,
+      "loss": 1.081,
+      "step": 3089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.315007003849403e-06,
+      "loss": 1.0644,
+      "step": 3090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3115276696277034e-06,
+      "loss": 1.0533,
+      "step": 3091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3080492580110268e-06,
+      "loss": 1.0847,
+      "step": 3092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.304571770900021e-06,
+      "loss": 1.0783,
+      "step": 3093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3010952101948403e-06,
+      "loss": 1.1012,
+      "step": 3094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2976195777951193e-06,
+      "loss": 1.1217,
+      "step": 3095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.294144875599998e-06,
+      "loss": 1.0926,
+      "step": 3096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.290671105508099e-06,
+      "loss": 1.0496,
+      "step": 3097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.28719826941754e-06,
+      "loss": 1.1187,
+      "step": 3098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2837263692259298e-06,
+      "loss": 1.0791,
+      "step": 3099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2802554068303595e-06,
+      "loss": 1.0553,
+      "step": 3100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2767853841274154e-06,
+      "loss": 1.0255,
+      "step": 3101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.273316303013164e-06,
+      "loss": 1.0987,
+      "step": 3102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.269848165383163e-06,
+      "loss": 1.0613,
+      "step": 3103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2663809731324477e-06,
+      "loss": 1.0885,
+      "step": 3104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2629147281555434e-06,
+      "loss": 1.0387,
+      "step": 3105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.259449432346454e-06,
+      "loss": 1.0692,
+      "step": 3106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2559850875986677e-06,
+      "loss": 1.1061,
+      "step": 3107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2525216958051474e-06,
+      "loss": 1.1145,
+      "step": 3108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2490592588583435e-06,
+      "loss": 1.0744,
+      "step": 3109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2455977786501762e-06,
+      "loss": 1.0936,
+      "step": 3110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.242137257072051e-06,
+      "loss": 1.0711,
+      "step": 3111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2386776960148413e-06,
+      "loss": 1.0683,
+      "step": 3112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2352190973689045e-06,
+      "loss": 1.1127,
+      "step": 3113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2317614630240623e-06,
+      "loss": 1.0911,
+      "step": 3114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.22830479486962e-06,
+      "loss": 1.057,
+      "step": 3115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2248490947943446e-06,
+      "loss": 1.0489,
+      "step": 3116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.221394364686482e-06,
+      "loss": 1.0922,
+      "step": 3117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.217940606433747e-06,
+      "loss": 1.0669,
+      "step": 3118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2144878219233177e-06,
+      "loss": 1.0677,
+      "step": 3119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2110360130418484e-06,
+      "loss": 1.0453,
+      "step": 3120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.207585181675451e-06,
+      "loss": 1.0526,
+      "step": 3121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2041353297097138e-06,
+      "loss": 1.0462,
+      "step": 3122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2006864590296794e-06,
+      "loss": 1.0255,
+      "step": 3123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.197238571519863e-06,
+      "loss": 1.0503,
+      "step": 3124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1937916690642356e-06,
+      "loss": 1.0671,
+      "step": 3125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1903457535462368e-06,
+      "loss": 1.0516,
+      "step": 3126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.186900826848759e-06,
+      "loss": 1.09,
+      "step": 3127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.183456890854163e-06,
+      "loss": 1.0769,
+      "step": 3128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.18001394744426e-06,
+      "loss": 1.0641,
+      "step": 3129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.176571998500325e-06,
+      "loss": 1.0366,
+      "step": 3130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1731310459030875e-06,
+      "loss": 1.0766,
+      "step": 3131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1696910915327307e-06,
+      "loss": 1.0844,
+      "step": 3132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.166252137268895e-06,
+      "loss": 1.078,
+      "step": 3133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.162814184990672e-06,
+      "loss": 1.0886,
+      "step": 3134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1593772365766107e-06,
+      "loss": 1.1,
+      "step": 3135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1559412939047048e-06,
+      "loss": 1.0485,
+      "step": 3136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.152506358852403e-06,
+      "loss": 1.0612,
+      "step": 3137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1490724332966016e-06,
+      "loss": 1.0785,
+      "step": 3138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.145639519113646e-06,
+      "loss": 1.1242,
+      "step": 3139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.142207618179327e-06,
+      "loss": 1.0846,
+      "step": 3140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1387767323688886e-06,
+      "loss": 1.0194,
+      "step": 3141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.135346863557009e-06,
+      "loss": 1.048,
+      "step": 3142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1319180136178212e-06,
+      "loss": 1.0879,
+      "step": 3143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.128490184424894e-06,
+      "loss": 1.0839,
+      "step": 3144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.125063377851244e-06,
+      "loss": 1.0447,
+      "step": 3145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1216375957693236e-06,
+      "loss": 1.0634,
+      "step": 3146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1182128400510312e-06,
+      "loss": 1.0916,
+      "step": 3147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1147891125676994e-06,
+      "loss": 1.0854,
+      "step": 3148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1113664151901036e-06,
+      "loss": 1.0957,
+      "step": 3149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.107944749788449e-06,
+      "loss": 1.0471,
+      "step": 3150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.104524118232387e-06,
+      "loss": 1.0812,
+      "step": 3151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1011045223909954e-06,
+      "loss": 1.0616,
+      "step": 3152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0976859641327893e-06,
+      "loss": 1.1008,
+      "step": 3153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.094268445325721e-06,
+      "loss": 1.047,
+      "step": 3154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0908519678371657e-06,
+      "loss": 1.0522,
+      "step": 3155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0874365335339396e-06,
+      "loss": 1.0336,
+      "step": 3156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0840221442822783e-06,
+      "loss": 1.0697,
+      "step": 3157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.080608801947858e-06,
+      "loss": 1.1086,
+      "step": 3158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0771965083957723e-06,
+      "loss": 1.061,
+      "step": 3159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0737852654905497e-06,
+      "loss": 1.0436,
+      "step": 3160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.070375075096139e-06,
+      "loss": 1.0498,
+      "step": 3161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0669659390759164e-06,
+      "loss": 1.0975,
+      "step": 3162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.063557859292684e-06,
+      "loss": 1.0677,
+      "step": 3163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.060150837608662e-06,
+      "loss": 1.0606,
+      "step": 3164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.056744875885496e-06,
+      "loss": 1.0638,
+      "step": 3165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0533399759842513e-06,
+      "loss": 1.0736,
+      "step": 3166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.049936139765412e-06,
+      "loss": 1.1098,
+      "step": 3167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0465333690888853e-06,
+      "loss": 1.0957,
+      "step": 3168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.043131665813988e-06,
+      "loss": 1.0678,
+      "step": 3169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.039731031799462e-06,
+      "loss": 1.0644,
+      "step": 3170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.036331468903463e-06,
+      "loss": 1.0588,
+      "step": 3171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.032932978983555e-06,
+      "loss": 1.0441,
+      "step": 3172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.029535563896726e-06,
+      "loss": 1.0732,
+      "step": 3173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.026139225499366e-06,
+      "loss": 1.0789,
+      "step": 3174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0227439656472878e-06,
+      "loss": 1.0831,
+      "step": 3175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0193497861957044e-06,
+      "loss": 1.0746,
+      "step": 3176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0159566889992465e-06,
+      "loss": 1.039,
+      "step": 3177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0125646759119466e-06,
+      "loss": 1.0656,
+      "step": 3178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0091737487872523e-06,
+      "loss": 1.0994,
+      "step": 3179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0057839094780104e-06,
+      "loss": 1.0705,
+      "step": 3180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.00239515983648e-06,
+      "loss": 1.062,
+      "step": 3181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9990075017143173e-06,
+      "loss": 1.0693,
+      "step": 3182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9956209369625914e-06,
+      "loss": 1.0426,
+      "step": 3183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9922354674317644e-06,
+      "loss": 1.0852,
+      "step": 3184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9888510949717083e-06,
+      "loss": 1.0755,
+      "step": 3185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9854678214316875e-06,
+      "loss": 1.0533,
+      "step": 3186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.982085648660374e-06,
+      "loss": 1.0783,
+      "step": 3187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9787045785058327e-06,
+      "loss": 1.0846,
+      "step": 3188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9753246128155276e-06,
+      "loss": 1.0542,
+      "step": 3189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.971945753436323e-06,
+      "loss": 1.0812,
+      "step": 3190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.968568002214469e-06,
+      "loss": 1.0218,
+      "step": 3191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.965191360995621e-06,
+      "loss": 1.0633,
+      "step": 3192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9618158316248224e-06,
+      "loss": 1.0664,
+      "step": 3193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9584414159465097e-06,
+      "loss": 1.0494,
+      "step": 3194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.95506811580451e-06,
+      "loss": 1.0467,
+      "step": 3195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9516959330420438e-06,
+      "loss": 1.0848,
+      "step": 3196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9483248695017164e-06,
+      "loss": 1.0878,
+      "step": 3197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9449549270255286e-06,
+      "loss": 1.0506,
+      "step": 3198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9415861074548593e-06,
+      "loss": 1.0866,
+      "step": 3199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9382184126304834e-06,
+      "loss": 1.0648,
+      "step": 3200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9348518443925526e-06,
+      "loss": 1.0705,
+      "step": 3201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.931486404580612e-06,
+      "loss": 1.0976,
+      "step": 3202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.92812209503358e-06,
+      "loss": 1.0678,
+      "step": 3203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.924758917589766e-06,
+      "loss": 1.0842,
+      "step": 3204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.921396874086856e-06,
+      "loss": 1.0655,
+      "step": 3205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9180359663619186e-06,
+      "loss": 1.0679,
+      "step": 3206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9146761962514014e-06,
+      "loss": 1.0999,
+      "step": 3207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9113175655911287e-06,
+      "loss": 1.1036,
+      "step": 3208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9079600762163042e-06,
+      "loss": 1.0708,
+      "step": 3209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.904603729961508e-06,
+      "loss": 1.0596,
+      "step": 3210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.901248528660694e-06,
+      "loss": 1.0999,
+      "step": 3211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8978944741471892e-06,
+      "loss": 1.0646,
+      "step": 3212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8945415682537025e-06,
+      "loss": 1.0783,
+      "step": 3213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8911898128123006e-06,
+      "loss": 1.0723,
+      "step": 3214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8878392096544388e-06,
+      "loss": 1.0551,
+      "step": 3215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8844897606109244e-06,
+      "loss": 1.0814,
+      "step": 3216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8811414675119526e-06,
+      "loss": 1.0608,
+      "step": 3217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.877794332187069e-06,
+      "loss": 1.0817,
+      "step": 3218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8744483564652037e-06,
+      "loss": 1.0357,
+      "step": 3219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.871103542174637e-06,
+      "loss": 1.0498,
+      "step": 3220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8677598911430293e-06,
+      "loss": 1.0654,
+      "step": 3221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.864417405197393e-06,
+      "loss": 1.0495,
+      "step": 3222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.861076086164115e-06,
+      "loss": 1.078,
+      "step": 3223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8577359358689323e-06,
+      "loss": 1.0862,
+      "step": 3224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8543969561369556e-06,
+      "loss": 1.0406,
+      "step": 3225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.851059148792648e-06,
+      "loss": 1.0826,
+      "step": 3226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.847722515659835e-06,
+      "loss": 1.0792,
+      "step": 3227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8443870585617003e-06,
+      "loss": 1.0622,
+      "step": 3228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8410527793207834e-06,
+      "loss": 1.083,
+      "step": 3229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.837719679758981e-06,
+      "loss": 1.0355,
+      "step": 3230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.834387761697548e-06,
+      "loss": 1.0615,
+      "step": 3231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8310570269570884e-06,
+      "loss": 1.0491,
+      "step": 3232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.827727477357564e-06,
+      "loss": 1.0621,
+      "step": 3233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.824399114718288e-06,
+      "loss": 1.0454,
+      "step": 3234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.821071940857923e-06,
+      "loss": 1.08,
+      "step": 3235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8177459575944844e-06,
+      "loss": 1.0751,
+      "step": 3236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.814421166745337e-06,
+      "loss": 1.0596,
+      "step": 3237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8110975701271913e-06,
+      "loss": 1.0534,
+      "step": 3238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8077751695561073e-06,
+      "loss": 1.0782,
+      "step": 3239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8044539668474968e-06,
+      "loss": 1.0548,
+      "step": 3240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.801133963816102e-06,
+      "loss": 1.0638,
+      "step": 3241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7978151622760297e-06,
+      "loss": 1.0454,
+      "step": 3242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7944975640407103e-06,
+      "loss": 1.0699,
+      "step": 3243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.791181170922932e-06,
+      "loss": 1.0606,
+      "step": 3244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.787865984734818e-06,
+      "loss": 1.0361,
+      "step": 3245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7845520072878308e-06,
+      "loss": 1.0514,
+      "step": 3246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7812392403927753e-06,
+      "loss": 1.0853,
+      "step": 3247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7779276858597936e-06,
+      "loss": 1.0225,
+      "step": 3248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.774617345498366e-06,
+      "loss": 1.0617,
+      "step": 3249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.771308221117309e-06,
+      "loss": 1.0687,
+      "step": 3250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7680003145247747e-06,
+      "loss": 1.0744,
+      "step": 3251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.764693627528249e-06,
+      "loss": 1.0541,
+      "step": 3252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7613881619345527e-06,
+      "loss": 1.0869,
+      "step": 3253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7580839195498397e-06,
+      "loss": 1.0662,
+      "step": 3254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7547809021795924e-06,
+      "loss": 1.0696,
+      "step": 3255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7514791116286276e-06,
+      "loss": 1.0557,
+      "step": 3256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7481785497010894e-06,
+      "loss": 1.0517,
+      "step": 3257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.744879218200452e-06,
+      "loss": 1.0561,
+      "step": 3258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.741581118929515e-06,
+      "loss": 1.0572,
+      "step": 3259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7382842536904077e-06,
+      "loss": 1.0684,
+      "step": 3260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.73498862428458e-06,
+      "loss": 1.0719,
+      "step": 3261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.731694232512818e-06,
+      "loss": 1.0545,
+      "step": 3262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7284010801752152e-06,
+      "loss": 1.1518,
+      "step": 3263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.725109169071204e-06,
+      "loss": 1.0665,
+      "step": 3264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7218185009995223e-06,
+      "loss": 1.072,
+      "step": 3265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7185290777582442e-06,
+      "loss": 1.073,
+      "step": 3266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7152409011447544e-06,
+      "loss": 1.0753,
+      "step": 3267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.711953972955759e-06,
+      "loss": 1.0499,
+      "step": 3268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.708668294987282e-06,
+      "loss": 1.073,
+      "step": 3269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.705383869034664e-06,
+      "loss": 1.0692,
+      "step": 3270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7021006968925613e-06,
+      "loss": 1.0885,
+      "step": 3271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6988187803549448e-06,
+      "loss": 1.0815,
+      "step": 3272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.695538121215101e-06,
+      "loss": 1.0538,
+      "step": 3273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6922587212656275e-06,
+      "loss": 1.0611,
+      "step": 3274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6889805822984348e-06,
+      "loss": 1.0717,
+      "step": 3275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.685703706104744e-06,
+      "loss": 1.0565,
+      "step": 3276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.682428094475086e-06,
+      "loss": 1.035,
+      "step": 3277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6791537491993013e-06,
+      "loss": 1.1235,
+      "step": 3278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6758806720665385e-06,
+      "loss": 1.016,
+      "step": 3279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6726088648652505e-06,
+      "loss": 1.046,
+      "step": 3280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.669338329383205e-06,
+      "loss": 1.0645,
+      "step": 3281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.666069067407461e-06,
+      "loss": 1.0452,
+      "step": 3282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6628010807243955e-06,
+      "loss": 1.0768,
+      "step": 3283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6595343711196764e-06,
+      "loss": 1.042,
+      "step": 3284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.656268940378287e-06,
+      "loss": 1.0547,
+      "step": 3285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6530047902844964e-06,
+      "loss": 1.0557,
+      "step": 3286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.64974192262189e-06,
+      "loss": 1.0876,
+      "step": 3287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.646480339173337e-06,
+      "loss": 1.0882,
+      "step": 3288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6432200417210204e-06,
+      "loss": 1.0759,
+      "step": 3289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.639961032046404e-06,
+      "loss": 1.0543,
+      "step": 3290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.636703311930263e-06,
+      "loss": 1.0443,
+      "step": 3291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.63344688315266e-06,
+      "loss": 1.0736,
+      "step": 3292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.630191747492952e-06,
+      "loss": 1.0946,
+      "step": 3293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6269379067297924e-06,
+      "loss": 1.0549,
+      "step": 3294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6236853626411248e-06,
+      "loss": 1.0781,
+      "step": 3295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6204341170041847e-06,
+      "loss": 1.0916,
+      "step": 3296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6171841715954993e-06,
+      "loss": 1.0551,
+      "step": 3297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.613935528190884e-06,
+      "loss": 1.055,
+      "step": 3298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.610688188565442e-06,
+      "loss": 1.0906,
+      "step": 3299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.607442154493568e-06,
+      "loss": 1.0753,
+      "step": 3300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.604197427748937e-06,
+      "loss": 1.0862,
+      "step": 3301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.6009540101045196e-06,
+      "loss": 1.0672,
+      "step": 3302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.597711903332556e-06,
+      "loss": 1.0884,
+      "step": 3303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.594471109204588e-06,
+      "loss": 1.0679,
+      "step": 3304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5912316294914232e-06,
+      "loss": 1.0685,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5879934659631658e-06,
+      "loss": 1.0589,
+      "step": 3306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5847566203891866e-06,
+      "loss": 1.073,
+      "step": 3307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.581521094538151e-06,
+      "loss": 1.0696,
+      "step": 3308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5782868901779888e-06,
+      "loss": 1.0738,
+      "step": 3309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5750540090759226e-06,
+      "loss": 1.0686,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.571822452998435e-06,
+      "loss": 1.0587,
+      "step": 3311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.568592223711302e-06,
+      "loss": 1.0607,
+      "step": 3312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5653633229795595e-06,
+      "loss": 1.0819,
+      "step": 3313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5621357525675304e-06,
+      "loss": 1.0697,
+      "step": 3314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.558909514238798e-06,
+      "loss": 1.0733,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5556846097562292e-06,
+      "loss": 1.0528,
+      "step": 3316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.552461040881957e-06,
+      "loss": 1.044,
+      "step": 3317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.549238809377382e-06,
+      "loss": 1.0331,
+      "step": 3318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5460179170031797e-06,
+      "loss": 1.0847,
+      "step": 3319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.542798365519289e-06,
+      "loss": 1.0998,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5395801566849187e-06,
+      "loss": 1.0677,
+      "step": 3321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.536363292258543e-06,
+      "loss": 1.0741,
+      "step": 3322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5331477739979026e-06,
+      "loss": 1.0391,
+      "step": 3323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.529933603660001e-06,
+      "loss": 1.0617,
+      "step": 3324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.526720783001107e-06,
+      "loss": 1.0583,
+      "step": 3325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5235093137767502e-06,
+      "loss": 1.0708,
+      "step": 3326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5202991977417237e-06,
+      "loss": 1.0974,
+      "step": 3327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.517090436650076e-06,
+      "loss": 1.1002,
+      "step": 3328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.513883032255127e-06,
+      "loss": 1.0759,
+      "step": 3329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5106769863094386e-06,
+      "loss": 1.053,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.507472300564848e-06,
+      "loss": 1.0751,
+      "step": 3331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5042689767724326e-06,
+      "loss": 1.0308,
+      "step": 3332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.501067016682541e-06,
+      "loss": 1.0612,
+      "step": 3333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4978664220447617e-06,
+      "loss": 1.0616,
+      "step": 3334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4946671946079505e-06,
+      "loss": 1.033,
+      "step": 3335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.491469336120208e-06,
+      "loss": 1.0695,
+      "step": 3336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.488272848328889e-06,
+      "loss": 1.0331,
+      "step": 3337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4850777329805997e-06,
+      "loss": 1.0393,
+      "step": 3338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4818839918211963e-06,
+      "loss": 1.0924,
+      "step": 3339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4786916265957835e-06,
+      "loss": 1.0296,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.475500639048714e-06,
+      "loss": 1.0268,
+      "step": 3341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4723110309235887e-06,
+      "loss": 1.1063,
+      "step": 3342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4691228039632548e-06,
+      "loss": 1.0761,
+      "step": 3343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.465935959909804e-06,
+      "loss": 1.0828,
+      "step": 3344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.462750500504573e-06,
+      "loss": 1.0966,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4595664274881404e-06,
+      "loss": 1.0373,
+      "step": 3346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4563837426003295e-06,
+      "loss": 1.1043,
+      "step": 3347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4532024475802046e-06,
+      "loss": 1.0658,
+      "step": 3348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4500225441660687e-06,
+      "loss": 1.0464,
+      "step": 3349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4468440340954664e-06,
+      "loss": 1.0328,
+      "step": 3350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4436669191051803e-06,
+      "loss": 1.0541,
+      "step": 3351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.440491200931229e-06,
+      "loss": 1.0457,
+      "step": 3352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4373168813088728e-06,
+      "loss": 1.0225,
+      "step": 3353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4341439619726033e-06,
+      "loss": 1.0481,
+      "step": 3354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.430972444656148e-06,
+      "loss": 1.0505,
+      "step": 3355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4278023310924676e-06,
+      "loss": 1.0215,
+      "step": 3356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4246336230137585e-06,
+      "loss": 1.0293,
+      "step": 3357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4214663221514457e-06,
+      "loss": 1.0373,
+      "step": 3358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.418300430236188e-06,
+      "loss": 1.0093,
+      "step": 3359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4151359489978733e-06,
+      "loss": 1.0466,
+      "step": 3360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4119728801656176e-06,
+      "loss": 1.0633,
+      "step": 3361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.408811225467767e-06,
+      "loss": 1.0932,
+      "step": 3362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4056509866318938e-06,
+      "loss": 1.0863,
+      "step": 3363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.402492165384798e-06,
+      "loss": 1.0739,
+      "step": 3364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3993347634525016e-06,
+      "loss": 1.0549,
+      "step": 3365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3961787825602555e-06,
+      "loss": 1.0621,
+      "step": 3366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3930242244325314e-06,
+      "loss": 1.045,
+      "step": 3367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3898710907930237e-06,
+      "loss": 1.0364,
+      "step": 3368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3867193833646494e-06,
+      "loss": 1.0603,
+      "step": 3369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3835691038695464e-06,
+      "loss": 1.0199,
+      "step": 3370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.380420254029068e-06,
+      "loss": 1.0415,
+      "step": 3371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.377272835563798e-06,
+      "loss": 1.0458,
+      "step": 3372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3741268501935212e-06,
+      "loss": 1.0319,
+      "step": 3373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3709822996372556e-06,
+      "loss": 1.0416,
+      "step": 3374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3678391856132203e-06,
+      "loss": 1.0732,
+      "step": 3375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.364697509838866e-06,
+      "loss": 1.0683,
+      "step": 3376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.361557274030839e-06,
+      "loss": 1.0399,
+      "step": 3377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.358418479905017e-06,
+      "loss": 1.0723,
+      "step": 3378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.355281129176473e-06,
+      "loss": 1.0958,
+      "step": 3379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3521452235595065e-06,
+      "loss": 1.0645,
+      "step": 3380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3490107647676168e-06,
+      "loss": 1.0381,
+      "step": 3381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3458777545135174e-06,
+      "loss": 1.0759,
+      "step": 3382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.342746194509129e-06,
+      "loss": 1.0202,
+      "step": 3383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.33961608646558e-06,
+      "loss": 1.0961,
+      "step": 3384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.336487432093205e-06,
+      "loss": 1.0666,
+      "step": 3385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3333602331015453e-06,
+      "loss": 1.0619,
+      "step": 3386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.330234491199346e-06,
+      "loss": 1.0489,
+      "step": 3387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3271102080945563e-06,
+      "loss": 1.0513,
+      "step": 3388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.323987385494329e-06,
+      "loss": 1.0263,
+      "step": 3389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.320866025105016e-06,
+      "loss": 1.0432,
+      "step": 3390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3177461286321783e-06,
+      "loss": 1.0476,
+      "step": 3391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.314627697780563e-06,
+      "loss": 1.0594,
+      "step": 3392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3115107342541336e-06,
+      "loss": 1.0563,
+      "step": 3393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.308395239756034e-06,
+      "loss": 1.0474,
+      "step": 3394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3052812159886216e-06,
+      "loss": 1.0671,
+      "step": 3395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3021686646534364e-06,
+      "loss": 1.0541,
+      "step": 3396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2990575874512264e-06,
+      "loss": 1.0446,
+      "step": 3397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2959479860819218e-06,
+      "loss": 1.0566,
+      "step": 3398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.292839862244659e-06,
+      "loss": 1.061,
+      "step": 3399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.289733217637753e-06,
+      "loss": 1.0303,
+      "step": 3400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.286628053958726e-06,
+      "loss": 1.0415,
+      "step": 3401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.283524372904276e-06,
+      "loss": 1.0643,
+      "step": 3402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2804221761703042e-06,
+      "loss": 1.0434,
+      "step": 3403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2773214654518887e-06,
+      "loss": 1.0618,
+      "step": 3404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2742222424433076e-06,
+      "loss": 1.0741,
+      "step": 3405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.271124508838013e-06,
+      "loss": 1.0417,
+      "step": 3406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.268028266328655e-06,
+      "loss": 1.0998,
+      "step": 3407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.264933516607063e-06,
+      "loss": 1.066,
+      "step": 3408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2618402613642516e-06,
+      "loss": 1.0244,
+      "step": 3409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.258748502290419e-06,
+      "loss": 1.1113,
+      "step": 3410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2556582410749458e-06,
+      "loss": 1.0785,
+      "step": 3411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2525694794063935e-06,
+      "loss": 1.0532,
+      "step": 3412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.249482218972506e-06,
+      "loss": 1.066,
+      "step": 3413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.246396461460205e-06,
+      "loss": 1.0534,
+      "step": 3414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2433122085555897e-06,
+      "loss": 1.043,
+      "step": 3415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2402294619439456e-06,
+      "loss": 1.0571,
+      "step": 3416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2371482233097213e-06,
+      "loss": 1.0499,
+      "step": 3417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.234068494336556e-06,
+      "loss": 1.0901,
+      "step": 3418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.23099027670725e-06,
+      "loss": 1.0735,
+      "step": 3419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2279135721037907e-06,
+      "loss": 1.0543,
+      "step": 3420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.224838382207328e-06,
+      "loss": 1.0779,
+      "step": 3421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.221764708698193e-06,
+      "loss": 1.0661,
+      "step": 3422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.218692553255879e-06,
+      "loss": 1.0895,
+      "step": 3423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2156219175590623e-06,
+      "loss": 1.0427,
+      "step": 3424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2125528032855727e-06,
+      "loss": 1.0567,
+      "step": 3425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.209485212112423e-06,
+      "loss": 1.0646,
+      "step": 3426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.206419145715788e-06,
+      "loss": 1.0481,
+      "step": 3427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2033546057710083e-06,
+      "loss": 1.0526,
+      "step": 3428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2002915939525917e-06,
+      "loss": 1.0576,
+      "step": 3429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.197230111934211e-06,
+      "loss": 1.07,
+      "step": 3430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1941701613887027e-06,
+      "loss": 1.0203,
+      "step": 3431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1911117439880673e-06,
+      "loss": 1.0732,
+      "step": 3432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.188054861403467e-06,
+      "loss": 1.046,
+      "step": 3433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1849995153052252e-06,
+      "loss": 1.0942,
+      "step": 3434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1819457073628254e-06,
+      "loss": 1.0523,
+      "step": 3435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1788934392449124e-06,
+      "loss": 1.0341,
+      "step": 3436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1758427126192877e-06,
+      "loss": 1.0748,
+      "step": 3437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.172793529152912e-06,
+      "loss": 1.0434,
+      "step": 3438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1697458905119005e-06,
+      "loss": 1.0474,
+      "step": 3439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1666997983615275e-06,
+      "loss": 1.0415,
+      "step": 3440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1636552543662187e-06,
+      "loss": 1.0911,
+      "step": 3441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.160612260189556e-06,
+      "loss": 1.0773,
+      "step": 3442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.157570817494276e-06,
+      "loss": 1.0205,
+      "step": 3443
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1545309279422656e-06,
+      "loss": 1.064,
+      "step": 3444
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.151492593194562e-06,
+      "loss": 1.0429,
+      "step": 3445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1484558149113544e-06,
+      "loss": 1.0405,
+      "step": 3446
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.145420594751981e-06,
+      "loss": 1.065,
+      "step": 3447
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.142386934374929e-06,
+      "loss": 1.0853,
+      "step": 3448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.139354835437833e-06,
+      "loss": 1.0637,
+      "step": 3449
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.136324299597474e-06,
+      "loss": 1.0488,
+      "step": 3450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.133295328509781e-06,
+      "loss": 1.0721,
+      "step": 3451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.130267923829824e-06,
+      "loss": 1.0321,
+      "step": 3452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.127242087211821e-06,
+      "loss": 1.1138,
+      "step": 3453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1242178203091316e-06,
+      "loss": 1.0695,
+      "step": 3454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.121195124774258e-06,
+      "loss": 1.0597,
+      "step": 3455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1181740022588425e-06,
+      "loss": 1.0544,
+      "step": 3456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.11515445441367e-06,
+      "loss": 1.0796,
+      "step": 3457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.112136482888663e-06,
+      "loss": 1.0327,
+      "step": 3458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1091200893328837e-06,
+      "loss": 1.0454,
+      "step": 3459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1061052753945334e-06,
+      "loss": 1.0352,
+      "step": 3460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1030920427209466e-06,
+      "loss": 1.0371,
+      "step": 3461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1000803929585957e-06,
+      "loss": 1.0442,
+      "step": 3462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.097070327753093e-06,
+      "loss": 1.0738,
+      "step": 3463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0940618487491736e-06,
+      "loss": 1.0879,
+      "step": 3464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.091054957590721e-06,
+      "loss": 1.076,
+      "step": 3465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0880496559207337e-06,
+      "loss": 1.0549,
+      "step": 3466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.085045945381357e-06,
+      "loss": 1.0677,
+      "step": 3467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.082043827613859e-06,
+      "loss": 1.0757,
+      "step": 3468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0790433042586393e-06,
+      "loss": 1.0279,
+      "step": 3469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.076044376955226e-06,
+      "loss": 1.041,
+      "step": 3470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0730470473422737e-06,
+      "loss": 1.0232,
+      "step": 3471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.070051317057567e-06,
+      "loss": 1.087,
+      "step": 3472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0670571877380146e-06,
+      "loss": 1.0873,
+      "step": 3473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0640646610196515e-06,
+      "loss": 1.0492,
+      "step": 3474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.061073738537635e-06,
+      "loss": 1.0944,
+      "step": 3475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.058084421926248e-06,
+      "loss": 1.0484,
+      "step": 3476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.055096712818895e-06,
+      "loss": 1.0507,
+      "step": 3477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.052110612848102e-06,
+      "loss": 1.0642,
+      "step": 3478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0491261236455158e-06,
+      "loss": 1.1001,
+      "step": 3479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0461432468419064e-06,
+      "loss": 1.0771,
+      "step": 3480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.043161984067153e-06,
+      "loss": 1.0686,
+      "step": 3481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.040182336950268e-06,
+      "loss": 1.0564,
+      "step": 3482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0372043071193637e-06,
+      "loss": 1.057,
+      "step": 3483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0342278962016852e-06,
+      "loss": 1.0717,
+      "step": 3484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.031253105823579e-06,
+      "loss": 1.0622,
+      "step": 3485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.028279937610518e-06,
+      "loss": 1.0667,
+      "step": 3486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0253083931870776e-06,
+      "loss": 1.0915,
+      "step": 3487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0223384741769567e-06,
+      "loss": 1.05,
+      "step": 3488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0193701822029556e-06,
+      "loss": 1.0356,
+      "step": 3489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.016403518886997e-06,
+      "loss": 1.0682,
+      "step": 3490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0134384858501e-06,
+      "loss": 1.083,
+      "step": 3491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0104750847124075e-06,
+      "loss": 1.0473,
+      "step": 3492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0075133170931573e-06,
+      "loss": 1.0328,
+      "step": 3493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0045531846107047e-06,
+      "loss": 1.0745,
+      "step": 3494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0015946888825078e-06,
+      "loss": 1.0401,
+      "step": 3495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9986378315251286e-06,
+      "loss": 1.062,
+      "step": 3496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9956826141542367e-06,
+      "loss": 1.0746,
+      "step": 3497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.992729038384604e-06,
+      "loss": 1.0493,
+      "step": 3498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9897771058301053e-06,
+      "loss": 1.07,
+      "step": 3499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9868268181037186e-06,
+      "loss": 1.0714,
+      "step": 3500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.983878176817523e-06,
+      "loss": 1.0547,
+      "step": 3501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9809311835826965e-06,
+      "loss": 1.0266,
+      "step": 3502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9779858400095188e-06,
+      "loss": 1.0642,
+      "step": 3503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9750421477073647e-06,
+      "loss": 1.0227,
+      "step": 3504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9721001082847137e-06,
+      "loss": 1.0742,
+      "step": 3505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9691597233491316e-06,
+      "loss": 1.0705,
+      "step": 3506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9662209945072923e-06,
+      "loss": 1.0641,
+      "step": 3507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.963283923364952e-06,
+      "loss": 1.0539,
+      "step": 3508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9603485115269743e-06,
+      "loss": 1.0588,
+      "step": 3509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9574147605973033e-06,
+      "loss": 1.045,
+      "step": 3510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9544826721789877e-06,
+      "loss": 1.0309,
+      "step": 3511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.951552247874155e-06,
+      "loss": 1.0692,
+      "step": 3512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9486234892840384e-06,
+      "loss": 1.0583,
+      "step": 3513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9456963980089444e-06,
+      "loss": 1.058,
+      "step": 3514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9427709756482827e-06,
+      "loss": 1.0816,
+      "step": 3515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9398472238005426e-06,
+      "loss": 1.043,
+      "step": 3516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9369251440633053e-06,
+      "loss": 1.0801,
+      "step": 3517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.934004738033234e-06,
+      "loss": 1.0372,
+      "step": 3518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.931086007306081e-06,
+      "loss": 1.0612,
+      "step": 3519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9281689534766813e-06,
+      "loss": 1.0326,
+      "step": 3520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9252535781389536e-06,
+      "loss": 1.0425,
+      "step": 3521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9223398828859014e-06,
+      "loss": 1.0888,
+      "step": 3522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.919427869309607e-06,
+      "loss": 1.0874,
+      "step": 3523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.916517539001237e-06,
+      "loss": 1.0648,
+      "step": 3524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.913608893551036e-06,
+      "loss": 1.0474,
+      "step": 3525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.910701934548329e-06,
+      "loss": 1.0291,
+      "step": 3526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.90779666358152e-06,
+      "loss": 1.0735,
+      "step": 3527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9048930822380885e-06,
+      "loss": 1.0613,
+      "step": 3528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9019911921045908e-06,
+      "loss": 1.0555,
+      "step": 3529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.899090994766667e-06,
+      "loss": 1.0607,
+      "step": 3530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.896192491809017e-06,
+      "loss": 1.0733,
+      "step": 3531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.893295684815431e-06,
+      "loss": 1.0799,
+      "step": 3532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8904005753687594e-06,
+      "loss": 1.0529,
+      "step": 3533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8875071650509342e-06,
+      "loss": 1.0564,
+      "step": 3534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8846154554429546e-06,
+      "loss": 1.07,
+      "step": 3535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8817254481248909e-06,
+      "loss": 1.0555,
+      "step": 3536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8788371446758846e-06,
+      "loss": 1.0284,
+      "step": 3537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8759505466741445e-06,
+      "loss": 1.0523,
+      "step": 3538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.873065655696949e-06,
+      "loss": 1.1085,
+      "step": 3539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.870182473320642e-06,
+      "loss": 1.0416,
+      "step": 3540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8673010011206362e-06,
+      "loss": 1.0446,
+      "step": 3541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8644212406714073e-06,
+      "loss": 1.043,
+      "step": 3542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8615431935464984e-06,
+      "loss": 1.0911,
+      "step": 3543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8586668613185132e-06,
+      "loss": 1.044,
+      "step": 3544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8557922455591216e-06,
+      "loss": 1.0618,
+      "step": 3545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.852919347839053e-06,
+      "loss": 1.0482,
+      "step": 3546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8500481697280991e-06,
+      "loss": 1.0693,
+      "step": 3547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8471787127951123e-06,
+      "loss": 1.0622,
+      "step": 3548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.844310978608005e-06,
+      "loss": 1.0459,
+      "step": 3549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8414449687337467e-06,
+      "loss": 1.0389,
+      "step": 3550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8385806847383646e-06,
+      "loss": 1.0326,
+      "step": 3551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8357181281869456e-06,
+      "loss": 1.0705,
+      "step": 3552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.832857300643628e-06,
+      "loss": 1.0728,
+      "step": 3553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.829998203671613e-06,
+      "loss": 1.0564,
+      "step": 3554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8271408388331458e-06,
+      "loss": 1.0179,
+      "step": 3555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8242852076895344e-06,
+      "loss": 1.0521,
+      "step": 3556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8214313118011356e-06,
+      "loss": 1.0663,
+      "step": 3557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8185791527273572e-06,
+      "loss": 1.109,
+      "step": 3558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8157287320266592e-06,
+      "loss": 1.0652,
+      "step": 3559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8128800512565514e-06,
+      "loss": 1.0746,
+      "step": 3560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8100331119735936e-06,
+      "loss": 1.0367,
+      "step": 3561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8071879157333932e-06,
+      "loss": 1.0685,
+      "step": 3562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.804344464090605e-06,
+      "loss": 1.0673,
+      "step": 3563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8015027585989314e-06,
+      "loss": 1.0374,
+      "step": 3564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.79866280081112e-06,
+      "loss": 1.0379,
+      "step": 3565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.795824592278964e-06,
+      "loss": 1.0616,
+      "step": 3566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7929881345533e-06,
+      "loss": 1.0941,
+      "step": 3567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.790153429184009e-06,
+      "loss": 1.0649,
+      "step": 3568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7873204777200132e-06,
+      "loss": 1.0185,
+      "step": 3569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7844892817092757e-06,
+      "loss": 1.0443,
+      "step": 3570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7816598426988069e-06,
+      "loss": 1.0535,
+      "step": 3571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7788321622346455e-06,
+      "loss": 1.08,
+      "step": 3572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7760062418618824e-06,
+      "loss": 1.0605,
+      "step": 3573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7731820831246328e-06,
+      "loss": 1.047,
+      "step": 3574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7703596875660645e-06,
+      "loss": 1.057,
+      "step": 3575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7675390567283673e-06,
+      "loss": 1.0236,
+      "step": 3576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7647201921527802e-06,
+      "loss": 1.0737,
+      "step": 3577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.761903095379563e-06,
+      "loss": 1.0534,
+      "step": 3578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590877679480238e-06,
+      "loss": 1.0347,
+      "step": 3579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562742113964914e-06,
+      "loss": 1.0788,
+      "step": 3580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753462427262339e-06,
+      "loss": 1.0394,
+      "step": 3581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7506524170819577e-06,
+      "loss": 1.0452,
+      "step": 3582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.747844182390782e-06,
+      "loss": 1.0404,
+      "step": 3583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7450377247232692e-06,
+      "loss": 1.0726,
+      "step": 3584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7422330456129066e-06,
+      "loss": 1.0559,
+      "step": 3585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7394301465922115e-06,
+      "loss": 1.0692,
+      "step": 3586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7366290291927262e-06,
+      "loss": 1.0549,
+      "step": 3587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7338296949450213e-06,
+      "loss": 1.0772,
+      "step": 3588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7310321453786922e-06,
+      "loss": 1.0664,
+      "step": 3589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7282363820223592e-06,
+      "loss": 1.0928,
+      "step": 3590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7254424064036645e-06,
+      "loss": 1.1071,
+      "step": 3591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7226502200492824e-06,
+      "loss": 1.0398,
+      "step": 3592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7198598244848942e-06,
+      "loss": 1.0492,
+      "step": 3593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7170712212352187e-06,
+      "loss": 1.046,
+      "step": 3594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7142844118239815e-06,
+      "loss": 1.0244,
+      "step": 3595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7114993977739409e-06,
+      "loss": 1.0554,
+      "step": 3596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7087161806068615e-06,
+      "loss": 1.0422,
+      "step": 3597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.705934761843538e-06,
+      "loss": 1.0425,
+      "step": 3598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7031551430037702e-06,
+      "loss": 1.0696,
+      "step": 3599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7003773256063882e-06,
+      "loss": 1.0314,
+      "step": 3600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.697601311169224e-06,
+      "loss": 1.0352,
+      "step": 3601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6948271012091367e-06,
+      "loss": 1.0521,
+      "step": 3602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6920546972419877e-06,
+      "loss": 1.0589,
+      "step": 3603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6892841007826637e-06,
+      "loss": 1.0116,
+      "step": 3604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6865153133450512e-06,
+      "loss": 1.0353,
+      "step": 3605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6837483364420592e-06,
+      "loss": 1.0415,
+      "step": 3606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6809831715856019e-06,
+      "loss": 1.0341,
+      "step": 3607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.678219820286603e-06,
+      "loss": 1.0571,
+      "step": 3608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6754582840549976e-06,
+      "loss": 1.0648,
+      "step": 3609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6726985643997273e-06,
+      "loss": 1.0253,
+      "step": 3610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6699406628287423e-06,
+      "loss": 1.0795,
+      "step": 3611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6671845808489983e-06,
+      "loss": 1.0176,
+      "step": 3612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.664430319966458e-06,
+      "loss": 1.0913,
+      "step": 3613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6616778816860868e-06,
+      "loss": 1.0332,
+      "step": 3614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6589272675118567e-06,
+      "loss": 1.0584,
+      "step": 3615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6561784789467423e-06,
+      "loss": 1.0649,
+      "step": 3616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6534315174927195e-06,
+      "loss": 1.0684,
+      "step": 3617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6506863846507653e-06,
+      "loss": 1.0697,
+      "step": 3618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6479430819208642e-06,
+      "loss": 1.0489,
+      "step": 3619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.645201610801988e-06,
+      "loss": 1.0577,
+      "step": 3620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.642461972792122e-06,
+      "loss": 1.0793,
+      "step": 3621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6397241693882372e-06,
+      "loss": 1.0563,
+      "step": 3622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6369882020863132e-06,
+      "loss": 1.0593,
+      "step": 3623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6342540723813143e-06,
+      "loss": 1.0476,
+      "step": 3624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6315217817672142e-06,
+      "loss": 1.045,
+      "step": 3625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6287913317369714e-06,
+      "loss": 1.0541,
+      "step": 3626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6260627237825428e-06,
+      "loss": 1.0442,
+      "step": 3627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6233359593948777e-06,
+      "loss": 1.047,
+      "step": 3628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6206110400639197e-06,
+      "loss": 1.0354,
+      "step": 3629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.617887967278602e-06,
+      "loss": 1.0585,
+      "step": 3630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6151667425268503e-06,
+      "loss": 1.0207,
+      "step": 3631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6124473672955798e-06,
+      "loss": 1.0194,
+      "step": 3632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6097298430706954e-06,
+      "loss": 1.0272,
+      "step": 3633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6070141713370906e-06,
+      "loss": 1.0823,
+      "step": 3634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6043003535786471e-06,
+      "loss": 1.0093,
+      "step": 3635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.601588391278232e-06,
+      "loss": 1.0629,
+      "step": 3636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.598878285917701e-06,
+      "loss": 1.1221,
+      "step": 3637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5961700389778923e-06,
+      "loss": 1.0642,
+      "step": 3638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.593463651938631e-06,
+      "loss": 1.0458,
+      "step": 3639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5907591262787247e-06,
+      "loss": 1.0335,
+      "step": 3640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5880564634759648e-06,
+      "loss": 1.007,
+      "step": 3641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5853556650071239e-06,
+      "loss": 1.0609,
+      "step": 3642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.582656732347956e-06,
+      "loss": 1.0638,
+      "step": 3643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5799596669731942e-06,
+      "loss": 1.0627,
+      "step": 3644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5772644703565564e-06,
+      "loss": 1.0402,
+      "step": 3645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5745711439707344e-06,
+      "loss": 1.0647,
+      "step": 3646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5718796892874e-06,
+      "loss": 1.0449,
+      "step": 3647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.569190107777201e-06,
+      "loss": 1.0525,
+      "step": 3648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.566502400909763e-06,
+      "loss": 1.0598,
+      "step": 3649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5638165701536866e-06,
+      "loss": 1.0429,
+      "step": 3650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5611326169765478e-06,
+      "loss": 1.0554,
+      "step": 3651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5584505428448965e-06,
+      "loss": 1.0494,
+      "step": 3652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5557703492242548e-06,
+      "loss": 1.0248,
+      "step": 3653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5530920375791192e-06,
+      "loss": 1.0484,
+      "step": 3654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5504156093729566e-06,
+      "loss": 1.0749,
+      "step": 3655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5477410660682047e-06,
+      "loss": 1.0081,
+      "step": 3656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5450684091262713e-06,
+      "loss": 1.0774,
+      "step": 3657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5423976400075347e-06,
+      "loss": 1.0406,
+      "step": 3658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5397287601713401e-06,
+      "loss": 1.0806,
+      "step": 3659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5370617710760004e-06,
+      "loss": 1.0323,
+      "step": 3660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5343966741787958e-06,
+      "loss": 1.0835,
+      "step": 3661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.531733470935976e-06,
+      "loss": 1.0172,
+      "step": 3662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5290721628027472e-06,
+      "loss": 1.0551,
+      "step": 3663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.526412751233292e-06,
+      "loss": 1.0523,
+      "step": 3664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5237552376807435e-06,
+      "loss": 1.0165,
+      "step": 3665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5210996235972103e-06,
+      "loss": 1.0365,
+      "step": 3666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5184459104337523e-06,
+      "loss": 1.0522,
+      "step": 3667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5157940996404008e-06,
+      "loss": 1.0641,
+      "step": 3668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5131441926661372e-06,
+      "loss": 1.0879,
+      "step": 3669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5104961909589117e-06,
+      "loss": 1.0262,
+      "step": 3670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.507850095965629e-06,
+      "loss": 1.0637,
+      "step": 3671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5052059091321513e-06,
+      "loss": 1.0548,
+      "step": 3672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.502563631903301e-06,
+      "loss": 1.0184,
+      "step": 3673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4999232657228546e-06,
+      "loss": 1.0659,
+      "step": 3674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4972848120335453e-06,
+      "loss": 1.0259,
+      "step": 3675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4946482722770622e-06,
+      "loss": 1.0566,
+      "step": 3676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4920136478940466e-06,
+      "loss": 1.0902,
+      "step": 3677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4893809403240945e-06,
+      "loss": 1.0933,
+      "step": 3678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4867501510057548e-06,
+      "loss": 1.0711,
+      "step": 3679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4841212813765255e-06,
+      "loss": 1.0418,
+      "step": 3680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4814943328728638e-06,
+      "loss": 1.076,
+      "step": 3681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4788693069301635e-06,
+      "loss": 1.0343,
+      "step": 3682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.476246204982783e-06,
+      "loss": 1.0639,
+      "step": 3683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4736250284640158e-06,
+      "loss": 1.0478,
+      "step": 3684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4710057788061155e-06,
+      "loss": 1.0391,
+      "step": 3685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4683884574402712e-06,
+      "loss": 1.0621,
+      "step": 3686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4657730657966296e-06,
+      "loss": 1.0415,
+      "step": 3687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4631596053042713e-06,
+      "loss": 1.0257,
+      "step": 3688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4605480773912344e-06,
+      "loss": 1.0486,
+      "step": 3689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4579384834844889e-06,
+      "loss": 1.0313,
+      "step": 3690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4553308250099585e-06,
+      "loss": 1.0388,
+      "step": 3691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4527251033924987e-06,
+      "loss": 1.077,
+      "step": 3692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4501213200559194e-06,
+      "loss": 1.0475,
+      "step": 3693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4475194764229577e-06,
+      "loss": 1.0818,
+      "step": 3694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4449195739153033e-06,
+      "loss": 1.0683,
+      "step": 3695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4423216139535735e-06,
+      "loss": 1.0827,
+      "step": 3696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4397255979573354e-06,
+      "loss": 1.0196,
+      "step": 3697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4371315273450864e-06,
+      "loss": 1.068,
+      "step": 3698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.434539403534263e-06,
+      "loss": 1.0333,
+      "step": 3699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4319492279412388e-06,
+      "loss": 1.0856,
+      "step": 3700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.429361001981322e-06,
+      "loss": 1.0582,
+      "step": 3701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4267747270687553e-06,
+      "loss": 1.0396,
+      "step": 3702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4241904046167166e-06,
+      "loss": 1.0486,
+      "step": 3703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4216080360373148e-06,
+      "loss": 1.0606,
+      "step": 3704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4190276227415923e-06,
+      "loss": 1.0185,
+      "step": 3705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4164491661395247e-06,
+      "loss": 1.0602,
+      "step": 3706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4138726676400138e-06,
+      "loss": 1.0445,
+      "step": 3707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4112981286509003e-06,
+      "loss": 1.0248,
+      "step": 3708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.408725550578942e-06,
+      "loss": 1.0574,
+      "step": 3709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4061549348298375e-06,
+      "loss": 1.0736,
+      "step": 3710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.403586282808202e-06,
+      "loss": 1.0227,
+      "step": 3711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4010195959175887e-06,
+      "loss": 1.0766,
+      "step": 3712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3984548755604655e-06,
+      "loss": 1.035,
+      "step": 3713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3958921231382372e-06,
+      "loss": 1.0424,
+      "step": 3714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3933313400512227e-06,
+      "loss": 1.0464,
+      "step": 3715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3907725276986733e-06,
+      "loss": 1.0511,
+      "step": 3716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3882156874787595e-06,
+      "loss": 1.0169,
+      "step": 3717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3856608207885746e-06,
+      "loss": 1.0438,
+      "step": 3718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3831079290241328e-06,
+      "loss": 1.0328,
+      "step": 3719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3805570135803709e-06,
+      "loss": 1.0285,
+      "step": 3720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.378008075851145e-06,
+      "loss": 1.0487,
+      "step": 3721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3754611172292299e-06,
+      "loss": 1.0597,
+      "step": 3722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.37291613910632e-06,
+      "loss": 1.0755,
+      "step": 3723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3703731428730271e-06,
+      "loss": 1.0486,
+      "step": 3724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3678321299188802e-06,
+      "loss": 1.04,
+      "step": 3725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.365293101632324e-06,
+      "loss": 1.1012,
+      "step": 3726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.36275605940072e-06,
+      "loss": 1.0584,
+      "step": 3727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3602210046103432e-06,
+      "loss": 1.0124,
+      "step": 3728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3576879386463831e-06,
+      "loss": 1.0366,
+      "step": 3729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3551568628929434e-06,
+      "loss": 1.0414,
+      "step": 3730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3526277787330388e-06,
+      "loss": 1.0782,
+      "step": 3731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3501006875485945e-06,
+      "loss": 1.0556,
+      "step": 3732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3475755907204523e-06,
+      "loss": 1.0717,
+      "step": 3733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3450524896283595e-06,
+      "loss": 1.0409,
+      "step": 3734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.342531385650973e-06,
+      "loss": 1.0813,
+      "step": 3735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3400122801658604e-06,
+      "loss": 1.0804,
+      "step": 3736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3374951745494957e-06,
+      "loss": 1.0352,
+      "step": 3737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3349800701772608e-06,
+      "loss": 1.0835,
+      "step": 3738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3324669684234442e-06,
+      "loss": 1.0476,
+      "step": 3739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.329955870661241e-06,
+      "loss": 1.0547,
+      "step": 3740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3274467782627486e-06,
+      "loss": 1.037,
+      "step": 3741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3249396925989715e-06,
+      "loss": 1.0612,
+      "step": 3742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.322434615039816e-06,
+      "loss": 1.0371,
+      "step": 3743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3199315469540923e-06,
+      "loss": 1.0712,
+      "step": 3744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3174304897095115e-06,
+      "loss": 1.0364,
+      "step": 3745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.314931444672687e-06,
+      "loss": 1.0485,
+      "step": 3746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.312434413209131e-06,
+      "loss": 1.0389,
+      "step": 3747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.309939396683258e-06,
+      "loss": 1.0325,
+      "step": 3748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3074463964583794e-06,
+      "loss": 1.0602,
+      "step": 3749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3049554138967052e-06,
+      "loss": 1.0322,
+      "step": 3750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3024664503593443e-06,
+      "loss": 1.0675,
+      "step": 3751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2999795072062988e-06,
+      "loss": 1.0361,
+      "step": 3752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2974945857964744e-06,
+      "loss": 1.0398,
+      "step": 3753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2950116874876617e-06,
+      "loss": 1.0752,
+      "step": 3754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.292530813636556e-06,
+      "loss": 1.0493,
+      "step": 3755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2900519655987365e-06,
+      "loss": 1.0435,
+      "step": 3756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2875751447286866e-06,
+      "loss": 1.0425,
+      "step": 3757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2851003523797695e-06,
+      "loss": 1.0489,
+      "step": 3758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2826275899042518e-06,
+      "loss": 1.036,
+      "step": 3759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2801568586532837e-06,
+      "loss": 1.0253,
+      "step": 3760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2776881599769074e-06,
+      "loss": 1.0257,
+      "step": 3761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2752214952240549e-06,
+      "loss": 1.045,
+      "step": 3762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.272756865742546e-06,
+      "loss": 1.0561,
+      "step": 3763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2702942728790897e-06,
+      "loss": 1.0543,
+      "step": 3764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2678337179792804e-06,
+      "loss": 1.0471,
+      "step": 3765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2653752023876008e-06,
+      "loss": 1.0411,
+      "step": 3766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2629187274474181e-06,
+      "loss": 1.0289,
+      "step": 3767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2604642945009843e-06,
+      "loss": 1.0558,
+      "step": 3768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2580119048894363e-06,
+      "loss": 1.0324,
+      "step": 3769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2555615599527948e-06,
+      "loss": 1.0647,
+      "step": 3770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.25311326102996e-06,
+      "loss": 1.0523,
+      "step": 3771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2506670094587232e-06,
+      "loss": 1.0474,
+      "step": 3772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2482228065757428e-06,
+      "loss": 1.0617,
+      "step": 3773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.245780653716573e-06,
+      "loss": 1.0424,
+      "step": 3774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2433405522156334e-06,
+      "loss": 1.0548,
+      "step": 3775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.240902503406236e-06,
+      "loss": 1.0423,
+      "step": 3776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2384665086205593e-06,
+      "loss": 1.0364,
+      "step": 3777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2360325691896712e-06,
+      "loss": 1.0695,
+      "step": 3778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2336006864435046e-06,
+      "loss": 1.0842,
+      "step": 3779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2311708617108792e-06,
+      "loss": 1.0929,
+      "step": 3780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2287430963194807e-06,
+      "loss": 1.0392,
+      "step": 3781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2263173915958797e-06,
+      "loss": 1.0748,
+      "step": 3782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2238937488655095e-06,
+      "loss": 1.0233,
+      "step": 3783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2214721694526887e-06,
+      "loss": 1.0422,
+      "step": 3784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.219052654680596e-06,
+      "loss": 1.0663,
+      "step": 3785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2166352058712927e-06,
+      "loss": 1.0952,
+      "step": 3786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2142198243457054e-06,
+      "loss": 1.0438,
+      "step": 3787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2118065114236327e-06,
+      "loss": 1.0334,
+      "step": 3788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.209395268423743e-06,
+      "loss": 1.0753,
+      "step": 3789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2069860966635732e-06,
+      "loss": 1.0344,
+      "step": 3790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2045789974595279e-06,
+      "loss": 1.071,
+      "step": 3791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.202173972126881e-06,
+      "loss": 1.046,
+      "step": 3792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1997710219797714e-06,
+      "loss": 1.0649,
+      "step": 3793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.197370148331204e-06,
+      "loss": 1.0497,
+      "step": 3794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1949713524930535e-06,
+      "loss": 1.0518,
+      "step": 3795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1925746357760503e-06,
+      "loss": 1.0362,
+      "step": 3796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1901799994898006e-06,
+      "loss": 1.0572,
+      "step": 3797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.18778744494276e-06,
+      "loss": 1.0166,
+      "step": 3798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.185396973442261e-06,
+      "loss": 1.0635,
+      "step": 3799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1830085862944851e-06,
+      "loss": 1.0727,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1806222848044851e-06,
+      "loss": 1.0565,
+      "step": 3801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1782380702761654e-06,
+      "loss": 1.078,
+      "step": 3802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.175855944012299e-06,
+      "loss": 1.0539,
+      "step": 3803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.173475907314508e-06,
+      "loss": 1.0774,
+      "step": 3804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.171097961483284e-06,
+      "loss": 1.0301,
+      "step": 3805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1687221078179634e-06,
+      "loss": 1.0692,
+      "step": 3806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1663483476167508e-06,
+      "loss": 1.0626,
+      "step": 3807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1639766821767007e-06,
+      "loss": 1.0502,
+      "step": 3808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1616071127937245e-06,
+      "loss": 1.0308,
+      "step": 3809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1592396407625883e-06,
+      "loss": 1.095,
+      "step": 3810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1568742673769117e-06,
+      "loss": 1.0811,
+      "step": 3811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1545109939291688e-06,
+      "loss": 1.0413,
+      "step": 3812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.152149821710684e-06,
+      "loss": 1.0314,
+      "step": 3813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1497907520116364e-06,
+      "loss": 1.026,
+      "step": 3814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1474337861210543e-06,
+      "loss": 1.0359,
+      "step": 3815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1450789253268174e-06,
+      "loss": 1.0332,
+      "step": 3816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.142726170915654e-06,
+      "loss": 1.0353,
+      "step": 3817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1403755241731423e-06,
+      "loss": 1.0696,
+      "step": 3818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1380269863837095e-06,
+      "loss": 1.0721,
+      "step": 3819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.135680558830629e-06,
+      "loss": 1.0752,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1333362427960204e-06,
+      "loss": 1.0604,
+      "step": 3821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1309940395608554e-06,
+      "loss": 1.0109,
+      "step": 3822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1286539504049416e-06,
+      "loss": 1.0587,
+      "step": 3823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1263159766069398e-06,
+      "loss": 1.0311,
+      "step": 3824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1239801194443507e-06,
+      "loss": 1.0343,
+      "step": 3825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1216463801935202e-06,
+      "loss": 1.0312,
+      "step": 3826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.119314760129635e-06,
+      "loss": 1.0473,
+      "step": 3827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1169852605267261e-06,
+      "loss": 1.0617,
+      "step": 3828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1146578826576648e-06,
+      "loss": 1.0426,
+      "step": 3829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1123326277941626e-06,
+      "loss": 1.0696,
+      "step": 3830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1100094972067715e-06,
+      "loss": 1.0573,
+      "step": 3831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1076884921648834e-06,
+      "loss": 1.0354,
+      "step": 3832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.105369613936727e-06,
+      "loss": 1.0505,
+      "step": 3833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1030528637893706e-06,
+      "loss": 1.0316,
+      "step": 3834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1007382429887192e-06,
+      "loss": 1.0639,
+      "step": 3835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.098425752799514e-06,
+      "loss": 1.063,
+      "step": 3836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0961153944853314e-06,
+      "loss": 1.0332,
+      "step": 3837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.093807169308585e-06,
+      "loss": 1.0211,
+      "step": 3838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0915010785305208e-06,
+      "loss": 1.046,
+      "step": 3839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0891971234112192e-06,
+      "loss": 1.0381,
+      "step": 3840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0868953052095938e-06,
+      "loss": 1.0282,
+      "step": 3841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.08459562518339e-06,
+      "loss": 1.0276,
+      "step": 3842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0822980845891845e-06,
+      "loss": 1.0598,
+      "step": 3843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0800026846823897e-06,
+      "loss": 1.045,
+      "step": 3844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0777094267172389e-06,
+      "loss": 1.0238,
+      "step": 3845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0754183119468053e-06,
+      "loss": 1.0513,
+      "step": 3846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0731293416229821e-06,
+      "loss": 1.0508,
+      "step": 3847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.070842516996498e-06,
+      "loss": 1.0516,
+      "step": 3848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0685578393169054e-06,
+      "loss": 1.0201,
+      "step": 3849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.066275309832584e-06,
+      "loss": 1.0034,
+      "step": 3850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0639949297907392e-06,
+      "loss": 1.0701,
+      "step": 3851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.061716700437404e-06,
+      "loss": 1.0343,
+      "step": 3852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0594406230174343e-06,
+      "loss": 1.0218,
+      "step": 3853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.057166698774511e-06,
+      "loss": 1.0432,
+      "step": 3854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0548949289511385e-06,
+      "loss": 1.0363,
+      "step": 3855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0526253147886427e-06,
+      "loss": 1.0609,
+      "step": 3856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0503578575271739e-06,
+      "loss": 1.0481,
+      "step": 3857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0480925584057022e-06,
+      "loss": 1.0607,
+      "step": 3858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0458294186620183e-06,
+      "loss": 1.0525,
+      "step": 3859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.043568439532735e-06,
+      "loss": 1.0325,
+      "step": 3860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0413096222532815e-06,
+      "loss": 1.0678,
+      "step": 3861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.039052968057907e-06,
+      "loss": 1.0609,
+      "step": 3862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.036798478179683e-06,
+      "loss": 1.0721,
+      "step": 3863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0345461538504887e-06,
+      "loss": 1.0409,
+      "step": 3864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.032295996301031e-06,
+      "loss": 1.0644,
+      "step": 3865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0300480067608232e-06,
+      "loss": 1.0616,
+      "step": 3866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0278021864582027e-06,
+      "loss": 1.0589,
+      "step": 3867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0255585366203124e-06,
+      "loss": 1.0572,
+      "step": 3868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.02331705847312e-06,
+      "loss": 1.0102,
+      "step": 3869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.021077753241394e-06,
+      "loss": 1.0463,
+      "step": 3870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.018840622148729e-06,
+      "loss": 1.061,
+      "step": 3871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.016605666417519e-06,
+      "loss": 1.0418,
+      "step": 3872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0143728872689785e-06,
+      "loss": 1.0836,
+      "step": 3873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0121422859231295e-06,
+      "loss": 1.0542,
+      "step": 3874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0099138635988026e-06,
+      "loss": 1.0476,
+      "step": 3875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.00768762151364e-06,
+      "loss": 1.0318,
+      "step": 3876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0054635608840912e-06,
+      "loss": 1.0512,
+      "step": 3877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0032416829254144e-06,
+      "loss": 1.027,
+      "step": 3878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0010219888516753e-06,
+      "loss": 1.065,
+      "step": 3879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.988044798757452e-07,
+      "loss": 1.0278,
+      "step": 3880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.965891572093023e-07,
+      "loss": 1.0296,
+      "step": 3881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.943760220628306e-07,
+      "loss": 1.0374,
+      "step": 3882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.921650756456164e-07,
+      "loss": 1.0761,
+      "step": 3883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.899563191657564e-07,
+      "loss": 1.0357,
+      "step": 3884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.877497538301406e-07,
+      "loss": 1.0331,
+      "step": 3885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.855453808444726e-07,
+      "loss": 1.0752,
+      "step": 3886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.833432014132471e-07,
+      "loss": 1.0858,
+      "step": 3887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.811432167397722e-07,
+      "loss": 1.0545,
+      "step": 3888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.78945428026144e-07,
+      "loss": 1.0308,
+      "step": 3889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.767498364732714e-07,
+      "loss": 1.0563,
+      "step": 3890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.745564432808507e-07,
+      "loss": 1.0575,
+      "step": 3891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.723652496473884e-07,
+      "loss": 1.0505,
+      "step": 3892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.701762567701777e-07,
+      "loss": 1.0324,
+      "step": 3893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.67989465845321e-07,
+      "loss": 1.03,
+      "step": 3894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.65804878067706e-07,
+      "loss": 1.0798,
+      "step": 3895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.636224946310275e-07,
+      "loss": 1.0497,
+      "step": 3896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.614423167277648e-07,
+      "loss": 1.0222,
+      "step": 3897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.592643455492024e-07,
+      "loss": 1.0817,
+      "step": 3898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.570885822854126e-07,
+      "loss": 1.0366,
+      "step": 3899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 1.0791,
+      "step": 3900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.527436842564147e-07,
+      "loss": 1.0628,
+      "step": 3901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.505745518653203e-07,
+      "loss": 1.0911,
+      "step": 3902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.484076321372237e-07,
+      "loss": 1.0728,
+      "step": 3903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.462429262561612e-07,
+      "loss": 1.0364,
+      "step": 3904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.440804354049576e-07,
+      "loss": 1.0533,
+      "step": 3905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.419201607652289e-07,
+      "loss": 1.0484,
+      "step": 3906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.397621035173793e-07,
+      "loss": 1.0714,
+      "step": 3907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.376062648406014e-07,
+      "loss": 1.0225,
+      "step": 3908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.354526459128766e-07,
+      "loss": 1.0446,
+      "step": 3909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.333012479109704e-07,
+      "loss": 1.0626,
+      "step": 3910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.311520720104411e-07,
+      "loss": 1.0601,
+      "step": 3911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.290051193856237e-07,
+      "loss": 1.0321,
+      "step": 3912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.26860391209648e-07,
+      "loss": 1.049,
+      "step": 3913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.247178886544189e-07,
+      "loss": 1.0848,
+      "step": 3914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.225776128906338e-07,
+      "loss": 1.0193,
+      "step": 3915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.204395650877679e-07,
+      "loss": 1.0651,
+      "step": 3916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.183037464140804e-07,
+      "loss": 1.0213,
+      "step": 3917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.161701580366128e-07,
+      "loss": 1.0441,
+      "step": 3918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.140388011211882e-07,
+      "loss": 1.0671,
+      "step": 3919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.119096768324098e-07,
+      "loss": 1.0625,
+      "step": 3920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.097827863336606e-07,
+      "loss": 1.0452,
+      "step": 3921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.07658130787104e-07,
+      "loss": 1.0511,
+      "step": 3922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.055357113536811e-07,
+      "loss": 1.039,
+      "step": 3923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.034155291931124e-07,
+      "loss": 1.0537,
+      "step": 3924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.01297585463895e-07,
+      "loss": 1.0292,
+      "step": 3925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.991818813233028e-07,
+      "loss": 0.9921,
+      "step": 3926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.97068417927387e-07,
+      "loss": 1.0491,
+      "step": 3927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.949571964309728e-07,
+      "loss": 1.0613,
+      "step": 3928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.928482179876619e-07,
+      "loss": 1.0089,
+      "step": 3929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.907414837498296e-07,
+      "loss": 1.0324,
+      "step": 3930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.886369948686252e-07,
+      "loss": 1.0852,
+      "step": 3931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.86534752493971e-07,
+      "loss": 1.0468,
+      "step": 3932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.844347577745616e-07,
+      "loss": 1.0561,
+      "step": 3933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.823370118578628e-07,
+      "loss": 1.0358,
+      "step": 3934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.802415158901145e-07,
+      "loss": 1.0323,
+      "step": 3935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.781482710163241e-07,
+      "loss": 1.0318,
+      "step": 3936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.760572783802701e-07,
+      "loss": 1.0115,
+      "step": 3937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.739685391245001e-07,
+      "loss": 1.076,
+      "step": 3938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.718820543903306e-07,
+      "loss": 1.063,
+      "step": 3939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.697978253178469e-07,
+      "loss": 1.0403,
+      "step": 3940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.677158530459001e-07,
+      "loss": 1.0457,
+      "step": 3941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.656361387121093e-07,
+      "loss": 1.019,
+      "step": 3942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.635586834528597e-07,
+      "loss": 1.0527,
+      "step": 3943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.614834884033029e-07,
+      "loss": 1.0254,
+      "step": 3944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.594105546973536e-07,
+      "loss": 1.0397,
+      "step": 3945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.573398834676922e-07,
+      "loss": 1.0477,
+      "step": 3946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.552714758457626e-07,
+      "loss": 1.0116,
+      "step": 3947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.532053329617724e-07,
+      "loss": 1.0871,
+      "step": 3948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.511414559446901e-07,
+      "loss": 1.0196,
+      "step": 3949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.490798459222477e-07,
+      "loss": 1.049,
+      "step": 3950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.470205040209362e-07,
+      "loss": 1.0384,
+      "step": 3951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.449634313660132e-07,
+      "loss": 1.0417,
+      "step": 3952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.429086290814875e-07,
+      "loss": 1.0794,
+      "step": 3953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.40856098290136e-07,
+      "loss": 1.0391,
+      "step": 3954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.388058401134857e-07,
+      "loss": 1.0375,
+      "step": 3955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.36757855671832e-07,
+      "loss": 1.0259,
+      "step": 3956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.347121460842172e-07,
+      "loss": 1.0274,
+      "step": 3957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.326687124684502e-07,
+      "loss": 1.0673,
+      "step": 3958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.30627555941087e-07,
+      "loss": 1.0186,
+      "step": 3959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.285886776174501e-07,
+      "loss": 1.0709,
+      "step": 3960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.265520786116055e-07,
+      "loss": 1.0294,
+      "step": 3961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.24517760036383e-07,
+      "loss": 1.0136,
+      "step": 3962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.224857230033628e-07,
+      "loss": 1.0802,
+      "step": 3963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.204559686228769e-07,
+      "loss": 1.0331,
+      "step": 3964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.18428498004013e-07,
+      "loss": 1.0751,
+      "step": 3965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.164033122546094e-07,
+      "loss": 1.0493,
+      "step": 3966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.143804124812549e-07,
+      "loss": 1.0222,
+      "step": 3967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.123597997892918e-07,
+      "loss": 1.0202,
+      "step": 3968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.103414752828104e-07,
+      "loss": 1.0622,
+      "step": 3969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.083254400646507e-07,
+      "loss": 1.0635,
+      "step": 3970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.063116952364065e-07,
+      "loss": 1.0257,
+      "step": 3971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.043002418984108e-07,
+      "loss": 1.0392,
+      "step": 3972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.022910811497558e-07,
+      "loss": 1.056,
+      "step": 3973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.002842140882694e-07,
+      "loss": 1.0469,
+      "step": 3974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.98279641810537e-07,
+      "loss": 1.0252,
+      "step": 3975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.96277365411881e-07,
+      "loss": 1.0374,
+      "step": 3976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.942773859863773e-07,
+      "loss": 1.0363,
+      "step": 3977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.922797046268382e-07,
+      "loss": 1.0554,
+      "step": 3978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.902843224248302e-07,
+      "loss": 1.0111,
+      "step": 3979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.882912404706527e-07,
+      "loss": 1.0637,
+      "step": 3980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.863004598533586e-07,
+      "loss": 1.0396,
+      "step": 3981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.843119816607342e-07,
+      "loss": 1.0567,
+      "step": 3982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.82325806979316e-07,
+      "loss": 1.0557,
+      "step": 3983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.803419368943738e-07,
+      "loss": 1.0075,
+      "step": 3984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.783603724899258e-07,
+      "loss": 1.0673,
+      "step": 3985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.763811148487227e-07,
+      "loss": 1.0325,
+      "step": 3986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.744041650522622e-07,
+      "loss": 1.0466,
+      "step": 3987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.724295241807761e-07,
+      "loss": 1.1083,
+      "step": 3988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.70457193313236e-07,
+      "loss": 1.0501,
+      "step": 3989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.684871735273514e-07,
+      "loss": 1.0479,
+      "step": 3990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.665194658995683e-07,
+      "loss": 1.0173,
+      "step": 3991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.645540715050698e-07,
+      "loss": 1.0436,
+      "step": 3992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.62590991417776e-07,
+      "loss": 1.0436,
+      "step": 3993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.606302267103405e-07,
+      "loss": 1.039,
+      "step": 3994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.586717784541531e-07,
+      "loss": 1.0513,
+      "step": 3995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.567156477193371e-07,
+      "loss": 1.0172,
+      "step": 3996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.547618355747493e-07,
+      "loss": 1.0738,
+      "step": 3997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.52810343087983e-07,
+      "loss": 1.0241,
+      "step": 3998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.508611713253566e-07,
+      "loss": 1.0436,
+      "step": 3999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.489143213519301e-07,
+      "loss": 1.0814,
+      "step": 4000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.469697942314846e-07,
+      "loss": 1.0642,
+      "step": 4001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.450275910265415e-07,
+      "loss": 1.0544,
+      "step": 4002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.430877127983438e-07,
+      "loss": 1.0479,
+      "step": 4003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.411501606068727e-07,
+      "loss": 1.0497,
+      "step": 4004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.392149355108297e-07,
+      "loss": 1.0397,
+      "step": 4005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.372820385676516e-07,
+      "loss": 1.0796,
+      "step": 4006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.353514708334997e-07,
+      "loss": 1.0694,
+      "step": 4007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.334232333632635e-07,
+      "loss": 1.0375,
+      "step": 4008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.314973272105591e-07,
+      "loss": 1.0329,
+      "step": 4009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295737534277286e-07,
+      "loss": 1.035,
+      "step": 4010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.276525130658396e-07,
+      "loss": 1.0438,
+      "step": 4011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.257336071746851e-07,
+      "loss": 1.0131,
+      "step": 4012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.23817036802782e-07,
+      "loss": 1.0334,
+      "step": 4013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.219028029973718e-07,
+      "loss": 1.0029,
+      "step": 4014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.199909068044181e-07,
+      "loss": 1.024,
+      "step": 4015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.180813492686084e-07,
+      "loss": 1.0591,
+      "step": 4016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.161741314333515e-07,
+      "loss": 1.0012,
+      "step": 4017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.142692543407775e-07,
+      "loss": 1.023,
+      "step": 4018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.123667190317396e-07,
+      "loss": 1.0413,
+      "step": 4019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.104665265458083e-07,
+      "loss": 1.0381,
+      "step": 4020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.085686779212758e-07,
+      "loss": 1.0633,
+      "step": 4021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.066731741951543e-07,
+      "loss": 1.0367,
+      "step": 4022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.047800164031726e-07,
+      "loss": 1.0432,
+      "step": 4023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.028892055797775e-07,
+      "loss": 1.0741,
+      "step": 4024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.010007427581378e-07,
+      "loss": 1.0494,
+      "step": 4025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.991146289701345e-07,
+      "loss": 1.0356,
+      "step": 4026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.972308652463666e-07,
+      "loss": 1.0551,
+      "step": 4027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.953494526161486e-07,
+      "loss": 1.0319,
+      "step": 4028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.93470392107512e-07,
+      "loss": 1.0341,
+      "step": 4029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.915936847472016e-07,
+      "loss": 1.0347,
+      "step": 4030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.89719331560676e-07,
+      "loss": 1.0462,
+      "step": 4031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.87847333572108e-07,
+      "loss": 1.0314,
+      "step": 4032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.859776918043848e-07,
+      "loss": 1.0318,
+      "step": 4033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.841104072791038e-07,
+      "loss": 1.0377,
+      "step": 4034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.822454810165763e-07,
+      "loss": 1.025,
+      "step": 4035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.803829140358237e-07,
+      "loss": 1.0593,
+      "step": 4036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.785227073545792e-07,
+      "loss": 1.0512,
+      "step": 4037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.766648619892857e-07,
+      "loss": 1.0764,
+      "step": 4038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.748093789550958e-07,
+      "loss": 1.0107,
+      "step": 4039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.729562592658717e-07,
+      "loss": 1.0571,
+      "step": 4040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.71105503934184e-07,
+      "loss": 1.0185,
+      "step": 4041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.692571139713111e-07,
+      "loss": 1.0578,
+      "step": 4042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.674110903872411e-07,
+      "loss": 1.0451,
+      "step": 4043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.655674341906643e-07,
+      "loss": 1.0532,
+      "step": 4044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.637261463889832e-07,
+      "loss": 1.0277,
+      "step": 4045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.618872279883004e-07,
+      "loss": 1.0353,
+      "step": 4046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.600506799934309e-07,
+      "loss": 1.0733,
+      "step": 4047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.582165034078853e-07,
+      "loss": 1.0255,
+      "step": 4048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.563846992338868e-07,
+      "loss": 1.0629,
+      "step": 4049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.545552684723583e-07,
+      "loss": 1.0203,
+      "step": 4050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.527282121229267e-07,
+      "loss": 1.0443,
+      "step": 4051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.509035311839207e-07,
+      "loss": 1.0707,
+      "step": 4052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.490812266523716e-07,
+      "loss": 1.0347,
+      "step": 4053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.47261299524013e-07,
+      "loss": 1.0417,
+      "step": 4054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.454437507932781e-07,
+      "loss": 1.0239,
+      "step": 4055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.436285814533006e-07,
+      "loss": 1.0753,
+      "step": 4056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.418157924959156e-07,
+      "loss": 1.0514,
+      "step": 4057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.400053849116556e-07,
+      "loss": 1.0839,
+      "step": 4058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.381973596897534e-07,
+      "loss": 1.0514,
+      "step": 4059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.363917178181389e-07,
+      "loss": 1.0526,
+      "step": 4060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.345884602834396e-07,
+      "loss": 0.9857,
+      "step": 4061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.327875880709832e-07,
+      "loss": 1.0419,
+      "step": 4062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.30989102164788e-07,
+      "loss": 1.0496,
+      "step": 4063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.291930035475763e-07,
+      "loss": 1.0419,
+      "step": 4064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.273992932007572e-07,
+      "loss": 1.0379,
+      "step": 4065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.256079721044434e-07,
+      "loss": 1.054,
+      "step": 4066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.23819041237434e-07,
+      "loss": 1.0418,
+      "step": 4067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.220325015772294e-07,
+      "loss": 1.044,
+      "step": 4068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.202483541000165e-07,
+      "loss": 1.0388,
+      "step": 4069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.184665997806832e-07,
+      "loss": 1.0847,
+      "step": 4070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.166872395927998e-07,
+      "loss": 1.0154,
+      "step": 4071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.149102745086382e-07,
+      "loss": 1.054,
+      "step": 4072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.131357054991538e-07,
+      "loss": 1.0231,
+      "step": 4073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.11363533533999e-07,
+      "loss": 1.0795,
+      "step": 4074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.095937595815104e-07,
+      "loss": 1.0702,
+      "step": 4075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.078263846087201e-07,
+      "loss": 1.0461,
+      "step": 4076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.06061409581345e-07,
+      "loss": 1.0318,
+      "step": 4077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.04298835463793e-07,
+      "loss": 1.0684,
+      "step": 4078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.025386632191588e-07,
+      "loss": 1.0345,
+      "step": 4079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.007808938092263e-07,
+      "loss": 1.0199,
+      "step": 4080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.99025528194464e-07,
+      "loss": 1.0804,
+      "step": 4081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.972725673340296e-07,
+      "loss": 1.0619,
+      "step": 4082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.955220121857647e-07,
+      "loss": 1.043,
+      "step": 4083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.937738637061979e-07,
+      "loss": 1.0547,
+      "step": 4084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.920281228505414e-07,
+      "loss": 1.0397,
+      "step": 4085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.902847905726922e-07,
+      "loss": 1.0519,
+      "step": 4086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.885438678252342e-07,
+      "loss": 1.0477,
+      "step": 4087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.868053555594283e-07,
+      "loss": 1.0425,
+      "step": 4088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.850692547252257e-07,
+      "loss": 1.0525,
+      "step": 4089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.833355662712525e-07,
+      "loss": 1.0764,
+      "step": 4090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.816042911448249e-07,
+      "loss": 1.0822,
+      "step": 4091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.798754302919312e-07,
+      "loss": 1.049,
+      "step": 4092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.781489846572496e-07,
+      "loss": 1.0384,
+      "step": 4093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.7642495518413e-07,
+      "loss": 1.0287,
+      "step": 4094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.747033428146115e-07,
+      "loss": 1.0249,
+      "step": 4095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.729841484894016e-07,
+      "loss": 1.0213,
+      "step": 4096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.712673731478962e-07,
+      "loss": 1.0409,
+      "step": 4097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.695530177281644e-07,
+      "loss": 1.0222,
+      "step": 4098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.678410831669545e-07,
+      "loss": 1.041,
+      "step": 4099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.661315703996905e-07,
+      "loss": 1.0184,
+      "step": 4100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.644244803604753e-07,
+      "loss": 1.0184,
+      "step": 4101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.627198139820866e-07,
+      "loss": 1.0371,
+      "step": 4102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.610175721959776e-07,
+      "loss": 1.0161,
+      "step": 4103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.593177559322776e-07,
+      "loss": 1.0444,
+      "step": 4104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.576203661197904e-07,
+      "loss": 1.0412,
+      "step": 4105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.559254036859929e-07,
+      "loss": 1.0529,
+      "step": 4106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.542328695570371e-07,
+      "loss": 1.0124,
+      "step": 4107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.525427646577464e-07,
+      "loss": 0.9953,
+      "step": 4108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.508550899116194e-07,
+      "loss": 1.0516,
+      "step": 4109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.49169846240824e-07,
+      "loss": 1.0499,
+      "step": 4110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.474870345662009e-07,
+      "loss": 1.058,
+      "step": 4111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.458066558072645e-07,
+      "loss": 1.0475,
+      "step": 4112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.441287108821936e-07,
+      "loss": 1.041,
+      "step": 4113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.424532007078448e-07,
+      "loss": 1.0424,
+      "step": 4114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.407801261997364e-07,
+      "loss": 1.0491,
+      "step": 4115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.391094882720632e-07,
+      "loss": 1.0797,
+      "step": 4116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.374412878376839e-07,
+      "loss": 1.0304,
+      "step": 4117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.357755258081266e-07,
+      "loss": 1.0685,
+      "step": 4118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.341122030935875e-07,
+      "loss": 1.0653,
+      "step": 4119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.324513206029292e-07,
+      "loss": 1.0375,
+      "step": 4120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.307928792436812e-07,
+      "loss": 1.0382,
+      "step": 4121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.291368799220392e-07,
+      "loss": 1.0432,
+      "step": 4122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.27483323542865e-07,
+      "loss": 1.0276,
+      "step": 4123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.258322110096842e-07,
+      "loss": 1.024,
+      "step": 4124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.241835432246888e-07,
+      "loss": 1.0581,
+      "step": 4125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.225373210887335e-07,
+      "loss": 1.0134,
+      "step": 4126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.208935455013369e-07,
+      "loss": 1.071,
+      "step": 4127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.192522173606818e-07,
+      "loss": 1.0516,
+      "step": 4128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.176133375636128e-07,
+      "loss": 1.057,
+      "step": 4129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.159769070056364e-07,
+      "loss": 1.0187,
+      "step": 4130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.143429265809218e-07,
+      "loss": 1.0774,
+      "step": 4131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.127113971822989e-07,
+      "loss": 1.0565,
+      "step": 4132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.110823197012571e-07,
+      "loss": 1.043,
+      "step": 4133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.0945569502795e-07,
+      "loss": 1.0257,
+      "step": 4134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.07831524051185e-07,
+      "loss": 1.0293,
+      "step": 4135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.062098076584348e-07,
+      "loss": 1.0601,
+      "step": 4136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.045905467358247e-07,
+      "loss": 1.0537,
+      "step": 4137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.029737421681446e-07,
+      "loss": 1.0206,
+      "step": 4138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.013593948388385e-07,
+      "loss": 1.0442,
+      "step": 4139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.997475056300078e-07,
+      "loss": 1.0403,
+      "step": 4140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.981380754224119e-07,
+      "loss": 1.0484,
+      "step": 4141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.965311050954663e-07,
+      "loss": 1.0808,
+      "step": 4142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.949265955272414e-07,
+      "loss": 1.0068,
+      "step": 4143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.933245475944648e-07,
+      "loss": 1.0563,
+      "step": 4144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.917249621725168e-07,
+      "loss": 1.0035,
+      "step": 4145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.901278401354348e-07,
+      "loss": 1.0357,
+      "step": 4146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.885331823559075e-07,
+      "loss": 1.0651,
+      "step": 4147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.86940989705279e-07,
+      "loss": 1.0167,
+      "step": 4148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.853512630535457e-07,
+      "loss": 1.0246,
+      "step": 4149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.837640032693558e-07,
+      "loss": 1.0126,
+      "step": 4150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.821792112200119e-07,
+      "loss": 1.0488,
+      "step": 4151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.805968877714645e-07,
+      "loss": 1.0554,
+      "step": 4152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.790170337883204e-07,
+      "loss": 1.0302,
+      "step": 4153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.774396501338302e-07,
+      "loss": 1.0629,
+      "step": 4154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.758647376699033e-07,
+      "loss": 1.0442,
+      "step": 4155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.742922972570885e-07,
+      "loss": 1.0409,
+      "step": 4156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7272232975459465e-07,
+      "loss": 1.0643,
+      "step": 4157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7115483602026945e-07,
+      "loss": 1.019,
+      "step": 4158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.695898169106178e-07,
+      "loss": 1.0292,
+      "step": 4159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.680272732807833e-07,
+      "loss": 1.014,
+      "step": 4160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6646720598456706e-07,
+      "loss": 1.0132,
+      "step": 4161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6490961587440707e-07,
+      "loss": 1.0609,
+      "step": 4162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6335450380139646e-07,
+      "loss": 1.0262,
+      "step": 4163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.618018706152666e-07,
+      "loss": 1.0474,
+      "step": 4164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.602517171644011e-07,
+      "loss": 1.0422,
+      "step": 4165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5870404429582404e-07,
+      "loss": 1.0534,
+      "step": 4166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5715885285520625e-07,
+      "loss": 1.0459,
+      "step": 4167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.556161436868617e-07,
+      "loss": 1.0418,
+      "step": 4168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5407591763374837e-07,
+      "loss": 1.0153,
+      "step": 4169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.52538175537468e-07,
+      "loss": 1.034,
+      "step": 4170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.510029182382636e-07,
+      "loss": 1.0273,
+      "step": 4171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.494701465750217e-07,
+      "loss": 1.006,
+      "step": 4172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.479398613852698e-07,
+      "loss": 1.0543,
+      "step": 4173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4641206350518e-07,
+      "loss": 1.0575,
+      "step": 4174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448867537695578e-07,
+      "loss": 1.0134,
+      "step": 4175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.433639330118589e-07,
+      "loss": 0.9866,
+      "step": 4176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.418436020641692e-07,
+      "loss": 1.0429,
+      "step": 4177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.403257617572232e-07,
+      "loss": 1.0444,
+      "step": 4178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.388104129203863e-07,
+      "loss": 1.0474,
+      "step": 4179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.372975563816706e-07,
+      "loss": 1.0254,
+      "step": 4180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3578719296771766e-07,
+      "loss": 1.023,
+      "step": 4181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3427932350381565e-07,
+      "loss": 1.0326,
+      "step": 4182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3277394881388104e-07,
+      "loss": 1.0801,
+      "step": 4183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3127106972047694e-07,
+      "loss": 1.0023,
+      "step": 4184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2977068704479266e-07,
+      "loss": 1.0216,
+      "step": 4185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.282728016066623e-07,
+      "loss": 1.0301,
+      "step": 4186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2677741422454786e-07,
+      "loss": 1.0288,
+      "step": 4187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.25284525715553e-07,
+      "loss": 1.0442,
+      "step": 4188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.237941368954124e-07,
+      "loss": 1.0511,
+      "step": 4189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2230624857849457e-07,
+      "loss": 1.06,
+      "step": 4190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2082086157780366e-07,
+      "loss": 1.0809,
+      "step": 4191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.193379767049749e-07,
+      "loss": 1.0632,
+      "step": 4192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1785759477027844e-07,
+      "loss": 1.0402,
+      "step": 4193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.16379716582615e-07,
+      "loss": 1.0197,
+      "step": 4194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1490434294951866e-07,
+      "loss": 1.0378,
+      "step": 4195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.134314746771534e-07,
+      "loss": 1.0461,
+      "step": 4196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1196111257031655e-07,
+      "loss": 0.9993,
+      "step": 4197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1049325743243327e-07,
+      "loss": 1.0815,
+      "step": 4198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0902791006556087e-07,
+      "loss": 1.103,
+      "step": 4199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0756507127038494e-07,
+      "loss": 1.0431,
+      "step": 4200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.061047418462233e-07,
+      "loss": 1.0261,
+      "step": 4201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.046469225910177e-07,
+      "loss": 1.0773,
+      "step": 4202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0319161430134423e-07,
+      "loss": 1.0167,
+      "step": 4203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.017388177724002e-07,
+      "loss": 1.0457,
+      "step": 4204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.0028853379801723e-07,
+      "loss": 1.0956,
+      "step": 4205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9884076317064813e-07,
+      "loss": 1.0333,
+      "step": 4206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.973955066813778e-07,
+      "loss": 1.059,
+      "step": 4207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.95952765119913e-07,
+      "loss": 1.0594,
+      "step": 4208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9451253927458855e-07,
+      "loss": 1.0373,
+      "step": 4209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.930748299323639e-07,
+      "loss": 1.0509,
+      "step": 4210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.916396378788234e-07,
+      "loss": 1.0084,
+      "step": 4211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.90206963898177e-07,
+      "loss": 1.0505,
+      "step": 4212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.887768087732574e-07,
+      "loss": 1.0365,
+      "step": 4213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.873491732855217e-07,
+      "loss": 1.1048,
+      "step": 4214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.859240582150492e-07,
+      "loss": 0.9907,
+      "step": 4215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.845014643405437e-07,
+      "loss": 1.0368,
+      "step": 4216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.830813924393301e-07,
+      "loss": 1.0669,
+      "step": 4217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8166384328735594e-07,
+      "loss": 1.0497,
+      "step": 4218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8024881765918955e-07,
+      "loss": 1.0285,
+      "step": 4219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.788363163280212e-07,
+      "loss": 1.0533,
+      "step": 4220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.774263400656614e-07,
+      "loss": 1.0555,
+      "step": 4221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.76018889642541e-07,
+      "loss": 1.058,
+      "step": 4222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7461396582771035e-07,
+      "loss": 1.0451,
+      "step": 4223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.732115693888394e-07,
+      "loss": 1.0221,
+      "step": 4224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.71811701092219e-07,
+      "loss": 1.0584,
+      "step": 4225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.704143617027545e-07,
+      "loss": 1.0398,
+      "step": 4226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.690195519839734e-07,
+      "loss": 1.0437,
+      "step": 4227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.67627272698019e-07,
+      "loss": 1.0352,
+      "step": 4228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6623752460565264e-07,
+      "loss": 1.0346,
+      "step": 4229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.648503084662508e-07,
+      "loss": 1.0173,
+      "step": 4230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6346562503780935e-07,
+      "loss": 1.053,
+      "step": 4231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.620834750769381e-07,
+      "loss": 1.0019,
+      "step": 4232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6070385933886343e-07,
+      "loss": 1.0224,
+      "step": 4233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.593267785774268e-07,
+      "loss": 1.0645,
+      "step": 4234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.579522335450841e-07,
+      "loss": 1.0425,
+      "step": 4235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.565802249929068e-07,
+      "loss": 1.0661,
+      "step": 4236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.55210753670579e-07,
+      "loss": 0.9945,
+      "step": 4237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.538438203263994e-07,
+      "loss": 1.0522,
+      "step": 4238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5247942570727947e-07,
+      "loss": 1.0455,
+      "step": 4239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511175705587433e-07,
+      "loss": 1.0503,
+      "step": 4240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4975825562492847e-07,
+      "loss": 1.0338,
+      "step": 4241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4840148164858313e-07,
+      "loss": 1.0613,
+      "step": 4242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.470472493710669e-07,
+      "loss": 1.0806,
+      "step": 4243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4569555953235466e-07,
+      "loss": 1.0385,
+      "step": 4244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.443464128710239e-07,
+      "loss": 1.0377,
+      "step": 4245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4299981012427205e-07,
+      "loss": 1.011,
+      "step": 4246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4165575202789735e-07,
+      "loss": 1.0165,
+      "step": 4247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.403142393163167e-07,
+      "loss": 1.0327,
+      "step": 4248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.389752727225465e-07,
+      "loss": 1.0149,
+      "step": 4249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3763885297822153e-07,
+      "loss": 1.0293,
+      "step": 4250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.363049808135771e-07,
+      "loss": 1.0564,
+      "step": 4251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.3497365695746075e-07,
+      "loss": 1.0841,
+      "step": 4252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.3364488213732727e-07,
+      "loss": 1.0408,
+      "step": 4253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.323186570792375e-07,
+      "loss": 1.0513,
+      "step": 4254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.30994982507859e-07,
+      "loss": 0.9926,
+      "step": 4255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.29673859146466e-07,
+      "loss": 1.044,
+      "step": 4256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.283552877169399e-07,
+      "loss": 1.0639,
+      "step": 4257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.270392689397661e-07,
+      "loss": 1.0385,
+      "step": 4258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.257258035340355e-07,
+      "loss": 1.0214,
+      "step": 4259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2441489221744446e-07,
+      "loss": 1.0297,
+      "step": 4260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2310653570629293e-07,
+      "loss": 1.029,
+      "step": 4261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.21800734715485e-07,
+      "loss": 1.0517,
+      "step": 4262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2049748995853116e-07,
+      "loss": 1.0465,
+      "step": 4263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.191968021475389e-07,
+      "loss": 1.0474,
+      "step": 4264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1789867199322533e-07,
+      "loss": 1.0428,
+      "step": 4265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1660310020490525e-07,
+      "loss": 1.0373,
+      "step": 4266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.153100874904991e-07,
+      "loss": 1.0685,
+      "step": 4267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.140196345565244e-07,
+      "loss": 1.0331,
+      "step": 4268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1273174210810553e-07,
+      "loss": 1.0634,
+      "step": 4269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1144641084896286e-07,
+      "loss": 1.0598,
+      "step": 4270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1016364148142244e-07,
+      "loss": 1.049,
+      "step": 4271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0888343470640347e-07,
+      "loss": 1.1015,
+      "step": 4272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.076057912234331e-07,
+      "loss": 1.0384,
+      "step": 4273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0633071173062966e-07,
+      "loss": 1.035,
+      "step": 4274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0505819692471797e-07,
+      "loss": 1.0326,
+      "step": 4275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.037882475010151e-07,
+      "loss": 1.0348,
+      "step": 4276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.025208641534422e-07,
+      "loss": 1.0416,
+      "step": 4277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.012560475745124e-07,
+      "loss": 1.0589,
+      "step": 4278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.999937984553419e-07,
+      "loss": 1.0445,
+      "step": 4279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9873411748563955e-07,
+      "loss": 1.0398,
+      "step": 4280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9747700535371347e-07,
+      "loss": 1.0314,
+      "step": 4281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.962224627464677e-07,
+      "loss": 1.0763,
+      "step": 4282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9497049034940107e-07,
+      "loss": 1.0251,
+      "step": 4283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.937210888466097e-07,
+      "loss": 1.048,
+      "step": 4284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9247425892078354e-07,
+      "loss": 1.0366,
+      "step": 4285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.912300012532082e-07,
+      "loss": 1.0469,
+      "step": 4286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.899883165237638e-07,
+      "loss": 1.0418,
+      "step": 4287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8874920541092423e-07,
+      "loss": 1.0503,
+      "step": 4288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8751266859175675e-07,
+      "loss": 1.0266,
+      "step": 4289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.862787067419237e-07,
+      "loss": 1.0358,
+      "step": 4290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850473205356774e-07,
+      "loss": 1.0381,
+      "step": 4291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8381851064586686e-07,
+      "loss": 1.0241,
+      "step": 4292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.825922777439283e-07,
+      "loss": 1.0379,
+      "step": 4293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.813686224998957e-07,
+      "loss": 1.0479,
+      "step": 4294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.801475455823882e-07,
+      "loss": 1.0376,
+      "step": 4295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7892904765862207e-07,
+      "loss": 1.0436,
+      "step": 4296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7771312939440023e-07,
+      "loss": 1.0598,
+      "step": 4297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.764997914541184e-07,
+      "loss": 1.0448,
+      "step": 4298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.752890345007608e-07,
+      "loss": 1.0374,
+      "step": 4299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7408085919590265e-07,
+      "loss": 1.008,
+      "step": 4300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.728752661997075e-07,
+      "loss": 1.0267,
+      "step": 4301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7167225617092784e-07,
+      "loss": 1.0246,
+      "step": 4302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.704718297669068e-07,
+      "loss": 1.0335,
+      "step": 4303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.692739876435729e-07,
+      "loss": 1.0595,
+      "step": 4304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6807873045544385e-07,
+      "loss": 1.0694,
+      "step": 4305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.66886058855626e-07,
+      "loss": 1.0509,
+      "step": 4306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6569597349581143e-07,
+      "loss": 1.0517,
+      "step": 4307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6450847502627883e-07,
+      "loss": 1.0585,
+      "step": 4308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.633235640958953e-07,
+      "loss": 1.0664,
+      "step": 4309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6214124135211185e-07,
+      "loss": 1.0221,
+      "step": 4310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6096150744096603e-07,
+      "loss": 1.0165,
+      "step": 4311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.597843630070823e-07,
+      "loss": 1.0263,
+      "step": 4312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.586098086936678e-07,
+      "loss": 1.0323,
+      "step": 4313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.574378451425158e-07,
+      "loss": 1.0163,
+      "step": 4314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5626847299400404e-07,
+      "loss": 1.0211,
+      "step": 4315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5510169288709363e-07,
+      "loss": 1.0383,
+      "step": 4316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.539375054593296e-07,
+      "loss": 1.0714,
+      "step": 4317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.527759113468403e-07,
+      "loss": 1.035,
+      "step": 4318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.516169111843375e-07,
+      "loss": 1.0213,
+      "step": 4319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.504605056051146e-07,
+      "loss": 1.0395,
+      "step": 4320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4930669524104835e-07,
+      "loss": 1.0527,
+      "step": 4321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.481554807225961e-07,
+      "loss": 1.0394,
+      "step": 4322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4700686267879917e-07,
+      "loss": 1.0299,
+      "step": 4323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.458608417372771e-07,
+      "loss": 1.064,
+      "step": 4324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.447174185242324e-07,
+      "loss": 1.0598,
+      "step": 4325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4357659366444795e-07,
+      "loss": 1.0681,
+      "step": 4326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.424383677812858e-07,
+      "loss": 1.0756,
+      "step": 4327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4130274149668896e-07,
+      "loss": 1.0405,
+      "step": 4328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4016971543117994e-07,
+      "loss": 1.0146,
+      "step": 4329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.390392902038596e-07,
+      "loss": 1.0446,
+      "step": 4330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3791146643240838e-07,
+      "loss": 1.0516,
+      "step": 4331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3678624473308486e-07,
+      "loss": 1.0467,
+      "step": 4332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.356636257207262e-07,
+      "loss": 1.0386,
+      "step": 4333
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.345436100087467e-07,
+      "loss": 1.0264,
+      "step": 4334
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3342619820914013e-07,
+      "loss": 1.0289,
+      "step": 4335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3231139093247367e-07,
+      "loss": 1.0394,
+      "step": 4336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.311991887878967e-07,
+      "loss": 1.0337,
+      "step": 4337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3008959238312867e-07,
+      "loss": 1.0504,
+      "step": 4338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2898260232447235e-07,
+      "loss": 1.0258,
+      "step": 4339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2787821921679887e-07,
+      "loss": 1.0713,
+      "step": 4340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2677644366356165e-07,
+      "loss": 1.0664,
+      "step": 4341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2567727626678527e-07,
+      "loss": 1.0595,
+      "step": 4342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.245807176270698e-07,
+      "loss": 1.0469,
+      "step": 4343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.234867683435915e-07,
+      "loss": 1.0939,
+      "step": 4344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2239542901409838e-07,
+      "loss": 1.0491,
+      "step": 4345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2130670023491396e-07,
+      "loss": 1.0136,
+      "step": 4346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2022058260093516e-07,
+      "loss": 1.0087,
+      "step": 4347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1913707670563223e-07,
+      "loss": 1.0237,
+      "step": 4348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1805618314104714e-07,
+      "loss": 1.0053,
+      "step": 4349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1697790249779638e-07,
+      "loss": 1.0264,
+      "step": 4350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1590223536506583e-07,
+      "loss": 1.0257,
+      "step": 4351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.148291823306181e-07,
+      "loss": 1.0507,
+      "step": 4352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1375874398078146e-07,
+      "loss": 1.0628,
+      "step": 4353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1269092090046128e-07,
+      "loss": 1.0112,
+      "step": 4354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1162571367312812e-07,
+      "loss": 1.0526,
+      "step": 4355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1056312288082914e-07,
+      "loss": 1.0356,
+      "step": 4356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0950314910417547e-07,
+      "loss": 1.0459,
+      "step": 4357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0844579292235546e-07,
+      "loss": 1.0747,
+      "step": 4358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0739105491312028e-07,
+      "loss": 1.0673,
+      "step": 4359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0633893565279618e-07,
+      "loss": 1.0309,
+      "step": 4360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0528943571627324e-07,
+      "loss": 1.0828,
+      "step": 4361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.042425556770161e-07,
+      "loss": 1.0228,
+      "step": 4362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.031982961070511e-07,
+      "loss": 1.0555,
+      "step": 4363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.021566575769801e-07,
+      "loss": 1.0094,
+      "step": 4364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0111764065596561e-07,
+      "loss": 1.032,
+      "step": 4365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0008124591174406e-07,
+      "loss": 1.0055,
+      "step": 4366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9904747391061298e-07,
+      "loss": 1.0264,
+      "step": 4367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9801632521744273e-07,
+      "loss": 1.0246,
+      "step": 4368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9698780039566657e-07,
+      "loss": 1.0277,
+      "step": 4369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9596190000728432e-07,
+      "loss": 1.0164,
+      "step": 4370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9493862461286318e-07,
+      "loss": 1.037,
+      "step": 4371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9391797477153474e-07,
+      "loss": 1.067,
+      "step": 4372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9289995104099623e-07,
+      "loss": 1.0567,
+      "step": 4373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9188455397751048e-07,
+      "loss": 1.0276,
+      "step": 4374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.908717841359048e-07,
+      "loss": 1.0202,
+      "step": 4375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8986164206957037e-07,
+      "loss": 1.0505,
+      "step": 4376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8885412833046402e-07,
+      "loss": 1.0414,
+      "step": 4377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784924346910427e-07,
+      "loss": 1.0101,
+      "step": 4378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.868469880345758e-07,
+      "loss": 1.0117,
+      "step": 4379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8584736257452273e-07,
+      "loss": 1.0707,
+      "step": 4380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8485036763515708e-07,
+      "loss": 1.0203,
+      "step": 4381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8385600376124856e-07,
+      "loss": 1.0255,
+      "step": 4382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8286427149613374e-07,
+      "loss": 1.0222,
+      "step": 4383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8187517138170684e-07,
+      "loss": 1.0567,
+      "step": 4384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.808887039584284e-07,
+      "loss": 1.0526,
+      "step": 4385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7990486976531608e-07,
+      "loss": 1.0639,
+      "step": 4386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7892366933995152e-07,
+      "loss": 1.0787,
+      "step": 4387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7794510321847703e-07,
+      "loss": 1.0554,
+      "step": 4388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7696917193559425e-07,
+      "loss": 1.0347,
+      "step": 4389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7599587602456558e-07,
+      "loss": 1.0495,
+      "step": 4390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7502521601721445e-07,
+      "loss": 1.0629,
+      "step": 4391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7405719244392218e-07,
+      "loss": 1.0391,
+      "step": 4392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7309180583363062e-07,
+      "loss": 1.0264,
+      "step": 4393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7212905671384118e-07,
+      "loss": 1.065,
+      "step": 4394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7116894561061303e-07,
+      "loss": 1.0271,
+      "step": 4395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7021147304856433e-07,
+      "loss": 1.0577,
+      "step": 4396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6925663955087156e-07,
+      "loss": 1.0069,
+      "step": 4397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6830444563926963e-07,
+      "loss": 1.0521,
+      "step": 4398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6735489183405008e-07,
+      "loss": 1.0372,
+      "step": 4399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.664079786540629e-07,
+      "loss": 1.0569,
+      "step": 4400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6546370661671473e-07,
+      "loss": 1.047,
+      "step": 4401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6452207623796902e-07,
+      "loss": 1.043,
+      "step": 4402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6358308803234469e-07,
+      "loss": 1.0523,
+      "step": 4403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6264674251292078e-07,
+      "loss": 1.0667,
+      "step": 4404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6171304019132695e-07,
+      "loss": 1.002,
+      "step": 4405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6078198157775225e-07,
+      "loss": 1.0321,
+      "step": 4406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5985356718094092e-07,
+      "loss": 1.0532,
+      "step": 4407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5892779750819098e-07,
+      "loss": 1.0497,
+      "step": 4408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5800467306535616e-07,
+      "loss": 1.0445,
+      "step": 4409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5708419435684463e-07,
+      "loss": 1.0398,
+      "step": 4410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.561663618856185e-07,
+      "loss": 0.9936,
+      "step": 4411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.552511761531944e-07,
+      "loss": 1.0465,
+      "step": 4412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5433863765964342e-07,
+      "loss": 1.0672,
+      "step": 4413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5342874690358888e-07,
+      "loss": 1.0953,
+      "step": 4414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.525215043822076e-07,
+      "loss": 1.0317,
+      "step": 4415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5161691059123018e-07,
+      "loss": 1.0314,
+      "step": 4416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5071496602493963e-07,
+      "loss": 1.0599,
+      "step": 4417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4981567117617e-07,
+      "loss": 1.0502,
+      "step": 4418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4891902653630942e-07,
+      "loss": 1.0358,
+      "step": 4419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4802503259529755e-07,
+      "loss": 1.0182,
+      "step": 4420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4713368984162423e-07,
+      "loss": 1.0062,
+      "step": 4421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.462449987623321e-07,
+      "loss": 1.0567,
+      "step": 4422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4535895984301495e-07,
+      "loss": 1.0362,
+      "step": 4423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.444755735678166e-07,
+      "loss": 1.0628,
+      "step": 4424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.435948404194304e-07,
+      "loss": 1.0272,
+      "step": 4425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4271676087910413e-07,
+      "loss": 1.0362,
+      "step": 4426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4184133542663014e-07,
+      "loss": 1.0699,
+      "step": 4427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4096856454035522e-07,
+      "loss": 1.0104,
+      "step": 4428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4009844869717183e-07,
+      "loss": 1.0755,
+      "step": 4429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3923098837252458e-07,
+      "loss": 1.0237,
+      "step": 4430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3836618404040603e-07,
+      "loss": 1.0347,
+      "step": 4431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.375040361733576e-07,
+      "loss": 1.0302,
+      "step": 4432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3664454524246806e-07,
+      "loss": 1.0087,
+      "step": 4433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3578771171737614e-07,
+      "loss": 1.0484,
+      "step": 4434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.349335360662668e-07,
+      "loss": 1.0594,
+      "step": 4435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.34082018755875e-07,
+      "loss": 1.0282,
+      "step": 4436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.332331602514808e-07,
+      "loss": 1.0432,
+      "step": 4437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3238696101691262e-07,
+      "loss": 1.0391,
+      "step": 4438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3154342151454615e-07,
+      "loss": 1.0512,
+      "step": 4439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.307025422053021e-07,
+      "loss": 1.0397,
+      "step": 4440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2986432354864964e-07,
+      "loss": 1.0537,
+      "step": 4441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2902876600260238e-07,
+      "loss": 1.0272,
+      "step": 4442
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.281958700237229e-07,
+      "loss": 1.0123,
+      "step": 4443
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2736563606711384e-07,
+      "loss": 1.0517,
+      "step": 4444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2653806458643004e-07,
+      "loss": 1.0717,
+      "step": 4445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.257131560338659e-07,
+      "loss": 1.0703,
+      "step": 4446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.248909108601648e-07,
+      "loss": 1.0667,
+      "step": 4447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2407132951461066e-07,
+      "loss": 1.015,
+      "step": 4448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2325441244503745e-07,
+      "loss": 1.0369,
+      "step": 4449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.22440160097817e-07,
+      "loss": 1.0352,
+      "step": 4450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2162857291787012e-07,
+      "loss": 1.0547,
+      "step": 4451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.208196513486587e-07,
+      "loss": 1.0704,
+      "step": 4452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.200133958321903e-07,
+      "loss": 1.0865,
+      "step": 4453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.192098068090114e-07,
+      "loss": 1.0382,
+      "step": 4454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1840888471821744e-07,
+      "loss": 1.0431,
+      "step": 4455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.176106299974411e-07,
+      "loss": 1.0265,
+      "step": 4456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1681504308286185e-07,
+      "loss": 1.0294,
+      "step": 4457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1602212440919858e-07,
+      "loss": 1.0302,
+      "step": 4458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1523187440971307e-07,
+      "loss": 1.0329,
+      "step": 4459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1444429351620933e-07,
+      "loss": 1.0421,
+      "step": 4460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.136593821590326e-07,
+      "loss": 1.051,
+      "step": 4461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.128771407670698e-07,
+      "loss": 1.0737,
+      "step": 4462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1209756976774844e-07,
+      "loss": 1.0382,
+      "step": 4463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1132066958703669e-07,
+      "loss": 1.0562,
+      "step": 4464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.105464406494433e-07,
+      "loss": 1.0507,
+      "step": 4465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0977488337801989e-07,
+      "loss": 1.0642,
+      "step": 4466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0900599819435365e-07,
+      "loss": 1.02,
+      "step": 4467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0823978551857628e-07,
+      "loss": 1.0266,
+      "step": 4468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0747624576935566e-07,
+      "loss": 1.0147,
+      "step": 4469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0671537936390198e-07,
+      "loss": 1.0686,
+      "step": 4470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0595718671796207e-07,
+      "loss": 1.0506,
+      "step": 4471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0520166824582456e-07,
+      "loss": 1.0434,
+      "step": 4472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0444882436031368e-07,
+      "loss": 0.9935,
+      "step": 4473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0369865547279534e-07,
+      "loss": 1.0152,
+      "step": 4474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0295116199317057e-07,
+      "loss": 1.0493,
+      "step": 4475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0220634432988263e-07,
+      "loss": 1.0524,
+      "step": 4476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0146420288990877e-07,
+      "loss": 1.0185,
+      "step": 4477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.007247380787657e-07,
+      "loss": 1.0674,
+      "step": 4478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.9987950300508e-08,
+      "loss": 1.0101,
+      "step": 4479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.925383995772641e-08,
+      "loss": 1.0469,
+      "step": 4480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.852240745154951e-08,
+      "loss": 1.0176,
+      "step": 4481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.779365318164147e-08,
+      "loss": 1.0744,
+      "step": 4482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.706757754620489e-08,
+      "loss": 1.0657,
+      "step": 4483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.634418094197684e-08,
+      "loss": 1.0754,
+      "step": 4484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.562346376423225e-08,
+      "loss": 1.0692,
+      "step": 4485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.490542640678057e-08,
+      "loss": 1.021,
+      "step": 4486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.419006926196738e-08,
+      "loss": 1.0467,
+      "step": 4487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.347739272067447e-08,
+      "loss": 1.0717,
+      "step": 4488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.276739717231753e-08,
+      "loss": 1.0486,
+      "step": 4489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.206008300484847e-08,
+      "loss": 1.0559,
+      "step": 4490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.135545060475481e-08,
+      "loss": 1.0399,
+      "step": 4491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.065350035705522e-08,
+      "loss": 1.0449,
+      "step": 4492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.995423264530733e-08,
+      "loss": 1.0418,
+      "step": 4493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.92576478515994e-08,
+      "loss": 1.069,
+      "step": 4494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.856374635655696e-08,
+      "loss": 1.0631,
+      "step": 4495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.787252853933504e-08,
+      "loss": 1.038,
+      "step": 4496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.718399477762707e-08,
+      "loss": 1.0755,
+      "step": 4497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.649814544765601e-08,
+      "loss": 1.0727,
+      "step": 4498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.581498092418094e-08,
+      "loss": 1.0303,
+      "step": 4499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.513450158049109e-08,
+      "loss": 1.0769,
+      "step": 4500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.44567077884112e-08,
+      "loss": 1.066,
+      "step": 4501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.378159991829615e-08,
+      "loss": 1.0411,
+      "step": 4502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.310917833903532e-08,
+      "loss": 1.0481,
+      "step": 4503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.243944341804921e-08,
+      "loss": 1.0383,
+      "step": 4504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.177239552129012e-08,
+      "loss": 1.0439,
+      "step": 4505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.11080350132426e-08,
+      "loss": 1.0259,
+      "step": 4506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.044636225692237e-08,
+      "loss": 1.014,
+      "step": 4507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.978737761387633e-08,
+      "loss": 1.0349,
+      "step": 4508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.913108144418369e-08,
+      "loss": 1.0544,
+      "step": 4509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.847747410645368e-08,
+      "loss": 1.0601,
+      "step": 4510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.782655595782674e-08,
+      "loss": 1.0454,
+      "step": 4511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.717832735397335e-08,
+      "loss": 1.0055,
+      "step": 4512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.653278864909464e-08,
+      "loss": 1.0254,
+      "step": 4513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.588994019592233e-08,
+      "loss": 1.0326,
+      "step": 4514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.524978234571767e-08,
+      "loss": 1.0384,
+      "step": 4515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.461231544827196e-08,
+      "loss": 1.0297,
+      "step": 4516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.397753985190603e-08,
+      "loss": 1.0015,
+      "step": 4517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.334545590347075e-08,
+      "loss": 1.0267,
+      "step": 4518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.271606394834486e-08,
+      "loss": 1.0723,
+      "step": 4519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.208936433043768e-08,
+      "loss": 1.0393,
+      "step": 4520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.146535739218585e-08,
+      "loss": 1.041,
+      "step": 4521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.084404347455609e-08,
+      "loss": 1.0489,
+      "step": 4522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.022542291704348e-08,
+      "loss": 1.0506,
+      "step": 4523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.960949605766987e-08,
+      "loss": 1.0511,
+      "step": 4524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.899626323298714e-08,
+      "loss": 1.0615,
+      "step": 4525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.838572477807448e-08,
+      "loss": 1.0358,
+      "step": 4526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.777788102653781e-08,
+      "loss": 1.006,
+      "step": 4527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.717273231051257e-08,
+      "loss": 1.0395,
+      "step": 4528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.657027896065982e-08,
+      "loss": 1.0269,
+      "step": 4529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.597052130616899e-08,
+      "loss": 1.0643,
+      "step": 4530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.537345967475516e-08,
+      "loss": 1.0468,
+      "step": 4531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.47790943926624e-08,
+      "loss": 1.0308,
+      "step": 4532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.41874257846592e-08,
+      "loss": 1.0633,
+      "step": 4533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.359845417404309e-08,
+      "loss": 1.039,
+      "step": 4534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.301217988263441e-08,
+      "loss": 1.0383,
+      "step": 4535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.242860323078414e-08,
+      "loss": 1.0395,
+      "step": 4536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.184772453736387e-08,
+      "loss": 1.0209,
+      "step": 4537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.126954411977692e-08,
+      "loss": 1.022,
+      "step": 4538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.069406229394615e-08,
+      "loss": 1.0381,
+      "step": 4539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.012127937432554e-08,
+      "loss": 1.0382,
+      "step": 4540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.955119567389034e-08,
+      "loss": 1.0249,
+      "step": 4541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.898381150414245e-08,
+      "loss": 1.0128,
+      "step": 4542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8419127175108894e-08,
+      "loss": 1.0721,
+      "step": 4543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.785714299534062e-08,
+      "loss": 1.0276,
+      "step": 4544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.7297859271914225e-08,
+      "loss": 1.0405,
+      "step": 4545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.674127631043025e-08,
+      "loss": 1.0205,
+      "step": 4546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.618739441501264e-08,
+      "loss": 1.0446,
+      "step": 4547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5636213888310975e-08,
+      "loss": 1.0151,
+      "step": 4548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5087735031498225e-08,
+      "loss": 1.0436,
+      "step": 4549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.454195814427021e-08,
+      "loss": 1.047,
+      "step": 4550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.399888352484783e-08,
+      "loss": 1.0411,
+      "step": 4551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.345851146997372e-08,
+      "loss": 1.0849,
+      "step": 4552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.292084227491612e-08,
+      "loss": 1.0729,
+      "step": 4553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.238587623346336e-08,
+      "loss": 1.0772,
+      "step": 4554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.185361363793051e-08,
+      "loss": 1.0217,
+      "step": 4555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.132405477915103e-08,
+      "loss": 1.0739,
+      "step": 4556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.079719994648513e-08,
+      "loss": 1.0427,
+      "step": 4557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0273049427813083e-08,
+      "loss": 1.0354,
+      "step": 4558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9751603509538007e-08,
+      "loss": 1.0571,
+      "step": 4559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9232862476584766e-08,
+      "loss": 1.0328,
+      "step": 4560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.871682661240218e-08,
+      "loss": 1.0316,
+      "step": 4561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.820349619895803e-08,
+      "loss": 1.0287,
+      "step": 4562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.769287151674407e-08,
+      "loss": 1.0266,
+      "step": 4563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7184952844772646e-08,
+      "loss": 1.072,
+      "step": 4564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6679740460577886e-08,
+      "loss": 1.015,
+      "step": 4565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.617723464021395e-08,
+      "loss": 1.0615,
+      "step": 4566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.567743565825844e-08,
+      "loss": 1.0763,
+      "step": 4567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5180343787807337e-08,
+      "loss": 1.058,
+      "step": 4568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4685959300479476e-08,
+      "loss": 1.0201,
+      "step": 4569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.419428246641377e-08,
+      "loss": 0.9989,
+      "step": 4570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.37053135542681e-08,
+      "loss": 1.0118,
+      "step": 4571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3219052831223185e-08,
+      "loss": 0.9883,
+      "step": 4572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.273550056297815e-08,
+      "loss": 1.0169,
+      "step": 4573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.22546570137522e-08,
+      "loss": 1.0209,
+      "step": 4574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.177652244628627e-08,
+      "loss": 1.0676,
+      "step": 4575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.130109712183916e-08,
+      "loss": 0.9893,
+      "step": 4576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.082838130018973e-08,
+      "loss": 1.0641,
+      "step": 4577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0358375239636374e-08,
+      "loss": 1.0156,
+      "step": 4578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.989107919699753e-08,
+      "loss": 1.0203,
+      "step": 4579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9426493427611177e-08,
+      "loss": 1.084,
+      "step": 4580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.896461818533148e-08,
+      "loss": 1.0709,
+      "step": 4581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8505453722535444e-08,
+      "loss": 1.0008,
+      "step": 4582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8049000290115156e-08,
+      "loss": 1.048,
+      "step": 4583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.759525813748499e-08,
+      "loss": 1.0498,
+      "step": 4584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.714422751257496e-08,
+      "loss": 1.0414,
+      "step": 4585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6695908661835146e-08,
+      "loss": 1.0363,
+      "step": 4586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.625030183023237e-08,
+      "loss": 1.0549,
+      "step": 4587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5807407261253533e-08,
+      "loss": 1.0549,
+      "step": 4588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.536722519690172e-08,
+      "loss": 1.0426,
+      "step": 4589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.492975587769898e-08,
+      "loss": 1.0235,
+      "step": 4590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.449499954268409e-08,
+      "loss": 1.0566,
+      "step": 4591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4062956429415374e-08,
+      "loss": 1.0699,
+      "step": 4592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3633626773965644e-08,
+      "loss": 1.072,
+      "step": 4593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.320701081092781e-08,
+      "loss": 1.0446,
+      "step": 4594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2783108773410396e-08,
+      "loss": 1.0574,
+      "step": 4595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.23619208930398e-08,
+      "loss": 1.0685,
+      "step": 4596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.194344739995803e-08,
+      "loss": 1.0306,
+      "step": 4597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.152768852282606e-08,
+      "loss": 1.0457,
+      "step": 4598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.11146444888194e-08,
+      "loss": 1.0434,
+      "step": 4599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0704315523631956e-08,
+      "loss": 1.0498,
+      "step": 4600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.029670185147271e-08,
+      "loss": 1.0324,
+      "step": 4601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.989180369506739e-08,
+      "loss": 1.0509,
+      "step": 4602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9489621275658465e-08,
+      "loss": 1.0153,
+      "step": 4603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9090154813004034e-08,
+      "loss": 1.0029,
+      "step": 4604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8693404525377834e-08,
+      "loss": 1.0693,
+      "step": 4605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8299370629570334e-08,
+      "loss": 1.0009,
+      "step": 4606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7908053340887087e-08,
+      "loss": 1.0284,
+      "step": 4607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.751945287314872e-08,
+      "loss": 1.0409,
+      "step": 4608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.713356943869261e-08,
+      "loss": 1.0287,
+      "step": 4609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6750403248370638e-08,
+      "loss": 1.0353,
+      "step": 4610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6369954511550334e-08,
+      "loss": 1.0473,
+      "step": 4611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.599222343611374e-08,
+      "loss": 1.0162,
+      "step": 4612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5617210228459643e-08,
+      "loss": 1.0302,
+      "step": 4613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5244915093499134e-08,
+      "loss": 1.0425,
+      "step": 4614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4875338234660596e-08,
+      "loss": 1.0153,
+      "step": 4615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.450847985388527e-08,
+      "loss": 1.0316,
+      "step": 4616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.414434015163003e-08,
+      "loss": 1.0451,
+      "step": 4617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3782919326865716e-08,
+      "loss": 1.0379,
+      "step": 4618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3424217577078245e-08,
+      "loss": 1.0296,
+      "step": 4619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3068235098266943e-08,
+      "loss": 1.035,
+      "step": 4620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2714972084945665e-08,
+      "loss": 1.0496,
+      "step": 4621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2364428730142773e-08,
+      "loss": 0.9979,
+      "step": 4622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2016605225398946e-08,
+      "loss": 1.0298,
+      "step": 4623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1671501760770487e-08,
+      "loss": 1.0704,
+      "step": 4624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1329118524827662e-08,
+      "loss": 1.0172,
+      "step": 4625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.098945570465194e-08,
+      "loss": 1.102,
+      "step": 4626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0652513485840963e-08,
+      "loss": 1.0236,
+      "step": 4627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0318292052503573e-08,
+      "loss": 1.0369,
+      "step": 4628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9986791587264243e-08,
+      "loss": 1.0605,
+      "step": 4629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9658012271259185e-08,
+      "loss": 1.0647,
+      "step": 4630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9331954284137476e-08,
+      "loss": 1.0443,
+      "step": 4631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9008617804062713e-08,
+      "loss": 1.0336,
+      "step": 4632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8688003007709677e-08,
+      "loss": 1.0411,
+      "step": 4633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.837011007026712e-08,
+      "loss": 1.0171,
+      "step": 4634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.80549391654361e-08,
+      "loss": 1.02,
+      "step": 4635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7742490465430528e-08,
+      "loss": 1.0255,
+      "step": 4636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7432764140977165e-08,
+      "loss": 1.0557,
+      "step": 4637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.712576036131397e-08,
+      "loss": 1.0278,
+      "step": 4638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6821479294192866e-08,
+      "loss": 1.039,
+      "step": 4639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.651992110587641e-08,
+      "loss": 1.0297,
+      "step": 4640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6221085961141693e-08,
+      "loss": 1.0396,
+      "step": 4641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5924974023274752e-08,
+      "loss": 1.064,
+      "step": 4642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5631585454076725e-08,
+      "loss": 1.0277,
+      "step": 4643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5340920413858818e-08,
+      "loss": 1.0755,
+      "step": 4644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5052979061443428e-08,
+      "loss": 1.0558,
+      "step": 4645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4767761554167481e-08,
+      "loss": 1.04,
+      "step": 4646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4485268047876866e-08,
+      "loss": 1.0558,
+      "step": 4647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4205498696930332e-08,
+      "loss": 1.0476,
+      "step": 4648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.392845365419726e-08,
+      "loss": 1.0693,
+      "step": 4649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3654133071059894e-08,
+      "loss": 1.0483,
+      "step": 4650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3382537097409998e-08,
+      "loss": 1.0373,
+      "step": 4651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3113665881651639e-08,
+      "loss": 1.0275,
+      "step": 4652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2847519570699518e-08,
+      "loss": 1.0541,
+      "step": 4653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.258409830998064e-08,
+      "loss": 1.0271,
+      "step": 4654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2323402243431536e-08,
+      "loss": 1.0153,
+      "step": 4655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.206543151350048e-08,
+      "loss": 1.0474,
+      "step": 4656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1810186261145273e-08,
+      "loss": 1.0481,
+      "step": 4657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1557666625837127e-08,
+      "loss": 1.0263,
+      "step": 4658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1307872745555115e-08,
+      "loss": 1.0551,
+      "step": 4659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1060804756790056e-08,
+      "loss": 1.0574,
+      "step": 4660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0816462794544513e-08,
+      "loss": 1.0329,
+      "step": 4661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0574846992329468e-08,
+      "loss": 1.0409,
+      "step": 4662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0335957482167091e-08,
+      "loss": 1.0515,
+      "step": 4663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0099794394590745e-08,
+      "loss": 1.0094,
+      "step": 4664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.866357858642206e-09,
+      "loss": 1.04,
+      "step": 4665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.635648001875552e-09,
+      "loss": 1.0275,
+      "step": 4666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.40766495035328e-09,
+      "loss": 1.0351,
+      "step": 4667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.182408828648514e-09,
+      "loss": 1.0563,
+      "step": 4668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.95987975984447e-09,
+      "loss": 1.0418,
+      "step": 4669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.740077865534435e-09,
+      "loss": 1.0241,
+      "step": 4670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.523003265820673e-09,
+      "loss": 1.0611,
+      "step": 4671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.308656079316635e-09,
+      "loss": 1.0342,
+      "step": 4672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.097036423143633e-09,
+      "loss": 1.0234,
+      "step": 4673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.888144412934174e-09,
+      "loss": 1.0431,
+      "step": 4674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.681980162830283e-09,
+      "loss": 1.0419,
+      "step": 4675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.47854378548185e-09,
+      "loss": 1.0504,
+      "step": 4676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.277835392050514e-09,
+      "loss": 1.0228,
+      "step": 4677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.079855092205212e-09,
+      "loss": 1.0705,
+      "step": 4678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.884602994126077e-09,
+      "loss": 1.0861,
+      "step": 4679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.692079204500546e-09,
+      "loss": 1.0536,
+      "step": 4680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5022838285272445e-09,
+      "loss": 1.0657,
+      "step": 4681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.315216969912663e-09,
+      "loss": 1.0521,
+      "step": 4682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.130878730872814e-09,
+      "loss": 1.0461,
+      "step": 4683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.949269212132125e-09,
+      "loss": 1.0312,
+      "step": 4684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.770388512925662e-09,
+      "loss": 1.0373,
+      "step": 4685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.594236730995795e-09,
+      "loss": 1.0341,
+      "step": 4686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4208139625944175e-09,
+      "loss": 1.0411,
+      "step": 4687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.250120302481842e-09,
+      "loss": 1.0292,
+      "step": 4688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.082155843927905e-09,
+      "loss": 1.0587,
+      "step": 4689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.916920678710857e-09,
+      "loss": 1.0509,
+      "step": 4690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7544148971173656e-09,
+      "loss": 1.0135,
+      "step": 4691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.594638587941957e-09,
+      "loss": 1.0215,
+      "step": 4692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.437591838490351e-09,
+      "loss": 1.0389,
+      "step": 4693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2832747345739056e-09,
+      "loss": 1.064,
+      "step": 4694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1316873605135034e-09,
+      "loss": 1.0622,
+      "step": 4695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.982829799139554e-09,
+      "loss": 1.0402,
+      "step": 4696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.836702131789216e-09,
+      "loss": 1.0245,
+      "step": 4697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6933044383091754e-09,
+      "loss": 1.0865,
+      "step": 4698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5526367970539765e-09,
+      "loss": 1.0569,
+      "step": 4699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.41469928488547e-09,
+      "loss": 1.0397,
+      "step": 4700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2794919771755864e-09,
+      "loss": 1.0187,
+      "step": 4701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1470149478035615e-09,
+      "loss": 1.018,
+      "step": 4702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0172682691559373e-09,
+      "loss": 1.0276,
+      "step": 4703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8902520121287804e-09,
+      "loss": 0.9997,
+      "step": 4704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.765966246124907e-09,
+      "loss": 1.0321,
+      "step": 4705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.64441103905666e-09,
+      "loss": 1.059,
+      "step": 4706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.525586457343132e-09,
+      "loss": 1.0333,
+      "step": 4707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4094925659112754e-09,
+      "loss": 1.0471,
+      "step": 4708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2961294281970136e-09,
+      "loss": 1.0004,
+      "step": 4709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.185497106143575e-09,
+      "loss": 1.0536,
+      "step": 4710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0775956602020476e-09,
+      "loss": 1.0438,
+      "step": 4711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9724251493308254e-09,
+      "loss": 1.0573,
+      "step": 4712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8699856309967177e-09,
+      "loss": 1.0615,
+      "step": 4713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.770277161173839e-09,
+      "loss": 1.0198,
+      "step": 4714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6732997943452733e-09,
+      "loss": 1.059,
+      "step": 4715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5790535835003006e-09,
+      "loss": 1.0384,
+      "step": 4716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.487538580136061e-09,
+      "loss": 1.0301,
+      "step": 4717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.398754834257554e-09,
+      "loss": 1.0383,
+      "step": 4718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3127023943781958e-09,
+      "loss": 1.0734,
+      "step": 4719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2293813075175964e-09,
+      "loss": 1.0362,
+      "step": 4720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1487916192037819e-09,
+      "loss": 1.0359,
+      "step": 4721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0709333734726379e-09,
+      "loss": 1.0464,
+      "step": 4722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.958066128656906e-10,
+      "loss": 1.0309,
+      "step": 4723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.234113784345467e-10,
+      "loss": 1.0423,
+      "step": 4724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.537477097364522e-10,
+      "loss": 1.0595,
+      "step": 4725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.868156448365139e-10,
+      "loss": 1.0379,
+      "step": 4726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.226152203076986e-10,
+      "loss": 1.0644,
+      "step": 4727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.611464712297233e-10,
+      "loss": 1.0471,
+      "step": 4728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.024094311901652e-10,
+      "loss": 1.0285,
+      "step": 4729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.464041322833513e-10,
+      "loss": 1.0763,
+      "step": 4730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.931306051120244e-10,
+      "loss": 1.0331,
+      "step": 4731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.425888787851218e-10,
+      "loss": 1.0556,
+      "step": 4732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9477898091944135e-10,
+      "loss": 1.0385,
+      "step": 4733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4970093763908585e-10,
+      "loss": 1.0562,
+      "step": 4734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0735477357490827e-10,
+      "loss": 1.046,
+      "step": 4735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.677405118661769e-10,
+      "loss": 1.0477,
+      "step": 4736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3085817415835487e-10,
+      "loss": 1.0458,
+      "step": 4737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9670778060421059e-10,
+      "loss": 1.0056,
+      "step": 4738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6528934986437263e-10,
+      "loss": 0.9984,
+      "step": 4739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.366028991062196e-10,
+      "loss": 1.063,
+      "step": 4740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.106484440038802e-10,
+      "loss": 1.0063,
+      "step": 4741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.742599874045355e-11,
+      "loss": 1.0293,
+      "step": 4742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.693557600356838e-11,
+      "loss": 1.0974,
+      "step": 4743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.917718699037899e-11,
+      "loss": 1.0331,
+      "step": 4744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4150841404789745e-11,
+      "loss": 1.0707,
+      "step": 4745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.185654745578969e-11,
+      "loss": 1.0097,
+      "step": 4746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.229431186244856e-11,
+      "loss": 1.0291,
+      "step": 4747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.464139849475913e-12,
+      "loss": 1.0712,
+      "step": 4748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3660351488864466e-12,
+      "loss": 1.0213,
+      "step": 4749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0674,
+      "step": 4750
+    },
+    {
+      "epoch": 1.0,
+      "step": 4750,
+      "total_flos": 1.573231370698752e+18,
+      "train_loss": 1.3342757062535537,
+      "train_runtime": 12719.0865,
+      "train_samples_per_second": 47.802,
+      "train_steps_per_second": 0.373
+    }
+  ],
+  "max_steps": 4750,
+  "num_train_epochs": 1,
+  "total_flos": 1.573231370698752e+18,
+  "trial_name": null,
+  "trial_params": null
+}